WATER VAPOR PROFILING USING A WIDELY TUNABLE AMPLIFIED
DIODE LASER DIFFERENTIAL ABSORPTION LIDAR (DIAL)
by
Michael Drew Obland
A dissertation submitted in partial fulllment
of the requirements for the degree
of
Doctor of Philosophy
in
Physics
MONTANA STATE UNIVERSITY
Bozeman, Montana
April 2007
c©COPYRIGHT by
Michael Drew Obland
2007
All Rights Reserved
ii
APPROVAL
of a dissertation submitted by
Michael Drew Obland
This dissertation has been read by each member of the dissertation committee and
has been found to be satisfactory regarding content, English usage, format, citations,
bibliographic style, and consistency, and is ready for submission to the Division of
Graduate Education.
Dr. Joseph A. Shaw
Dr. John L. Carlsten
Approved for the Department of Physics
Dr. William A. Hiscock
Approved for the Division of Graduate Education
Dr. Carl A. Fox
iii
STATEMENT OF PERMISSION TO USE
In presenting this dissertation in partial fulllment of the requirements for a doc-
toral degree at Montana State University, I agree that the Library shall make it
available to borrowers under rules of the Library. I further agree that copying of this
dissertation is allowable only for scholarly purpose, consistent with fair use as pre-
scribed in the U.S. Copyright Law. Requests for extensive copying or reproduction
of this dissertation should be referred to Bell & Howell Information and Learning,
300 North Zeeb Road, Ann Arbor, Michigan 48106, to whom I have granted the
exclusive right to reproduce and distribute my dissertation in and from microform
along with the non-exclusive right to reproduce and distribute my abstract in any
format in whole or in part.
Michael Drew Obland
April 2007
iv
To my family,
Thank you for teaching me that I can do anything that I set my mind to, and for
making this all possible with your unfailing guidance, support, and sacrice through
the many years it took to achieve this dream.
To my friends,
Thank you for your unselsh understanding and support through all of the Friday
and Saturday nights when I was too busy studying, working, or running
experiments to hang out.
I could not have done this without you. To you, I dedicate this dissertation.
vACKNOWLEDGMENTS
If I have seen a little further it is by standing on the shoulders of Giants. - Sir
Isaac Newton
Thank you to my co-chairs, Dr. Joe Shaw and Dr. John Carlsten, for the oppor-
tunity to do this work and for being so much fun to work for. Thank you for teaching
me, for leading me to nd my own answers, for patiently answering my many, many
questions, and for giving me space to nd my own path. Dr. Kevin Repasky, thank
you for your patience and uncanny ability for getting hardware to work. Thank you
for putting your students (including me) rst and for being a leading example of how
the education system should work. Dr. Dave Klumpar, thank you for the many satel-
lite adventures we have had over the years, and for putting your faith and the fate of
a grand project like Montana's rst satellite in the hands of someone who really had
no idea what he was doing! Dr. Bill Hiscock, thank you for bringing me to MSU,
and for your ceaseless support of anything I wanted to do to further my education.
Above all, thank you to all of you for not only being mentors, but for being friends.
Thank you to: the Physics Department sta, Margaret Jarret, Jeannie Gunderson,
Rose Waldon, Glenda Winslow, Sarah Barutha, Jeremy Gay, and Norm Williams;
Wilma Ang and Muriel Holmquist in the Electrical and Computer Engineering de-
partment; Dale Huls in the MSU Oce of Sponsored Programs; the Montana Space
Grant Consortium and Clarice Koby; NASA and the NASA Graduate Student Re-
searchers Program for supporting my research; my NASA mentors, Dr. Jonathan A.
R. Rall, Mr. Joe Kujawski, and Dr. Doug O'Handley; the students of the Space
Science and Engineering Laboratory (SSEL), especially Dr. Brian Larsen; Dr. Lei
Meng, Dr. Gregg Switzer, and the students of the Optical Remote Sensor Laboratory,
especially Nathan Seldomridge, Dr. Nathan Pust, Amin Nehrir, and Nick Jurich.
vi
TABLE OF CONTENTS
1. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
The Importance of Atmospheric Water Vapor Monitoring . . . . . . . . . . 1
Using Lidar to Measure Atmospheric Water Vapor . . . . . . . . . . . . . 2
The Montana State University Water Vapor Dierential Absorption Lidar 6
2. THEORY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Earth's Atmosphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Remote Sensing With Lidar . . . . . . . . . . . . . . . . . . . . . . . . . . 8
The Lidar Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Radiometric Derivation of the Lidar Equation . . . . . . . . . . . . . . . . 13
Full Overlap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Partial Overlap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Comparison With the Lidar Equation . . . . . . . . . . . . . . . . . . 18
DIAL Equation Derivation . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3. HORIZONTAL TUNING EXPERIMENTS . . . . . . . . . . . . . . . . . . 24
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Steam Tunnel Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Roofport Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
System Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
External Cavity Diode Laser (ECDL) . . . . . . . . . . . . . . 31
Tapered Amplier . . . . . . . . . . . . . . . . . . . . . . . . . 36
Acousto-optic Modulator (AOM) . . . . . . . . . . . . . . . . 41
Reference Power Meters and Beam Splitter . . . . . . . . . . . 43
Telescope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
Narrowband Filter . . . . . . . . . . . . . . . . . . . . . . . . 45
Avalanche Photodiode (APD) Detector . . . . . . . . . . . . . 46
Multi-channel Scalar (MCS) . . . . . . . . . . . . . . . . . . . 48
Targets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
System Diagnostics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
P/I Curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Linewidth and Tuning . . . . . . . . . . . . . . . . . . . . . . 51
Reference Power Measurements . . . . . . . . . . . . . . . . . 52
CW Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Preliminary Pulsed Measurements . . . . . . . . . . . . . . . . . . . . 58
Pulsed Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4. WATER VAPOR ABSORPTION LINE SELECTION . . . . . . . . . . . . 64
Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
vii
Line Selection Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Temperature Sensitivity . . . . . . . . . . . . . . . . . . . . . . . . . 65
Optical Depth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Nearby Absorption Features . . . . . . . . . . . . . . . . . . . . . . . 79
Line Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5. DIAL SYSTEM AND EXPERIMENT . . . . . . . . . . . . . . . . . . . . 81
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
System Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Cascaded Tapered Ampliers . . . . . . . . . . . . . . . . . . . . . . 88
Multi-channel Scalar Card . . . . . . . . . . . . . . . . . . . . . . . . 91
Narrowband Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
Extended Tuning System . . . . . . . . . . . . . . . . . . . . . . . . . 93
Data Acquisition and Analysis Software . . . . . . . . . . . . . . . . . 100
Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Noise and Background Signal Levels . . . . . . . . . . . . . . . . . . . 101
Beam Shape . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Spectral Purity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
P/I Curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Linewidth and Tuning . . . . . . . . . . . . . . . . . . . . . . . . . . 114
On- and O-line Tuning Characterization . . . . . . . . . . . . 116
Absorption Cell Measurements . . . . . . . . . . . . . . . . . . 119
Reference Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
Data Runs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
6. CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
APPENDIX A: PRELIMINARY LIDAR EXPERIMENTS . . . . . . . . . 150
APPENDIX B: MAXIMUM PERMISSABLE EXPOSURE (MPE) LIMITS
AND FEDERAL AVIATION ADMINISTRATION (FAA) APPROVALS168
APPENDIX C: DIAL OPERATING GUIDE . . . . . . . . . . . . . . . . . 180
APPENDIX D: LIST OF ACRONYMS . . . . . . . . . . . . . . . . . . . . 187
viii
LIST OF TABLES
Table Page
3.1 Estimated pathlengths needed to obtain measurable absorptions due
to water vapor in the atmosphere for three candidate water vapor ab-
sorption lines within tuning range of MSU ECDL's, ordered left to
right from strongest to weakest. . . . . . . . . . . . . . . . . . . . . 25
4.1 Values for E ′′, α, and γL as determined by the linear t of Figure 4.1. 69
4.2 Optical Depth values for the water vapor absorption line centered at
12074.5689 cm−1 using a custom MATLAB analysis and HiTRAN-PC
simulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5.1 Laser transmitter requirements for water-vapor DIAL measurements
with an error due to individual laser properties of < 3% compared to
the Montana State University DIAL transmitter specications. . . . . 89
5.2 Ordering specications for the narrowband lter. . . . . . . . . . . . . 94
5.3 Results of testing for background light leakage into the APD. . . . . . 103
A.1 A comparison between the simulated and measured amplier output
powers for Construct 1 at the wavelengths of 1064.6 nm and 975 nm. 160
A.2 A comparison between the simulated and measured amplier output
powers for Construct 2 at the wavelength of 1064.6 nm. . . . . . . . . 163
A.3 A comparison between the simulated and measured amplier output
powers for Construct 3 in dierent congurations at the wavelength of
1064.6 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
B.4 Contact information for the FAA that I used in 2004. . . . . . . . . . 172
B.5 Contact information for the FAA, current as of February, 2007. . . . . 172
B.6 Specications of the FAA's four ight zones. (*Note that minimum altitude
requirements for these irradiance levels also apply. NM=Nautical Mile) . . . . . 173
ix
B.7 Values used for the maximum permissable exposure calculations. . . . 174
B.8 Contact information for the CDRH. . . . . . . . . . . . . . . . . . . . 177
B.9 Contact information for airports in Bozeman and Salt Lake City, Utah. 178
xLIST OF FIGURES
Figure Page
2.1 Schematics of Earth's atmospheric layers and composition. Images
courtesy of NASA
(http://lifto.msfc.nasa.gov/academy/space/atmosphere.html). . . . . 9
2.2 A cartoon outline of the basic principle behind lidar. See the text for
a detailed explanation of each step. . . . . . . . . . . . . . . . . . . 11
2.3 The most basic lidar conguration, consisting of a transmitting laser,
a scattering target, and a receiving telescope. The transmitted beam
laser is depicted with solid lines, while the telescope FOV is shown
with dotted lines. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.4 Partial overlap of the transmitted beam area (solid lines) with the
receiver's FOV (dotted lines). . . . . . . . . . . . . . . . . . . . . . . 17
3.1 A picture of the steam tunnel experiment. . . . . . . . . . . . . . . . 27
3.2 A schematic diagram of the horizontal tuning experiment. A tunable
ECDL is coupled into a tapered ared amplier. The amplied out-
put is sent out into the atmosphere and a 28-cm telescope is used to
collect the backscattered light. The light collected by the telescope is
measured using an avalanche photodiode operating in the Geiger mode. 29
3.3 A plot showing absorption features of atmospheric constituents of cur-
rent scientic interest in the visible and near-infrared optical spectrum,
compared to the availability of diodes with wavelengths in this region. 32
3.4 A schematic of a tunable external cavity diode laser in a Littman-
Metcalf conguration. The collimated light from a laser diode is inci-
dent on the diraction grating. The zeroth order reection is used as
the output from the external cavity laser while the rst order reection
is used to spatially separate the spectral output from the diode. The
prism serves as a retroreector to provide optical feedback to the diode
laser via a second reection from the diraction grating and is used to
control the operating frequency of the external cavity laser. Tuning is
achieved by rotating the prism. . . . . . . . . . . . . . . . . . . . . . 34
xi
3.5 A picture of an ECDL built at Montana State University. This laser
can be tuned from 824 nm to 841 nm. . . . . . . . . . . . . . . . . . . 35
3.6 A picture of the Sacher Lasertechnik TA830 tapered amplier. The
input mirror used for seeding the TA is visible in the center of the
picture. The collimating optics are on the far left. . . . . . . . . . . . 37
3.7 A picture of the Sacher Lasertechnik Pilot driver used to control the
tapered amplier in the horizontal tuning experiments. . . . . . . . . 38
3.8 A plot of the tuning of the laser system. This is shown by three over-
layed plots of the optical power of the injection seeded amplier (atten-
uated to avoid damaging the OSA). The ECDL was tuned mechanically
by adjusting the retroreective prism to 824 nm (832 nm, 841 nm) and
was used to seed the amplier. The spectral output of the amplier is
controlled by the spectral properties of the injection seeded ECDL laser. 38
3.9 A plot of the horizontal path transmission calculated using HiTRAN-
PC through a 1 km path length as a function of wavelength accessible
by the tunable ECDL shown in gure 3.8. The absorption features are
due to water vapor. . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.10 A plot of the pulsing characteristics of the Sacher Lasertechnik tapered
amplier. Faster pulses also have reduced output amplitudes, disap-
pearing almost completely below 4 µs, much too slow for use in the
DIAL system. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.11 A picture of the acousto-optic modulator (AOM) used to pulse the cw
laser beam. The entrance aperture is visible on the side, below the 5
in 1205C-2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.12 A plot of the pulsing characteristics of the AOM. Note the very clean
square-wave shape of the output pulses. . . . . . . . . . . . . . . . . . 42
3.13 A picture of broadband beam samplers. Image courtesy of Newport.com
(http://www.newport.com). . . . . . . . . . . . . . . . . . . . . . . . 43
3.14 A picture of the Newport 1830c (top) and 1930c (bottom), used for
diagnostics and reference power measurements in the horizontal tuning
and vertical DIAL experiments. . . . . . . . . . . . . . . . . . . . . . 44
xii
3.15 A schematic of a Schmidt-Cassegrain telescope. Image courtesy of
Celestron.com (http://www.celestron.com). . . . . . . . . . . . . . . . 45
3.16 A transmission plot for the Starbright coating on the telescope used
in the horizontal lidar tuning experiments. Image courtesy of Cele-
stron.com (http://www.celestron.com). . . . . . . . . . . . . . . . . . 46
3.17 A picture of the type of narrowband lter used in the horizontal tuning
experiments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.18 A picture of the avalanche photodiode (APD) detector. . . . . . . . . 48
3.19 A picture of the Stanford Research Systems Multi-channel Scalar (MCS)
with display screen used in the horizontal tuning experiments. . . . . 49
3.20 A picture of the three hard targets and their ranges used in the hor-
izontal tuning experiments, taken from the roof of Cobleigh Hall at
Montana State University, just above the roofport room where the
measurements were made. . . . . . . . . . . . . . . . . . . . . . . . . 49
3.21 Early results from the horizontal tuning experiments. The data should
show a water vapor absorption line centered at 829.022 nm with the
counts rising to a steady value in the wings of the line. . . . . . . . . 50
3.22 P/I curves for the ECDL and TA used in the horizontal tuning exper-
iments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.23 Optical spectrum analyzer traces for the ECDL and TA used in the
horizontal tuning experiments. . . . . . . . . . . . . . . . . . . . . . . 52
3.24 A plot of the ratio output wavelength of the amplier to the output
wavelength of the ECDL. The wavelengths agree to within the error of
the wavemeters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.25 A plot showing the power modes of the tapered amplier. As the ECDL
tunes, the power output of the TA changes drastically, necessitating
normalization of the return data with transmit power. . . . . . . . . . 54
xiii
3.26 A plot showing how the reference power measurement actually tracks
the transmit power of the horizontal system. Note that the Newport
1830c is more accurate. . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.27 A plot of the transmission through the atmosphere as a function of
wavelength near 829.02 nm. The closed (open) circles represent mea-
surements made for a 1.71 km (0.35 km) path length. The solid
(dashed) line is a theoretical calculation using HiTRAN-PC with the
measured temperature and humidity used in the modeling. . . . . . . 57
3.28 A plot of the transmission through the atmosphere as a function of
wavelength near 831.62 nm. The closed circles represent measurements
made for a 1.67 km path length. The solid line is a theoretical calcula-
tion using HiTRAN-PC with the measured temperature and humidity
used in the modeling. . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.29 A plot of the atmospheric transmission as a function of wavelength.
The open circles represent measurements made with a cw laser trans-
mitter while the lled circles represent measurements made with a
pulsed laser transmitter. The solid line represents the results of a
HiTRAN-PC calculation for a 1.67 km horizontal path calculation. . . 59
3.30 A plot of the relative transmission through the atmosphere as a func-
tion of wavelength near 829.022 nm. The open (closed) circles represent
measurements made for a 0.35 km (1.71 km) path length. The solid
and dashed lines are theoretical calculations using HiTRAN-PC with
the measured temperature, humidity, and path length. . . . . . . . . 62
3.31 A plot of the relative transmission through the atmosphere as a func-
tion of wavelength near 831.850 nm. The open (closed) circles represent
measurements made for a 0.35 km (1.67 km) path length. The solid
and dashed lines are theoretical calculations using HiTRAN-PC with
the measured temperature, humidity, and path length. . . . . . . . . 63
4.1 The top graphs are Figures 2 and 3 from Browell et al. (1991) showing
linear t relationships between the linewidth temperature dependence
parameter α, the Lorentz linewidth γL, and the ground state energy
E” in the 720nm region. These graphs are repeated here to show
consistency between my method and that of Browell et al. (1991). The
bottom graphs show the linear t relationships between these same
parameters in the 824nm to 841nm region. . . . . . . . . . . . . . . 70
xiv
4.2 The MATLAB integration function could not solve the Voigt integral
at low temperatures, sending the value of the integral to zero at a
certain point. A custom integration method needed to be written to
solve the integral down to 100K. . . . . . . . . . . . . . . . . . . . . 71
4.3 These gures show the number density error produced across a range
of temperatures, pressures, and E” values, generated using my MAT-
LAB Voigt prole calculator. They are nearly identical to Figures 4
and 5 from Browell et al. (1991), showing that the calculation method
is valid. The top left graph shows the number density temperature
sensitivity on a larger scale. The other graphs show the number den-
sity temperature sensitivity at 1.0 atm, 0.5 atm, and 0.25 atm pressures
scaled to a ±0.10%/K error. . . . . . . . . . . . . . . . . . . . . . . . 72
4.4 These gures show the mixing ratio error produced across a range of
temperatures, pressures, and E” values, generated using my MAT-
LAB Voigt prole calculator. They are nearly identical to Figure 6
from Browell et al. (1991). The top left graph shows the mixing ratio
temperature sensitivity on a larger scale. The other graphs show the
mixing ratio temperature sensitivity at 1.0 atm, 0.5 atm, and 0.25 atm
pressures scaled to a ±0.10%/K error. . . . . . . . . . . . . . . . . . 73
4.5 Deviation from the Voigt prole in the number density error calcula-
tions at 1.0 atm that would occur if a Lorentz prole were used. . . . 74
4.6 Deviation from the Voigt prole in the number density error calcula-
tions at 0.25 atm that would occur if a Lorentz prole were used. . . . 75
4.7 The average monthly water vapor density in Bozeman, Montana. The
water vapor density was calculated using temperature, pressure, and
relative humidity data collected by the Optical Remote Sensor Labo-
ratory's weather station on the roof of Cobleigh Hall at Montana State
University between August 2005 and July 2006. . . . . . . . . . . . . 77
4.8 A HiTRAN-PC plot at default values of 1 atm and 296K across a
1.5 km horizontal path length, showing all absorption lines between
12059.8167 cm−1 and 12086.0527 cm−1 (829.2-827.4nm). The nal on-
and o-line wavelengths selected for the water vapor DIAL are indicated. 80
xv
5.1 A plot of the maximum allowable error in pointing a bistatic laser
beam to keep it within the eld of view of the detector, as a function
of altitude. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5.2 A plot produced by Zemax showing the design for the DIAL receiver. 87
5.3 A plot of the reection for the Thorlabs E03 coating, used in all of the
mirrors in the DIAL system. Image courtesy of Thorlabs.com. . . . . 89
5.4 A picture of the Sacher Lasertechnik Pilot driver used to control the
cascaded tapered amplier in the vertical DIAL experiments. . . . . . 92
5.5 A plot of the DIAL's target water vapor absorption line at 828.187
nm (vacuum wavelength, n = 1.0) using HiTRAN-PC. The wavelength
shift due to two values of the index of refraction of air is shown. A value
of n = 1.00022 was calculated internally by the Burleigh wavemeter
using ambient temperature and relative humidity measurements. A
value of n = 1.0002896 was calculated using Cauchy's formula. The
red box signies a hypothetical rectangular band pass region of a 250
pm narrowband lter centered at 828.0069 nm, the line center location
using the wavemeter index of refraction. . . . . . . . . . . . . . . . . 94
5.6 A plot showing the transmission curve for the narrowband lter, with
the DIAL on-line (828.0069 nm) and o-line (828.1069 nm) labeled.
Data for the transmission was provided courtesy of Barr Associates. . 95
5.7 A schematic of the extended tuning system for the ECDL. . . . . . . 97
5.8 A schematic of the gas absorption cell experimental setup. Note that
the laser makes 36 passes within the gas absorption cell for a total path
length of 19.8 meters. . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
5.9 A picture of the DIAL instrument in its operational form inside the
roofport room of Cobleigh Hall. . . . . . . . . . . . . . . . . . . . . . 102
5.10 A BeamView gure showing the intensity of the ECDL beam after
traveling through the anamorphic prism pair, optical isolators, half
wave plate, and PBS cube. . . . . . . . . . . . . . . . . . . . . . . . 104
xvi
5.11 These BeamView gures show the tapered amplier beam shape at
3 dierent distances from the amplier, unseeded (left column) and
seeded by the ECDL (right column). The top row is at a distance of
5 inches, the middle row is at a distance of 10 inches, and the bottom
row is at a distance of 30 inches. The red lines are cross sections of the
intensity in two dimensions. The yellow ellipse shows the ellipticity of
the beam. Notice that the beam is much more uniform when seeded
by the ECDL, and becomes more circular as it propagates. . . . . . . 105
5.12 A plot of a water vapor absorption line at low pressure (high altitude)
or at high pressure, such as in a pressurized multi-pass, gas absorption
cell. A hypothetical Gaussian laser line with a 40 MHz linewidth is
overplotted for comparison. *See Wulfmeyer , 1998 for a description
of using a multi-pass gas absorption cell for making spectral purity
measurements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
5.13 A linear plot of the power output of the DIAL transmitter with a
narrowband lter in place, measured before and after the AOM through
a multi-mode ber on an OSA. . . . . . . . . . . . . . . . . . . . . . 109
5.14 These OSA traces show that the ASE structure in the DIAL transmit-
ter output is relatively independent of the type of ber used to make
the measurement, as the structure is similar before (left) and after
(right) the AOM for both single-mode and multi-mode bers. . . . . 109
5.15 These OSA traces show that the ASE structure in the DIAL trans-
mitter output is relatively independent of the presence of the AOM.
The structure is similar before and after the AOM for both multi-mode
(top row) and single-mode (bottom row) bers, with (right column) or
without (left column) a narrowband lter in place. . . . . . . . . . . . 111
5.16 These OSA traces show that the ASE structure in the DIAL transmit-
ter output is relatively independent of the length of ber optic cable
being used. The structure is similar before (left column) and after
(right column) the AOM for both multi-mode (top row) and single-
mode (bottom row) bers, with and without a narrowband lter in
place and for varying ber lengths. . . . . . . . . . . . . . . . . . . . 112
5.17 OSA traces showing the eect of using a narrow band lter in the
receiver of the DIAL. The spectral purity is drastically improved. . . 113
xvii
5.18 The latest P/I curve for the ECDL is plotted against the P/I curve per-
formed almost a year earlier during the horizontal tuning experiments
described in Chapter 3. Aging of the diode is evident as it requires
more current to reach the same power output. . . . . . . . . . . . . . 114
5.19 P/I curves for the two ampliers used in the DIAL experiment. Notice
the large dierence between ampliers seeded, but not saturated, and
seeded at nearly full seed power. . . . . . . . . . . . . . . . . . . . . . 115
5.20 An OSA trace of the ECDL output, showing its narrow linewidth and
large sidemode suppression. . . . . . . . . . . . . . . . . . . . . . . . 115
5.21 OSA traces of the outputs of the DIAL ampliers. The large increase
in amplied spontaneous emission due to higher drive current and seed
power for the second amplier is prominent, but disappears as expected
when the amplier is seeded properly. . . . . . . . . . . . . . . . . . . 116
5.22 A ratio of the laser wavelength measured after the AOM and after the
ECDL, showing that the wavelengths agree. . . . . . . . . . . . . . . 117
5.23 A plot of tuning time necessary to lock the laser system to the on-
line wavelength, for dierent initial starting PZT voltage settings, and
hence starting wavelengths, of the laser. . . . . . . . . . . . . . . . . . 118
5.24 A plot showing stable on- and o-line tuning of the laser system at
approximately one-hour intervals over a span of ve hours. . . . . . . 119
5.25 An expanded view of the second segment of Fig. 5.24 displaying the
computer-controlled feedback loop of the laser system, ne-tuning and
holding the laser output to the on-line wavelength, 828.187 nm. . . . 120
5.26 A plot of the water vapor number density in the absorption cell com-
pared to the number density in the atmosphere as measured by a
weather station on the roof of the building where the absorption cell
measurements were performed. Notice that the number density within
the absorption cell is always higher than that of the atmosphere due
to the presence of the liquid water reservoir connected to the cell. The
eect of adding humid air to the absorption cell can be easily seen. . 121
xviii
5.27 A plot of the normalized power transmitted through the 19.8 meter
path length of the gas absorption cell. Absorption by water vapor
molecules within the cell is responsible for the reduced on-line signal. 122
5.28 A plot of the relative transmission through the gas absorption cell as
a function of wavelength. The closed circles represent measurements
made by the laser system. The solid line is a theoretical prediction of
the absorption by water vapor in the absorption cell using HiTRAN-PC
with the in situ measurements of temperatures and humidity. . . . . 124
5.29 Reference power test measurements showing the relaxation eect of the
AOM on the left and the dierence between pulsed and cw measure-
ments of reference power on the right. . . . . . . . . . . . . . . . . . . 126
5.30 An example of initial DIAL data runs, in which the saturation and
afterpulse of the APD is apparent, and the striping eect caused by
the AOM can be seen. . . . . . . . . . . . . . . . . . . . . . . . . . . 127
5.31 The high-count correction factor for the APD detector. Data is cour-
tesy of Perkin-Elmer. . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
5.32 Example DIAL data in which the background and afterpulsing eects
were attempted to be removed, without success. . . . . . . . . . . . . 129
5.33 Evidence of alignment of the transmit beam with the receiver FOV.
The initial laser pulse is 75 meters wide while the ash seen by the
detector is over 225 meters wide, indicating return from atmospheric
molecules and particles in the near eld. Clouds can be seen at about
2.4 km altitude and decrease to about 1.8 km above the ground at the
conclusion of the experiment. . . . . . . . . . . . . . . . . . . . . . . 131
5.34 Time-averaged raw counts (right) and background-subtracted counts
(left) showing the initial laser pulse, atmospheric returns, and cloud
returns. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
5.35 A comparison between a radiosonde launched at MSU and one launched
in Great Falls around the same time. . . . . . . . . . . . . . . . . . . 133
xix
5.36 A comparison of an aligned (01/09/07) transmit beam and an un-
aligned (01/18/07) transmit beam. The left graph shows time-averaged
on-line counts and the right graph shows time-averaged o-line counts.
Notice the steep fallo in the unaligned (green) beam after the initial
laser pulse, whereas the aligned (red) beam is considerably longer due
to atmospheric returns. The spike above 1500 meters is probably due
to multiple scattering of the laser beam within the cloud. . . . . . . . 134
5.37 A water vapor prole from the DIAL compared to a MSU radiosonde. 135
A.1 Simulated output signal of the Ytterbium gain ber versus length. . . 157
A.2 A schematic of Construct 1. . . . . . . . . . . . . . . . . . . . . . . . 158
A.3 A schematic of Construct 2. . . . . . . . . . . . . . . . . . . . . . . . 158
A.4 A schematic of Construct 3. This schematic represents the nal ber
optic laser amplier design used for testing. . . . . . . . . . . . . . . 159
A.5 P-I curve for the seed laser. . . . . . . . . . . . . . . . . . . . . . . . 160
A.6 Spectrum of the seed laser. . . . . . . . . . . . . . . . . . . . . . . . . 161
A.7 P-I curve for the pump laser. A neutral density lter was used to
obtain the data in pink. . . . . . . . . . . . . . . . . . . . . . . . . . 161
A.8 Spectrum of the pump laser. . . . . . . . . . . . . . . . . . . . . . . . 162
A.9 Simulated and measured ASE of Construct 2. . . . . . . . . . . . . . 163
A.10 Simulated and measured results of the system. . . . . . . . . . . . . . 165
A.11 Measured gain curve of the ber amplier. . . . . . . . . . . . . . . . 166
xx
ABSTRACT
Water vapor is one of the most signicant constituents of the atmosphere because
of its role in cloud formation, precipitation, and interactions with electromagnetic
radiation, especially its absorption of longwave infrared radiation. Some details of
the role of water vapor and related feedback mechanisms in the Earth system need to
be characterized better if local weather, global climate, and the water cycle are to be
understood. Water vapor proles are currently obtained with several remote sensing
techniques, such as microwave radiometers, passive instruments like the Atmospheric
Emitted Radiance Interferometer (AERI) and Atmospheric Infrared Sounder (AIRS),
and Raman lidar. Each of these instruments has some disadvantage, such as only pro-
ducing column-integrated water vapor amounts or being large, overly customized, and
costly, making them dicult to use for deployment in networks or onboard satellites
to measure water vapor proles.
This thesis work involved the design, construction, and testing of a highly-tunable
Dierential Absorption Lidar (DIAL) instrument utilizing an all-semiconductor trans-
mitter. It was an attempt to take advantage of semiconductor laser technology to
obtain range-resolved water vapor proles with an instrument that is cheaper, smaller,
and more robust than existing eld instruments. The eventual goal of this project was
to demonstrate the feasibility of this DIAL instrument as a candidate for deployment
in multi-point networks or satellite arrays to study water vapor ux proles.
This new DIAL instrument transmitter has, for the rst time in any known DIAL
instrument, a highly-tunable External Cavity Diode Laser (ECDL) as a seed laser
source for two cascaded commercial tapered ampliers. The transmitter has the
capability of tuning over a range of ∼17 nm to selectively probe several available
water vapor absorption lines, depending on current environmental conditions. This
capability has been called for in other recent DIAL experiments. Tests of the DIAL
instrument to prove the validity of its measurements are presented. Initial water
vapor proles, taken in the Bozeman, MT, area, were taken, analyzed, and compared
with co-located radiosonde measurements. Future improvements and directions for
the next generation of this DIAL instrument are discussed.
1CHAPTER 1
INTRODUCTION
The Importance of Atmospheric Water Vapor Monitoring
Water vapor plays an enormous role in Earth's atmospheric dynamics through
cloud formation, precipitation, and interactions with electromagnetic radiation, es-
pecially its absorption of longwave infrared radiation (Harries, 1997). It is widely
agreed that water vapor is one of the most important gasses in the atmosphere with
regards to its role in local weather, global climate, and the water cycle. It is the pri-
mary driver of Earth's atmospheric heat engine and would have played a signicant
role on Mars if an environment suitable to the emergence of life ever developed there.
Especially with the growing concern for understanding and predicting global climate
change, detailed data of water vapor distribution and ux and related feedback mech-
anisms in the lowest 3 km of the troposphere, where most of the atmospheric water
vapor resides, are required to aid in climate models (Kington, 2000). Radiosondes
are the current standard method used to obtain routine water vapor proles, but
this technique can only provide information at one location at one point in time, is
not very well distributed globally, and cannot easily be used to monitor the spatial
and temporal changes of the water vapor concentrations (Turner et al., 2000). This
has led to the exploration of passive and active remote sensing instruments for water
vapor prole monitoring, such as microwave radiometers (Han and Westwater , 1995),
passive instruments like the Atmospheric Emitted Radiance Interferometer (AERI)
(Feltz et al., 2003) and Atmospheric Infrared Sounder (AIRS) (Divakarla et al., 2006),
2and Raman LIDAR (Turner et al., 2000). While each of these instruments shows
great utility, they also all have some disadvantage, such as only producing column-
integrated water vapor amounts or being large, overly customized, and costly, making
them dicult to use for deployment in networks or onboard satellites to measure wa-
ter vapor proles. Improved capabilities to monitor range-resolved tropospheric water
vapor proles continuously in time at many locations are needed (Rycroft , 2000).
Using Lidar to Measure Atmospheric Water Vapor
Light Detection and Ranging (Lidar) systems have been used to probe the atmo-
sphere since shortly after the invention of the laser itself (Schotland , 1966). Lidars
can be thought of as the optical equivalent to radars, since they actively illuminate
the atmosphere with radiation (from a laser in the case of a lidar) and measure the
subsequent scattered radiation. Many dierent types of lidars exist. Doppler lidars
are used to infer atmospheric wind speeds based on Doppler shifting of the radiation
scattered o of moving particles (McGill et al., 1997; Souprayen et al., 1999; Gentry
et al., 2000). High Spectral Resolution Lidar (HSRL) takes advantage of the dieren-
tial spectral broadening of lighter air molecules compared to heavier aerosol particles
to map and characterize aerosol types in the atmosphere (Grund and Eloranta, 1991;
Hair et al., 2001). Aerosol is a term used to describe dust, smoke, pollen, or gener-
ally anything in the atmosphere that is not air. Polarization lidars can be used to
distinguish between ice and water clouds (Sassen, 1991; Seldomridge et al., 2006), as
well as for more novel uses such as detecting sh (Churnside and Wilson, 2004) and
land mines (Shaw et al., 2005).
Two other types of lidars have been used to retrieve water vapor proles. Raman
lidars can locate certain gasses by detecting the very faint characteristic Stokes-shifted
3backscattered radiation caused by inelastic scattering. Dierential Absorption Lidar
(DIAL) is a technique that takes advantage of the dierential absorption of a laser
signal tuned on and o of a atmospheric constituent's absorption line. Signicant
progress has been demonstrated in developing and using both Raman and DIAL
high-performance lidars for proling atmospheric water vapor at isolated research sites
and in detailed process studies, from both the ground (Rall , 1994; Goldsmith et al.,
1998; Wulfmeyer , 1998; Wulfmeyer and Bösenberg , 1998; Bösenberg , 1998; Turner
et al., 2000; Wulfmeyer and Walther , 2001; Turner et al., 2002) and the air (Ehret
et al., 1993, 1998; Browell et al., 1998; Ismail et al., 2000). However, more routine
deployment of lidars for water vapor proling at multiple sites will require smaller,
mostly autonomous, lower-cost systems that are eye-safe. The trend in recent years
toward smaller and more robust lidar systems has resulted in a signicant increase in
the use of lidars for aerosol and cloud studies (Spinhirne, 1993; Rall and Abshire, 1996;
Shaw et al., 2001; Campbell et al., 2002; Intrieri et al., 2002), and a similar trend is
beginning to emerge for water vapor lidars. For example, previously large and complex
Raman lidar systems are now being packaged in moderately sized trailers and operated
routinely in long-term, largely unattended eld deployments (Rall and Abshire, 1996;
Goldsmith et al., 1998; Turner et al., 2000, 2002). However, even small Raman lidars
require high-power pulsed lasers because the Raman backscattering cross section is ∼
4 orders of magnitude below that of both Rayleigh and Mie backscatter (Measures,
1984; Grant , 1991). Raman lidar systems also require external calibration.
Dierential absorption lidar (DIAL) (Measures, 1984; Grant , 1991; Bösenberg ,
1998; Kovalev and Eichinger , 2004) systems also are moving toward smaller size and
lower-cost, which may someday allow deployment in unattended networks (Reagan
et al., 1993; Rall , 1994;Reagan et al., 1996; Prasad et al., 2000; Little and Papen, 2001;
Penchev et al., 2003; Machol et al., 2004). DIAL measurements require measuring
4backscattered light at wavelengths on and o an absorption line, with wavelengths
suciently close to each other that aerosol scattering and other systematic features
of the measurement cancel in a ratio. The ratio of the on- and o-line measurements
can be used to determine a vertical prole of the constituent concentration. One
advantage of a DIAL system compared to a Raman lidar is that the DIAL system
does not require as much transmit power, making the necessary laser smaller and the
eye-safe requirement easier to achieve. Another advantage is that the DIAL technique
is self-calibrating when careful attention is taken to account for all sources of error
in the measurement. The details of the DIAL technique will be explained further in
Chapter 2. Current systems have largely used solid state lasers, dye lasers, and optical
parametric oscillators, but these systems are not as small as what can be achieved
with diode laser transmitters.
One promising avenue of research toward compact water vapor DIAL instruments
is to use semiconductor laser transmitters (Reagan et al., 1993; Rall , 1994; Reagan
et al., 1996; Oh et al., 1999; Switzer , 1999; Prasad et al., 2000; Little and Papen, 2001;
Penchev et al., 2003; Machol et al., 2004). Diode lasers are compact, inexpensive,
can be tuned readily, and have good spectral coverage in the near infrared spectral
region where appropriate water vapor absorption lines exist. In the early 1990s,
the increased availability of high-power diode lasers and photon-counting avalanche
photodiode (APD) detectors led to the proposal of diode DIAL systems for boundary
layer water vapor proling (Reagan et al., 1993; Rall , 1994; Reagan et al., 1996).
However, remaining challenges included low laser power and hence low signal return,
spectrally broad laser pulses, and insuciently precise or stable laser tuning.
Several variations on the theme of diode laser transmitters with APD detectors
in photon-counting mode have been studied numerically (Reagan et al., 1993, 1996;
Oh et al., 1999; Penchev et al., 2003), but few systems have been implemented. Rall
5(1994) built a DIAL system using an externally modulated AlGaAs laser near 811.6
nm and achieved mean water vapor number density measurements that agreed with
measured humidity values to within 6.5% and 20% in integrated path (∼ 5 km hor-
izontal one-way path) and range-resolved (4 km horizontal one-way path) modes,
respectively. Oh et al. (1999) reported initial experimental results from a diode-
pumped Cr:LiSAF laser operating near 824.6 nm with a APD detector. They showed
on- and o-line water vapor absorption proles from the ground up to approximately
3 km, but the long time delay (∼ 1 hour) between measurements prevented the re-
trieval of a DIAL prole. Prasad et al. (2000) describe a Cr:LiSAF laser for DIAL
operation on a Unpiloted Airborne Vehicle (UAV), and show measurements from a
breadboard system that have loose agreement with a regional radiosonde prole up
to approximately 1.4 km.
Little and Papen (2001) reported nighttime data with multi-hour averaging times
from a ber-based lidar, showing reasonable agreement between their data, simula-
tions, and a regional radiosonde prole up to an altitude of 2 km. Most recently,
Machol et al. (2004) reported initial water vapor DIAL measurements from a system
based on a distributed feedback (DFB) diode laser used as a seed for a diode ared
amplier, operating at 823 nm with 0.8 nm of tuning. This system provides 0.15
µJ pulse energy at pulse repetition frequencies of 6-10 kHz. Their horizontal-path
measurements showed good agreement between the DIAL retrieval and surface in
situ sensor water vapor data; they also showed zenith measurements that agreed well
with a radiosonde prole between 800 and approximately 2500 m altitude. Machol
et al. note that new laser transmitter designs are needed for better spectral coverage
and larger tuning ranges, calling for a new laser [with] a tuning range that accesses a
larger selection of good water-vapor lines.... They also note that their DFB laser is no
longer available from the vendor, making it particularly important to have alternate
6laser transmitter sources.
The Montana State University Water Vapor Dierential Absorption Lidar
The research performed for this dissertation involves leveraging expertise in the
Montana State University (MSU) laser source development group to build a DIAL
system that is cheaper, smaller, and more robust than existing eld instruments and
able to access a large selection of water vapor lines using a widely tunable laser
transmitter. This transmitter, an External Cavity Diode Laser (ECDL), has the
ability to tune across a 17 nm spectrum near 830 nm, allowing it access to multiple
water vapor absorption lines of varying strengths. Because of this wide tunability, the
optimal absorption line for the DIAL technique in this region can be selectively probed
based upon existing atmospheric conditions. The DIAL uses an all-semiconductor
transmitter and for the rst time in any known DIAL instrument uses the ECDL as
a seed laser source for two cascaded commercial tapered ampliers to increase the
output power. The receiver uses a ber-coupled APD detector. Mostly commercial-
o-the-shelf components are employed so that the system, including the transmitter,
can be repaired quickly and relatively easily should a part fail, demonstrating a step
towards the robustness needed for eld deployment in multi-point arrays. The DIAL
is low-power, compact, with a desktop-sized footprint, and has the ability to be made
eye-safe. The goal of this project is to demonstrate that low-power DIAL instruments
using widely tunable diode laser transmitters, which can be designed at multiple
wavelengths, can achieve useful water vapor proles and are acceptable candidates
for use in multi-point lidar networks or satellite arrays to study water vapor ux
proles.
This dissertation describes the design, construction, characterization, and result-
7ing measurements of a water vapor DIAL using a widely tunable amplied ECDL
transmitter, and is summarized as follows. Chapter 2 discusses the background and
theory behind lidar, and especially DIAL, measurements. An introduction to the
atmosphere is given with particular emphasis on the terms, structure, and behaviors
relevant to this research. The mathematical framework for the rest of the dissertation
is presented. Signal-to-noise ratio calculations and model results are shown. Chapter
3 describes horizontal tuning measurements that were performed to test and validate
the laser transmitter. Chapter 4 explains the analysis undertaken to select a suitable
water vapor absorption line for use in the DIAL experiments, which are described in
detail with results in Chapter 5. Initial water vapor proles, taken in the Bozeman,
MT, area are analyzed. Necessary improvements to the DIAL system and concluding
remarks are given in Chapter 6. Appendix A describes preliminary lidar experiments
and the lessons learned from them. Appendix B gives detailed information required
to approve lidars for outdoor operation according to FAA rules. A user manual for
operating the DIAL instrument is found in Appendix C. Finally, a list of acronyms
is included as Appendix D.
8CHAPTER 2
THEORY
Earth's Atmosphere
Earth's atmosphere is a mixture of gases and a variety of small gas and liquid
particles (called aerosols) that extends over 300 miles or almost 500 km above the
ground. The composition of the atmosphere is primarily nitrogen, oxygen, and ar-
gon, with trace amounts of water vapor, ozone, carbon dioxide, and other gases, as
shown in gure 2.1. It comprises several distinct layers, distinguished by their heating
properties. The troposphere is the layer nearest to the ground, and is where nearly
all weather occurs. It is characterized by a decrease in temperature with increasing
altitude of ∼ 6.5◦C/km on average, called a temperature lapse rate. Simply because
of gravity, most of the atmosphere by mass is located in the bottom few kilometers
of the troposphere. The boundaries of the layers are very dynamic, but typically the
stratosphere will begin around 10 km above the ground, and is dened by a heating
of the atmosphere by absorption of ultraviolet radiation. The ozone layer is located
in the stratosphere. Above the stratosphere are two other layers, the mesosphere,
or middle atmosphere, and thermosphere, or upper atmosphere. The thermosphere
contains the ionosphere, where Aurora occur, and is where the atmosphere merges
with interplanetary gases or space (Battan, 1984).
9Figure 2.1: Schematics of Earth's atmospheric layers and composition. Images cour-
tesy of NASA
(http://lifto.msfc.nasa.gov/academy/space/atmosphere.html).
10
Remote Sensing With Lidar
One way to probe the atmosphere is through the use of LIght Detection And
Ranging, or lidar. Lidar is a form of active remote sensing, where laser radiation is sent
into the atmosphere to interact with it, and the results of this interaction are studied
to learn more about the atmosphere. A schematic of the basic principles behind lidar
is shown in gure 2.2. Laser pulses are emitted from a laser transmitter (1) with
some pulse length, τ (2). The laser interacts with atmospheric gasses and aerosols
through scattering and absorption (3) characterized by a atmospheric transmission
factor, T . When the laser photons scatter o of an atmospheric constituent (4), the
scatter strength, probability, and direction are governed by a scattering function, β.
Some of the scattered photons then may travel back through the atmosphere (5) again
aected by a transmission factor, T . If a lidar receiver is aimed such that its eld
of view (FOV) contains a portion of the scattered radiation path (6) according to an
overlap function, ξ(R), its entrance aperture will collect a fraction of these photons
according to the solid angle that the telescope subtends as seen from the scattering
source, or Ar/R2 (7). This receiver will then have some spectral eciency as well
as a detector eciency (8 and 9) that will cause some of the received photons to be
lost, ξ(λ). Finally, whatever photons do reach the detector are converted to electrons
through some mechanism and are counted as a function of time elapsed from the
laser pulse, which is directly related to the altitude of the scattering source in the
atmosphere (10).
From this illustration, it can be seen that an equation for lidar can be built from
this simple understanding, relating the counted signal photons or received power to
a modication of the transmitted photons or power by the atmosphere. In fact,
most derivations of the lidar equation do build the equation term-by-term through
11
Figure 2.2: A cartoon outline of the basic principle behind lidar. See the text for a
detailed explanation of each step.
12
these physical arguments based on a basic lidar setup and the typical response of the
environment. However, the lidar equation is also derivable simply by understanding
the radiometry of the situation. This derivation is rarely seen, and thus is shown
here, displaying that the lidar equation is in agreement with the basics of radiometry.
Through this process, the denitions of the unit volume backscatter coecient, β,
and the lidar geometrical compression form factor, or overlap function ξ(R), are
elucidated.
The Lidar Equation
The lidar equation is used in a number of dierent laser ranging experiments to
solve for or estimate various atmospheric parameters, using techniques as varied as
elastic (Mie or Rayleigh) scattering, inelastic (Raman) scattering, dierential absorp-
tion, and uorescence. The lidar equation is stated most often in a form similar to
(see, for instance, Measures, 1984, or Kovalev and Eichinger , 2004)
Pr(λL, R) = Pt
Ar
R2β∆zT
2ξ(λL)ξ(R), (2.1)
where Pr(λL, R) [W ] is the power received by the photodetector at the laser wave-
length, λL [m], from range R [m], Pt [W ] is the power transmitted by the laser, Ar [m2]
is the entrance pupil area of the receiver (typically a telescope), β [km−1sr−1] is
the unit volume backscatter coecient, ∆z = cτ/2 [m] is the range bin with c =
2.998× 108 m/s being the speed of light and τ [s] the laser pulse width, T 2 [unitless]
is the round-trip transmission factor through the atmosphere, ξ(λL) [unitless] is the
spectral transmission of the system at λL, and ξ(R) [unitless] is the lidar geometrical
compression form factor, also known as the overlap function. This function is equal
to 1 when the transmitted laser beam area lies completely within the eld-of-view
13
(FOV) of the system, known as full overlap, and is a complicated function otherwise
(see Measures, 1984, for example).
Since many low-power lidar applications require photon-counting detectors, the
lidar equation is often written in terms of transmitted and detected photons using
the conversion Pt = nt(hc)/(λτ), where nt is the number of transmitted photons and
h = 6.626× 10−34 J·s is Planck's constant. T 2, which will not be discussed in detail
here, is given by the Beer's Law relationship
T 2 = exp
(
−2
∫ R
0
α(r, λL)dr
)
, (2.2)
where α [m−1] is the atmospheric extinction coecient including scattering and ab-
sorption (Stephens, 1994). T 2 and ξ(λL) can be viewed in their simplest form as
eciency factors that modify the output power of the laser, and so will be folded
into one term that ranges between 0 and 1, and will be taken to be constant for the
purposes of this derivation. T 2 is of course not a constant, and in fact contains most
of the interesting physics involved with doing lidar experiments, especially in the case
of Dierential Absorption Lidar (DIAL). Yet, as will be shown below, it is irrelevant
to the most basic radiometric derivation of the lidar equation, and would only com-
plicate the derivation if a non-constant atmosphere were to be included. With this in
mind, the lidar equation can be rewritten as
Pr = Pt
Ar
R2β∆zξ(R)[eciency factors]. (2.3)
14
Radiometric Derivation of the Lidar Equation
The radiometric derivation of the lidar equation will be examined in two situations:
rst, where the lidar system is in full overlap, and second, where the transmitted
laser beam area overlls the FOV of the system or lies partially outside of it (partial
overlap).
Full Overlap
Consider the most basic lidar setup, in which laser light from a laser transmitter is
backscattered by atmospheric constituents, and a fraction of this backscattered light
is collected by the receiver telescope, shown in Figure 2.3. It is assumed that the entire
area being illuminated, Ai, is within the full FOV of the telescope. For a thorough
treatment of the full FOV see Stelmaszczyk et al., 2005. The laser is transmitting
power Pt with perfect eciency into the atmosphere. In reality, Pt would be modied
by the eciency of the transmission optics, which is contained within the system's
spectral transmission factor ξ(λ) as dened above, and assumed to be 1 here for
simplicity. The receiver will be represented simply as a focusing lens of area Ar with
a detector of area Ad located at the focal point of the telescope, a distance f behind
the entrance pupil. To further simplify the situation, it is assumed that the light is
scattered entirely within a scattering plane at altitude R, and parallel to the receiving
lens plane. Multiple scattering is not taken into consideration.
In general, the radiance [W/(m2 · sr)] at a scattering plane is given by
L = PA · Ω , (2.4)
where P is the power being scattered from an area A into a total scattering solid
15
Figure 2.3: The most basic lidar conguration, consisting of a transmitting laser, a
scattering target, and a receiving telescope. The transmitted beam laser is depicted
with solid lines, while the telescope FOV is shown with dotted lines.
angle Ω (Schott , 1997). For the lidar situation described, this becomes
Ltarget =
Pt
Ai · Ωtarget
, (2.5)
where Ai is the illuminated area, and Ωtarget is pi if the scattering plane is Lambertian,
or could be a complicated function of wavelength and angle, as is common in the
atmosphere. It is assumed for the moment that Ai is perfectly scattering in two
ways. First, the entire area, not just a fraction of Ai, is scattering the incoming light,
so that
Ascatter
Ai
= 1, (2.6)
where Ascatter is the area actually causing scattering. Second, all of the incoming
light is scattered, so that Ai has a perfect reectivity ρ = 1. In actual lidar mea-
surements, these two eciencies must be taken into account, as will be seen in the
lidar-comparison section below.
The power at the receiver is given by multiplying the target radiance, Ltarget, by
16
the receiver throughput, or A · Ω product,
Pr = LtargetAΩ. (2.7)
The throughput consists of an area and a projected solid angle as viewed from that
area. In this case, there are two equivalent ways to describe the throughput. If
the area selected is the receiver area, Ar, then the solid angle that must be used is
the solid angle subtended by the scattering area as seen from the receiver, Ai/R2.
Conversely, if the area selected is the area of the scattering source, Ai, then the solid
angle that must be used is the solid angle subtended by the receiver area as seen from
the source, Ar/R2. It is assumed that A  R in both cases so that the small-angle
approximation of the projected solid angle may be used. Both methods of dening
the throughput are identical and can now be used in equation (2.7) to write the power
at the receiver,
Pr = LtargetAi
Ar
R2 . (2.8)
Substituting for Ltarget with equation (2.5) , the received power at the telescope in
the case of full overlap is
Pr = Pt
Ar
R2
1
Ωtarget
. (2.9)
Partial Overlap
Now consider the situation for which the divergence and/or pointing angle of the
transmitting laser causes the illuminated area to be only partially contained within
the FOV of the system, known as partial overlap. The same assumptions are made
for the scattering target. The situation is depicted in Figure 2.4.
17
Figure 2.4: Partial overlap of the transmitted beam area (solid lines) with the re-
ceiver's FOV (dotted lines).
The radiance at the scattering plane for this situation is given again by equation
(2.5). The power at the receiver, however, will now be written as
Pr = LtargetAr
Aoverlap
R2 . (2.10)
Notice that Ai has in this case been replaced by Aoverlap, the area of overlap between
the area seen by the receiver and Ai. This can be understood in the above equation
as the receiver pupil only receiving light from the solid angle subtended by the area
illuminated within its FOV, Aoverlap/R2. Equivalently, by reversing the areas, it
indicates that only the light scattering from area Aoverlap will be collected by the solid
angle subtended by the receiver Ar/R2 as seen from the scattering plane. Combining
equations (2.5) and (2.10) gives the power at the receiver while in partial overlap,
Pr = Pt
Ar
R2
1
Ωtarget
(Aoverlap
Ai
)
. (2.11)
The power received at the receiver in the case of partial overlap diers from the
full overlap case (equation (2.9)) only by the term in parentheses. This term is the
18
overlap factor,
Aoverlap
Ai
= ξ(R), (2.12)
as discussed above (Stelmaszczyk et al., 2005). When Aoverlap contains the entire
illuminated area, Ai, the system is in full overlap and ξ(R) = 1.
Comparison With the Lidar Equation
Equation (2.11) is similar to equation (2.3), the lidar equation. In fact, it is
identical if it can be shown that the term β∆z is equal to Ω−1target. β generally is
dened as
β = σ ·N (2.13)
where σ [m2sr−1] is the dierential scattering cross section of the target and N [m−3]
is the number of particles per unit volume involved with the scattering. In lidar
setups where the transmitter is close to the receiver compared to R, σ is taken to
be the backscattering cross section, σpi. It is assumed that the range bin ∆z is small
compared to R, such that σpi is constant across the entire range bin. In simplest
terms, σpi depends on the eective scattering area and backscattering solid angle of
the particle,
σpi =
A
e, particle
Ωparticle
. (2.14)
The particles involved with the scattering process, n, are only those that are
illuminated by the laser, contained within a volume, Vi, dened by the laser pulse,
N = nVi
, (2.15)
19
where
Vi = Ai ·∆z. (2.16)
Combining σ and N in equation (2.13) using equations (2.14), (2.15), and (2.16)
gives the unit volume backscatter coecient of the illuminated volume,
β = Ae, particle · nΩparticle · Ai ·∆z
, (2.17)
and,
β∆z = Ae, particle · nΩparticle · Ai
. (2.18)
Notice that for an atmosphere constant in composition over the time scale of a laser
pulse, n ∝ Ai such that as Ai increases, n increases proportionally, making n/Ai con-
stant. A
e, particle and Ωparticle are simply properties of the atmospheric constituent
causing the scattering, and β∆z is therefore constant for a given wavelength, li-
dar setup, and set of atmospheric conditions. Ignoring particle shadowing eects,
A
e, particle · n is the total area of the scattering particles in the volume Ai · ∆z, or
simply the area from which the incoming light is being scattered, Ascatter, as dened
in section 2. Ωparticle is the single-particle solid angle response averaged over many
particles, and therefore the solid angle of a group of particles will average to this
same quantity when the receiver is close to the transmitter, so that Ωparticle = Ωtarget,
assuming that σ is equal to σpi for each particle in the volume. Taking these points
together, or more rigorously, integrating the dierential scattering cross section σ over
the illuminated volume, gives
β∆z = 1Ωtarget
Ascatter
Ai
, (2.19)
20
where Ωtarget is the same quantity rst introduced in equation (2.5). If the scattering
medium is completely opaque and perfectly reecting, as was assumed for the targets
in the Full Overlap and Partial Overlap sections above, then equation (2.6) holds true
and equation (2.19) reduces to
β∆z = 1Ωtarget
, (2.20)
as desired. In realistic lidar measurements the atmosphere is neither completely
opaque nor perfectly reecting, such that Ascatter < Ai and ρ < 1. These actual scat-
tering eciencies then have to be considered and will reduce the power backscattered
to the receiver.
Returning to the simple lidar situation, equation (2.11), the power at the receiver
for a lidar system in partial overlap, can now be rewritten with the results found in
equations (2.12) and (2.20):
Pr = Pt
Ar
R2β∆zξ(R). (2.21)
Including the eciency factors that modify the transmitted power in a real lidar
system gives equation (2.3), the lidar equation:
Pr = Pt
Ar
R2β∆zξ(R)[eciency factors]. (2.22)
Therefore, the radiometric derivation for the power received in a partially or fully
overlapped lidar system is equivalent to the lidar equation itself, showing that the
lidar equation can be derived through simple radiometric arguments. All terms in the
lidar equation originate from terms dened by a simple radiometric understanding of
the basic lidar setup. This derivation elucidates that the overlap function, or lidar
21
geometrical compression form factor ξ(R), is fundamentally a ratio of the area of the
transmitted beam that overlaps with the receiver FOV. It is also shown that the unit
volume backscatter coecient multiplied by the range bin in the lidar equation, β∆z,
is equivalent to the inverse of the radiometric target backscatter solid angle, Ω−1target.
Both cases of full overlap and partial overlap of the transmitted laser beam and the
receiver FOV are consistent with their radiometric counterparts.
DIAL Equation Derivation
It is clear from inspection of the lidar equation, equation 2.1, that many terms can
be directly measured in the laboratory. However, there are some terms that cannot
be easily measured or estimated, such as the unit volume backscatter coecient, β, or
the atmospheric extinction coecient, α, inside the atmospheric transmission factor,
T , which can be expanded as
α(r, λL) = κ(r, λL) + σ(r, λL)N(r), (2.23)
where κ(r, λL) [km−1] is the atmospheric extinction factor due to all extinction ex-
cluding absorption and σ(r, λL)N(r) [km−1] is the atmospheric extinction due to ab-
sorption, where σ(r, λL) [m2] is now the absorption cross section of an atmospheric
constituent and N(r) [m−3] is the number density of absorbing molecules. One di-
culty with all lidar measurements is that there are more unknowns in the lidar equa-
tion, meaning estimations have to be made of the value of some terms in order to
determine others, leading to uncertainties. One technique that is used to circumvent
some of these uncertainties is called Dierential Absorption Lidar (DIAL).
The theory behind DIAL is to form the dierence in the logarithm of Pr(λL, R)
22
evaluated at ranges of R and R + ∆R at two wavelengths: λon located on the line
center of an absorption line of some atmospheric constituent such as ozone or water
vapor, and λoff located some spectral distance away from the absorption line center,
in another part of the absorption continuum for the case of ozone or completely clear
of the absorption line in the case of water vapor (Schotland , 1974). From equation
2.1 then,
lnPon,off (R)− lnPon,off(R + ∆R) = ln
{
Pt,on,off
Ar
R2βon,off (R)∆z
·exp
(
−2
∫ R
0
α(r, λL)dr
)
· ξ(λL)on,offξ(R)
}
−ln
{
Pt,on,off
Ar
(R + ∆R)2βon,off (R + ∆R)∆z
·exp
(
−2
∫ R+∆R
0
α(r, λL)dr
)
·ξ(λL)on,offξ(R + ∆R)} (2.24)
Since Pt,on,off , Ar, ξ(λL), and ∆z are the same at R and R + ∆R, they will
cancel out of the equation. Evaluating the integrals, combining the range terms, and
simplifying leaves
lnPon,off (R)− lnPon,off(R + ∆R) = lnβon,off (R)− lnβon,off (R + ∆R)
+ln
(
1 + 2∆RR +
(∆R
R
)2
)
+lnξ(R)− lnξ(R + ∆R)
+2κon,off(∆R) ·∆R
+2σon,off(∆R) ·N(∆R) ·∆R. (2.25)
23
Now, if the o-line terms are subtracted from the on-line terms,
(lnPon(R)− lnPon(R + ∆R)) − (lnPoff (R)− lnPoff (R + ∆R))
= ln βon(R)βon(R + ∆R)
− ln βoff (R)βoff (R + ∆R)
+2 (κon(∆R)− κoff (∆R)) ·∆R
+2 (σon,(∆R) ·N(∆R)− σoff (∆R) ·N(∆R))
·∆R. (2.26)
Notice that the overlap factor is completely removed from the equation. If the
assumption is made that the on-line and o-line wavelengths are so spectrally close
that the unit volume backscatter coecient, β, and atmospheric extinction factor, κ,
are unchanged at the two wavelengths, the equation simplies and N(R + ∆R) can
be solved for, forming the DIAL equation,
N(R + ∆R) = 12(σon − σoff )∆R
ln
[P (R)r,onP (R + ∆R)r,off
P (R + ∆R)r,onP (R)r,off
]
. (2.27)
Using the DIAL equation allows a measurement of the number density as a function of
range to be solved for, without having to know many other atmospheric parameters.
σon, off can be calculated using HiTRAN 2000, a radiative transfer database (Rothman
et al., 2003), and inputs from a colocated radiosonde for increased accuracy. The other
quantities are known.
24
CHAPTER 3
HORIZONTAL TUNING EXPERIMENTS
Introduction
Prior to attempting vertically pointing water vapor DIAL experiments, horizon-
tally pointing lidar experiments were completed to test and verify various components
similar to those that were used in the vertical water vapor DIAL system, including
in particular, the actual tuning of the laser. The rst step towards constructing a
water vapor DIAL with a widely tunable diode laser transmitter is to verify that
the transmitter has the ability to tune on and o of water vapor absorption lines
while operating in a lidar conguration. To accomplish this task, the laser beam
must be tuned to a water vapor absorption line and transmitted across a pathlength
containing water vapor, and a measurement must be made of how much laser power
is absorbed by the water vapor. The laser must then be tuned completely o of the
absorption line, and the power measurement should verify that the absorption due to
water vapor is no longer present.
Several experiments at dierent locations were attempted without success before
the horizontal lidar was moved to the roofport room of Cobleigh Hall on the MSU
campus. There, continuous wave (cw) measurements were made as an initial test
of the system before pulsed measurements were attempted. The horizontal lidar
transmitter used the same External Cavity Diode Laser (ECDL) and injection-seeded
tapered amplier, operating near 830 nm, that was used in the vertical water vapor
DIAL system. It was aimed at hard targets, increasing the return signal by orders of
25
magnitude over a vertically-pointing lidar that only receives backscattered photons
from the atmosphere, allowing the tuning to be tested without needing to optimize
a weak return signal. The steps taken to arrive at these measurements are described
in this chapter. These early measurements, as well as a full description of the ECDL
and lidar setups, are described in the literature (Obland et al., 2005, 2006a,b).
Steam Tunnel Experiments
An estimate of the pathlength needed to measure absorption by water vapor was
needed before any absorption experiments could be designed. Modifying equation
2.2 for a one-way, homogeneous pathlength gives T = e−αR. Using this equation
and assuming that the absorption extinction coecient is completely due to water
vapor absorption such that there is no scattering allows for a simple estimation to be
made of the range needed to measure water vapor absorption. Table 3.1 shows the
estimated pathlengths needed for three candidate water vapor absorption lines within
the tuning range of ECDL's that have been built at MSU, ordered left to right from
strongest to weakest. The absorption coecients were taken from HiTRAN 2000, a
radiative transfer database (Rothman et al., 2003), calculations using a US Standard
Atmosphere, which tends to be more moist than the atmosphere around Bozeman,
meaning that these absorption results are probably overestimates.
Wavelength (nm) 834.459 831.615 850.818
Distance to 10% Absorption 387 m 450 m 2564 m
Absorption at 440 m 11.30% 8.90% 1.80%
Absorption at 880 m 21.30% 16.90% 3.60%
Table 3.1: Estimated pathlengths needed to obtain measurable absorptions due to
water vapor in the atmosphere for three candidate water vapor absorption lines within
tuning range of MSU ECDL's, ordered left to right from strongest to weakest.
26
It is immediately apparent that in the 830 nm region of the spectrum, even a rela-
tively strong line such as the one at 834.459 nm is still too weak to give a measurable
absorption in transmitted power over a range of less than several hundred meters.
This fact makes a lab measurement of absorption very dicult to accomplish without
access to a gas absorption cell with a very long pathlength. A longer pathlength,
preferably within a controlled space, was required.
The rst idea conceived for accomplishing the long-pathlength absorption exper-
iments was to perform the experiments in the steam tunnels running underneath the
Montana State University campus. Access to the steam tunnels is tightly controlled,
greatly minimizing the chance of someone being injured by stepping into the laser
beam. Tours of the tunnels showed that the water vapor content of the air nearly
matched the ambient level of the outside air. The longest one-way pathlength was
about 220 meters, but could be increased to 440 meters or 880 meters by using mirrors
to reect the beam back to the detector. Performing experiments in the steam tunnels
was a dicult endeavor due to the lack of elevator access, meaning all equipment had
to be carried by hand down several ights of stairs. Also, compatible power outlets
were hard to nd, limiting the experiment location to one spot within the tunnels.
Several data runs were performed in the steam tunnels with a ECDL centered
around a wavelength of 850.818 nm. The laser and associated instruments were set
up on heavy carts and optical tables. The ECDL output was expanded, collimated,
and sent down the length of the tunnel. Two, 2-inch mirrors were mounted on a
portable optical table at the other end of the tunnel to make the rst and third
reections. After the third reection, the laser returned to the starting point of the
tunnel, where the beam was focused onto a detector. A reference power measurement
of the original transmitted beam was made as well, to remove uctuations in the laser
power from the nal results. A picture of the experiment is shown in gure 3.1.
27
Figure 3.1: A picture of the steam tunnel experiment.
None of the data runs in the steam tunnels succeeded in measuring absorption in
the laser beam. Tests with a Helium-Neon laser, visible at 633 nm, showed that me-
chanical vibrations within the tunnels were signicant, making it dicult to keep the
laser beam aligned on three mirrors and a detector simultaneously. Even after placing
all mirrors on sand bags to dampen these mechanical vibrations, thermal variations
along the length of the tunnels (measured to be ±10◦ at times) still caused unaccept-
able beam distortion and misalignment due to air turbulence. Even if alignment was
not an issue, trying to nd absorption on the level of a few percent would still be
dicult or impossible to measure due to the power variations in the laser. A dierent
laser with access to stronger water vapor absorption lines would be required, as well
as a new experiment location with less vibration and thermal variation.
A suggested possible solution to these problems was to perform horizontal ab-
sorption measurements on the roof of Cobleigh Hall on the MSU campus, using other
buildings around campus or in the city of Bozeman as hard reection targets. How-
28
ever, because the weather could not be controlled and an adequate, accurate pointing
mechanism for the laser would be dicult to design, the decision was made to perform
horizontal, hard target absorption measurements through the window in the roofport
room, on the sixth oor of Cobleigh Hall. The environment and beam-pointing could
be well controlled within this room. These horizontal absorption experiments from
the roofport room are described in the rest of this chapter.
Roofport Experiments
The ECDL used in the steam tunnel experiments was replaced with a tunable
ECDL transmitter with a center wavelength near 830 nm, which was operated with
a lidar receiver to measure atmospheric transmission across water vapor absorption
lines centered at 829.022 nm and 831.615 nm. To focus on the tuning ability of this
transmitter, backscatter from three distant hard targets was measured with the laser
operating initially in cw mode. Following the cw measurements, the transmitter was
operated in a pulsed mode, but with pulses of longer temporal duration than would be
employed in the actual atmospheric DIAL system. While the pulse length prohibited
range resolution, the experiment was performed as an initial demonstration that the
transmitter can maintain its performance characteristics when pulsed. Finally, pulses
with widths of 500 ns (range resolution of 75 m) were used to demonstrate that the
system could perform the necessary tuning under realistic DIAL conditions.
System Description
The experimental setup is shown in gure 3.2. The horizontal tuning experi-
ments were built around several key components, which are described in more detail
individually below.
29
Isolator l/2ECDL
NBFilter
Isolator PBSIris Iris
Iris Iris
To OSA
To OSA
Tapered
Amplifier
l/2
PBS
Isolator
Collimating
Lenses
Light received by telescope
Telescope
Mount
4 Foot
2 Foot
4% Beam
Splitter
To Reference Detector
To APD
Detector
Telescope
(Above Optics)
AOM
Figure 3.2: A schematic diagram of the horizontal tuning experiment. A tunable
ECDL is coupled into a tapered ared amplier. The amplied output is sent out into
the atmosphere and a 28-cm telescope is used to collect the backscattered light. The
light collected by the telescope is measured using an avalanche photodiode operating
in the Geiger mode.
30
The output of the ECDL passes through two Faraday isolators to prevent optical
feedback from aecting the performance of the ECDL or damaging the diode laser.
After the isolators, the light is incident on a half-wave plate and polarizing beam
splitter (PBS). One polarization output of the PBS is launched into an optical ber
for monitoring the ECDL wavelength on a wavemeter with a resolution of 10 MHz,
while the second polarization output of the PBS was free-space coupled into the
tapered amplier by using two irises for alignment. By rotating the half wave plate,
the amount of light sent to the amplier and ber can be adjusted. The output of the
amplier was collimated and sent through a Faraday isolator preventing damage to
the tapered amplier from optical feedback. Light was next incident on a second half-
wave plate and PBS. One polarization output of the second PBS was launched into an
optical ber for monitoring the output of the tapered amplier on an optical spectrum
analyzer (OSA) with a resolution of 0.1 nm. For the cw and initial pulsed experiments,
the second polarization output of this PBS was transmitted o of the optical table
after being directed through two widely-spaced irises for precision alignment with the
chosen target. For the faster pulsing experiments, the second polarization output of
the PBS was rst sent through an acousto-optic modulator (AOM) that pulsed the cw
beam, as described below, and then transmitted o of the optical table through the
two irises. In all experiments, the transmit beam was passed through a ∼4%-reective
beam splitter before leaving the table. The reected signal was sent to a reference
detector to monitor changes in the transmit power as the transmitter tuned. These
uctuations were normalized out of the nal data. The remaining 96% of the light
was transmitted into the atmosphere. Alignment was achieved by visually aiming
the instrument such that the target area dened by the irises was in full view of the
telescope. Care was taken to avoid any visible steam vents or exhaust plumes along
the horizontal path length. The nal transmit power from this system was typically
31
between 80 mW and 120 mW, although this was again reduced by several percent
because of reection o of the room window.
Light scattered from the atmosphere and the distant hard target was collected
by an optical receiver that employed a Schmidt-Cassegrain telescope, sent through a
10-nm-wide interference lter centered at about 830 nm and a 650-µm core-diameter
optical ber, to a photon-counting avalanche photodiode (APD) detector module
operating in photon-counting mode. The telescope eld of view is about 13.5 mrad,
while the transmitted beam divergence was 1.43 mrad, ensuring that the beam spot
at the target was always completely within full overlap with the receiver. A multi-
channel scalar (MCS) was used to count the APD pulses and bin them in time.
Laser tuning and data acquisition were operated via computer control and Lab-
VIEW software. The control program simultaneously measured both the wavelength
of transmitted laser light and the reference beam power, initialized the MCS and
other instruments used to operate the experiment, and began data collection. Since
distance resolving did not need to be considered for the cw and preliminary pulse
measurements, the mean count in all of the bins was calculated. For the faster puls-
ing experiments, counts in each bin were averaged for some amount of time, and
data analysis used the counts from the bin or bins containing signal reected from
the target. The reference power, mean count, and laser frequency at one wavelength
were recorded to a data le after which the computer tuned the laser by adjusting
the voltage applied to the piezo-electric tuner. The MCS counter was cleared and
the data acquisition process reinitiated until a scan across an absorption feature was
completed. The temperature of the ECDL was adjusted manually when necessary to
shift the mode-hop-free region across the absorption, allowing for a scan of greater
than 50 GHz.
32
400 600 800 1000 1200 1400 1600 1800 2000
H2O
CO2
O2
Diode Availability
Wavelength Regions of Absorption (nm)
Figure 3.3: A plot showing absorption features of atmospheric constituents of current
scientic interest in the visible and near-infrared optical spectrum, compared to the
availability of diodes with wavelengths in this region.
External Cavity Diode Laser (ECDL) The key to the entire water vapor DIAL de-
scribed herein is the ECDL built at MSU. External cavity diode lasers (ECDL's) have
found applications in a variety of areas, including molecular spectroscopy (Nguyen
et al., 1994; Aumiler et al., 2004) and as seed sources for dierential absorption li-
dar systems (Machol et al., 2004; Repasky et al., 2004; Obland et al., 2006a). Diode
lasers oer wide spectral coverage and tunability, especially in wavelength regions
containing absorption features of scientically interesting atmospheric constituents,
as shown in gure 3.3. ECDL congurations can narrow the spectral bandwidth and
raise the spectral purity of the diode output. The laser source development group
at Montana State University has extensive experience building high-quality ECDL's,
which was leveraged to develop the highly tunable transmitter for the water vapor
DIAL. A detailed description of how the ECDL was designed and built can be found
in Switzer , 1999.
Several dierent type of external cavity congurations exist for diode lasers, but
33
placing the diode in a Littman-Metcalf external cavity conguration (Littman and
Metcalf , 1978) such that the output of the diode laser is incident on a diraction
grating allows for a stable tuning method to be employed. Another advantage of
this conguration is that the output of the ECDL remains pointed in the same di-
rection, unlike other congurations. A schematic of a ECDL in a Littman-Metcalf
conguration built at Montana State University is shown in gure 3.4. The output
of a diode laser is collimated and strikes a diraction grating at a grazing incidence
angle, spatially separating the wavelengths of the diode's broad spectral output. The
zeroth-order reection is used as the output for the ECDL. A retro-reecting roof
prism is tilted at the correct angle to direct one spectral component of the rst-order
reection back to the diode laser via a second reection from the diraction grating,
forcing the laser to run single-mode at the chosen wavelength. The spectral charac-
teristics of the ECDL are dened by the frequency of the light fed back into the diode
laser and the resonant condition that requires an integer number of half wavelengths
to t within the external cavity. Tuning the ECDL is accomplished by rotating the
roof prism around a pivot point (McNicholl and Metcalf , 1985; de Labachelerie and
Passedat , 1993) to change the frequency of light fed back to the diode laser while
simultaneously changing the external cavity length, so that a constant integer num-
ber of half wavelengths is maintained within the external cavity (Meng et al., 2000;
Repasky et al., 2001). If the optical cavity length changes as the roof prism is rotated
so that the same number of half wavelengths is maintained within the optical cavity
as the wavelength changes, continuous tuning will result. If this is not the case, a
mode hop will occur.
A picture of an ECDL that was built at Montana State University is shown in
gure 3.5. A 150-mW diode laser with a center wavelength of 830 nm (SDL-5421)
is collimated using an aspheric lens with a focal length of 4.5 mm and a numerical
34
Figure 3.4: A schematic of a tunable external cavity diode laser in a Littman-Metcalf
conguration. The collimated light from a laser diode is incident on the diraction
grating. The zeroth order reection is used as the output from the external cavity
laser while the rst order reection is used to spatially separate the spectral output
from the diode. The prism serves as a retroreector to provide optical feedback to the
diode laser via a second reection from the diraction grating and is used to control
the operating frequency of the external cavity laser. Tuning is achieved by rotating
the prism.
aperture of 0.55 (Thor Labs 350230-B). The collimated light is next incident on a
1600 line/mm grating, 15 mm wide by 60 mm long by 10 mm thick (Spectrogon), at
a grazing angle of 3 degrees. The rst-order reection from the diraction grating is
found from cosθout = cosθin − λ/d to be 109◦, where θin (θout) is the angle between
the incoming (outgoing) beam and the plane of the diraction grating, λ is the wave-
length, and d is the line spacing of the diraction grating. The rst-order reection is
incident on a roof prism that directs the light back into the diode laser via a second
reection from the diraction grating, providing optical feedback to the diode laser.
The advantage of using the prism over a mirror is that it is easier to align since it acts
like a corner cube in the non-dispersive direction. The 3.8-cm-long external cavity
has a free spectral range of 3.9 GHz.
The roof prism rotates so that the cavity length changes in concert with the
wavelength of light fed back to the diode laser, allowing for mode-hop-free tuning
35
Figure 3.5: A picture of an ECDL built at Montana State University. This laser can
be tuned from 824 nm to 841 nm.
(McNicholl and Metcalf , 1985; de Labachelerie and Passedat , 1993). The roof prism
can be rotated mechanically by a 3/16-100 screw for coarse tuning. Fine rotation of
the roof prism is achieved by applying a voltage to a piezo-electric stack (Thor Labs
AE0505D16), giving mode-hop-free tuning of over 20 GHz at a xed temperature.
The roof prism can be rotated by a piezoelectric tuner (PZT) allowing the output of
the ECDL to be tuned electronically (Littman and Metcalf , 1978).
The ECDL is placed on a thermo-electric cooler (TEC) for temperature stabi-
lization, and monitored and controlled by a commercial temperature and current
controller (ILX LDD3722) to within 0.1 degrees Celsius. The same controller is used
to supply a drive current to the diode laser, which is operated in a cw mode. The
output of the ECDL is sent through a Faraday isolator to prevent unwanted feedback
from aecting its performance.
The ECDL performance is summarized as follows. The coarse tuning varies from
824 nm to 841 nm by mechanically changing the angle of the retroreective prism. By
adjusting the piezo-electric stack, we can also obtain a mode-hop-free tuning range
greater than 20 GHz. The beginning and ending wavelengths of this tuning range
36
can be altered by adjusting the diode temperature, typically between 19.2 and 20.2
degrees Celsius. The diode current is locked at about 37 mA. The full-width at
half-maximum line-width is less than 200 kHz, as determined by beating experiments
(Repasky et al., 2002). The maximum output power is 20 mW, with a side-mode
suppression of greater than 45 dB, as measured on an OSA. ECDLs have been built
at Montana State University with similar performance at center wavelengths of 790,
808, 830, 850, 935, 1050, 1160, 1330, and 1540 nm. Other wavelengths such as
950 nm, where water vapor has strong absorption features, can be reached easily
through simple modications of the ECDL design. Commercial tunable ECDLs are
also available (New Focus Product Guide, 2007). The advantage of building them at
Montana State University is that wavelengths specic to DIAL applications can be
easily achieved.
Tapered Amplier In spite of its narrow linewidth and broad tunability, the
ECDL has low output power that is limited to < 20 mW to extend the diode's lifetime
and improve the tuning characteristics. The ECDL is therefore used to injection seed
a semiconductor tapered amplier (Sacher Lasertechnik TA830, gure 3.6) to obtain
higher optical power, up to 500 mW, while maintaining the spectral properties of the
seed laser. This power was never achieved in practice, however, as the seed power
from the ECDL was only about 5 mW after traveling through two Faraday isolators,
and therefore was unable to saturate the TA. Even with low seed power, the spectral
characteristics of the ECDL, including linewidth and tunability, are transferred to the
output of the tapered amplier (Repasky et al., 2001). Note that when the amplier
is not powered, about 2 µW of power is still transmitted from the ECDL through
the amplier. Therefore, the ratio of the output of the amplier when it is on to
the transmitted power of the seed laser with the amplier o is about 50 dB. The
37
Figure 3.6: A picture of the Sacher Lasertechnik TA830 tapered amplier. The input
mirror used for seeding the TA is visible in the center of the picture. The collimating
optics are on the far left.
amplier's temperature and current are controlled and monitored by a commercial
laser diode controller (Sacher Lasertechnik Pilot P3000, gure 3.7). The amplier
temperature and current are set at about 21 degrees Celsius and 2.5 A. Figure 3.8
shows an OSA trace of the optical power at the amplier output, plotted as a function
of wavelength and attenuated to avoid damaging the OSA. A tuning range of 17 nm,
from 824 to 841 nm, is obtained when the ECDL is tuned. The side-mode suppres-
sion ratio of the amplier output reaches greater than 45 dB, which is identical to the
ECDL output. For spatial quality, M2 values of the amplier output beam have the
typical values between 1.4 and 2.5, as claimed by the manufacturer. Figure 3.9 shows
a plot from HiTRAN-PC, a atmospheric-modeling software code described in the cw
measurements section below, of the water vapor absorption lines within the tuning
range of the ECDL/TA transmitter system. The wide tunability of this transmitter
allows it to access any of the water vapor absorption lines in this region, giving it an
unprecedented ability to choose the best absorption line for the given environmental
conditions, which is a signicant improvement over existing low-power water vapor
DIAL systems.
The biggest advantages to using these commercial tapered ampliers is that they
are easily available (although expensive) and work well at the necessary wavelength,
38
Figure 3.7: A picture of the Sacher Lasertechnik Pilot driver used to control the
tapered amplier in the horizontal tuning experiments.
820 822 824 826 828 830 832 834 836 838 840 842 844
-60
-50
-40
-30
-20
-10
0
17 nm
O
pt
ic
al
 P
ow
er
 (d
Bm
)
Wavelength (nm)
Figure 3.8: A plot of the tuning of the laser system. This is shown by three overlayed
plots of the optical power of the injection seeded amplier (attenuated to avoid dam-
aging the OSA). The ECDL was tuned mechanically by adjusting the retroreective
prism to 824 nm (832 nm, 841 nm) and was used to seed the amplier. The spectral
output of the amplier is controlled by the spectral properties of the injection seeded
ECDL laser.
39
820 824 828 832 836 840 844
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
820 824 828 832 836 840 844
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
 
831.615 nm
829.022 nm
 
W
at
er
 V
ap
or
 T
ra
ns
m
is
si
on
Wavelength (nm)
Figure 3.9: A plot of the horizontal path transmission calculated using HiTRAN-PC
through a 1 km path length as a function of wavelength accessible by the tunable
ECDL shown in gure 3.8. The absorption features are due to water vapor.
∼ 830 nm. The rst major drawback is that they are typically dicult to align with
the seed laser beam. Alignment requires that the seed beam be tightly collimated and
horizontally polarized with respect to the TA input. Correct seeding requires precise
alignment with the amplier gain region, and typically required two independent
mirrors for beam steering. New TA models, such as the second TA used in the DIAL
system described in chapter 5, have a 1 mm longer gain region and thus are more
dicult to align.
The second, and more important, major drawback to these tapered ampliers is
that pulsing them at the speeds necessary for DIAL measurements has so far proven
unreliable. Several techniques were tried in an attempt to pulse the tapered amplier.
The TA itself has a pulsing option using external modulation, but the maximum rate
that the amplier could be cycled on and o was only 10 kHz, corresponding to a
pulse length of 100µs (range resolution of 15000 meters), much too long for the range
resolutions required in the DIAL experiment. A cycling rate of at least 1 MHz was
40
required to achieve pulse lengths of 1µs (range resolution of 150 meters) or less.
An attempt was made to pulse the TA using a dierent current driver (Directed
Energy PCX-7410) in place of the Pilot driver. The Directed Energy driver has the
capability to create sub-microsecond pulses at 50 kHz and beyond. Unfortunately,
the electronics within the TA could not be changed, and apparently are not optimized
for pulses shorter than about 10 µs. The results of these tests are shown in gure
3.10. The output pulse shape created by this method is not temporally square and
is in fact unpredictable pulse-to-pulse, which would make data inversion of a DIAL
signal more dicult. The power of the output pulse was seen to decrease rapidly as
the pulse length was made shorter than 10 µs, disappearing almost completely below
4 µs. The pulse with a width of 4 µs is probably uniform enough temporally and
would still provide reasonable power for an atmospheric experiment, but the pulses
are not short enough to provide scientically meaningful spatial resolutions. Another
method of pulsing the cw beam would have to be found, and is described in the next
section.
It should be noted that frequency chirp experiments following the method of
Repasky and Carlsten (2000) were done with the PCX-7410 pulsing the TA at 10
kHz with 10 µs pulses to determine if the amplier output was changing in frequency
while being pulsed. Such a frequency change, or chirp, if too large, would render the
transmitter useless for water vapor DIAL work, since frequency accuracy is critical to
the DIAL technique. These experiments showed that the upper limit for frequency
chirp in the DIAL transmitter was measured to be 110.637±5.636 MHz. The allowable
chirp for a DIAL transmitter to keep errors below 3% for laser properties according to
Bösenberg (1998) is ±161 MHz at 830 nm, so at least for these pulsing conditions the
DIAL transmitter's frequency chirp is within allowable tolerances. Frequency chirp
at shorter pulse lengths should not be an issue because the amplier is being seeded
41
0 5 10 15
0
1
2
3
4
9/01/05 Backup Sacher Amp Pulsing Tests
2.0 A Drive Current, 2 KHz Rep. Rate
Pu
ls
e 
Am
pl
itu
de
 (m
V)
Pulse Width (microseconds)
 10 ms
 5 ms
 4 ms
 3 ms
 2 ms
 1 ms
Figure 3.10: A plot of the pulsing characteristics of the Sacher Lasertechnik tapered
amplier. Faster pulses also have reduced output amplitudes, disappearing almost
completely below 4 µs, much too slow for use in the DIAL system.
by a cw ECDL source and its optical properties follow that of the ECDL. Still, the
experiments should be repeated in the future if a way to pulse the TA directly with
1 µs or shorter pulses is found, since the chirping characteristics might change as the
pulse length gets shorter and the repetition rate increases.
Acousto-optic Modulator (AOM) When pulsing the TA directly failed, the next
option was to use a mechanical chopper wheel located after the TA. Unfortunately, no
commercial chopper could be found that would allow pulse widths of 1µs or less. The
next option, and ultimately the option that was chosen, was to use an acousto-optic
modulator (AOM) made by Isomet (1205C-2, gure 3.11). The advantages to using
an AOM include that there is no frequency chirp in the transmitted beam and each
pulse is a near-perfect square shape in time, as shown in gure 3.12. The drawback
is that only about 66% of the light input to the AOM is transmitted through it into
a pulsed beam, reducing the transmit power of an already low-power system.
42
Figure 3.11: A picture of the acousto-optic modulator (AOM) used to pulse the cw
laser beam. The entrance aperture is visible on the side, below the 5 in 1205C-2.
0 1 2
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
11/08/05 Horizontal WV Lidar Testing
AOM Pulsing Tests, 828.979 nm, 2.9 A on Amplifier
Am
pl
itu
de
 (V
)
Time (microseconds)
10 KHz Rep. Rate
 500 nanosecond width
 1 microsecond width
 10 microsecond width
 CW
Figure 3.12: A plot of the pulsing characteristics of the AOM. Note the very clean
square-wave shape of the output pulses.
43
Figure 3.13: A picture of broadband beam samplers. Image courtesy of Newport.com
(http://www.newport.com).
Reference Power Meters and Beam Splitter To make the reference power measure-
ments, the transmit beam was passed through a Newport Broadband Beam Sampler
(gure 3.13) that reected ∼ 4% of the light to a detector. The beam sampler was
wedged in shape and anti-reection-coated for light at 830 nm to avoid interference
eects in the beam.
Several dierent types of detectors were used to make the reference power mea-
surements before nally settling on the Newport 1830c optical power meter (gure
3.14). Newport 2001 detectors were used initially, but the active area was very small,
making alignment a critical issue. Also, the Newport 2001 provided a voltage pro-
portional to the light incident on its detector, which then had to be converted to
a power measurement, and the Newport 1830c power meter provided a direct, cali-
brated measure of optical power. The 1830c was picked over the newer 1930c for a
few reasons. First, it has a larger detector head allowing for more beam movement
while still taking a valid power measurement. Second, it had a better response than
the 1930c in the sense that the measurements did not vary much across the detector
head, whereas they did on the 1930c. Third, the 1830c responded more consistently
44
Figure 3.14: A picture of the Newport 1830c (top) and 1930c (bottom), used for
diagnostics and reference power measurements in the horizontal tuning and vertical
DIAL experiments.
to remote command and control. Being certain that the number being returned from
the power meter is the correct measurement is critical for the DIAL measurements,
and seemed to be more reliable with the 1830c. The 1930c was used primarily as a
diagnostic tool, testing transmit powers along the beam path.
Telescope A commercial Schmidt-Cassegrain telescope (Celestron CGE1100) with
a 28 cm diameter was used to collect the return photons. The telescope uses a folded
design, with a primary mirror at the bottom of the telescope housing that reects light
to a smaller secondary mirror located in the middle of the entrance aperture. The
advantage of this telescope conguration is that it is compact, but one disadvantage
is that the secondary mirror blocks some of the aperture opening, eectively reducing
the receiver area, Ar. A schematic of the telescope is given in gure 3.15. According to
45
Figure 3.15: A schematic of a Schmidt-Cassegrain telescope. Image courtesy of Cele-
stron.com (http://www.celestron.com).
the lidar equation, equation 2.1, the lidar return signal is directly proportional to the
area of the receiving telescope. Therefore, larger telescope diameters allow for greater
return signals. Basically, the telescope acts as a light bucket collecting as many
photons as possible and focusing them for transport to a detector. While commercial
telescopes are easy to purchase and use, one disadvantage is that their optical surfaces
tend to be coated to maximize light transmission in the visible spectrum, and not in
the near infrared where the DIAL experiments operate. This lowers the return signal,
and cannot be avoided without putting custom coatings on the telescope optical
surfaces, greatly increasing the purchase price. A plot of the Celestron Starbright
coating is shown in gure 3.16.
Narrowband Filter An interference lter (Thorlabs, gure 3.17) with a 10-nm-
wide band pass centered on 830 nm was used to exclude a large portion of background
light being collected by the lidar receiver. It was placed directly on the entrance
of the detector ber to attempt to ensure that all light that would travel to the
detector would rst have to pass through the lter. 50% of the signal is lost due
to the transmission of the lter, but with a photon-counting detector, as in these
46
Figure 3.16: A transmission plot for the Starbright coating on the telescope
used in the horizontal lidar tuning experiments. Image courtesy of Celestron.com
(http://www.celestron.com).
experiments, blocking background light from room lights, street lights, and other
sources is critical for obtaining an acceptable signal-to-noise ratio. An even narrower
interference lter would be necessary for the vertical DIAL experiments.
Avalanche Photodiode (APD) Detector The detector (Perkin-Elmer SPCM-AQR-
13-FC) is a self-contained single photon counting module avalanche photodiode (APD)
requiring only a 5 V input and no external cooling. It is very compact and easy to
use. The output of the detector is a TTL pulse for every detected photon over a range
of 400 to 1100 nm. The advantage of using APD's in low-power DIAL experiments is
their high peak photon detection eciency of ∼ 74% at 700 nm. The other primary
detector that would work in this wavelength region is a photomultiplier tube (PMT),
but their detection eciency is much lower, perhaps ∼ 20% at most. The advantage
of PMT's is that they typically have large detector active areas, whereas the APD
active area is 170µm in diameter, making alignment critical. A GRIN lens is glued
to the interior of the APD housing that projects an image of the ber magnied by a
factor of 1.4 onto the detector area. Since a 650µm core diameter optical ber carried
47
Figure 3.17: A picture of the type of narrowband lter used in the horizontal tuning
experiments.
the received photons to the APD in these horizontal experiments, a large percentage
of signal was lost due to overlling of the detector active area. Again, this was not
considered a problem as the hard targets produced more than enough return signal
for these experiments. This would not be the case in a vertically-pointing lidar, how-
ever. The maximum count rate of the APD is nominally 15 Mc/s, with a dead time
of 50 ns between pulses. If the count rate exceeds about 1 Mc/s, photons will go un-
detected because the APD does not have adequate time to reset between detections,
and therefore a correction factor must be applied to the output.
APD's in photon counting mode are amazingly sensitive to all sources of light.
Simply switching room lights on while the APD was active and exposed would risk
damaging it. For this reason, one of the most critical design aspects of lidar systems
using this detector, a lesson learned through many frustrating data runs, is that the
APD must be isolated from all light sources to allow for accurate signal counts. If
uncovered, even with room lights o, the APD is able to detect light from instrument
48
Figure 3.18: A picture of the avalanche photodiode (APD) detector.
displays, or light entering the room through door jams or cracks in walls. Great
care was taken in the lidar experiments to completely isolate the APD in a sealed
housing. Even the ber carrying photons to the APD had to be wrapped in electrical
tape because the APD was able to detect background photons from the room that
penetrated the ber cladding at some point between the ber launch and the sealed
detector.
Multi-channel Scalar (MCS) The TTL logic pulses that originate from the APD
detector are then counted and binned in time by a multi-channel scalar (MCS, Stan-
ford Research Systems SR 430, gure 3.19), which makes range resolution possible.
This model of MCS was chosen because it is very easy to use and adjust and has
a front-panel screen so that the return counts per bin can be observed in real time.
The disadvantage to this type of MCS is that the display limits the speed of the
instrument. The fastest rate at which the MCS could be triggered, and therefore
the fastest rate at which we could pulse the laser, was 2272 Hz. For low-power lidar
systems, one way to increase the return signal is to increase the laser repetition rate.
For this reason, the vertical DIAL system would need a faster scalar card.
49
Figure 3.19: A picture of the Stanford Research Systems Multi-channel Scalar (MCS)
with display screen used in the horizontal tuning experiments.
Figure 3.20: A picture of the three hard targets and their ranges used in the horizontal
tuning experiments, taken from the roof of Cobleigh Hall at Montana State University,
just above the roofport room where the measurements were made.
Targets
Three buildings at dierent distances from the lidar were selected as hard targets
for the tuning experiments. All distances to the targets were determined by GPS.
The closest target was a metal exhaust stack on the roof of the MSU heating plant
building, located 0.175 km away. The next two targets were similar in range: the
upper-level of the MSU football stadium at a distance of 0.835 km away, and the
second-story wall of the Museum of the Rockies at 0.855 km away. Each target was
reasonably reective so that the laser spot could be easily seen on the targets at night.
Steam exhaust plumes from the Engineering and Physical Sciences (EPS) Building
and the heating plant prohibited data collection on several occasions.
50
828.96 828.98 829.00 829.02 829.04 829.06 829.08 829.10
0
10
20
30
40
50
60
70
80
90
100
Wavelength (nm)
Ca
lib
ra
te
d 
Re
tu
rn
 C
ou
nt
s 
(m
V-
1 )
10/27/04 Horizontal WV Lidar
829.022 nm line, Heating Plant
Figure 3.21: Early results from the horizontal tuning experiments. The data should
show a water vapor absorption line centered at 829.022 nm with the counts rising to
a steady value in the wings of the line.
System Diagnostics
Early results with the horizontal experiments showed signicant problems, as seen
in gure 3.21. Months of iterating through system components, software versions, and
testing was required to obtain accurate results.
P/I Curves The rst tests that needed to be done characterized the health and
behavior of the ECDL and TA. Figure 3.22 shows P/I curves for the ECDL and TA
used in the horizontal experiments. These graphs show the increase in output power
of each laser with increasing drive current. The ECDL was typically at a drive current
37 mA. The maximum output power of 8 mW was reduced by transmission through
two optical isolators and a pick-o for the wavelength measurement, so typically only
51
0 10 20 30 40
0
2
4
6
8
Drive Current (mA)
O
ut
pu
t P
ow
er
 (m
W
)
10/31/05 Experiment ECDL PI Curve
828.956 nm, Newport 1930c Power Meter
0 500 1000 1500 2000 2500 3000
0
50
100
150
200
10/31/05 Experiment Sacher Amp PI Curve
828.956 nm, Newport 1930c Power Meter
Drive Current (mA)
O
ut
pu
t P
ow
er
 (m
W
)  Seeded Unseeded
Figure 3.22: P/I curves for the ECDL and TA used in the horizontal tuning experi-
ments.
5-6 mW was used to seed the TA. The P/I curve for the TA shows both the seeded
and unseeded responses. Notice that when the TA is being seeded with only 5-6 mW,
the maximum power output is only around 200 mW, well below the 500 mW output
supposedly achievable if the TA is saturated with a seed power of ∼ 20 mW. The TA
was typically driven with a current of 2500-2800 mA. Drive currents above that value
drastically reduced the ability of the TA to follow the spectral qualities of the ECDL.
Linewidth and Tuning The next tests studied the ability of the TA to preserve
the spectral qualities of the ECDL. Specically, for the DIAL system to work, it had
to be shown that the TA would maintain the narrow linewidth of the ECDL, and
tune across wavelengths as the ECDL tuned. Figure 3.23 shows traces of the ECDL
and TA output spectrum taken on an OSA. The high quality of the ECDL can be
seen, with a narrow linewidth and sidemode suppression of greater than 30 dBm. The
broad output spectrum of the TA is evident in the unseeded case. The power in this
broad spectrum is forced into a narrow linewidth when seeded by the ECDL.
52
815 820 825 830 835 840 845
-50
-40
-30
-20
-10
0
Wavelength (nm)
Po
w
er
 (d
Bm
)
11/08/05 Horizontal WV Lidar Testing
828.977 nm
 ECDL
815 820 825 830 835 840 845
-50
-40
-30
-20
-10
0
11/08/05 Horizontal WV Lidar Testing
Seeded at 828.977 nm
Wavelength (nm)
Po
w
er
 (d
Bm
)
 Sacher Amp, Pre-AOM, Seeded
 Sacher Amp, Pre-AOM, Unseeded
Figure 3.23: Optical spectrum analyzer traces for the ECDL and TA used in the
horizontal tuning experiments.
Testing the tuning of the system was done by measuring the wavelength of the
ECDL output and comparing that to the wavelength of the TA output as the ECDL
tuned. A ratio of the amplier wavelength and the ECDL wavelength was computed
and should be very close to 1 if the TA is following the tuning of the ECDL. Figure
3.24 shows this to be the case. The ratio between the two wavelengths were identical
to within the uncertainty of the wavemeter, ±0.1 pm.
Reference Power Measurements The hardest problem to solve with the horizontal
tuning experiments was accurately measuring and accounting for the uctuations in
the system's transmitted power. The optical power transmitted from the amplier
uctuated by perhaps as much as 30% as the gain region heated up and changed
size, changing the mode structure of the output radiation. These uctuations were
very pronounced when the amplier was initially powered on, but calmed down con-
siderably as the temperature came to an equilibrium. The power still uctuated
predictably as the ECDL tuned. Tests were done to verify that the ECDL did not
53
62.6 62.7 62.8 62.9 63.0 63.1
0.9995
0.9996
0.9997
0.9998
0.9999
1.0000
1.0001
11/08/05 Horizontal WV Lidar Testing
ECDL/Amp Tuning Tests Pre-AOM
37 mA on ECDL, 2.995 A on Amp
19.4 deg. C, 10.2 V
 Run #1 (Tuning 20)
 Run #2 (Tuning 22)
ECDL Wavenumber (relative to 12,000 cm-1)
Am
p/
EC
DL
 W
av
el
en
gt
h 
Ra
tio
Figure 3.24: A plot of the ratio output wavelength of the amplier to the output
wavelength of the ECDL. The wavelengths agree to within the error of the wavemeters.
show the same power uctuations. The results of the tests are shown in gure 3.25,
showing the relatively constant power of the ECDL versus the power hops of the TA
as the ECDL is tuned.
These power uctuations are not a problem for the DIAL experiment as long as
they can be monitored and normalized out of the return signal to ensure that any
changes in return signal are due to atmospheric, and not laser, eects. As described in
the reference power meters section above, the reference power detector was changed
from a Newport 2001 solid-state detector to a Newport 1830c optical power meter to
give a better response. One problem with the 1830c power meter, though, is that it
does not seem to have the capability to clear the memory buer completely. When
a power measurement is made, it is stored in an internal memory buer until read
remotely by the laptop. Even after a clear memory command is sent to the meter,
it will not erase the last power measurement that was made. Thus, after gaps in data
taking, such as during laser tunings, the rst reference power reading was wrong and
had to be manually removed from the nal data.
54
0 5 10 15 20 25 30
1.4
1.5
1.6
1.7
1.8
1.9
2.0
2.1
Data Number (Increasing Wavelength)
11/07/05 Horizontal WV Lidar Testing
Amplifier Power Mode Tests
Po
w
er
 (W
)
 ECDL Power
 Amplifier Output Power, Pre-AOM
Figure 3.25: A plot showing the power modes of the tapered amplier. As the ECDL
tunes, the power output of the TA changes drastically, necessitating normalization of
the return data with transmit power.
Ways to minimize the error in reference power measurements were found through
several iterations of the operational software and experiment. Initially, the reference
power was only measured once, before each data point was taken. This was found to
be inadequate as the power would drift across the time needed to take a measurement,
especially during the pulsed experiments where each data point was averaged for 100
seconds. To account for this, the reference power was measured before and after the
data collection time to average out any output power drifts that might occur while
data are being collected. Both reference powers, the return signal counts in the target
bin, and the laser frequency were recorded to a data le at one wavelength after which
the computer tuned the laser by adjusting the voltage applied to a piezo-electric tuner
within the ECDL. Another software x was to build in a settle time to allow the
lasers to equilibrate for 5 seconds after each tuning period. This helped alleviate many
of the power drifts. A hardware x involved installing metal heat sinks underneath
the amplier to help dissipate heat. This, too, made a large dierence in the amount
55
62.8 62.9 63.0 63.1 63.2 63.3 63.4
0
5
10
15
20
25
30
35
40
11/09/05 Horizontal WV Lidar Testing
Reference Power Meter Tests
Wavenumber Relative to 12000 cm-1
TX
 P
ow
er
/R
ef
. P
ow
er
ECDL at 19.2 deg. C, 0 V, Amp at 2.9 A
 1930c Ref. Detector (+ 2.13%)
 1830c Ref. Detector (+ 1.24%)
Figure 3.26: A plot showing how the reference power measurement actually tracks
the transmit power of the horizontal system. Note that the Newport 1830c is more
accurate.
of uctuations that were measured. Many tests were done to ensure that the power
uctuations were indeed being normalized out of the return signal. The results of one
such test are shown in gure 3.26. This graph shows the results of using the 1830c
or the 1930c as the reference power meter, and comparing what they measure to the
the actual measured transmit power through a ratio. Each line should be at if the
power uctuations are fully removed. Both meters do a decent job, but the 1830c
was consistently more accurate.
CW Measurements
The rst water vapor measurements were done with the laser transmitter operating
in a cw mode, aimed at the hard targets described above. The outside air temper-
ature and relative humidity were measured using a digital psychrometer-hygrometer
(Mannix SAM990DW). About one hour was needed to fully tune and collect data
across a water vapor line, taking a transmission measurement at typically over one
56
hundred wavelength points. Vertical DIAL systems will only need to collect return
counts at two wavelengths, instead of scanning across wavelengths as in these tuning
experiments. The LabVIEW software that controlled the experiment rst recorded
the laser wavelength from a wavemeter, then recorded the reference power of the
transmitted beam, followed by triggering the MCS and recording the photon counts
from it, and nally tuning the diode wavelength by adjusting the PZT voltage.
Measured and calculated relative atmospheric transmission spectra near 829.02
nm are plotted as a function of wavelength in gure 3.27 (normalized to 100% in the
wings, where water vapor absorption is essentially zero). The measured transmission
is determined from a ratio of the return signal with the transmitted power to compen-
sate for the changing optical power of the laser transmitter as it is tuned. The closed
(open) circles represent the measured transmission as a function of wavelength for a
total path length of 1.71 km (0.35 km), and the solid (dashed) line represents the theo-
retical spectral transmittance calculated with HiTRAN-PC, a atmospheric-modeling
software code that uses the HiTRAN 2000 database. HiTRAN-PC was developed
by Dr. Dennis Killinger's group at the University of South Florida Laboratory for
Laser Atmospheric Studies and is available through Ontar Corporation at 9 Village
Way, North Andover, MA 01845-2000, USA (ph. 1-508-689-9622). The measured
temperature of 12◦ C (4◦ C) and relativity humidity of 47% (48%) were used in these
calculations. The laser was tuned over 50 GHz for both of these measurements and an
absorption feature at 829.022 nm is clearly evident, while a second weaker absorption
feature at 829.055 nm is easily resolved for the 1.71 km path length.
Measured and calculated relative atmospheric transmission spectra near 831.615
nm are plotted as a function of wavelength in gure 3.28. The solid circles represent
the measured atmospheric spectral transmission wavelength, while the solid line rep-
resents the theoretical predictions. The measured temperature of 7.3◦ C and relative
57
-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
829.08 829.06 829.04 829.02 829.00 828.98 828.96
 
Tr
an
sm
is
si
on
Wavenumber relative to 12,062.41 cm-1
 CW 0.35 km pathlength
 CW 1.71 km pathlength
 HiTRAN 0.35 km pathlength
 HiTRAN 1.71 km pathlength
 Wavelength (nm)
Figure 3.27: A plot of the transmission through the atmosphere as a function of
wavelength near 829.02 nm. The closed (open) circles represent measurements made
for a 1.71 km (0.35 km) path length. The solid (dashed) line is a theoretical calcu-
lation using HiTRAN-PC with the measured temperature and humidity used in the
modeling.
58
-0.4 -0.2 0.0 0.2 0.4 0.6
0.70
0.75
0.80
0.85
0.90
0.95
1.00
0.70
0.75
0.80
0.85
0.90
0.95
1.00
1.67 km pathlength
 CW Data
 HiTRAN
 
Tr
an
sm
is
si
on
Wavenumber Relative to 12,024.79 cm-1
831.64 831.62 831.60 831.58
 Wavelength (nm)
Figure 3.28: A plot of the transmission through the atmosphere as a function of
wavelength near 831.62 nm. The closed circles represent measurements made for a
1.67 km path length. The solid line is a theoretical calculation using HiTRAN-PC
with the measured temperature and humidity used in the modeling.
humidity of 50% were used in these calculations, made for a path length of 1.67 km.
Preliminary Pulsed Measurements
A comparison of the cw and pulsed operation of the laser transmitter was carried
out for the absorption line near 831.615 nm. A plot of the relative atmospheric
spectral transmission is shown in gure 3.29.
The open circles represent the cw measurements as described above while the lled
circles represent the pulsed measurements. For both sets of measurements the ECDL
laser was operated in a cw mode and pulsing was achieved by supplying a square
pulse of current to the tapered amplier. The pulse duration was 10 µs, limited by
59
24.5 24.6 24.7 24.8 24.9 25.0
0.70
0.75
0.80
0.85
0.90
0.95
1.00
1.05
0.70
0.75
0.80
0.85
0.90
0.95
1.00
1.05
831.635 831.630 831.625 831.620 831.615 831.610 831.605 831.600
1.67 km pathlength
 CW Data
 Pulse Data
 HiTRAN
 
 Wavelength
Tr
an
sm
is
si
on
Wavenumber Relative to 12,000 cm-1
 
Figure 3.29: A plot of the atmospheric transmission as a function of wavelength. The
open circles represent measurements made with a cw laser transmitter while the lled
circles represent measurements made with a pulsed laser transmitter. The solid line
represents the results of a HiTRAN-PC calculation for a 1.67 km horizontal path
calculation.
60
the bandwidth of the current-driver electronics, which corresponds to a 3-km light
path in air. Therefore, range resolving was not possible for the 1.67-km path in
these preliminary pulsed measurements. The goal was to demonstrate the validity of
pulsing the tapered amplier as a DIAL transmitter.
Although these initial measurements employed a pulse repetition rate of 10 kHz
and average optical power of only 18 mW, it was still possible to collect enough return
signal by accumulating about 2000 pulses at each frequency. Good agreement between
the pulsed and cw data is seen in gure 3.29, in which the solid line represents the
radiative transfer calculation for a 1.67 km horizontal path at a temperature of 7.8◦
C and a relative humidity of 52%. This conrms the capability of this transmitter to
maintain its high-quality characteristics during pulsed operation.
Pulsed Measurements
After the cw and preliminary pulsed measurements, faster pulsed absorption mea-
surements were taken by this system on three water vapor absorption lines, 829.022
nm, 831.615 nm, and 831.850 nm, at varying distances using the AOM. Data taken
on the 831.615 nm absorption line proved to be too weak to be reliable and so the
lidar was coarse-tuned to the slightly stronger 831.850 nm line instead, demonstrating
the powerful exibility of a system able to tune to several absorption lines. Pulses
500 ns wide, leading to 75-m-long rangebins and average pulse energies of 50 nJ per
pulse, were transmitted at a repetition rate of 500 Hz. The repetition rate was limited
by the data collection speed of the MCS. Signal averaging of typically 100 seconds
was used to increase the return signal and smooth out short-timescale variations. An
average of the o-line data was taken and used to normalize the entire data spectrum.
Data sets from two water vapor lines compared to HiTRAN predictions illustrating
61
the system tunability are shown in gures 3.30 and 3.31. Figure 3.30 shows tuning
data taken across the 829.022 nm water vapor line at pathlengths of 1.71 km and
0.35 km. Figure 3.31 shows tuning data taken across the 831.850 nm water vapor
absorption line at pathlengths of 1.67 km and 0.35 km. The varying pathlengths can
be thought of as equivalent to the system response to varying relative humidities, as
the absorption lines would become more or less pronounced depending on the water
vapor density present in the atmosphere. The temperature and relative humidity
measurements were made by a weather station located about 5 meters above the
lidar and were averaged over the typically 1 hour or more needed to take all of the
data points.
Conclusions
Taken together, the gures from the pulsed measurements show the capability of
the ECDL/TA transmitter to selectively probe dierent water vapor absorption lines
depending on current atmospheric conditions. The experimental results, taken over
many months and 46 night data runs, show that a DIAL system utilizing this ECDL
and tapered amplier combination will have an unprecedented ability to select which
water vapor lines to scan depending on the prevailing atmospheric conditions. With
the transmitter fully tested and veried, the next step was to select a water vapor
absorption line to use in the DIAL experiments and begin designing, building, and
testing the DIAL.
62
Horizontal WV Lidar
829.022 nm line
500ns pulses, 640ns bins, 500 Hz rep. rate, 100 sec. ave.
61.5 62.0 62.5 63.0
0.0
0.2
0.4
0.6
0.8
1.0
829.10 829.08 829.06 829.04 829.02 829.00 828.98 828.96
0.0
0.2
0.4
0.6
0.8
1.0
Wavelength (nm)
 HiTRAN: 10.58 deg. C, 82.9% RH
 1.71 km pathlength
 HiTRAN: 3.75 deg. C, 57.9% RH
 0.35 km pathlength
Wavenumber Relative to 12000 cm
-1
T
r
a
n
s
m
i
s
s
i
o
n
Figure 3.30: A plot of the relative transmission through the atmosphere as a function
of wavelength near 829.022 nm. The open (closed) circles represent measurements
made for a 0.35 km (1.71 km) path length. The solid and dashed lines are theoretical
calculations using HiTRAN-PC with the measured temperature, humidity, and path
length.
63
Horizontal WV Lidar
831.850 nm line
500ns pulses, 640ns bins, 500 Hz rep. rate, 100 sec. ave.
20.6 20.8 21.0 21.2 21.4 21.6 21.8 22.0 22.2
0.0
0.2
0.4
0.6
0.8
1.0
831.90 831.88 831.86 831.84 831.82 831.80
0.0
0.2
0.4
0.6
0.8
1.0
1.2
 HiTRAN: -5.60 deg. C, 58.0% RH
 1.67 km pathlength
 HiTRAN: 4.62 deg. C, 50.8% RH
 0.35 km pathlength
Wavenumber Relative to 12000 cm
-1
T
r
a
n
s
m
i
s
s
i
o
n
Wavelength (nm)
Figure 3.31: A plot of the relative transmission through the atmosphere as a function
of wavelength near 831.850 nm. The open (closed) circles represent measurements
made for a 0.35 km (1.67 km) path length. The solid and dashed lines are theoretical
calculations using HiTRAN-PC with the measured temperature, humidity, and path
length.
64
CHAPTER 4
WATER VAPOR ABSORPTION LINE SELECTION
Theory
Selection of a water vapor absorption line for use in DIAL measurements is of
critical importance to the accuracy of the measurement. Three sources of error need
to be considered when selecting a line. First, absorption line strengths, and therefore
line shapes and absorption cross sections, can be strong functions of temperature,
causing errors in the measurement due to the uncertainty in temperature as the laser
pulse propagates through the atmosphere. Second, the optical depth produced by
the absorption line must be considered, as absorption lines can either be too strongly
absorbing, resulting in too low of a return signal, or too weak, resulting in not enough
dierence between the on- and o-line return signals. Third, the presence of other
absorption features near to the absorption line being used can cause errors to be made
in the absorption cross section measurement of this line, unless these other features
are properly accounted for. A balance must be struck between the three criteria of
temperature sensitivity, optical depth, and nearby absorption features when nding
an appropriate water vapor absorption line for DIAL measurements.
The procedure for selecting a water vapor absorption line described below closely
follows the methods of Browell et al., 1991. Their analysis covered absorption lines
in the 720nm region, and I have applied a similar analysis to absorption lines in the
824-841nm region. More recently, Ambrico et al., 2000 showed results for temperature
sensitivity analysis in much of the near infrared for a greater number of molecules
65
than just water vapor. I compare my results to theirs below.
Line Selection Criteria
Temperature Sensitivity
Water vapor absorption line cross sections in the lower troposphere can be de-
scribed by one of three line shape proles, depending on the dominant broadening
process, which is a function of pressure and hence altitude in the atmosphere. In any
case, the temperature-dependent line strength, S, is given by (Browell et al., 1991)
S(T ) = S0
(T0
T
)3/2{ 1− exp[−hcν0/(kT )]
1− exp[−hcν0/(kT0)]
}
exp(
(hc
k
[ 1
T0
− 1T
]
E”
)
, (4.1)
where S0 [cm−1/(mol · cm−2)] and T0 [Kelvin] are reference values of line strength
and temperature, h [6.626 × 10−34Joule · s] is Planck's constant, c [2.998 × 108m/s]
is the speed of light in vacuum, ν0 [cm−1] is the line center wavenumber, k [1.38 ×
10−23J/(mol · K)] is Boltzmann's constant, and E” [cm−1] is the energy needed to
transition between the ground state and a specic excited state. As is typical in
spectroscopy, the energy E” is normalized by hc and given in wavenumbers. For
absorption lines close to the surface (< 2 km above sea level), atmospheric pressure, or
collisions between molecules, is the dominant broadening process and the absorption
line prole can be described by the Lorentzian prole. The Lorentz linewidth, dened
as the half width of the line at half of its maximum amplitude (HWHM, in cm−1), is
given by
γL = γ0
( P
P0
)(T0
T
)α
, (4.2)
where P [atm] is the pressure and α [unitless] is the linewidth temperature depen-
66
dence parameter (Browell et al., 1991; Liou, 2002). Again, P0, T0, and γ0 are initial
values. The cross section σ0 [cm2] at line center is (Browell et al., 1991)
σ0 =
S
piγL
. (4.3)
Here, the subscript 0 denotes that the cross section value is taken at line center, ν0,
as opposed to some other wavenumber ν [cm−1].
In the opposite case of high altitudes (> 50 km above sea level) where the pressure
is low and intermolecular collisions are infrequent, the dominant broadening process
will be Doppler broadening. This process arises due to the Maxwell-Boltzmann dis-
tribution of the molecular velocities, which creates Doppler shifts in the observed
radiation frequencies. In this case, the absorption line prole can be described by
the Gaussian prole and the Doppler linewidth (HWHM) is given by (Browell et al.,
1991)
γD =
(ν0
c
)
(2kT ln2
m
)1/2
, (4.4)
where m [kg/molecule] is the mass of the water molecule. The absorption cross section
is dened by a Gaussian function,
σ(ν) = SγD
( ln2
pi
)1/2
exp
[
− ln2(ν − ν0)
2
γ2D
]
, (4.5)
where ν [cm−1] is the wavenumber at which the cross section is being calculated. At
line center (ν = ν0) this cross section reduces to
σ0 =
S
γD
( ln2
pi
)1/2
. (4.6)
While pressure broadening is dominant at low altitudes (< 2 km above sea level)
67
and Doppler broadening is dominant at much higher altitudes (> 50 km above sea
level), eects of both broadening mechanisms can be seen in between these two ex-
tremes. In this case, a convolution of the Lorentz and Gaussian proles is used, which
is called a Voigt prole. The Voigt prole is given by
V (x, y) = σ(x, y)K =
y
pi
∫ ∞
−∞
exp(−t2)
y2 + (x− t)2dt, (4.7)
where σ(x, y) [cm2] is the absorption cross section, x = [(ν − ν0)/γD](ln2)1/2, y =
(γL/γD)(ln2)1/2, and K = (S/γD)(ln2/pi)1/2.
Analyzing the temperature sensitivity of an absorption line then involves using
these denitions to quantify how the measurement being made changes with temper-
ature. With number density measurements, the number density error is given as the
percent change in absorption cross section per unit temperature change, or
1
dT
dσ0
σ0
≈ 1T − T ′
σ0(T )− σ0(T ′)
[σ0(T )+σ0(T ′)]
2
. (4.8)
For mixing ratio measurements, Cahen et al. (1982) showed that the error depends
on σ/T instead of just σ, and therefore the mixing ratio error is calculated by
1
dT
d(σ0/T )
(σ0/T )
≈ 1T − T ′
σ0(T )/T − σ0(T ′)/T ′
[σ0(T )/T+σ0(T ′)/T ′]
2
. (4.9)
This also has the implication that absorption lines suitable for number density mea-
surements are not suitable for mixing ratio measurements, and vice versa. For this
reason, and knowing that the narrowband lters needed for each absorption line are
expensive (∼ $5000), I chose to only select a line for number density measurements,
leaving line selection for mixing ratio measurements to the interested reader.
A MATLAB program (voigt.m) was written to calculate the number density and
68
mixing ratio errors with respect to changes in temperature, pressure, and ground
state transitional energies, according to equations 4.8 and 4.9. The goal of the error
calculations was to repeat the analysis and gures of Browell et al. (1991) to verify
that the algorithm used was correct before continuing with line selection. Reference
values for the program were taken from the HiTRAN 2000 database and were: S0 =
2.75 × 10−23 cm−1/(mol · cm−2), P0 = 1.0 atm, and To = 296.0K. Values in the
HiTRAN 2000 database are specied for this temperature. All calculations were
performed at line center, such that ν0 = ν = 13785.0 cm−1. The pressure was input
at the command line from the user. Calculations were performed at 1.0 atm, 0.5 atm,
and 0.25 atm. The values for α and γL were taken from the HiTRAN 2000 database
and plotted against E” to determine a linearly varying relationship. Figure 4.1 shows
Figures 2 and 3 from Browell et al. (1991), as repeated by me using HiTRAN 2000
data for water vapor lines in the 720nm region. The data used originally in these
gures was measured, and therefore known to be reasonably accurate. I repeated the
same graph techniques for water vapor lines between 824nm and 841nm, the tuning
range of the lidar transmitter. Unfortunately, the values of α, γL, and E” are not
known for all or maybe even most of the water vapor lines in this region, according
to the HiTRAN 2000 database. Some values were missing, while others appeared to
be just average values. An example of this is that γL = 0.68 cm−1 was used for a
majority of absorption lines throughout the database. Guessing that many of these
were averages and not actual values, I only plotted values for water vapor absorption
lines where γL 6= 0.68 cm−1. As can be seen in the lower graphs of Figure 4.1, the
linear ts for the 824nm to 841nm region were both higher than for the 720nm
region. Since I could not verify that these linear ts were correct or were merely due
to my bias in data selection, I used the ts from Browell et al. (1991) to calculate
values for α and γL based on E” values of 0 to 500 cm−1 in increments of 50 cm−1.
69
The values used are specied in table 4.1.
E” [cm−1] 0 50 100 150 200 250
α 0.79806 0.77542 0.7284 0.73015 0.70751 0.68488
γL [cm−1] 0.10235 0.09996 0.0976 0.09518 0.09280 0.090403
E” [cm−1] 250 300 350 400 450 500
α 0.68488 0.66224 0.63960 0.61697 0.59433 0.57170
γL [cm−1] 0.090403 0.088014 0.085625 0.083235 0.080846 0.078457
Table 4.1: Values for E ′′, α, and γL as determined by the linear t of Figure 4.1.
With the initial values specied, the values for α, γL , and E” were hard-coded into
the program and incremented until data were taken at each E” value. The powerful
matrix capability of MATLAB was used to do each computation at temperatures
from 100K to 500K in 1K segments. The program rst calculates the Lorentz
cross section, using equation 4.3, followed by the number density and mixing ratio
errors for the Lorentz case using equations 4.8 and 4.9. Next, the pressure-dependent
Lorentz line width and the Doppler line width, as dened in equations 4.2 and 4.4
were computed, allowing the Voigt function parameters x and y to be calculated. The
Voigt function integral was then evaluated using a step-wise procedure, adding up the
value of the integral in steps from one side of the function to the other. Surprisingly,
the int() and quad() functions in MATLAB could not solve the seemingly simple
Voigt function integral. The results of the custom step-wise procedure versus the
MATLAB int() function are shown in Figure 4.2.
After the Voigt function integral is computed, the values for the line strength
S, the Voigt function parameter K, and the Voigt function itself from equation 4.7
were computed. Then, using the relation in equation 4.7 of V = σ/K, the Voigt
function cross section σ was found. Finally, this cross section was again used to
nd the number density and mixing ratio errors using equations 4.8 and 4.9. The
results, combined into graphs separated by pressure, are shown in Figures 4.3 and
70
0 100 200 300 400 500 600 700 800 900 1000
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
 
 
Al
ph
a
E" (cm-1)
Browell, et al., 1991
 Linear Fit
0 100 200 300 400 500 600 700 800 900 1000
0.04
0.05
0.06
0.07
0.08
0.09
0.10
0.11
0.12
0.13
0.14
 
 
Li
ne
w
id
th
 (c
m
-1
)
E" (cm-1)
Browell, et al., 1991
 Linear Fit
0 100 200 300 400 500 600 700 800 900 1000
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
 
 
Al
ph
a
E" (cm-1)
824 nm to 841 nm
 Linear Fit
0 100 200 300 400 500 600 700 800 900 1000
0.04
0.05
0.06
0.07
0.08
0.09
0.10
0.11
0.12
0.13
0.14
 
 
Li
ne
w
id
th
 (c
m
-1
)
E" (cm-1)
824 nm to 841 nm
 Linear Fit
Figure 4.1: The top graphs are Figures 2 and 3 from Browell et al. (1991) showing
linear t relationships between the linewidth temperature dependence parameter α,
the Lorentz linewidth γL, and the ground state energy E” in the 720nm region.
These graphs are repeated here to show consistency between my method and that of
Browell et al. (1991). The bottom graphs show the linear t relationships between
these same parameters in the 824nm to 841nm region.
71
100 200 300 400 500
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
 
 
 Custom Integration Method
 Matlab Integration Function
Vo
ig
t I
nt
eg
ra
l V
al
ue
Temperature (K)
Figure 4.2: The MATLAB integration function could not solve the Voigt integral at
low temperatures, sending the value of the integral to zero at a certain point. A
custom integration method needed to be written to solve the integral down to 100K.
4.4. The curves are in excellent agreement with those published in Browell et al.
(1991), diering slightly because slightly dierent values of α and γL (directly from
the linear ts described above), and ν0 were probably used, since the authors did
not specify their exact values. Examining the three smaller-scale graphs of Figure
4.3 shows that the optimal E” limits for number density measurements that give
error values of ±0.10%/K between 200K and 300K for pressures changing from
1.0 atm to 0.25 atm are about 125 − 225 cm−1. My temperature sensitivity results,
therefore, agree with those of Browell et al. (1991). Water vapor absorption lines with
ground state transitional energies within this range will be acceptably temperature
independent for number density measurements.
It should be noted here that the limits commonly stated in the literature for
where pressure broadening or Doppler broadening are important are estimations at
best. Unless it is known with certainty that measurements are being made well
within either of those two limits, a Voigt function should be used for calculations
72
100 200 300 400 500
-0.4
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
0.4
E"=300
E"=250
 
 
1.0 Atm
E"=200
E"=50
E"=100
E"=150
E"=0
De
ns
ity
 E
rr
or
 (%
/K
)
Temperature (K)
100 200 300 400 500
-0.10
-0.05
0.00
0.05
0.10
 
 
 
1.0 Atm
E"=300
E"=250
E"=200
E"=150
E"=100
E"=50
De
ns
ity
 E
rr
or
 (%
/K
)
Temperature (K)
100 200 300 400 500
-0.10
-0.05
0.00
0.05
0.10
0.5 Atm
E"=300
E"=250
E"=200
E"=150
E"=100
E"=50
 
 
De
ns
ity
 E
rr
or
 (%
/K
)
Temperature (K)
100 200 300 400 500
-0.10
-0.05
0.00
0.05
0.10
0.25 Atm
E"=300
E"=250
E"=200
E"=150
E"=100
E"=50
 
 
De
ns
ity
 E
rr
or
 (%
/K
)
Temperature (K)
Figure 4.3: These gures show the number density error produced across a range of
temperatures, pressures, and E” values, generated using my MATLAB Voigt prole
calculator. They are nearly identical to Figures 4 and 5 from Browell et al. (1991),
showing that the calculation method is valid. The top left graph shows the number
density temperature sensitivity on a larger scale. The other graphs show the number
density temperature sensitivity at 1.0 atm, 0.5 atm, and 0.25 atm pressures scaled to
a ±0.10%/K error.
73
100 200 300 400 500
-0.4
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
0.4
E"=100
E"=500
E"=450
E"=400
E"=350
E"=300
E"=250
E"=200
E"=150
 
 
1.0 Atm
Temperature (K)
M
ix
in
g 
Ra
tio
 E
rr
or
 (%
/K
)
100 200 300 400 500
-0.10
-0.05
0.00
0.05
0.10
 
 
1.0 Atm
E"=250
E"=300
E"=350
E"=400
E"=450
E"=500
M
ix
in
g 
Ra
tio
 E
rr
or
 (%
/K
)
Temperature (K)
100 200 300 400 500
-0.10
-0.05
0.00
0.05
0.10
0.5 Atm
E"=500
E"=450
E"=400
E"=350
E"=300
E"=250
 
 
M
ix
in
g 
Ra
tio
 E
rr
or
 (%
/K
)
Temperature (K)
100 200 300 400 500
-0.10
-0.05
0.00
0.05
0.10
E"=500
E"=450
E"=400
E"=350
E"=250
E"=300
0.25 Atm
 
 
M
ix
in
g 
Ra
tio
 E
rr
or
 (%
/K
)
Temperature (K)
Figure 4.4: These gures show the mixing ratio error produced across a range of
temperatures, pressures, and E” values, generated using my MATLAB Voigt prole
calculator. They are nearly identical to Figure 6 from Browell et al. (1991). The
top left graph shows the mixing ratio temperature sensitivity on a larger scale. The
other graphs show the mixing ratio temperature sensitivity at 1.0 atm, 0.5 atm, and
0.25 atm pressures scaled to a ±0.10%/K error.
74
100 200 300 400 500
-0.4
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
0.4
1.0 Atm
 
 
E"=300
E"=250
E"=200
E"=150
E"=100
E"=50
E"=0
De
ns
ity
 E
rr
or
 (%
/K
)
Temperature (K)
 Voigt Profile
 Lorentz Profile
Figure 4.5: Deviation from the Voigt prole in the number density error calculations
at 1.0 atm that would occur if a Lorentz prole were used.
if possible. Examples of errors that could be incurred if a Lorentzian line prole is
incorrectly assumed are given in Figures 4.5 and 4.6. Figure 4.5 shows the deviation
from the Voigt prole in the number density error calculations at 1.0 atm that would
occur if a Lorentz prole were used. The error in the calculated density error could
change by about 0.01%/K depending on the atmospheric temperature. Figure 4.6
shows the deviation from the Voigt prole in the number density error calculations
at 0.25 atm, where the error becomes much more apparent, changing by 0.1%/K or
more. Using a Voigt prole becomes especially important at high altitude locations,
such as Bozeman, Montana, where the ground level air pressure is already less than
1.0 atm.
Optical Depth
After temperature sensitivity, the next limiting factor in the absorption line se-
lection procedure is the strength of the line, or more specically, the optical depth it
75
100 200 300 400 500
-0.4
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
0.4
E"=300E"=250E"=200
E"=150
E"=100
E"=50
E"=0
0.25 Atm
 Voigt Profile
 Lorentz Profile
 
 
De
ns
ity
 E
rr
or
 (%
/K
)
Temperature (K)
Figure 4.6: Deviation from the Voigt prole in the number density error calculations
at 0.25 atm that would occur if a Lorentz prole were used.
produces. Optical depth (OD), τ [unitless], is dened as
τ =
∫ R
0
Nσdr, (4.10)
where R [m] is the range of the measurement and N [molecules/cm3] is the number
density of water vapor molecules. If the absorption line is too strong, σ will be
relatively large compared to the o-line absorption, causing τ to be too large and
the on-line return signal to be too weak, since the signal goes as e−τ . Conversely,
if the absorption line is too weakly absorbing, σ will be relatively small compared
to the o-line absorption, causing τ to be too small and the on-line return signal to
be undierentiable from the o-line signal return. Dierential absorption lidar error
analysis has shown that the optimal one-way value for τ is 1.1 at the desired range
(Remsberg and Gordley , 1978).
Because the optical depth is a function of the number density of water vapor
76
molecules, the optimal strength absorption line will be dependent on the geographic
location of the measurement. Water vapor lines that are suitable for Bozeman, Mon-
tana, a relatively dry climate, would be unusable in a highly humid tropical region.
Some a priori knowledge of the behavior of water vapor in the region of interest is
required. For my analysis, I used archived temperature, pressure, and relative humid-
ity data collected by the Optical Remote Sensor Laboratory's weather station located
on the roof of Cobleigh Hall on the Montana State University campus from August
2005 through July 2006 (Shaw , 2006).
The number density of water vapor is calculated by rst calculating the partial
pressure of water vapor using the equation
pH2O = RH ×
( 6.11
1013.25
)
× 10( aTb+T ), (4.11)
where RH [unitless] is the relative humidity and ranges between 0 and 1, a = 7.5,
and b = 237.3. The factors of 6.11, a, and b relate to saturated vapor with respect to
liquid water, and would be slightly dierent if ice produced the vapor. The factor of
1/1013.25 converts the partial pressure from millibars to atmospheres (Kyle, 1991).
From the partial pressure, the number density of water vapor can be calculated using
the ideal gas law, pH2Ov = nkT , where v [m−3] is the volume and n [unitless] is the
number or molecules. Solving for the number density, N = n/v,
N =
(pH2O
kT
)
×
(1.01325× 105
1.0× 106
)
[cm−3]. (4.12)
Using equations 4.11 and 4.12, a monthly average water vapor mass density was
computed for one year spanning the range of dates given above. The results are
plotted in Figure 4.7.
77
1 2 3 4 5 6 7 8 9 10 11 12
0
2
4
6
8
10
 
 
Average Monthly Water Vapor Density
W
at
er
 V
ap
or
 D
en
si
ty
 (g
/m
3 )
Month
Figure 4.7: The average monthly water vapor density in Bozeman, Montana. The
water vapor density was calculated using temperature, pressure, and relative humidity
data collected by the Optical Remote Sensor Laboratory's weather station on the roof
of Cobleigh Hall at Montana State University between August 2005 and July 2006.
The average density of water vapor was then used in a MATLAB program to
compute the monthly average OD using equation 4.10. The desired range was in-
put by the user at the command line; 2.0 km was used in my analysis. Water vapor
lines being analyzed were limited to those within the acceptable bounds for E” as
stated in section 4. The center wavenumbers of the most promising lines, which pro-
duced OD's between 0.70 and 2.00 during ve or more months of the year include:
12060.1078 cm−1(8months), 12074.5689 cm−1(9months), 12078.817 cm−1(7months),
12082.223 cm−1(5months), and 12085.602 cm−1(8months). The nal two absorption
lines listed were listed as usable in analysis of the near-infrared wavelength region by
Ambrico et al. (2000). Again, it must be emphasized that absorption line analysis
must be done for the environment where the measurements are being made, and can-
not be universally generalized. These lines, as well as the others picked for the 830nm
78
region from Ambrico et al. (2000) were not chosen for my measurements because they
were determined to be either too weak or too strong for the Bozeman atmosphere.
One limitation of the custom MATLAB OD analysis is that it assumes uniform,
constant water vapor distribution throughout the entire range being studied, which is
not true and will cause the optical depths to be larger than they would be in reality.
Therefore, I wanted to check these numbers with a more realistic simulator, namely
HiTRAN-PC. I performed vertical (slant path 90 degrees from horizontal) simulations
using the monthly average weather station data for T , P , and pH2O as ground-level
initial values. The simulation included 10 atmospheric layers of 200m height each
starting from 1.524 km (the GPS-measured altitude of the DIAL instrument) and
ending at 3.524 km. The atmospheric model used was either the Arctic winter, U.S.
Standard, or Mid-latitude summer model, depending on the time of year being sim-
ulated. While using these models corrected the problem of the water vapor being
constant throughout the range of measurement, they were still not exactly correct for
Bozeman, tending to be too wet in the winter months and too dry in the summer
months.
The HiTRAN-PC simulation showed remarkable similarity to the MATLAB anal-
ysis, and basically conrmed the results that one absorption line had the optimal line
strength throughout most of the year: 12074.5689 cm−1. OD values for both analyses
are shown in Table 4.2.
Month Jan. Feb. Mar. Apr. May June
OD (MATLAB) 0.93 0.71 0.96 1.38 1.56 ≥ 2.00
(HiTRAN-PC) 0.94 0.93 0.95 1.00 1.03 1.16
Month July Aug. Sept. Oct. Nov. Dec.
OD (MATLAB) ≥ 2.00 ≥ 2.00 1.64 1.58 1.14 0.87
(HiTRAN-PC) 1.17 1.10 1.03 1.02 0.98 0.94
Table 4.2: Optical Depth values for the water vapor absorption line centered at
12074.5689 cm−1 using a custom MATLAB analysis and HiTRAN-PC simulation.
79
Nearby Absorption Features
The nal criterion for selecting a suitable water vapor absorption line is to ensure
that the absorption line is adequately isolated from other absorption features of wa-
ter vapor or other gasses. Figure 4.8 produced by HiTRAN-PC shows all absorption
lines in the 12059.8167-12086.0527 cm−1 (829.2-827.4nm) range, where the best tar-
get water vapor lines reside according the the previous temperature sensitivity and
OD analysis. All absorption features in this region are due to water vapor. It is
evident that several of the candidate water vapor absorption lines listed in the pre-
vious section are isolated from and do not overlap with other absorption lines. It is
interesting to note that the absorption line at 12062.4098 cm−1 (829.022nm), used for
the horizontal tuning experiments described in Chapter 3 is actually unsuitable for
DIAL measurements because it overlaps with two weaker absorption lines that are not
suciently temperature insensitive. As a DIAL pulse propagates upwards through
the atmosphere, the absorption cross sections of these weaker absorption lines would
change quickly with temperature, rendering a DIAL measurement on this line useless
even though it ts the other two line selection criteria quite well. It is, however,
adequate for the horizontal tuning experiments where the temperature and pressure
can be assumed to be nearly constant throughout the entire horizontal pathlength,
and where the entire absorption feature is scanned across.
Line Selection
The water vapor absorption line at 12074.5689 cm−1 (828.187nm) was ultimately
selected for use in the MSU water vapor DIAL. It meets the temperature sensitivity
requirements, has a line strength that gives nearly optimal optical depth during most
months of the year in Bozeman, Montana, and is suciently isolated from other
80
12060 12065 12070 12075 12080 12085
0.0
0.2
0.4
0.6
0.8
1.0
829.2 829.0 828.8 828.6 828.4 828.2 828.0 827.8 827.6 827.4
0.0
0.2
0.4
0.6
0.8
1.0
Off-line
On-line
Tr
an
sm
is
si
on
Wavenumber (cm-1)
 Wavelength (nm)
Figure 4.8: A HiTRAN-PC plot at default values of 1 atm and 296K across a 1.5 km
horizontal path length, showing all absorption lines between 12059.8167 cm−1 and
12086.0527 cm−1 (829.2-827.4nm). The nal on- and o-line wavelengths selected for
the water vapor DIAL are indicated.
nearby absorption features. The o-line wavelength was chosen to be 12073.1099 cm−1
(828.287nm) to minimize interference with other absorption features, in particular
the absorption line located at 12075.8654 cm−1 (828.098nm). Figure 4.8 indicates
the nal on-line and o-line wavelengths.
81
CHAPTER 5
DIAL SYSTEM AND EXPERIMENT
Introduction
After it was shown through the horizontal experiments that the ECDL and TA
transmitter system would be able to successfully tune across a number of dierent
water vapor absorption features, and taking into account the lessons learned from
these experiments, a vertical DIAL system was ready to be designed and built. Since
only the transmitter was being tested and hard targets were used for scattering in
the horizontal experiments, many of the design ineciencies could be ignored. This
was not the case with a vertical DIAL system. The scattering cross section for at-
mosphere, especially in a clean-air environment such as Bozeman where aerosols are
relatively minimal, is much smaller than that for a hard target. Thus, careful at-
tention had to be paid to the DIAL design to ensure that the maximum number of
photons could be transmitted and then counted. Several components used in the
horizontal experiments were changed or upgraded. This chapter describes the design
considerations that drove the selection of components and subsequent design of the
DIAL system. The system is described in detail, along with its characterization and
testing. The experimental results of the nal DIAL system are shown.
Design
The layout for the DIAL experiment followed the layout of the horizontal tuning
82
system closely, with several key improvements. Increasing a lidar's transmit power is
the quickest way to obtain better signal-to-noise ratio (SNR), so a second tapered am-
plier was added to the DIAL. The second way to greatly improve SNR is to decrease
background radiation, or in the DIAL case, obtain a narrow-band lter with a very
small bandwidth to block as much light outside the laser line as possible. For compact,
low-power systems, increasing the average energy output of the laser transmission by
using high laser repetition rates increases the SNR. This was accomplished by using a
much faster MCS card in the receiver, allowing for higher repetition rates. A fast, sta-
ble tuning system that would tune the laser between the on- and o-line wavelengths
and hold it stable at the wavelength for the duration of the measurement was devel-
oped for the DIAL. This tuning system, along with improvements in the operational
and analysis software, allowed the DIAL to become a mostly autonomous instrument,
a radical improvement over the horizontal tuning system. These improvements are
described in detail later in this section.
The original DIAL design called for placing all transmitter optics, including the
laser and ampliers, on the horizontal surface of a 2′ × 4′ optical breadboard to
keep the instrument footprint to a minimum. The laser beam would then be sent
to a vertical breadboard for transmission into the atmosphere. Placing all of the
receiver optics on the opposite side of this vertical breadboard then would allow for
isolation and separation of the receiver, especially the APD, from transmitted laser
light, greatly reducing background signal in the nal data sets. This design would
also easily facilitate a bistatic lidar approach, where the receiver eld of view (FOV)
and transmit beam are not co-located when they leave the system. The limited space
of the 2′×4′ optical breadboard made the optical layout challenging because not only
did all of the parts need to t, but space also had to be factored in to account for the
many adjustments and repairs that would be done to the system over time, which,
83
after all, is still is a prototype.
The rst major deviation from the design plan occurred when it was discovered
that vertical breadboards with holes drilled and tapped on both surfaces were no
longer available for purchase in the size that was needed (1′× 3′). The nal transmit
optics and receiver optics would have to be overlapped. This design change also made
a bistatic approach more dicult, as the transmit beam would have to be directed
around the telescope.
Multiple discussions were had about the advantages of a bistatic DIAL system
versus a coaxial system, where the transmit beam is sent through the receiver tele-
scope. The advantage of the co-axial system is that in low-power lidar systems where
the FOV is very tight to limit background radiation from reaching the detector, the
transmit beam and receiver FOV will always be in alignment. Vibrations in the
system should aect both the transmit and receive beams identically, and using the
telescope to expand and transmit the laser beam ensures that the beam is within the
receiver FOV. Bistatic systems are much harder to align with such narrow FOV's,
and are subsequently harder to keep in alignment through vibrations and thermal
variations. Another problem with bistatic lidars is the overlap function must be ac-
curately accounted for, and care must be taken to ensure the system is in overlap at
the altitude where data are desired. The major disadvantage to the coaxial system
is that there is not an easy, compact, or inexpensive way to avoid the leading edge
of the full-power transmit pulse from being back-reected into the photon-counting
detector, temporarily blinding it at best or damaging it at worst. Pockell cells are
one solution for isolating the detector from these back-reections, but that option was
not very feasible in a DIAL receiver as compact as the one that was being designed
for this work. Another disadvantage to the coaxial approach is that, as explained in
chapter 3, the telescope optics are not optimized for near-infrared light, which would
84
lower the overall system eciency since a portion of the transmit beam would be lost
as it traveled through the telescope on both the transmit and receive transits. The
decision between bistatic and coaxial systems comes down to if the initial blinding
and resulting afterpulsing in the detector can be removed through post-processing.
If that is the case, then the coaxial technique can be used, as is currently done with
Micropulse Lidars (MPL's) (Campbell et al., 2002). However, if the afterpulsing ef-
fects cannot be eectively and consistently removed, then a bistatic approach must
be used, which has also been used with success (Machol et al., 2004).
Simple calculations were performed to determine how accurate the pointing of
a bistatic system would need to be to keep the transmit beam in overlap with the
receiver FOV as a function of altitude. The results of these calculations are shown
in gure 5.1. If a mirror or other optic in the transmit path caused the beam to
deviate by just 200 µrad, for example, the transmit beam would leave overlap with the
receiver FOV at around 800 meters. Primarily because of stability concerns about the
transmit beam pointing, the decision was made to make the DIAL a coaxial system.
System Description
To achieve accurate DIAL data with a low-power vertical system, collecting every
return photon is important. For this reason, much care was put into the design of
the DIAL system, especially the receiver. Optical design software, Zemax, was used
to design the receiver optics. Since the APD active area is 170µm in diameter, its
size dictated the type of ber optic cable that could feed photons to it, to avoid
over-lling the detector area. The ber that was used was a custom multi-mode
105 µm core diameter ber with a numerical aperture (NA) of 0.22. This ber size
and NA made it the eld stop of the receiver system, and dened the full FOV
85
Figure 5.1: A plot of the maximum allowable error in pointing a bistatic laser beam
to keep it within the eld of view of the detector, as a function of altitude.
86
of the system to be 150 µrad. The rest of the receiver optics were responsible for
taking the photons collected and focused through the telescope, collimating them for
passage through the narrowband lter, and focusing them into the ber to be carried
to the APD for detection. Trial-and-error with commercially available optics led to
the Zemax design shown in gure 5.2. The vertical line on the left represents the
focal plane of the telescope. 8.89 cm (3.5 inches) from there a Newport PAC037
converging lens collimated the beam at a diameter of 9 mm. The collimated space
then could contain several optics, including a quarter-wave plate used for allowing
return photons, but not transmit photons, to reach the detector, a beamsplitter cube
for the future addition of a near-eld channel, a PBS where the transmit beam is
coaligned with the receiver line, and the narrowband lter. Finally, two Thorlabs
lenses (AC127-075-B and AC127-019-B) focused the beam down into the ber optic
cable. The receiver was originally designed to be 33.02 cm (13 inches) from the
telescope focal point to the ber launch.
With the receiver fully designed, the rest of the DIAL could be put together. The
ECDL output was circularized by an anamorphic prism pair before passing through
two optical isolators to prevent feedback into the laser diode, as in the horizontal
tuning experiments. Again a half-wave plate and PBS combination was used to send
part of the ECDL beam to a wavemeter for monitoring, and the other part of the
beam on to seed the rst tapered amplier. The rst TA output was sent through
an optical isolator and second half-wave plate and PBS for monitoring. Two mirrors,
widely-spaced irises, and a collimating lens were used to seed the second tapered
amplier. The output of this TA was similarly sent through an optical isolator and
half-wave plate and PBS. The transmit light was focused and incident on a mirror
that turned the beam vertically onto the vertical breadboard. All mirrors and optics
were coated for infrared light near 830 nm to minimize scattering and raise system
87
Figure 5.2: A plot produced by Zemax showing the design for the DIAL receiver.
88
eciency. A plot of the reection properties of the mirrors is shown in gure 5.3.
The transmit beam was sent through the AOM and then through an iris to block
all light except the rst-order diraction pulsed beam. Several ipper mirrors and a
ber launch were included before and after the AOM for use in measuring the spectral
purity, as described in the testing section below. The transmit beam was collimated
to a diameter of roughly 9 mm after the iris and brought up to a piggy-backed optical
breadboard that contained the receiver optics. The transmit beam was brought into
the side of the PBS in the receiver line. The transmit light was vertically polarized
with respect to the plane of the vertical breadboard, and therefore was reected out of
the PBS towards the telescope in the receiver line. It passed through the quarter-wave
plate, causing it to become circularly polarized. After scattering o of a molecule or
particle in the atmosphere, the transmit photons would be collected by the telescope,
focused, collimated, and sent through the quarter-wave plate now circularly polarized
in the opposite direction with respect to the quarter wave plate. This would produce
horizontally polarized light with respect to the plane of the vertical breadboard, which
would now pass straight through the PBS, through the narrowband lter, and into
the ber. Photons were detected by the APD and counted by a new MCS card. More
detailed descriptions of the key components of the DIAL are included below.
The requirements for DIAL measurements with an error due to individual laser
properties of < 3% are stringent (Bösenberg , 1998), yet were still met or exceeded by
the nal MSU DIAL instrument. These requirements are compared to the current
MSU DIAL specications in Table 5.1.
Cascaded Tapered Ampliers
Transmit power is probably the single most critical parameter for lidar perfor-
89
Figure 5.3: A plot of the reection for the Thorlabs E03 coating, used in all of the
mirrors in the DIAL system. Image courtesy of Thorlabs.com.
Parameter Measured Value Requirement (at 830 nm)
On-/O-line Wavelength (nm, vacuum) 828.187/828.287
Repetition Rate (kHz) 20
Pulse Width (µs) 1.0
Pulse Power (µJ) ∼ 0.25
Linewidth (FWHM; MHz) < 0.300 < 298
Frequency Stability (MHz) ±88 ±160
Spectral Purity 0.995 > 0.995
Telescope Diameter (cm) 28
Far-eld Full Field of View (µrad) 150
Filter Bandwidth (pm) ∼ 250
Table 5.1: Laser transmitter requirements for water-vapor DIAL measurements with
an error due to individual laser properties of < 3% compared to the Montana State
University DIAL transmitter specications.
90
mance. In an attempt to increase the transmit power above the ~ 120 mW maximum
power of the horizontal lidar system, a second, cascaded tapered amplier was added
to the DIAL, which to my knowledge is the rst time cascaded tapered ampliers
have been used in a DIAL system. The ECDL transmit power is too low to fully
saturate the TA that it seeds, but this TA can be used to seed and fully saturate a
second TA, allowing it to operate at its maximum output power of ~ 400 mW.
The initial transmit optics are very similar to the setup used in the horizontal
tuning experiments. The ECDL is used to seed the rst tapered amplier. The align-
ment procedure was identical to that of the horizontal tuning experiment. Greater
care was taken to provide adequate heat sinks for the TA. However, since the out-
put power of the rst TA was kept to ∼ 20 mW for seeding the second TA, heating
was no longer a problem. The rst TA was driven with currents around 2 A, and
the temperature was held stable at 21.0◦ Celsius. The second amplier was a newer
model, and was much harder to align because of a narrower gain region, making it
more dicult to send the seed beam completely through the gain region. Because
of the diculty with alignment, the second amplier was placed on the opposite end
of the optical breadboard with relation to the seeding amplier, allowing for close
to a meter of path for the laser to travel between ampliers. This made it possible
to use widely-spaced irises and focusing optics to decrease the beam diameter and
more accurately aim it through the gain region of the second amplier. The second
amplier was also placed on a z-axis translation stage to allow for ner adjustments
in seeding.
Eort was put into adequately heat-sinking the second amplier with mixed re-
sults. A heat sink that attached to the amplier while simultaneously allowing it to be
attached to the z-axis translation stage could not be found, and should be machined
for future experiments. Smaller heat sinks were attached to the amplier where pos-
91
sible, but in the end the amplier had to be cooled by a fan installed into the wall
of the enclosure box close to the second amplier. Before installation of the fan, the
TEC on the second amplier could not cool it quickly enough, causing the entire
system to overheat after only about 1 hour of operation. The amplier would have to
be cooled overnight before another data run was attempted. The other problem with
overheating is it causes the spectral quality of the transmit beam to degrade. After
installation of the fan, the TEC was able to stabilize the amplier's temperature at
the nominal value of 21.0◦ Celsius without diculty. The second amplier could be
operated at full power basically indenitely without overheating. Tests were run for
at least 5 hours straight at full power without the TEC current increasing to continue
cooling the amplier. Experimental tests showed that the fan was balanced enough
so that it did not vibrate the entire system.
Overall, as will be seen by test results in the forthcoming test section, the cascaded
amplier design gave a much-needed boost in transmit power, up to an average cw
power of ∼ 250 mW. The increase in output power was again not as high as we had
hoped because of the optics that came after the second amplier, such as another
optical isolator, and mirrors and lenses which each scattered some of the transmit
beam. The output power was eectively doubled from the horizontal experiments,
though.
A picture of the current and temperature controller for the second amplier (Pilot
3000 PC) is shown in gure 5.4.
Multi-channel Scalar Card
To increase the allowable repetition rate of the DIAL, the MCS used for the
horizontal tuning experiments was replaced by a much simpler and faster single MCS
92
Figure 5.4: A picture of the Sacher Lasertechnik Pilot driver used to control the
cascaded tapered amplier in the vertical DIAL experiments.
card (ASRC Aerospace AMCS-USB). The MCS card was built into a user-friendly
enclosure by Nick Jurich, an electrical engineering undergraduate. The MCS box
operated basically the same as the old MCS, except that without a display screen it
would operate much faster than the DIAL would be pulsed at, allowing the DIAL
repetition rate to be increased from the horizontal tuning limit of 2272 Hz to 20 kHz,
increasing the average output power by nearly a factor of 10.
Narrowband Filter
To increase the SNR of the DIAL, an extremely narrowband (NB) lter was desired
for the receiver to block as much background light outside of the laser wavelength as
possible from reaching the detector. Barr Associates manufactures high-quality NB
lters with bandpasses of 250 pm that are routinely used in lidar work. With such
a narrow band pass, though, care has to be taken to center the lter exactly at the
desired wavelength. Water vapor absorption line wavelengths measured in vacuum
are separated from wavelengths of the same lines in the atmosphere by more than
half of this bandwidth, meaning that a lter designed to be centered at the vacuum
wavelength of an absorption line would be rendered useless, as the line would be
outside of the band pass in the actual atmosphere. This eect is shown in gure 5.5.
An accurate method of calculating the wavelength shift due to the index of refraction
93
of air needs to be used to ensure that the NB lter is designed correctly. One method
is to use an equation, Cauchy's theorem, to theoretically calculate index of refraction
of air. This equation is given by
nair = 1.000287566 +
1.3412× 10−18
λ2 +
3.777× 10−32
λ4 , (5.1)
where λ as usual is the laser wavelength (Born and Wolf , 1999). This method is only
an approximation as it does not take into account changing atmospheric conditions
which can radically aect the index of refraction. At λ = 828.187nm, the index of
refraction is computed to be 1.00022.
Another, more experimental, method is to use the wavemeter to measure the
index of refraction. The ECDL was tuned to the target on-line vacuum wavelength
of 828.187 nm, and then the wavemeter was switched o of vacuum mode and into
air mode. It will then use ambient temperature and pressure readings to internally
calculate the new air wavelength, from which the index of refraction of air can be
computed. Using this method, the index of refraction was computed to be 1.0002896.
Because the wavemeter method took into account actual atmospheric conditions it
was deemed to be more accurate, and a NB lter was ordered from Barr with a center
wavelength of 828.0069 nm. This specication was correct, as the transmission curve
for the NB lter shows in gure 5.6. The on- and o-line wavelengths are marked
for reference. Note that the transmission for the o-line wavelength is considerably
reduced compared to the on-line wavelength, due to the wide spectral separation of
the two wavelengths. A full list of specications for the narrowband lter are given
in table 5.2.
94
827.7 827.8 827.9 828.0 828.1 828.2 828.3 828.4
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
250 pm NB Filter
Tr
an
sm
is
si
on
Wavelength (nm)
 n=1.0
 n=1.00022 (wavemeter)
 n=1.0002896 (theory)
Figure 5.5: A plot of the DIAL's target water vapor absorption line at 828.187 nm
(vacuum wavelength, n = 1.0) using HiTRAN-PC. The wavelength shift due to two
values of the index of refraction of air is shown. A value of n = 1.00022 was calcu-
lated internally by the Burleigh wavemeter using ambient temperature and relative
humidity measurements. A value of n = 1.0002896 was calculated using Cauchy's
formula. The red box signies a hypothetical rectangular band pass region of a 250
pm narrowband lter centered at 828.0069 nm, the line center location using the
wavemeter index of refraction.
Parameter Value
Center Wavelength (nm) 828.01± 0.05
FWHM (nm) 0.25± 0.05
Peak Transmission > 50%
Diameter (inches) 1.0
Angle of Incidence (
◦
C) 0
Operating Temperature (
◦
C) 23.0
Thickness (mm) < 7.1
Table 5.2: Ordering specications for the narrowband lter.
95
827.50 827.75 828.00 828.25 828.50
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Tr
an
sm
is
si
on
Wavelength (nm)
 NB Filter Transmission
 Target Wavelengths
Figure 5.6: A plot showing the transmission curve for the narrowband lter, with
the DIAL on-line (828.0069 nm) and o-line (828.1069 nm) labeled. Data for the
transmission was provided courtesy of Barr Associates.
96
Extended Tuning System
Stable tuning of ECDL's is important for scanning across spectroscopic features,
or for quickly changing between frequencies on and o of a spectroscopic feature,
as in DIAL. To meet the frequency stability requirements described above, and to
ensure that mode hops do not occur in the laser wavelength (or at the very least are
accounted for so that they can be removed from the data sets in post-processing),
a fast, stable tuning method needed to be developed. Fortunately, a method for
extending the continuous tuning range of a ECDL, created by Repasky et al., 2006 for
use in oxygen absorption experiments, was able to be modied to provide the fast,
stable tuning needed by the DIAL system. It makes use of an electronic feedback loop
to suppress mode hops and extend the continuous tuning range of the ECDL. The
system and associated tests are also described in detail by Obland et al. (accepted for
publication, 2007).
A schematic of the tuning setup is shown in gure 5.7. A low-noise current con-
troller was built based on the design of reference Libbrecht and Hall , 1993 to serve
three purposes. First, it provides a stable dc current to the diode laser. Second, a
small RF modulation can be added to the dc current to modulate the wavelength
of the ECDL. The current controller converts an RF voltage signal into a modu-
lated current and adds this to the dc set point. Third, a small signal current that
is proportional to the voltage applied to the small signal input port can be added
to the dc current. The small signal correction current will change the temperature
and carrier density of the gain region of the laser diode and thus change the optical
path length of the ECDL. In this way, small corrections to the optical path length
of the ECDL can be made to synchronize the cavity mode and the frequency of the
optical feedback so that mode hops can be suppressed. A function generator is used
to provide an RF signal to the current controller to modulate the operating frequency
97
BeamSplitter
Detector
ECDL
Differential
Amplifier
Current
Controller
Function
Generator
Temperature
Controller
Reference
Voltage
PZT Voltage
Controller
Lock-in Amplifier
Figure 5.7: A schematic of the extended tuning system for the ECDL.
of the ECDL, which in turn will create a small modulation in the output power of
the ECDL as seen by the fast silicon photo-detector. By comparing the phase of the
modulated optical power with the phase of the RF signal created by the function
generator through the use of a lock-in amplier, an error signal is created. The error
signal created by the lock-in amplier is sent to a dierential amplier that allows
rst for the subtraction of a reference voltage and second for amplication of the
resulting error signal. Finally, the conditioned error signal is fed into the small signal
input port of the current controller to allow corrections to the external cavity length
to be made as the laser tunes to synchronize the cavity mode and the frequency of
the optical feedback. The result of this electronic feedback scheme is the suppression
of mode hops and the extension of the continuous tuning range of the ECDL. The
cost of the tuning system, excluding the ECDL and its controllers, is probably less
than 10,000 USD and comprised of commercial-o-the-shelf parts. Deploying such a
system in the eld would require reasonable environmental controls, but is probably
less complicated than deploying the laser system in the eld.
The optical setup for studying the tuning of the ECDL is shown in gure 5.8.
The elliptical output beam of the ECDL is rst sent through an anamorphic prism
pair to create a nearly circular beam. Next, the beam passes through two Faraday
98
isolators to prevent back reections from aecting the performance of the ECDL.
After the isolators, a portion of the beam is picked-o using a half wave plate and
polarizing beam splitter (PBS) cube. The picked-o beam is next incident on a
wedged beam splitter. The light reected from the wedged beam splitter is sent to
a detector used for monitoring the output power of the ECDL, as shown in gure
5.7 for the electronic feedback loop. The light transmitted through the wedged beam
splitter is ber coupled and sent to a Burleigh wavemeter that monitors the operating
wavelength of the ECDL. The primary optical beam that passes through the rst pick-
o is coupled into a commercial solid-state ared amplier (Sacher Lasertechnik) that
provides an amplied collimated circular output beam. The amplied output beam is
sent through a Faraday isolator to prevent back reections from damaging the optical
amplier. A second pick-o using a half wave plate and PBS cube is used to send
a portion of the amplied beam to a ber coupler. The ber-coupled light can then
be used with either a wavemeter or optical spectrum analyzer to monitor the optical
spectrum of the amplied beam. The light that passes through the second pick-o
is next incident on a wedged beam splitter. The wedged beam splitter reects about
four percent of the laser beam onto a reference power meter that accounts for changes
in the power output of the laser system, while the remainder of the amplied beam
that passes through the wedged beam splitter is coupled into a sealed gas absorption
cell. Light that passes through the gas absorption cell is incident on a second power
meter used to measure the transmission through the gas absorption cell.
A second computer-controlled feedback loop to control the ECDL operating wave-
length was created as well. The operating wavelength of the ECDL is monitored by
the wavemeter and read every 2 seconds by a computer, which then compares the
measured wavelength with a user dened desired wavelength. The computer will out-
put a voltage relative to the dierence between the actual and desired wavelength
99
Isolator l/2ECDL
Multi-passGas Absorption Cell
Isolator PBS
Anamorphic
Prism Pair
Transmission Detector
4% Beam
Splitter
To Wavemeter
To OSA
Tapered
Amplifier
l/2
PBS
Isolator
Collimating
Lenses
4% Beam
Splitter
Reference Detector
To Lock-in
Detector
Figure 5.8: A schematic of the gas absorption cell experimental setup. Note that the
laser makes 36 passes within the gas absorption cell for a total path length of 19.8
meters.
100
which is then sent to the piezoelectric transducer used for ne tuning of the operating
wavelength of the ECDL. This second computer controlled feedback loop is used to
lock the operating wavelength of the ECDL to a desired set point.
Data Acquisition and Analysis Software
LabVIEW code from the horizontal tuning experiments was modied to create the
operational code for the water vapor DIAL, and operated in the following manner.
The software (DIAL_v10.vi, described in appendix D), operating via GPIB connec-
tions, initialized the MCS card and all instruments. It then grabbed the date and
time for the purposes of le labeling, and for time-stamping every second (20,000 laser
pulses) of data. A custom waveform was selected on the arbitrary waveform generator
(AWG) that created the desired pulse, a 1 µs pulsewidth at a repetition rate of 20
kHz. The AWG triggered the APD to begin collecting data but the laser pulse was
not activated until 5 µs later, allowing for a background light measurement to be
made. The ECDL PZT voltage was set to zero to initialize it for tuning. The laser
was allowed to settle for 5 seconds at the zero PZT voltage and the MCS memory
buer was cleared and readied for data collection. The laser was then tuned to the
on-line wavelength and allowed to equilibrate for 5 seconds. The wavemeter feedback
loop described in the tuning tests below tuned the laser wavelength until the on-line
wavelength was reached. This wavelength and the reference power were measured.
The MCS collected data for one second across 500 bins of 50 ns each. A time stamp,
the wavelength and reference power, and these bins were recorded to a data le. This
procedure of recording wavelength, reference power, and photon counts was repeated
as long as desirable at this wavelength, typically 60 seconds, before the laser was
tuned to the o-line wavelength and the procedure was repeated there, typically for
101
90 seconds. The dierence in averaging times helped alleviate the lower transmission
of the o-line wavelength through the NB lter. This on- and o-line tuning contin-
ued as long as was desired, typically for 3-5 hours per data run. The LabVIEW code
allowed for real-time display of the wavelength for monitoring tuning, the reference
power, and the bin counts.
After the data were taken, MATLAB code was used to analyze it. The anal-
ysis code (described in appendix D) scanned the data set and removed data that
did not t within the wavelength stability requirement of ±160 MHz of the chosen
wavelength. It then averaged the data spatially, binning the counts according to the
resolution dened by the pulse width, or 150 meters for 1 µs pulses. After spatial
averaging, the software normalized the counts with power and subtracted the back-
ground measurement from the overall data on a second-by-second basis to increase
accuracy. Temporal averaging was typically performed over an hour of data to collect
enough signal without allowing the atmosphere to change drastically. This averag-
ing produced diculties on nights when the atmosphere was changing rapidly, such
as on windy nights or when a storm system was entering the area. The averaged
and background subtracted counts were then used with absorption cross section val-
ues generated from radiosonde temperature and pressure measurements to calculate
water vapor proles using the DIAL equation, equation 2.27.
Testing
Many tests and diagnostics were run to verify that the DIAL instrument, shown
in its complete form in gure 5.9, would be able to meet the stringent requirements
for accurate DIAL data, as previously dened.
102
Figure 5.9: A picture of the DIAL instrument in its operational form inside the
roofport room of Cobleigh Hall.
Noise and Background Signal Levels
Tests were run to determine how well the APD was isolated from background
light. Results from these tests are shown in table 5.3. Note that for these tests
the telescope receiver was uncovered, so any background light present in the room
would have a path to reach the detector. The room lights were kept o. When
only the ECDL was on, the APD was counting very near to its stated noise oor of
∼ 150 counts per second. Activating the rst tapered amplier slightly increased the
number of photons being counted. Activating the second tapered amplier increased
the background level more, even with the AOM inactive and completely blocked. This
is an indication of light from the second TA nding a path into the APD independent
of the transmit beam. When the AOM was active but still blocked, the background
level increased dramatically, indicating either that light being pulsed through the
AOM is reaching the APD independent of the transmit beam path, or that reections
o of the beam block were reaching the detector, which would not be a problem in
103
Conguration Counts/Second
APD Dark Count (from manufacturer) ∼ 150
ECDL Only 168.97
ECDL and TA 1 170.77
ECDL, TA 1, TA 2, AOM o and blocked 218.87
ECDL, TA 1, TA 2, AOM on and blocked 321.62
ECDL, TA 1, TA 2, AOM o and unblocked 1898.79
Viewing Daylight Sky 2.86× 106
Table 5.3: Results of testing for background light leakage into the APD.
the actual experiment. Leaving the AOM o but unblocked immediately illustrates
the problem of light leaking through the AOM crystal and increasing the background
light level. This is support for replacing the AOM and nding a way to pulse the
ampliers directly. Finally, as a test of overall ltering of the receiver, the roofport
was opened and the detector was allowed to view the sky during daylight hours. As
can be seen, the count rate jumped to almost 3 million counts, indicating that much
more work has to be done with ltering before daytime measurements will be possible
with this system.
Beam Shape
A camera (Big Sky Laser TM-745) and software program called BeamView were
used to image the output of the ECDL and rst amplier, both seeded and unseeded,
to better understand the beam quality of the laser system. The laboratory room lights
were shut o during the image captures to avoid saturating the camera. The camera
was calibrated via software before and between each measurement. The ECDL was
run at the typical current of 37 mA. For an image of the ECDL output, the camera was
placed after the anamorphic prism pair, optical isolators, half wave plate, and PBS
cube. Neutral density lters were placed in front of the camera to avoid damaging
it. The front of the camera lens was located a distance of 4.5 inches after the edge of
104
Figure 5.10: A BeamView gure showing the intensity of the ECDL beam after
traveling through the anamorphic prism pair, optical isolators, half wave plate, and
PBS cube.
the PBS cube. The ECDL output image is shown in gure 5.10, and shows that the
beam is relatively circular shortly before it seeds the rst amplier.
The TA was seeded by the ECDL and run at full power of ∼ 230 mW with a
drive current of 2.997 A. An optical isolator was not used after the tapered amplier,
but again, neutral density lters were placed in front of the camera. For images of
the TA output, the front of the camera lens was placed at distances of 5 inches, 10
inches, and 30 inches from the edge of the TA collimating lens. Figure 5.11 shows
the resulting images when the TA was both seeded and unseeded. The images show
that the TA output is much more circular when seeded by the ECDL, and becomes
more circular in shape with distance.
Spectral Purity
Spectral purity is dened as the fraction of laser power within the laser linewidth
105
Figure 5.11: These BeamView gures show the tapered amplier beam shape at
3 dierent distances from the amplier, unseeded (left column) and seeded by the
ECDL (right column). The top row is at a distance of 5 inches, the middle row is at
a distance of 10 inches, and the bottom row is at a distance of 30 inches. The red
lines are cross sections of the intensity in two dimensions. The yellow ellipse shows
the ellipticity of the beam. Notice that the beam is much more uniform when seeded
by the ECDL, and becomes more circular as it propagates.
106
with respect to the entire output power of the laser. It can also be thought of as
basically how much laser power exists outside of the laser linewidth versus how much
laser power exists at the desired wavelength. The higher the spectral purity, the less
power is in the sidebands of the laser line. It is a critical parameter for accurate DIAL
measurements because the assumption is made that the laser is nearly monochromatic,
and therefore all of the laser light is absorbed uniformly by an absorption line. If,
however, the laser power is spread across a larger spectrum, the laser power will be
absorbed dierently at each wavelength, and the functional dependence of the laser
output power with wavelength must be well known to invert the DIAL equation.
A technique for measuring spectral purity is given by Wulfmeyer , 1998. In that
paper, the authors describe saturating the air inside of a multi-pass gas absorption
cell with water vapor. To measure spectral purity, they then pass the laser beam
from their lidar system through this gas cell and measure the amount of power that is
transmitted completely through the cell. The idea is that, with the laser wavelength
set to line center of the water vapor absorption line, all laser power that is contained
within the line width will be totally absorbed by the water vapor. Any power that
is not absorbed in the cell must be spectrally located at wavelengths outside of the
laser line, giving a measure of spectral purity. An example gure showing a saturated
water vapor absorption line at a pressure of 15 mbar plotted with a theoretical 40
MHz wide laser line and a pressure-broadened absorption line is shown in gure 5.12.
One argument about why this approach may not be an entirely accurate method
for measuring spectral purity is simply that a saturated water vapor line is not an
accurate representation of the shape of water vapor absorption lines in the actual
atmosphere. Typically, the absorption line is far from saturated, perhaps only ab-
sorbing 70 % of the laser power. Because the line is not saturated, it will also be much
thinner. Therefore, laser power that may lie outside of the laser linewidth and still be
107
726.74 726.75 726.76 726.77 726.78
0.0
0.2
0.4
0.6
0.8
1.0
Wulfmeyer Spectral Purity Test*
Tr
an
sm
is
si
on
Wavelength (nm)
HITRAN Water Vapor 
Absorption Spectrum
 291 K, 15 mbar
 291 K, 850 mbar
 Gaussian Laser Line
(40 MHz linewidth)
Figure 5.12: A plot of a water vapor absorption line at low pressure (high altitude)
or at high pressure, such as in a pressurized multi-pass, gas absorption cell. A hypo-
thetical Gaussian laser line with a 40 MHz linewidth is overplotted for comparison.
*See Wulfmeyer , 1998 for a description of using a multi-pass gas absorption cell for
making spectral purity measurements.
108
absorbed by a broadened, saturated water vapor absorption line, may not be absorbed
by a much thinner absorption line that would be used to take DIAL measurements,
lowering the spectral purity. Perhaps a better technique for measuring spectral purity
is to measure the power in the laser linewidth directly from OSA traces, and compare
that to the overall power output of the laser beam. This would at least provide a
lower limit for the spectral purity of the transmitter. This technique was used to
measure the spectral purity of the MSU DIAL.
Initial spectral purity measurements of the DIAL transmitter showed that the
system had a spectral purity of 0.872. This value is not surprising since diode lasers
and ampliers based on semiconductor technology tend to have broad spectrums and
large amplied spontaneous emission (ASE), hence the use of external cavities to
force the diode laser output to be more monochromatic. Adding a narrowband lter
to the receiver of the DIAL drastically improves the spectral purity of the system,
as only the light that passes through the lter is measured and a large amount of
the ASE is blocked. Figure 5.13 shows OSA traces of the power output of the DIAL
transmitter both before and after the AOM, on a linear scale with a NB lter in place.
Before spectral purity measurements could be made, several questions needed to be
answered to ensure that the spectral purity being measured was real and not a relic
of the measurement technique itself.
The rst question to be answered was whether the spectral purity measurements
needed to be made with multi-mode (MM) or single-mode (SM) bers. Results of
measurements made show that the ASE structure is relatively consistent between
MM and SM bers, both before and after the AOM, as seen in gure 5.14. Note that
these and subsequent spectral purity measurements are not normalized to each other,
so amplitude dierences are simply a result of dierences in alignment.
The next question to be answered was whether the spectral purity is signicantly
109
810 820 830 840 850
0.0
0.5
1.0
1.5
2.0
2.5
3.0
06/19/06 Spectral Purity Test
Po
w
er
 (m
W
)
Wavelength (nm)
MM Fiber, 3 m length 
(125 micron core)
 Post-AOM, NB filter
 Pre-AOM, NB filter
*Note that ASE has 
 been magnified X20
Figure 5.13: A linear plot of the power output of the DIAL transmitter with a nar-
rowband lter in place, measured before and after the AOM through a multi-mode
ber on an OSA.
810 820 830 840 850
-60
-50
-40
-30
-20
-10
0
10
06/19/06 Spectral Purity Test
Po
w
er
 (d
Bm
)
Wavelength (nm)
Pre-AOM
 MM Fiber, 3 m length (125 micron core)
 SM Fiber, 5 m length (820 nm)
810 820 830 840 850
-60
-50
-40
-30
-20
-10
0
10
06/19/06 Spectral Purity Test
Po
w
er
 (d
Bm
)
Wavelength (nm)
Post-AOM
 MM Fiber, 3 m length (125 micron core)
 SM Fiber, 5 m length (820 nm)
Figure 5.14: These OSA traces show that the ASE structure in the DIAL transmitter
output is relatively independent of the type of ber used to make the measurement, as
the structure is similar before (left) and after (right) the AOM for both single-mode
and multi-mode bers.
110
aected by the presence of the AOM. Figure 5.15 shows the results of OSA traces for
both MM and SM bers, with and without the NB lter, before and after the AOM.
The ASE structure is consistent in these comparisons, and shows that the AOM does
not seem to have an eect on the spectral purity of the transmit beam.
Finally, the question was posed as to whether the spectral purity measurements
depended on the length of ber optic cable that was carrying the transmit beam to
the OSA. The graphs in gure 5.16 show comparisons of MM and SM bers of varying
lengths, before and after the AOM and with and without a NB lter. As before, the
ASE structure does not seem to be aected by longer ber optic cable lengths.
After the method for measuring spectral purity was thoroughly tested and un-
derstood, the nal OSA measurements were made, and are shown in gure 5.17. As
stated before, the spectral purity before use of a narrowband lter is 0.872. With a
NB lter in place, the spectral purity of the transmitter improves to >0.995 at the
on-line wavelength and >0.992 at the o-line wavelength. Note that the spectral pu-
rity for the o-line wavelength does not meet the spectral purity requirements stated
in table 5.1, but this should not matter because the o-line wavelength is far removed
from any water vapor absorption lines, and therefore the uniform absorption assump-
tion still holds. These tests show that the spectral purity of the DIAL is adequate
for accurate water vapor retrievals.
It should be noted that while performing the spectral purity measurements, an
interesting correlation between the amount of ASE in the transmit beam and the
measured reference power of the beam was observed. It was repeatedly shown by
measuring the power modes of the seeded tapered amplier while simultaneously
watching the mode structure of the amplier output, that the mode of highest power
corresponded to the mode with the lowest ASE. Stated another way, the higher the
quality of the transmit beam, the higher its power. A possible explanation for this is
111
810 820 830 840 850
-70
-60
-50
-40
-30
-20
-10
0
10
06/19/06 Spectral Purity Test
Po
w
er
 (d
Bm
)
Wavelength (nm)
MM Fiber, 3 m length (125 micron core)
 Pre-AOM
 Post-AOM
810 820 830 840 850
-70
-60
-50
-40
-30
-20
-10
0
10
06/19/06 Spectral Purity Test
Po
w
er
 (d
Bm
)
Wavelength (nm)
MM Fiber, 3 m length (125 micron core)
 Pre-AOM, NB filter
 Post-AOM, NB filter
810 820 830 840 850
-70
-60
-50
-40
-30
-20
-10
0
10
06/19/06 Spectral Purity Test
SM Fiber, 5 m length (820 nm)
 Pre-AOM
 Post-AOM
Po
w
er
 (d
Bm
)
Wavelength (nm)
810 820 830 840 850
-70
-60
-50
-40
-30
-20
-10
0
10
06/19/06 Spectral Purity Test
SM Fiber, 5 m length (820 nm)
 Pre-AOM, NB filter
 Post-AOM, NB filter
Po
w
er
 (d
Bm
)
Wavelength (nm)
Figure 5.15: These OSA traces show that the ASE structure in the DIAL transmitter
output is relatively independent of the presence of the AOM. The structure is similar
before and after the AOM for both multi-mode (top row) and single-mode (bottom
row) bers, with (right column) or without (left column) a narrowband lter in place.
112
810 820 830 840 850
-70
-60
-50
-40
-30
-20
-10
0
10
20
06/19/06 Spectral Purity Test
MM Fiber, 3 m length 
(125 micron core)
 Pre-AOM
 Pre-AOM, NB filter
 Pre-AOM, 
  +2 m MM fiber (600 micron core)
 Pre-AOM, NB filter, 
  +2 m MM fiber (600 micron core)
Po
w
er
 (d
Bm
)
Wavelength (nm)
810 820 830 840 850
-70
-60
-50
-40
-30
-20
-10
0
10
20
06/19/06 Spectral Purity Test
Po
w
er
 (d
Bm
)
Wavelength (nm)
MM Fiber, 3 m length 
(125 micron core)
 Post-AOM
 Post-AOM, NB filter
 Post-AOM, 
  +2 m MM fiber (600 micron core)
 Post-AOM, NB filter, 
  +2 m MM fiber (600 micron core)
810 820 830 840 850
-70
-60
-50
-40
-30
-20
-10
0
10
20
06/19/06 Spectral Purity Test
Po
w
er
 (d
Bm
)
SM Fiber, 5 m fiber length (820 nm)
 Pre-AOM
 Pre-AOM, NB filter
 Pre-AOM, +5 m SM fiber (850 nm) 
 Pre-AOM, NB filter, +5 m SM fiber (850 nm)
Wavelength (nm)
810 820 830 840 850
-70
-60
-50
-40
-30
-20
-10
0
10
20
06/19/06 Spectral Purity Test
SM Fiber, 5 m length (820 nm)
 Post-AOM
 Post-AOM, NB filter
 Post-AOM, 
  +5 m SM fiber (850 nm)
 Post-AOM, NB filter, 
  +5 m SM fiber (850 nm)
Po
w
er
 (d
Bm
)
Wavelength (nm)
Figure 5.16: These OSA traces show that the ASE structure in the DIAL trans-
mitter output is relatively independent of the length of ber optic cable being used.
The structure is similar before (left column) and after (right column) the AOM for
both multi-mode (top row) and single-mode (bottom row) bers, with and without a
narrowband lter in place and for varying ber lengths.
113
810 815 820 825 830 835 840 845 850
-80
-70
-60
-50
-40
-30
-20
-10
0
11/14/2006 DIAL Spectral Purity
Po
w
er
 (d
Bm
)
Wavelength (nm)
On-line (828.187 nm)
 No Filter
 With Narrowband Filter
810 815 820 825 830 835 840 845 850
-80
-70
-60
-50
-40
-30
-20
-10
0
11/14/2006 DIAL Spectral Purity
Po
w
er
 (d
Bm
)
Wavelength (nm)
Off-line (828.287 nm)
 No Filter
 With Narrowband Filter
Figure 5.17: OSA traces showing the eect of using a narrow band lter in the receiver
of the DIAL. The spectral purity is drastically improved.
that when the ampliers are correctly seeded the ASE is forced into the primary laser
linewidth, increasing the overall power. If the ampliers are not seeded correctly,
power that could be in the laser linewidth is spread to the sidemodes, decreasing the
overall measured power. This observation indicated that paying careful attention to
correctly seeding the ampliers before each data run would increase the overall power
of the experiment, improving the results.
P/I Curves
P/I curves were again performed for the ECDL and two ampliers used for the
vertical DIAL experiments, and are shown in gures 5.18 and 5.19. The ECDL
was now being driven at a drive current of 40 mA because of the requirements of the
extended tuning system. Typically only ∼ 6 mW was used to seed the rst TA, which
was driven at ∼ 2 A to seed the second TA with 20-22 mW. The second amplier
driven just below full current, ∼ 2.8 A. Higher seed power makes an obvious dierence
114
0 5 10 15 20 25 30 35 40
0
1
2
3
4
5
6
7
 9/14/06
 10/31/05
DIAL ECDL P/I Curve
Po
w
er
 (m
W
)
Drive Current (mA)
Figure 5.18: The latest P/I curve for the ECDL is plotted against the P/I curve
performed almost a year earlier during the horizontal tuning experiments described
in Chapter 3. Aging of the diode is evident as it requires more current to reach the
same power output.
in the output power of the second TA compared to the rst. The second amplier
produced nearly 450 mW output power on a consistent basis.
Linewidth and Tuning
As in the horizontal tuning tests, optical spectrum analyzer traces of the ECDL
and ampliers were taken, and are shown in gures 5.20 and 5.21. The high quality
linewidth and sidemode suppression of the ECDL and TA's is still evident, even
though the amplied spontaneous emission of the second amplier is very evident
because it is being run at nearly full drive current and full seed power.
Testing to ensure that the wavelength of the transmit beam after being passed
through the AOM was the same as that of the ECDL was performed as in the hor-
izontal tuning experiments: by measuring both wavelengths and forming their ratio
for comparison as the ECDL tuned. As expected from previous measurements, the
115
0.0 0.5 1.0 1.5 2.0 2.5 3.0
0
50
100
150
200
250
300
350
400
450
DIAL Amplifier P/I Curves
10/31/06
Po
w
er
 (m
W
)
Drive Current (A)
 Amp 1 Seeded (21.0 C, 6.1 mW seed power)
 Amp 1 Unseeded (21.0 C)
 Amp 2 Seeded (21.0 C, 19.8 mW seed power)
 Amp 2 Unseeded (21.0 C)
Figure 5.19: P/I curves for the two ampliers used in the DIAL experiment. Notice
the large dierence between ampliers seeded, but not saturated, and seeded at nearly
full seed power.
810 815 820 825 830 835 840 845 850
-60
-50
-40
-30
-20
-10
ECDL OSA Trace
11/02/06
Po
w
er
 (d
Bm
)
Wavelength (nm)
Figure 5.20: An OSA trace of the ECDL output, showing its narrow linewidth and
large sidemode suppression.
116
810 815 820 825 830 835 840 845 850
-60
-50
-40
-30
-20
-10
0
 Seeded
 Unseeded
DIAL Amp 1 OSA Trace
10/31/06
Po
w
er
 (d
Bm
)
Wavelength (nm)
810 815 820 825 830 835 840 845 850
-60
-50
-40
-30
-20
-10
0
DIAL Amp 2 OSA Trace
10/31/06
Po
w
er
 (d
Bm
)
Wavelength (nm)
 Seeded
 Unseeded
Figure 5.21: OSA traces of the outputs of the DIAL ampliers. The large increase
in amplied spontaneous emission due to higher drive current and seed power for
the second amplier is prominent, but disappears as expected when the amplier is
seeded properly.
results shown in gure 5.22 show that the wavelength output of the DIAL agrees the
wavelength being measured from the ECDL to within the uncertainty of the waveme-
ter, ±0.1 pm.
The speed and stability of the extended tuning system that was added to the
DIAL was tested in two ways. First, the speed and stability of the tuning system was
characterized through direct measurement of the laser wavelength as it locked onto
the target on- or o-line wavelength. Second, a dierential absorption measurement
of water vapor contained in a multi-pass gas absorption cell was used to verify that
the system was indeed tuning on and o of a water vapor absorption line. These tests
are described in two sections below.
On- and O-line Tuning Characterization Initial testing and characterization of
the tuning of the ECDL were made in the following manner. A computer program
was developed that begins by setting the voltage of the PZT used to electronically
117
45450 45500 45550 45600 45650 45700 45750
0.999990
0.999992
0.999994
0.999996
0.999998
1.000000
1.000002
1.000004
1.000006
1.000008
1.000010
Off- to On-line Tune
On- to Off-line Tune
11/30/06 DIAL Experiment
AOM/ECDL Tuning Test
AO
M
/E
CD
L 
W
av
el
en
gt
h 
Ra
tio
Time (seconds)
Figure 5.22: A ratio of the laser wavelength measured after the AOM and after the
ECDL, showing that the wavelengths agree.
tune the ECDL to zero. The voltage is then changed to a value pre-selected by
the user that tunes the ECDL to approximately 828.187 nm (12074.5677 cm−1), the
center wavelength (in vacuum) of a water vapor absorption line. The computer then
polls the Burleigh wavemeter every 2 seconds and compares the actual operating
wavelength of the ECDL with the desired operating wavelength. If the frequency
dierence between the actual wavelength and desired wavelength is larger than the
resolution of the wavemeter, ± 88 MHz, a voltage signal is generated and sent to
the PZT to tune the ECDL towards the desired operating wavelength and ultimately
hold the laser at that wavelength. A plot of the operating wavelength as a function of
time is shown in gure 5.23 for dierent initial starting operating wavelengths of the
ECDL. Fine-tuning of the ECDL to the precise wavelength may take a few seconds
up to minutes, depending on how close the initial PZT voltage setting, and hence
the initial laser wavelength, was to the correct value. By simply checking the voltage
settings before starting the experiment, the uncertainty in PZT voltage can easily be
reduced to ± 0.01 V, reducing ne-tuning times to 7-8 seconds between the selected
118
0 10 20 30 40 50 60 70 80 90 100
828.180
828.182
828.184
828.186
828.188
828.190
828.192
828.187 nm (On-line)
V
on
+0.01
V
on
+0.05
V
on
+0.11
V
on
-0.12
V
on
-0.08
V
on
-0.04
V
on
W
a
v
e
l
e
n
g
t
h
 
(
n
m
)
Time (seconds)
Figure 5.23: A plot of tuning time necessary to lock the laser system to the on-line
wavelength, for dierent initial starting PZT voltage settings, and hence starting
wavelengths, of the laser.
wavelengths. This gure indicates that this computer-controlled feedback loop using
the Burleigh wavemeter to monitor the operating wavelength and provide a voltage
signal to the PZT in the ECDL can robustly reach the desired wavelength.
The ability to tune on- and o-line for dierential absorption measurements was
studied next. As described in the previous paragraph, the computer program auto-
matically tunes the laser to the desired on-line absorption wavelength of 828.187 nm
(12074.5677 cm−1) and uses the computer-controlled feedback mechanism to maintain
this wavelength for an amount of time pre-set by the user. The program then tunes
the laser to the o-line wavelength, 828.287 nm (12073.1099 cm−1, vacuum) and the
computer feedback holds this wavelength for another pre-set amount of time. At the
end of the o-line measurement period, the PZT voltage is taken to zero before being
switched back to the on-line voltage setting to avoid hysteresis in the PZT.
A plot of the operating wavelength as a function of time is shown in gure 5.24,
in which the ECDL is quickly tuned about 44 GHz automatically between the on-
119
0.00 1.00 2.00 3.00 4.00 5.00
828.14
828.16
828.18
828.20
828.22
828.24
828.26
828.28
828.30
828.187 nm (On-line)
828.287 nm (Off-line)
W
a
v
e
l
e
n
g
t
h
 
(
n
m
)
Time (Hours)
Figure 5.24: A plot showing stable on- and o-line tuning of the laser system at
approximately one-hour intervals over a span of ve hours.
and o-line wavelengths over a period of 5 hours without mode hopping. Figure 5.25
shows a close up view of the rst on-line tuning segment, displaying the ability of
the tuning system to tune and hold the laser wavelength to within ± 88 MHz of the
correct on-line wavelength. The frequency stability of ± 88 MHz (±0.00020nm) is
well within the frequency stability requirement of < 200 MHz (< 0.00046 pm) for
accurate water vapor DIAL measurements (Bösenberg , 1998). The ECDL was held
at each wavelength for one hour so the long-term stability of the operating wavelength
could be determined. This tuning system has been tested for on- and o-line data
accumulation times ranging from tens of seconds to over an hour with similar results.
Without the extending continuous tuning range due to the electronic feedback loop
described above, the ECDL would likely mode hop making the tuning between the
on- and o-line dicult and unrepeatable.
Absorption Cell Measurements A multi-pass gas absorption cell was set up to
contain a known amount of water vapor so that a dierential absorption measurement
120
1.16 1.18 1.20 1.22 1.24 1.26
828.180
828.182
828.184
828.186
828.188
828.190
828.187 nm (On-line)
W
a
v
e
l
e
n
g
t
h
 
(
n
m
)
Time (Hours)
Figure 5.25: An expanded view of the second segment of Fig. 5.24 displaying the
computer-controlled feedback loop of the laser system, ne-tuning and holding the
laser output to the on-line wavelength, 828.187 nm.
could be made. It was evacuated using a vacuum pump, and then allowed to come
into a steady state with a large volume of air above a reservoir containing liquid water.
Salt solutions could be used instead of pure water in the reservoir to produce known
relative humidities, but were deemed to be unnecessary for these experiments. Fully
sealing the container to control the relative humidity was desirable but ultimately
too dicult with the available reservoir. Because the reservoir was loosely sealed,
the relative humidity obtained in the absorption cell tracked that of the ambient air,
but was always higher due to the presence of the volume of water, as illustrated in
gure 5.26. The mirrors of the absorption cell were aligned to allow the laser beam
to make 36 passes, for a total path length of 19.8 meters. A digital psychrometer
(Mannix EM8716) with a stated relative humidity accuracy of ±3% at 25◦ Celsius
and temperature accuracy of ± 0.6◦ Celsius was placed inside the absorption cell to
make an in situ measurement of the relative humidity and temperature. Care was
taken to ensure the psychrometer did not interrupt the laser beam path.
121
1 2 3 4 5 6 7 8
1.0
1.5
2.0
2.5
3.0
3.5
1-hour average
2-hour average
1-minute average
Water Vapor added 
after measurement
1-minute average
5-hour average
W
at
er
 V
ap
or
 N
um
be
r D
en
si
ty
 (x
10
17
 c
m
-3
)
Days
 Weather Station
 Absorption Cell
Figure 5.26: A plot of the water vapor number density in the absorption cell compared
to the number density in the atmosphere as measured by a weather station on the
roof of the building where the absorption cell measurements were performed. Notice
that the number density within the absorption cell is always higher than that of the
atmosphere due to the presence of the liquid water reservoir connected to the cell.
The eect of adding humid air to the absorption cell can be easily seen.
The reference and transmission powers were recorded every two seconds using the
computer programs described above, and the normalized transmission power calcu-
lated by taking a ratio of the transmission power to the reference power at every data
collection point. Figure 5.27 shows the normalized transmission power as a function
of time, exactly corresponding to the wavelength tuning shown in gure 5.24. A clear
change in transmission between the on-line and o-line normalized transmitted pow-
ers is visible. Because the on- and o-line laser beams travel an identical path and are
spectrally very close, the assumption is made that all scattering along the path will
be identical for both wavelengths, and therefore, the reduction in on-line normalized
transmitted power compared to o-line power can be attributed solely to water vapor
absorption. To verify that the drop in power is not due to an instrumental eect, tests
were run at several nearby and widely spaced pairs of o-line wavelengths, showing
no change in normalized transmitted power between two o-line wavelengths. Also,
122
0.00 1.00 2.00 3.00 4.00 5.00
4.0
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
4.9
5.0
828.187 nm (On-line)
828.287 nm (Off-line)
N
o
r
m
a
l
i
z
e
d
 
T
r
a
n
s
m
i
s
s
i
o
n
 
P
o
w
e
r
 
(
m
W
*
m
W
-
1
)
Time (Hours)
Figure 5.27: A plot of the normalized power transmitted through the 19.8 meter path
length of the gas absorption cell. Absorption by water vapor molecules within the
cell is responsible for the reduced on-line signal.
since the laser linewidth has been previously measured through beating experiments
to be ∼ 300 kHz (FWHM), much smaller than the water vapor linewidth of ∼ 5GHz
(FWHM), it was tuned across the water vapor line to verify the smooth transition
between on- and o-line powers as predicted by HiTRAN-PC.
Using a dierential absorption calculation, which is basically a simplied one-way,
single-range bin DIAL technique, the number density of water vapor molecules in
the absorption cell can be determined by knowing the on- and o-line normalized
powers, the absorption cell path length, and the dierential absorption cross section
of the water vapor absorption line. The number density, N [cm−3] is calculated using
a one-way, single-range bin DIAL equation,
N = 1σdiff ·∆R
ln
[〈Ptx, off/Pref, off 〉
〈Ptx, on/Pref, on〉
]
, (5.2)
where σdiff = (σon − σoff ) [cm2] is the dierence between the on- and o-line ab-
sorption cross sections due to water vapor, ∆R [m] is the total path length, and
123
〈Ptx, off (on)/Pref, off (on)〉 [mW/mW ] is the o-line (on-line) normalized transmission
power averaged over the measurement time (Schotland , 1974). Values from the Hi-
TRAN 2000 database (Rothman et al., 2003) were used in custom calculations based
on those of Browell et al., 1991 to determine the absorption prole of the water va-
por line and therefore σon = 5.685 × 10−23 cm2 and σoff = 3.716× 10−25 cm2. Note
that σoff is non-zero because the wings of nearby water vapor lines still contribute to
absorption at the o-line wavelength. Contributions to the absorption cross sections
from the 22 nearest water vapor lines on each side of the absorption line of interest at
12074.5677 cm−1 were taken into account. The temperature reading for the absorp-
tion cell number density calculation was taken from the psychrometer inside the cell,
and a pressure measurement with an accuracy of ±0.5 mbar was taken by a weather
station located on the roof of the building in which the measurements were made.
The slight change in pressure between the roof and the laboratory did not make an
appreciable dierence in the calculated values.
The psychrometer readings averaged over the measurement period were 48.8±3.0%
for relative humidity and 21.65±0.6 degrees Celsius for temperature. These readings
can be used to calculate a partial pressure for water vapor, which can then be con-
verted to a number density (see, for example, (Kyle, 1991)). The atmospheric parame-
ters and the calculated partial pressure for water vapor were used with HiTRAN-PC to
produce a prediction of the water vapor absorption inside the gas absorption cell. This
prediction is compared with the measurements made by the laser system, which have
been normalized to the o-line transmission measurement, in gure 5.28. The number
density of water vapor molecules in the gas absorption cell determined by the laser sys-
tem and simplied DIAL technique was N = 2.8564±0.4237×1017 cm−3. The number
density measured by the in situ psychrometer was N = 3.1129± 0.3015× 1017 cm−3.
The error in the psychrometer measurement of N is due to a slight temperature drift
124
828.16 828.18 828.20 828.22 828.24 828.26 828.28 828.30
0.95
0.96
0.97
0.98
0.99
1.00
1.01
1.02
 Absorption Cell Average Measurements
 HiTRAN Predicted Absorption
T
r
a
n
s
m
i
s
s
i
o
n
Wavelength (nm)
Figure 5.28: A plot of the relative transmission through the gas absorption cell as a
function of wavelength. The closed circles represent measurements made by the laser
system. The solid line is a theoretical prediction of the absorption by water vapor in
the absorption cell using HiTRAN-PC with the in situ measurements of temperatures
and humidity.
across the measurement period, and the uncertainties in the relative humidity and
temperature readings, the latter of which is nonlinear in the conversion from partial
pressure to number density. The error in the dierential absorption measurement
of N is primarily due to the deviation in the power measurements, which are then
enlarged by the natural logarithm in equation 5.2. Both the calculated values for N
and the predicted and measured transmissions agree within their error values.
Reference Power
Testing the reference power measurement was done in the same manner as for the
horizontal tuning experiments. Initially, the DIAL was designed to run similarly to
the horizontal tuning experiment, in that the laser would be pulsed for data collection,
but would then be switched to cw mode for taking reference power measurements.
As will be seen in the data runs section, this method produced a striping eect in
125
the data. A possible solution to this problem was to move the reference power meter
to the cw side of the AOM, where it could take a cw measurement of the laser at
any time while the laser could remain pulsing. The result of this method is shown in
the left graph panel in gure 5.29. It was discovered that there is a relaxing eect
present in the AOM in which the power transmitted through the AOM decays over
many minutes to a steady value, which probably led to the striping eect in the
previous conguration. This eect nullied the technique of measuring the reference
power before the AOM, because the transmit power was decreasing through the AOM
while the reference power was not. The next, and ultimately nal, solution was to
leave the reference power meter on the transmit side of the AOM and measure pulse
power instead of cw power. Tests were done to show that the power measured by the
reference power meter, even while it was pulsing at 20 kHz, was still directly related
to the transmit power, as shown by the right graph panel in gure 5.29. Pulsed
measurements were noisier when compared to making cw measurements, but were
acceptable for the DIAL experiment.
Data Runs
With the operational tests nished, DIAL data runs commenced on November
14th, 2006. As expected, the coaxial design approach created a large afterpulse in
the APD detector, as shown in gure 5.30. The graph shows about 30 minutes of
data. The y-axis is the time, starting from the top. Data were taken on 1-second
intervals, so every 1-second horizontal row is the number of photons measured by the
APD as a function of range. The x-axis is the range in meters. The intensity shows
the number of counts in a 50-ns bin. Due to the averaging of the data, the counts
represented by the color bar multiplied by a factor of 1000 give the counts per second.
126
51400 51600 51800 52000 52200 52400 52600
0
10
20
30
40
50
On-/Off-line Tuning
11/30/06 DIAL Experiment
TX/Ref Power Test (CW)
TX
/R
ef
 P
ow
er
 R
at
io
 (m
W
*m
W
-1
)
Time (Seconds)
37000 37500 38000 38500 39000 39500
0
5
10
15
20
25
30
35
40
45
50
On-/Off-line Tuning
12/04/06 DIAL Experiment
TX/Ref Power Test
TX
/R
ef
 P
ow
er
 R
at
io
 (m
W
*m
W
-1
)
Time (Seconds)
 Pulsed
 CW
Figure 5.29: Reference power test measurements showing the relaxation eect of the
AOM on the left and the dierence between pulsed and cw measurements of reference
power on the right.
Therefore, the deep red stripe at about 18000 on the color bar, which represents the
initial blinding of the APD as the laser res, is showing a count rate of over 18 million
counts per second, which basically means that the APD is saturated. The subsequent
vertical stripes located around 750 m, 1500 m, 2250 m, and 2800 m are the ringing
of the detector after being blasted by the laser. The horizontal striping is due to
the relaxation eect in the AOM after switching the laser to cw mode to measure
reference power, as described above. Clearly, there were many data relics that needed
to be removed before a small return signal could be extracted from these data.
Because the APD was counting so many photons in every bin, the counts now
had to be corrected for missed counts. A correction factor was supplied by the
manufacturer and the counts in each bin were multiplied by the factor corresponding
to their count rate. The correction factor is shown in gure 5.31. Measuring the
reference power by leaving the transmitter in pulse mode, as described in the reference
power test section above, eliminated the striping in the time domain. Many attempts
127
Range (m)
Ti
m
e 
(se
co
nd
s)
11/14/06 Raw Counts
1000 ns pulse width
0  375 750 1125 1500 1875 2250 2625 3000 3375
0
200
400
600
800
1000
1200
1400
1600
1800
2000
4000
6000
8000
10000
12000
14000
16000
18000
Figure 5.30: An example of initial DIAL data runs, in which the saturation and
afterpulse of the APD is apparent, and the striping eect caused by the AOM can be
seen.
were made to then remove the afterpulsing eects and background signal from the
data. A typical result of such attempts is shown in gure 5.32, where the x-axis is
now time of the experiment and the y-axis is range. Negative ranges correspond to
the time before the laser res, which is used to measure the background light level
without the laser pulse present. The initial pulse where the APD is saturated cannot
be corrected, since counting information is lost in that situation, thus data in those
bins were typically set to zero and ignored. To remove the afterpulsing stripes, a
time average was taken of the DIAL signal with the roofport hatch open at ∼ 45◦,
allowing the afterpulse to be measured while ensuring that no atmospheric returns
were being collected. This method has been used by other lidars that utilize a coaxial
approach (Campbell et al., 2002). It was hoped that the afterpulsing would be uniform
enough in time that a time-average could be subtracted from subsequent atmospheric
128
DIAL data to completely remove the afterpulse stripes. This turned out to not be
the case. The afterpulses showed considerable variability and could not be removed
consistently. Explanations for this include that the other lidars that use the coaxial
technique tend to have higher spatial resolution and higher power. Therefore, the
afterpulse is less of a percentage of their data in the spatial domain, and can be
more easily ignored if subtraction is not an option. Their designs also use techniques
such as blocking the central portion of the secondary telescope mirror, which was not
done in the MSU DIAL system to avoid damaging the telescope, and slightly angling
optical elements to avoid large back-reections. After spatially averaging the DIAL
data to the typical value of 150 meters, the afterpulse stripes were present in over a
quarter of the data bins, meaning they could not be ignored. Also, because of the
low power of this DIAL system, the background and afterpulse striping had to be
removed to within an accuracy of a few counts per bin, or the level of the expected
return signal. The subtraction routine had to be able to remove the highest counts,
perhaps around 15,000 counts per bin, to a level of around 15 counts per bin, or 1 in
1000. That level of accuracy was not reliably achievable. For these reasons, it became
apparent that the coaxial approach would not work for the DIAL, and the system
was redesigned to be bistatic.
The DIAL was redesigned to be bistatic by removing the PBS and quarter-wave
plate in the receiver line and no longer bringing the transmit beam up to that level.
Instead, the transmit beam was sent around the outside of the telescope, and a 45-
degree mirror mount was epoxied to the top of the receiver telescope's secondary
mirror housing. Sending the transmit beam o of the back of the secondary telescope
mirror allowed the system to be coaligned at ground level, and took some of the
alignment uncertainty out of the bistatic approach. Still, it was well known that
redesigning for a bistatic lidar was simply trading the coaxial afterpulsing problem
129
0 2000 4000 6000 8000 10000 12000 14000 16000
1
2
3
4
5
6
7
8
9
Co
rr
ec
tio
n 
Fa
ct
or
Count Rate (103/s)
Figure 5.31: The high-count correction factor for the APD detector. Data is courtesy
of Perkin-Elmer.
Time (s)
R
an
ge
 (m
)
12/11/06 Afterpulse−corrected On−line Data
7.3 7.35 7.4 7.45 7.5 7.55
x 104
−250
250
750
1250
1750
2250
2750
Time (s)
R
an
ge
 (m
)
12/11/06 Afterpulse−corrected Off−line Data
7.35 7.4 7.45 7.5 7.55 7.6 7.65
x 104
−250
250
750
1250
1750
2250
2750
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
x 104
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
x 104
Figure 5.32: Example DIAL data in which the background and afterpulsing eects
were attempted to be removed, without success.
130
for probably an equally dicult alignment problem.
Alignment did turn out to be very dicult. Because the range bins were so
large, it was very easy to have the transmit beam misaligned to the point where
it was completely out of the receiver FOV within one range bin. The APD still
detected a ash from the initial pulse of the laser, but at a much lower level than the
coaxial case, and if the transmit beam was not aligned above 150 meters, a return
signal would never be seen. Also, the system does not put out enough power to
measure an appreciable signal without at least minutes of time averaging. For these
reasons, peaking up the alignment of the transmit beam could not easily be done by
hand simply by moving the mirrors and watching for increasing return signal in the
software. Another method was needed.
Simple calculations showed that the farther out the transmit beam could be reli-
ably aligned with the receiver FOV, the higher in altitude it would remain in alignment
before falling out of overlap. Aligning the beam at 1 meter was not sucient but 30
meters might be. A system was designed where a mirror was placed at roughly 45
degrees over the roofport hatch, and the transmit beam was sent across the roof of
the building, hitting the edge about 30 meters away. A visible helium-neon beam was
sent backwards through the receiver ber optic cable to simulate the FOV, and the
transmit beam was able to be aligned within the FOV across a 30 meter path length.
This method was successful, and the rst bistatic DIAL data were taken on De-
cember 21st, 2006, as shown in gure 5.33. The graph again shows time on the y-axis
and altitude on the x-axis. The afterpulsing striping disappears in the bistatic ap-
proach, and the initial ash of the laser pulse is about 500 times lower, located at
around 200 m due to the electronics lag between ring the laser the pulse exiting the
system. The initial ash of the laser is so weak that a correction factor for the APD
counts was no longer need, as the APD was not anywhere near saturation levels. It
131
Range (m)
Ti
m
e 
(se
co
nd
s)
12/21/06 Raw Counts
500 ns pulse width
−750 −375 0 375 750 1125 1500 1875 2250 2625 3000
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
11000
5
10
15
20
25
30
35
40
45
50
55
Figure 5.33: Evidence of alignment of the transmit beam with the receiver FOV. The
initial laser pulse is 75 meters wide while the ash seen by the detector is over 225
meters wide, indicating return from atmospheric molecules and particles in the near
eld. Clouds can be seen at about 2.4 km altitude and decrease to about 1.8 km
above the ground at the conclusion of the experiment.
is immediately apparent that the system is at least in partial alignment rst because
the pulse would only produce return o of the transmit optics for a range of 75 m
as it exited the system, but the laser ash produces visible returns for a range that
is two or three times longer, and second because cloud returns are visible starting at
about 2.4 km above the ground and decreasing in altitude as the night went on and a
storm system entered the area. At the end of the experiment, the clouds were located
at around 1.875 km above the ground and the data run had to be terminated because
of rain.
Examples of data containing clouds averaged over time is shown in gure 5.34. In
the left graph, the eect of background subtracting can be seen as the bins used for
background determination, located below zero altitude, are very near to zero. The
132
−500 0 500 1000 1500 2000 2500 3000
−500
0
500
1000
1500
2000
2500
3000
3500
4000
Range (m)
B
ac
kg
ro
un
d−
su
bt
ra
ct
ed
 C
ou
nt
s
12/27/06 DIAL Counts
500 ns pulse width, 2 min. tunes
15.12 min. on−line (green), 13.20 min. off−line (red)
−500 0 500 1000 1500 2000 2500 3000
2000
2500
3000
3500
4000
4500
5000
Range (m)
R
aw
 C
ou
nt
s
Figure 5.34: Time-averaged raw counts (right) and background-subtracted counts
(left) showing the initial laser pulse, atmospheric returns, and cloud returns.
ash of the laser pulse is seen at zero, and lasts much longer than the 75 meters that
would be expected if the beam were not scattering o of atmosphere in the near eld
of the receiver. Clouds are visible from 500 meters in altitude and up. As seen in
gure 5.33 as well, the o-line signal is stronger at the clouds because it has not been
absorbed by water vapor during its transit from the ground to the cloud and back.
The right graph shows the raw counts before background subtraction. The averaged
counts such as these are used in the DIAL equation to calculate water vapor density.
With the system in alignment, analysis of the data for producing water vapor
proles could be commenced. To verify that the water vapor DIAL was measuring
the accurate amount of water vapor, radiosondes were launched with each DIAL data
run, to give an in situ measurement of temperature and relative humidity, which were
then converted to water vapor number density. To check the validity of the MSU
radiosondes, and to satisfy curiosity, a comparison was performed between an MSU
133
0 1 2 3 4 5 6 7 8 9 10
x 1016
0
200
400
600
800
1000
1200
1400
1600
1800
2000
Number Density (cm−3)
R
an
ge
 (m
)
01/09/07 Radiosonde Comparison
Great Falls radiosonde (green) vs. MSU radiosonde (red)
Figure 5.35: A comparison between a radiosonde launched at MSU and one launched
in Great Falls around the same time.
radiosonde and a radiosonde launched in Great Falls around the same time. While
Bozeman and Great Falls are not located very close to each other, the atmospheric
parameters should roughly line up, especially during quiet weather periods. This
turned out to be the case, as seen in gure 5.35.
Many nights of data were taken to attempt to produce water vapor proles. The
DIAL was very dicult to keep aligned. The transmit beam would move outside of
the receiver FOV easily overnight due to vibrations or thermal changes, meaning the
system had to be realigned almost nightly, which it was not built for. The eects
of misalignment were very apparent in return signals, as shown in gure 5.36. The
system was aligned on January 9th, 2007, and out of alignment on January 18th,
2007. The left graph shows time-averaged on-line counts and the right graph shows
time-averaged o-line counts. Notice the steep fallo in the unaligned (green) beam
after the initial laser pulse, whereas the aligned (red) beam is considerably longer
134
−1000 −500 0 500 1000 1500 2000 2500 3000
0
1
2
3
4
5
6
7
8
9
10
11
12 x 10
5
Range (m)
Co
un
ts
On−line total counts
01/18/07 (green, 28.45 mins.) and
 01/09/07 (red, 22.82 mins.)
−1000 −500 0 500 1000 1500 2000 2500 3000
0
1
2
3
4
5
6
7 x 10
5
Range (m)
Co
un
ts
Off−line total counts
01/18/07 (green, 41.28 mins.) and 
01/09/07 (red, 19.17 mins.)
Figure 5.36: A comparison of an aligned (01/09/07) transmit beam and an unaligned
(01/18/07) transmit beam. The left graph shows time-averaged on-line counts and
the right graph shows time-averaged o-line counts. Notice the steep fallo in the
unaligned (green) beam after the initial laser pulse, whereas the aligned (red) beam
is considerably longer due to atmospheric returns. The spike above 1500 meters is
probably due to multiple scattering of the laser beam within the cloud.
due to atmospheric returns. The spike above 1500 meters is probably due to multiple
scattering of the laser beam within the cloud.
Several water vapor proles were taken while the DIAL was in alignment that
agreed reasonably well with the results from a colocated radiosonde. One example is
shown in gure 5.37. The data were averaged over about 1 hour with pulse widths
of 1 µs. Below 500 meters, the data does not agree within the error bars, probably
due to a background light leakage into the detector, which has the eect in the DIAL
equation of reducing the amount of water vapor. The error bars were produced by a
simple dierential analysis of the DIAL equation.
This gure shows the ability of the low-power DIAL system to achieve meaningful
135
0 1 2 3 4 5
0
500
1000
1500
2000
 DIAL Measurements
 MSU Radiosonde
01/09/07 DIAL Measurements 7:04pm - 8:04pm
1000 ns pulse width (averaged over 150 m)
60 sec. on-line, 60 sec. off-line tuning
22.82 min. total online, 19.17 min. total offline
Water Vapor Number Density (x1017 cm-3)
Al
tit
ud
e 
ab
ov
e 
Co
bl
ei
gh
 H
al
l (
m
)
Figure 5.37: A water vapor prole from the DIAL compared to a MSU radiosonde.
136
water vapor proles up to about 2 km. While the DIAL returns are not exactly
accurate yet, the ability to produce such a close result is a major step towards proving
the viability of this system for eld deployments, and will only improve with future
versions of the instrument.
137
CHAPTER 6
CONCLUSION
A compact, low-power dierential absorption lidar (DIAL) using a widely tunable
diode laser was built, tested, and used to produce proles of atmospheric water vapor
up to an altitude of 2 km above ground level. The transmitter for the DIAL used
a external cavity diode laser (ECDL) that was built through the expertise of the
laser source development group at Montana State University (MSU) coupled with a
commercial tapered amplier (TA). The ECDL has the capability of tuning across a
17 nm spectrum near 830 nm, giving it access to numerous water vapor absorption
lines.
The tunability of the transmitter was rst shown through horizontally pointing
lidar experiments in which the laser was tuned across water vapor absorption lines at
wavelengths of 829.022 nm, 831.615 nm, and 831.850 nm, in both continuous wave
(cw) and pulsed modes. The scans were compared and were in excellent agreement
with absorption values given by HiTRAN-PC atmospheric modeling software.
This transmitter was then coupled for the rst time in any known DIAL instrument
with a second, cascaded TA, and used in a vertically-pointing DIAL. The DIAL used
an acousto-optic modulator (AOM) to pulse the cw beam from the transmitter. The
receiver made use of a commercial Schmidt-Cassegrain telescope with a diameter of
28 cm, an extremely narrow band (NB) lter with a band pass of ∼ 250 pm, and
a ber-coupled, photon counting avalanche photodiode (APD) detector, and had a
narrow, 150-µrad eld of view (FOV). Both coaxial and bistatic congurations were
attempted, with the bistatic approach producing successful results. The DIAL system
was almost completely autonomously controlled using LabVIEW software and a novel
138
tuning system that quickly tuned the ECDL between the on-line vacuum wavelength
of 828.187 nm and o-line vacuum wavelength of 828.287 nm. The tuning mechanism
was extensively tested, showing that the laser wavelength could be held stable to
within ±88 MHz, well within the requirement of ±160 MHz for accurate water vapor
proles. The spectral purity was measured to be >0.995, within allowable tolerances.
Pulses with widths of 1.0 µs and energies of ∼0.25 µJ , at a repetition rate of
20 kHz, were used to probe the lower troposphere up to 2 km, resulting in water
vapor proles that were compared to co-located radiosonde measurements. The mea-
surements agreed with the radiosonde measurements to within an order of magnitude,
with the discrepancies being explainable and potentially xable in future DIAL modi-
cations. A number of potential improvements to the system are listed below. Making
these changes to the DIAL would create a second-generation instrument capable of
accurate nighttime water vapor proles up to at least 2 km, and potentially capable
of daylight proles.
• It became clear during the DIAL tuning tests and experiments that the PZT
within the ECDL housing needs to be replaced. It was noted by monitoring the
ECDL wavelength that the PZT no longer seems to be relaxing to its initial
position after its external voltage is removed. This eect is denitely growing
worse over time and will eventually reach the point where the PZT does not
relax at all and the ECDL will not be able to be quickly tuned.
• The commercial tapered ampliers should be replaced with better versions,
probably built at MSU. With custom TA's, like those already being designed
and built at MSU, the power output could be increased, heat sinks could be
incorporated directly into the TA housing to improve temperature control, and
designs could be implemented that would ease the alignment procedures.
139
• Even if new TA's are not built, more should be done to eectively cool the second
tapered amplier. A heat sink should be machined to pull heat eectively
away from the gain region of the amplier, while still allowing the amplier
to be placed on a translation stage for easier seed alignment. The current
fan placement and mounting should probably be improved to ensure against
introducing vibration into the system.
• A technique that has been discussed for improving the SNR of the DIAL is to
use preferential illumination in which multiple transmit beams are brought into
overlap at varying altitudes. In this manner, more laser light could be placed
at high altitudes, where it is needed most, smoothing out the dynamic range
problem that most lidars face.
• An important step towards a better DIAL system is to nd a way to pulse the
tapered ampliers directly and remove the AOM. While this may be dicult
unless custom TA's are used, this change immediately removes the problems
directly associated with the use of an AOM: a > 33% loss in transmit beam
power and cw leakage light passing through AOM and raising the background
signal level. The drawback to replacing the AOM is that there will probably be
less control over the pulse shape, and that may need to be accounted for while
processing the data.
• To further reduce the possibility of background light reaching the detector,
the DIAL should be redesigned so that the receiver and transmit beam are on
opposite sides of the vertical optical breadboard. The receiver optics need to
be more isolated from all sources of laser light.
• A complete enclosed box for the receiver box needs to be machined and built.
140
Even a few background photons passing through the receiver box and reaching
the detector is enough to seriously skew the resulting data. If the box was
built of metal with fewer seams, it would allow for better isolation. Also, a
sturdier receiver box with a top that could be removed would allow easier and
quicker access to the receiver optics, which is an improvement over the current
need to remove and replace electrical tape over all seams after every receiver
adjustment.
• A new technique should be found to reliably align the receiver and transmitter,
and the transmit mirror currently epoxied to the top of the secondary telescope
mirror housing should probably be permanently attached with a more stable
mount.
• One avenue for drastically reducing the amount of background light that reaches
the detector is to use a Fabry-Perot etalon as a extremely narrowband lter on
the receiver. If the etalon had a band pass that was not much larger than the
laser linewidth itself and could be tuned between the on- and o-line wave-
lengths, this would reduce background light and increase the SNR by a large
amount.
• The receiver telescope should be custom coated with anti-reection coatings for
operation near 830 nm. This would increase the eciency of the receiver and
help send more photons to the detector.
• The operational and analysis software codes can always be improved to be made
more module for future modications, more user-friendly, and easier to use so
that extensive training is not required to take water vapor proles.
This DIAL instrument has demonstrated that low-power DIAL instruments using
141
widely tunable diode laser transmitters, which can be designed at multiple wave-
lengths, can achieve useful water vapor proles. The system is robust, can be re-
paired quickly and relatively easily, compact, and can be made eye-safe, all necessary
requirements for an autonomous eld or satellite instrument. It is hoped that this
DIAL will lead to the development of next-generation, widely tunable DIAL instru-
ments that in the future may be acceptable candidates for use in multi-point lidar
networks or satellite arrays to study water vapor ux proles.
142
REFERENCES
Ambrico, P. F., A. Amodeo, P. di Girolamo, and N. Spinelli, Sensitivity Analy-
sis of Dierential Absorption Lidar Measurements in the Mid-Infrared Region,
Appl. Opt., 39 , 68476865, 2000.
Aumiler, D., T. Ban, and G. Pichler, High-resolution measurements of the pres-
sure broadening and shift of the rubidium 52S1/262P3/2 line by argon and helium,
Phys. Rev. A, 70 , 032,723+, 2004.
Battan, L. J., Fundamentals of Meteorology , Englewood Clis: Prentice-Hall, 1984,
2nd ed., 1984.
Becker, P. C., N. A. Olsson, and J. R. Simpson, Erbium-doped ber ampliers :
fundamentals and technology , Erbium-doped ber ampliers : fundamentals and
technology, by P. C. Becker, N. A. Olsson, and J. R. Simpson. San Diego, CD:
Academic Press, 1999, xv, 460 p., 1 diskette (3 1/2 in.) Optics and photonics. ISBN
0120845903., 1999.
Born, M., and E. Wolf (Eds.), Principles of optics : electromagnetic theory of propa-
gation, interference and diraction of light , 1999.
Bösenberg, J., Ground-Based Dierential Absorption Lidar for Water-Vapor and
Temperature Proling: Methodology, Appl. Opt., 37 , 38453860, 1998.
Browell, E. V., S. Ismail, and B. E. Grossmann, Temperature sensitivity of dierential
absorption lidar measurements of water vapor in the 720-nm region, Appl. Opt.,
30 , 15171524, 1991.
Browell, E. V., S. Ismail, and W. B. Grant, Dierential absorption lidar (DIAL)
measurements from air and space, Applied Physics B: Lasers and Optics, 67 , 399
410, 1998.
Cahen, C., G. Megie, and P. Flamant, Lidar Monitoring of the Water Vapor Cycle in
the Troposphere, Journal of Applied Meteorology , 21 , 15061515, 1982.
Campbell, J. R., D. L. Hlavka, E. J. Dutton, C. J. Flynn, D. D. Turner, J. D.
Spinhirne, V. S. Scott III, and I. H. Hwang, Full-time, eye-safe cloud and aerosol
lidar observation at atmospheric radiation measurement program sites: Instruments
and data processing, J. Atmos. Ocean. Technol., 19 , 431442, 2002.
Churnside, J. H., and J. J. Wilson, Airborne lidar imaging of salmon, AO , 43 , 1416
1424, 2004.
Crutzen, P. J., and F. Arnold, Nitric acid cloud formation in the cold antarctic
stratosphere - A major cause for the springtime 'ozone hole', Nature, 324 , 651
655, 1986.
143
de Labachelerie, M., and G. Passedat, Mode-hop suppression of Littrow grating-tuned
lasers, Appl. Opt., 32 , 269274, 1993.
Divakarla, M. G., C. D. Barnet, M. D. Goldberg, L. M. McMillin, E. Maddy, W. Wolf,
L. Zhou, and X. Liu, Validation of Atmospheric Infrared Sounder temperature
and water vapor retrievals with matched radiosonde measurements and forecasts,
Journal of Geophysical Research (Atmospheres), 111 , 9+, 2006.
Ehret, G., C. Kiemle, W. Renger, and G. Simmet, Airborne remote sensing of tro-
pospheric water vapor with a near-infrared dierential absorption lidar system,
Appl. Opt., 32 , 45344551, 1993.
Ehret, G., A. Fix, V. Weiss, G. Poberaj, and T. Baumert, Diode-laser-seeded optical
parametric oscillator for airborne water vapor DIAL application in the upper tropo-
sphere and lower stratosphere, Applied Physics B: Lasers and Optics, 67 , 427431,
1998.
Fahey, D. W., K. K. Kelly, S. R. Kawa, A. F. Tuck, and M. Loewenstein, Observations
of denitrication and dehydration in the winter polar stratospheres, Nature, 344 ,
321324, 1990.
Feltz, W. F., W. L. Smith, H. B. Howell, R. O. Knuteson, H. Woolf, and H. E.
Revercomb, Near-Continuous Proling of Temperature, Moisture, and Atmospheric
Stability Using the Atmospheric Emitted Radiance Interferometer (AERI)., Journal
of Applied Meteorology , 42 , 584597, 2003.
Gentry, B. M., H. Chen, and S. X. Li, Wind measurements with 355-nm molecular
Doppler lidar, Optics Letters, 25 , 12311233, 2000.
Goldsmith, J. E. M., F. H. Blair, S. E. Bisson, and D. D. Turner, Turn-Key Raman
Lidar for Proling Atmospheric Water Vapor, Clouds, and Aerosols, Appl. Opt.,
37 , 49794990, 1998.
Grant, W. B., Dierential absorption and Raman lidar for water vapor prole mea-
surements - A review, Optical Engineering , 30 , 4048, 1991.
Grund, C. J., and E. W. Eloranta, University of Wisconsin High Spectral Resolution
Lidar, Optical Engineering , 30 , 612, 1991.
Hair, J. W., L. M. Caldwell, D. A. Krueger, and C.-Y. She, High-Spectral-Resolution
Lidar with Iodine-Vapor Filters: Measurement of Atmospheric-State and Aerosol
Proles, Appl. Opt., 40 , 52805294, 2001.
Han, Y., and E. R. Westwater, Remote sensing of tropospheric water vapor and cloud
liquid water by integrated ground-based sensors, J. Atmos. Ocean. Technol., 12 ,
10501059, 1995.
144
Harries, J. E., Atmospheric radiation and atmospheric humidity, Quarterly Journal
of the Royal Meteorological Society , 123 , 21732186, 1997.
Intrieri, J. M., M. D. Shupe, T. Uttal, and B. J. McCarty, An annual cycle of Arctic
cloud characteristics observed by radar and lidar at SHEBA, Journal of Geophysical
Research (Oceans), 107 , 51, 2002.
Ismail, S., E. V. Browell, R. A. Ferrare, S. A. Kooi, M. B. Clayton, V. G. Brackett,
and P. B. Russell, LASE measurements of aerosol and water vapor proles during
TARFOX, J. Geophys. Res., 105 , 99039916, 2000.
Kington, J. A., THE ATMOSPHERIC SCIENCES ENTERING THE TWENTY-
FIRST CENTURY, Board on Atmospheric Sciences and Climate Commission on
Geosciences, Environment and Resources, National Research Council, USA, Na-
tional Academy Press, Washington, DC, 1998. No. of pages: xvi+364. Price:
#34.95. ISBN 0-309-06415-5 (hardback)., International Journal of Climatology ,
20 , 935936, 2000.
Kovalev, V. A., and W. E. Eichinger, Elastic Lidar: Theory, Practice, and Analysis
Methods, New Jersey, Wiley-Interscience, 615 p., 2004.
Kyle, T. G., Atmospheric Transmission, Emission, and Scattering , Pergamon Press,
1991.
Libbrecht, K. G., and J. L. Hall, A low-noise high-speed diode laser current controller,
Review of Scientic Instruments, 64 , 21332135, 1993.
Liou, K. N., An Introduction to Atmospheric Radiation, Second Edition, San Diego,
Elsevier Science, 583 p., 2002.
Little, L. M., and G. C. Papen, Fiber-Based Lidar for Atmospheric Water-Vapor
Measurements, Appl. Opt., 40 , 34173427, 2001.
Littman, M. G., and H. J. Metcalf, Spectrally narrow pulsed dye laser without beam
expander, Appl. Opt., 17 , 22242227, 1978.
Machol, J. L., T. Ayers, K. T. Schwenz, K. W. Koenig, R. M. Hardesty, C. J. Sen,
M. A. Krainak, J. B. Abshire, H. E. Bravo, and S. P. Sandberg, Preliminary Mea-
surements with an Automated Compact Dierential Absorption Lidar for the Pro-
ling of Water Vapor, Appl. Opt., 43 , 31103121, 2004.
McGill, M. J., W. R. Skinner, and T. D. Irgang, Validation of wind proles measured
with incoherent Doppler lidar, AO , 36 , 19281932, 1997.
McNicholl, P., and H. J. Metcalf, Synchronous cavity mode and feedback wavelength
scanning in dye laser oscillators with gratings, Appl. Opt., 24 , 27572761, 1985.
145
Measures, R. M., Laser remote sensing: Fundamentals and applications, New York,
Wiley-Interscience, 521 p., 1984.
Meng, L. S., K. S. Repasky, P. A. Roos, and J. L. Carlsten, Widely tunable
continuous-wave Raman laser in diatomic hydrogen pumped by an external-cavity
diode laser, Optics Letters, 25 , 472474, 2000.
New Focus Product Guide, 15, 20-39, 2584 Junction Avenue, San Jose, CA 95134-
1902, USA (ph. 1-866-683-6287), 2007.
Nguyen, Q.-V., R. W. Dibble, and T. Day, High-resolution oxygen absorption spec-
trum obtained with an external-cavity continuously tunable diode laser, Optics
Letters, 19 , 21342136, 1994.
Obland, M. D., L. S. Meng, K. S. Repasky, J. A. Shaw, and J. L. Carlsten, Progress
toward a water-vapor dierential absorption lidar (DIAL) using a widely tunable
amplied diode laser source, in Lidar Remote Sensing for Environmental Monitor-
ing VI. Proceedings of the SPIE, Volume 5887 , edited by U. N. Singh, pp. 205215,
2005.
Obland, M. D., K. S. Repasky, J. A. Shaw, and J. L. Carlsten, Preliminary testing of
a water-vapor dierential absorption lidar (DIAL) using a widely tunable amplied
diode laser source, in Proceedings of the 23rd International Laser Radar Conference,
edited by C. Nagasawa and N. Sugimoto, pp. 249252, 2006a.
Obland, M. D., K. S. Repasky, J. A. Shaw, and J. L. Carlsten, Preliminary testing of
a water-vapor dierential absorption lidar (DIAL) using a widely tunable amplied
diode laser source, in Proceedings of the 2006 IEEE International Geoscience and
Remote Sensing Symposium, 2006b.
Obland, M. D., A. R. Nehrir, K. S. Repasky, J. L. Carlsten, and J. A. Shaw, Ap-
plication of extended tuning range for external cavity diode lasers to water vapor
dierential absorption measurements, Optical Engineering , accepted for publica-
tion, 2007.
Oh, C., C. R. Prasad, V. A. Fromzel, I. H. Hwang, J. A. Reagan, H. Fang, and
M. Rubio, Water vapor micropulse dierential absorption lidar, in Proc. SPIE
Vol. 3707, p. 177-186, Laser Radar Technology and Applications IV, Gary W.
Kamerman; Christian Werner; Eds., edited by G. W. Kamerman and C. Werner,
pp. 177186, 1999.
Penchev, S., S. Naboko, V. Naboko, V. Pencheva, T. Donchev, L. Pavlov, and P. Sime-
onov, DIAL monitoring of atmospheric climate-determining gases employing high-
power pulsed laser diodes, in 12th International School on Quantum Electronics:
Laser Physics and Applications Edited by P. A. Atanasov, A. A. Serafetinides, and
146
I. N. Kolev. Proceedings of the SPIE, Volume 5226 , edited by U. N. Singh, pp.
290294, 2003.
Poole, L. R., and M. P. McCormick, Polar stratospheric clouds and the Antarctic
ozone hole, J. Geophys. Res., 93 , 84238430, 1988.
Prasad, C. R., V. A. Fromzel, J. S. Smucz, I. H. Hwang, and W. E. Hasselbrack,
A diode-pumped Cf:LiSAF laser for UAV-based water vapor dierential absorp-
tion lidar (DIAL), in Proceedings of the 2000 IEEE International Geoscience and
Remote Sensing Symposium, pp. 14651467, 2000.
Rall, J. A. R., Dierential Absorption LIDAR Measurements of Atmospheric Water
Vapor Using a Pseudonoise Code Modulated Aluminum Gallium Arsenide Laser.,
Ph.D. Thesis, 1994.
Rall, J. A. R., and J. B. Abshire, Antarctic Miniature Lidar, in Proceedings of the 1996
IEEE International Geoscience and Remote Sensing Symposium, pp. 15381540,
1996.
Rall, J. A. R., J. Kujawski, M. Obland, and M. Ott, Multi spectral lidar (MSL), in
Remote Sensing for Environmental Monitoring, GIS Applications, and Geology V.
Edited by Ehlers, Manfred; Michel, Ulrich. Proceedings of the SPIE, Volume 5984 ,
edited by U. N. Singh, pp. 132139, 2005.
Reagan, J. A., T. W. Cooley, and J. A. Shaw, Prospects for an economical, eye-safe
water vapor LIDAR, in Proceedings of the 1993 IEEE International Geoscience and
Remote Sensing Symposium, pp. 872874, 1993.
Reagan, J. A., H. Liu, and J. F. McCalmont, Laser diode based new generation lidars,
in Proceedings of the 1996 IEEE International Geoscience and Remote Sensing
Symposium, pp. 15351537, 1996.
Remsberg, E. E., and L. L. Gordley, Analysis of dierential absorption lidar from the
Space Shuttle, Appl. Opt., 17 , 624630, 1978.
Repasky, K. S., and J. L. Carlsten, Simple Method for Measuring Frequency Chirps
with a Fabry Perot Interferometer, Appl. Opt., 39 , 55005504, 2000.
Repasky, K. S., P. Roos, L. S. Meng, and J. L. Carlsten, Amplied output of a
frequency chirped diode source via injection locking, Optical Engineering , 40 , 2505
2509, 2001.
Repasky, K. S., G. W. Switzer, and J. L. Carlsten, Design and performance of a
frequency chirped external cavity diode laser, Review of Scientic Instruments,
73 , 31543159, 2002.
147
Repasky, K. S., J. A. Shaw, J. L. Carlsten, M. D. Obland, L. S. Meng, and D. S.
Homan, Diode laser transmitter for water vapor DIAL measurements, in Proceed-
ings of the 2004 IEEE International Geoscience and Remote Sensing Symposium,
vol. 3, pp. 19471950, 2004.
Repasky, K. S., A. R. Nehrir, J. T. Hawthorne, G. W. Switzer, and J. L. Carlsten,
Extending the continuous tuning range of an external-cavity diode laser, Appl. Opt.,
45 , 90139020, 2006.
Rothman, L. S., A. Barbe, D. C. Benner, L. R. Brown, C. Camy-Peyret, M. R.
Carleer, K. Chance, C. Clerbaux, V. Dana, V. M. Devi, A. Fayt, J.-M. Flaud,
R. R. Gamache, A. Goldman, D. Jacquemart, K. W. Jucks, W. J. Laerty, J.-Y.
Mandin, S. T. Massie, V. Nemtchinov, D. A. Newnham, A. Perrin, C. P. Rinsland,
J. Schroeder, K. M. Smith, M. A. H. Smith, K. Tang, R. A. Toth, J. Vander Auwera,
P. Varanasi, and K. Yoshino, The HITRAN molecular spectroscopic database: edi-
tion of 2000 including updates through 2001, Journal of Quantitative Spectroscopy
and Radiative Transfer , 82 , 544, 2003.
Rycroft, M. J., Global environmental change: research pathways for the next decade
- by Committee on Global Change Research of the US National Research Council,
National Academy Press, 1999, 595 pp. ISBN: 0 309 06420 1, (hb) @$39.95, Journal
of Atmospheric and Terrestrial Physics, 62 , 711712, 2000.
Sassen, K., The Polarization Lidar Technique for Cloud Research: A Review and
Current Assessment., Bulletin of the American Meteorological Society, vol. 72, Issue
12, pp.1848-1866 , 72 , 18481866, 1991.
Schotland, R. M., Some observations of the vertical prole of water vapor by means of
a ground based optical radar, in Proceedings of the Fourth Symposium on Remote
Sensing of the Environment , pp. 273283, University of Michigan, Ann Arbor,
Mich., 1966.
Schotland, R. M., Errors in the lidar measurment of atmospheric gases by dierential
absorption, J. Appl. Meteor., 13 , 7177, 1974.
Schott, J. R., Remote Sensing: the image chain approach, New York, Oxford Univer-
sity Press, 1997, 424 p., 1997.
Seldomridge, N. L., J. A. Shaw, and K. S. Repasky, Dual-polarization lidar using a
liquid crystal variable retarder, Optical Engineering , 45 , 106,202, 2006.
Shaw, J. A., http://www.coe.montana.edu/ee/weather/, 2006.
Shaw, J. A., J. J. Bates, W. L. Eberhard, R. J. Alvarez II, S. P. Sandberg, and
L. Larrabee Strow, Atmospheric measurements with unattended FTIR and Lidar,
148
in Technical Digest, Optical Society of America Topical Meeting on Optical Remote
Sensing of the Atmosphere, pp. 7779, 2001.
Shaw, J. A., N. L. Seldomridge, D. L. Dunkle, P. W. Nugent, L. H. Spangler, J. J.
Bromenshenk, C. B. Henderson, J. H. Churnside, and J. J. Wilson, Polarization
lidar measurements of honey bees in ight for locating land mines, Optics Express,
13 , 5853+, 2005.
Souprayen, C., A. Garnier, A. Hertzog, A. Hauchecorne, and J. Porteneuve, Rayleigh-
Mie Doppler wind lidar for atmospheric measurements. I. Instrumental setup, val-
idation, and rst climatological results, AO , 38 , 24102421, 1999.
Spinhirne, J. D., Micro pulse lidar, IEEE Transactions on Geoscience and Remote
Sensing , 31 , 4855, 1993.
Stelmaszczyk, K., M. dell'Aglio, S. Chudzynski, T. Stacewicz, and L. Wöste, Analyt-
ical function for lidar geometrical compression form-factor calculations, Appl. Opt.,
44 , 13231331, 2005.
Stephens, G. L., Remote Sensing of the Lower Atmosphere: an introduction, New
York, Oxford University Press, 544 p., 1994.
Switzer, G. W., Semiconductor laser transmitter for water vapor lidar on Mars,
Ph.D. Thesis, 1999.
Turner, D. D., W. F. Feltz, and R. A. Ferrare, Continuous Water Vapor Proles from
Operational Ground-Based Active and Passive Remote Sensors., Bulletin of the
American Meteorological Society, vol. 81, Issue 6, pp.1301-1318 , 81 , 13011318,
2000.
Turner, D. D., R. A. Ferrare, L. A. Heilman Brasseur, W. F. Feltz, and T. P. Tooman,
Automated retrievals of water vapor and aerosol proles from an operational Raman
lidar, J. Atmos. Ocean. Technol., 19 , 3750, 2002.
Wulfmeyer, V., Ground-Based Dierential Absorption Lidar for Water-Vapor and
Temperature Proling: Development and Specications of a High-Performance
Laser Transmitter, Appl. Opt., 37 , 38043824, 1998.
Wulfmeyer, V., and J. Bösenberg, Ground-Based Dierential Absorption Lidar for
Water-Vapor Proling: Assessment of Accuracy, Resolution, and Meteorological
Applications, Appl. Opt., 37 , 38253844, 1998.
Wulfmeyer, V., and C. Walther, Future Performance of Ground-Based and Airborne
Water-Vapor Dierential Absorption Lidar. I. Overview and Theory, Appl. Opt.,
40 , 53045320, 2001.
149
APPENDICES
150
APPENDIX A
PRELIMINARY LIDAR EXPERIMENTS
151
Other Lidar Experiments
Besides the experiments directly related to making DIAL measurements of water
vapor, and the development that went into creating those experiments, many other
projects were undertaken to learn about the intracacies of lidar systems and devel-
opment, or to advance other technologies. This appendix describes one such project,
involving the design, construction, and testing of a ber optic laser amplier at NASA
Goddard Space Flight Center (GSFC) for use in an Antarctic lidar. The opportunity
to go to GSFC and work on this project was funded by the NASA Graduate Student
Researchers Program (GSRP). My advisors at GSFC were Dr. Jonathan Rall and
Mr. Joe Kujawski. Part of this work was reported by Rall et al., 2005.
NASA Goddard Space Flight Center (GSFC) Fiber Optic Laser Amplier
Introduction
Polar stratospheric clouds (PSCs) are optically thin clouds that have been shown
to have an important role in Antarctic ozone destruction. They form in the lower
stratosphere during the cold winter months, and can be composed of water ice or
Nitrogen compounds, depending on the formation temperature. PSCs provide a sur-
face for chemical reactions to take place that lead to chlorine free radicals, which are
destructive to ozone (Crutzen and Arnold , 1986). PSCs also prolong these chemical
reactions through denitrication, and thus play an important part in explaining mas-
sive ozone loss in the Antarctic (Fahey et al., 1990). Understanding PSC formation,
distribution, and movement is central to comprehending and predicting ozone loss
(Poole and McCormick , 1988).
152
One way that has been proposed to study Polar stratospheric clouds is by using
backscatter lidar. Lidar can detect the presence of PSCs and gather data used to
infer particle size, which can then be used to study radiative eects of the clouds. A
collaborative group of researchers from NASA Goddard Space Flight Center (GSFC)
and Montana State University (MSU) have proposed to build an autonomous, ground-
based backscatter lidar and install it on the Antarctic plateau to study PSCs. The
lidar would be self-contained with enough battery power to last through the Antarctic
winter. Clever power budgeting would allow it to take data on the order of minutes
and relay that data via satellite back to ground stations in the United States. The
entire system would have to be robust enough to operate in the harsh Antarctic
environment, where ambient temperatures are expected to range between -20 and
-80 degrees Celsius. To conserve power, very little active thermal control will be
done, though passively controlled electronics boxes will be available at temperatures
between -20 and -40 degrees Celsius.
The proposed lidar will take advantage of a ber optic laser amplier to produce
the powers necessary from the transmitter. Fiber optic laser ampliers typically use
a single-mode glass ber that has been doped with one or more rare-Earth elements.
Dierent elements are used to produce light at dierent wavelengths. Erbium is
typically used in telecom components, since it provides gain at 1550 nm, in the
low-loss window of silica bers. Ytterbium has a large gain bandwidth between
880 nm and 1000 nm, with a large peak around 975 nm. Emission can occur most
eciently near 975 nm or between 1.02 microns and 1.07 microns. Gain is achieved
by emission from the doped Ytterbium ions when the amplier is pumped into a state
of population inversion. Ytterbium can be classied as having a three-level energy
scheme. The doped ions are excited to a higher energy state by absorption of pump
photons and then relax rapidly to a lower, stable energy state with lifetimes on the
153
order of milliseconds. This stored energy can then be released through stimulated
emission by a seed source, which eectively transfers energy from the pump to the
seed.
Fiber ampliers have several advantages over conventional free-space amplication
of light as is typically used in operational lidars. Fiber ampliers allow high gain in a
small package, since the bers, which may be up to several meters long, can be coiled
into bundles. The only limit to size is the bend radius, which needs to be sized so as to
not allow light leakage or cause attenuation eects within the amplier. This radius
is typically around 10 to 20 cm. Fiber ampliers do not use free-space optics, such as
mirrors, lenses, or lters, as all of the necessary optical components can be contained
within the ber optic itself. In this manner, the transmitter always remains aligned,
regardless of the direction and movement of the ber. This makes a ber amplier-
based lidar very vibration resistant, and therefore ideal for eld deployments.
Another advantage of ber optic laser ampliers is that they are a mature technol-
ogy. Fiber ampliers have been in worldwide use in the telecommunications industry
for over a decade now. Optical ber ampliers were actually invented in 1964 by E.
Snitzer, who created a Neodymium doped amplier operating at 1.06 mm, and were
looked at for transmission systems as early as the 1970's (Becker et al., 1999). Fiber
amplier components are widely available and have low failure rates. The Technology
Readiness Level (TRL) for these devices is high, making them a good candidate for
eld instruments.
This section describes a prototype Ytterbium-doped ber optic laser amplier
that was built at GSFC during the summer of 2005, and is organized as follows.
The next section discusses the lidar equation calculations used to set the initial lidar
system parameters. Then a description of the simulation software is presented, with
an explanation of how the gain ber length was determined. Next, the construction
154
of the ber amplier is presented. Following that, simulated and measured results for
the system are presented, as well as a discussion of possible reasons for discrepancies
between the two. Finally, suggestions for future work are oered. The section is
concluded with concluding remarks.
Lidar Equation Calculations
The rst step in designing the GSFC Antarctic polar stratospheric cloud lidar was
to use the lidar equation (equation 2.1) to determine what system parameters would
be needed to produce scientically valuable data from the desired target. Questions
that needed to be answered before design and construction could begin included de-
termining what type of detector to use, if frequency-doubling was required or desired,
what laser frequency was most advantageous to operate at, and what power, pulse
repetition frequency (PRF), pulse width, averaging time, and receiver area to use.
A spreadsheet was created (Lidar_equation.xls) that allows the user to input sys-
tem and atmospheric parameters and calculates the expected daytime and nighttime
signal-to-noise ratio (SNR) for a specied altitude. Through use of this spreadsheet,
it was determined that the desired wavelength to use is 1064 nm, due to the large
number of well-tested and easily available parts for this common wavelength. Run-
ning through the calculation showed that a nighttime SNR of 37.7 and daytime SNR
of 1.83 could be achieved at a range of 20 km using a PRF of 7500, 1 µs pulse widths,
a 200 pm bandpass lter, 20 µJ of energy per pulse, and averaging for 5 minutes. 5
inch diameter athermal telescopes already eld tested by GSFC scientists will most
likely be used on the receiver side. A detector eciency of 2% was assumed, because
high-eciency detectors for 1064 nm are dicult to manufacture and therefore not
easily available. Several communications with Perkin-Elmer occurred discussing the
155
possibility purchasing a custom-made detector optimized for this wavelength. Perkin-
Elmer was responsible for a similar detector used on the Geoscience Laser Altimeter
System (GLAS) launched into orbit aboard ICESat in 2003. With the wavelength
picked and calculations completed on the feasibility of the lidar system, design work
began on the ber amplier.
Gain Fiber Length Determination
Once the requirements of the Antarctic lidar were known from lidar equation
calculations, a model of the ber amplier's performance was desired to expedite and
optimize construction. Liekki, the company supplying the highly-doped Ytterbium
gain ber, provides software for modeling the performance of their products in ber
ampliers. This software, Liekki Application Designer v2.0 (LAD) was used to design
the ber amplier described in this paper.
LAD employs a graphical user interface that allows the user to place dierent
components onto a workspace, connect them together with ber optic cables, run
simulations, and view results at any component-ber juncture. Components include
optical isolators, seed and pump diode lasers, lters, couplers, wavelength division
multiplexers (WDM's), reectors, gratings, ber bundles, and Ytterbium and Er-
bium highly-doped single and double-cladding gain bers. Results available are seed
power, pump power, and Amplied Spontaneous Emission (ASE) power, all given in
decibels relative to 1 mW (dBm). Properties for each component can be loaded from
pre-stored les or input manually and changed between every simulation. Simula-
tions typically take between a few seconds and a few minutes depending on model
complexity. All simulations assume that the lasers are operated in CW mode. It has
been reported that future versions of LAD will contain the option of pulsing lasers
156
(See http://www.liekki.com).
LAD was initially used to determine through simulation the optimal gain ber
length to be used in the nal ber amplier. Three ber amplier models were built
and simulated: the ber amplier design that was actually built, the actual design
minus the 1064 nm isolator, and a version of the amplier utilizing 4 W unstabilized
Liekki pump diodes instead of the Bragg grating-stabilized 200 mW QPhotonics pump
diode lasers. The length of the gain ber was the only variable being changed between
simulations.
The data for the ber amplier model that was actually built are shown in gure
A.1. Results for the other two modeled ampliers are contained in the spreadsheet
le
YDFA_Gain_Fiber_Length.xls for future reference. Parameters for each component
were taken from the actual data sheets of the parts used. It was assumed that the
pump lasers and seed laser were being run at maximum power (150 mW and 200 mW,
respectively). The simulation showed that the output of the ber amplier reached
a maximum of 25.453 dBm at a gain ber length of about 8.5 cm. The simulation
also indicated that the ber amplier output did not decrease with further increases
of gain ber length, which is a puzzling result, as the high absorption and losses in
the ber should begin to attenuate the output signal. Nonetheless, care was taken in
the construction of the ber amplier to keep the gain ber length above this limit
of 8.5 cm, with less attention trying to achieve an actual length of exactly 8.5 cm.
It should be noted that simulations were performed for optimal gain ber length
of a second stage amplier, which adds two more pump diode lasers and a second
stage of gain ber to the output of the rst stage. These results are also contained in
the le YDFA_Gain_Fiber_Length.xls, to be used if the decision is made to add a
second stage to this amplier in the future.
157
Figure A.1: Simulated output signal of the Ytterbium gain ber versus length.
Fiber Amplier Construction
After the gain ber length simulations, the Liekki Application Designer was used
to simulate the three phases of construction of the actual ber amplier. The rst
phase, Construct 1, spliced a Bragg grating-stabilized, 150 mW pump laser (QPho-
tonics QFBGLD-980-150) running at 975 nm to a 980/1060 WDM. This acts as the
forward pump. A Bragg grating-stabilized, 200 mW seed laser (Lumics LU106M200)
running at 1064.6 nanometers was spliced to a polarization insensitive single stage iso-
lator (Advanced Fiber Resources PSSI-06-P-N-B-1), which was subsequently spliced
to the 1060 nm input of the WDM. Construct 1 is shown schematically in gure A.2.
All ber splicing was performed using a Vytran FFS-2000 Automated Fusion Splic-
ing Workstation. Most of the bers were spliced by the author after being trained
on this apparatus. The Vytran workstation automates all of the steps needed to cre-
ate a robust, high-quality ber splice: stripping the ber coating to reveal the bare
glass ber, cleaning the ber of all particulates, accurately cleaving the ber end,
158
Figure A.2: A schematic of Construct 1.
Figure A.3: A schematic of Construct 2.
computer-aided aligning and fusion of the bers using a Tungsten lamp, and recoat-
ing the splice with a UV-cured coating material. With practice, several splices can
be completed within one hour.
The second phase, Construct 2, spliced 21.0 cm of highly-doped Yb gain ber to
the output of the WDM. The length of the ber was chosen to allow for multiple
splices to be made while still keeping the length of the gain ber above 8.5 cm, since
each splice removes ~ 2 cm from the ber due to cleaving. Construct 2 is shown
schematically in gure A.3.
The third and nal phase, Construct 3, spliced an identical QPhotonics pump
laser diode to the 980 nm input of an identical WDM as used in Construct 1. This
acts as the reverse pump. The output of the WDM was then spliced to the free
end of the gain ber, which was reduced to 19.0 cm total length in the process. A
159
Figure A.4: A schematic of Construct 3. This schematic represents the nal ber
optic laser amplier design used for testing.
single-mode FC connector was spliced to the 1060 nm input of the WDM, making this
the output of the ber amplier. The FC connector was used to make measurement
acquisition easier. Construct 3 is shown schematically in gure A.4.
Simulated and Measured Results
Construct 1 The Liekki Application Designer was used to simulate the three
phases of construction of the actual ber amplier to compare its expected perfor-
mance with actual measured results. The simulated and measured results for Con-
struct 1 are shown in Table A.1 below. Both lasers are being run at optimum operating
current, 450 mA for the seed laser (corresponding to 200 mW) and 250 mA for the
pump laser (corresponding to 150 mW).
P-I curves and spectra were taken at this point for both the seed laser and the
forward pump laser. The results are shown in gures A.5, A.6, A.7, and A.8, re-
spectively. Only one laser was running at a time when the P-I curves and spectra
were taken. The lasers were placed in 14-pin diode mounts (Thorlabs LM14S2 for the
seed and reverse pump lasers and ILX LDM-4980 for the forward pump). Current
160
Amplier Output Power at 1064.6 nm dBm mW
Simulated 21.24030004 133.0546337
Measured 20.51152522 112.5
Amplier Output Power at 975 nm dBm mW
Simulated 21.54091495 142.5907964
Measured 21.58663981 144.1
Table A.1: A comparison between the simulated and measured amplier output pow-
ers for Construct 1 at the wavelengths of 1064.6 nm and 975 nm.
Figure A.5: P-I curve for the seed laser.
was provided by Thorlabs LD3000 current drivers. Temperature was controlled by
ILX LDT-5100 temperature controllers. Both laser TEC's were held steady at 25 de-
grees Celsius. The bare ber of the forward WDM output was aligned with a power
detector for the P-I measurements and input into an OSA to measure the spectra.
From the P-I curves, it was determined that the power from the seed and pump
lasers after the WDM was 112.5 mW and 144.1 mW, respectively. The seed power
was considerably lower than the 200 mW that was presumably being fed into the
161
Figure A.6: Spectrum of the seed laser.
Figure A.7: P-I curve for the pump laser. A neutral density lter was used to obtain
the data in pink.
162
Figure A.8: Spectrum of the pump laser.
amplier, and was probably attenuated heavily by the isolator. Since these powers
were experimentally measured to be the input powers to the gain bers, the LAD
model was modied for simulations beyond this point to reect these measured pow-
ers. To accomplish this in the model, the power parameter of the seed laser was set to
169 mW (down from 200 mW for Construct 1) and the pump laser power parameter
was left unchanged at 150 mW. These parameters give the correct answer for input
powers into the gain ber in the forward direction. A P-I curve was not taken for the
reverse pump diode, as it was assumed to be identical to the forward pump diode.
Construct 2 The simulated and measured results for Construct 2 are shown in
Table A.2 below. Again, both lasers are being run at optimum operating current and
the TEC's were held steady at 25 degrees Celsius. All pump power above the noise
oor was absorbed in the gain ber.
The measured power at this point shows gain but not nearly as much as the LAD
163
Amplier Output Power at 1064.6 nm dBm mW
Simulated 23.58927244 228.5215939
Measured 20.79181246 120
Table A.2: A comparison between the simulated and measured amplier output pow-
ers for Construct 2 at the wavelength of 1064.6 nm.
Figure A.9: Simulated and measured ASE of Construct 2.
simulation predicted. One diculty in making this measurement was that the power
meter used seemed to have a position-sensitive detector head. Dierent powers were
measured depending on where the ber was aimed on the detector. 120 mW was a
consistently repeatable value, so it was used, even though it was on the lower end
of all power values observed. The predicted ASE did not match the measured ASE
either, as shown in gure A.9.
Construct 3 The simulated and measured results for Construct 3 are shown in
Table A.3 below. All three lasers were run at optimum operating current and the
TEC's of the pump lasers were held steady at 25 degrees Celsius. The seed laser
164
required tuning to remain single-mode at 1064.6 nm, and a voltage of 557.4 mV was
monitored on the TEC. All pump power was again absorbed into the gain ber.
Conguration Simulated Power (mW) Measured Power (mW) % dierence
Seed Laser Only 67.78330154 61.2 9.7
Seed and Forward Pump 177.5027186 101.2 43.0
Seed and Reverse Pump 194.580914 113.3 41.8
Seed and both pumps 283.0100864 160.4 43.3
Table A.3: A comparison between the simulated and measured amplier output pow-
ers for Construct 3 in dierent congurations at the wavelength of 1064.6 nm.
Power measurements were easier to make at this point because of the use of a
FC connector versus the bare ber, as in Construct 2. The amplier showed gain
when operated with both pumps or individual pumps. It showed slightly more gain
when operated with the individual reverse pump than with the individual forward
pump, and displayed about 4 dB of gain over the seed laser only with both pumps
at full power. However, the wide dierence between the LAD-simulated gain and the
measured gain is disturbing. The measured performance does not seem to be even
close to the performance that was expected based on simulations. Figure A.10 shows
a spectrum of the predicted performance versus the measured performance.
An attempt was made to pulse the seed laser and obtain pulse energy measure-
ments. Unfortunately, a suitable pulse driver was not immediately available to drive
the seed laser. Custom pulse driver cards have been built and used successfully with
other diode lasers at GSFC, but there were no cards available that could drive the seed
laser with enough current. Therefore, no pulse energy measurements were obtained.
Several problems have been suggested for explaining the discrepancies between the
simulated results and the measured results. First, there may be a problem with the
LAD simulation not including an important loss or nonlinear eect that is reducing
the overall power output. As stated before, it is troubling that LAD simulations show
165
Figure A.10: Simulated and measured results of the system.
no loss of signal when using any length of gain ber between 8.5 cm and 1.5 m. It
may be likely that signicant loss is present immediately above 8.5 cm and using 19.0
cm of highly-doped gain ber is signicantly attenuating the signal. A problem in the
gain stage of the ber amplier seems to be indicated by looking at the measured gain
curve, shown in gure A.11. The curve does not have a step-function characteristic as
expected of ber ampliers. As of this writing, the author has been in contact with
software technicians at Liekki to insure that the simulation parameters used match
the real situation.
Second, there are indications that the pump lasers are interfering with each other.
When one pump is already operating at full power and the power of the other pump
is slowly increased, an interaction is visible through an IR viewer. The light leaking
out of the bers from one pump immediately decreases in a discontinuous manner
when the other pump laser's power is increased just above threshold. It is possible
that the two pumps and their respective gratings are creating a laser cavity, reducing
166
Figure A.11: Measured gain curve of the ber amplier.
the pumping eciency and therefore reducing the overall amplier gain. Evidence of
this is seen when the pump lasers are running at full power and the input seed laser
power is just above threshold. The amplier seems to be trying to lase at about 1023
nm. This behavior appears whenever one pump is running at full power, and the
other is at any power above threshold.
Third, there may be a currently undiscovered source of excessive loss in the ber
amplier. Each ber splice was veried to be of high quality during the construction
process. Yet, damage may have occurred to a component, ber splice, or to the ber
itself. Loss may be occurring due to the ber being wound too tightly in some places.
More investigation is needed to nd the source of these problems.
Future Work
Three distinct action items need to be completed to move this project forward.
First, 975 nm isolators should be purchased and installed on the pump laser lines. This
167
should eliminate any interference between the pump lasers. Second, the optimized
gain ber length needs to be understood. Conversations with Liekki and further
LAD simulations should show unequivocally whether there is a performance dierence
between using 8.5 cm or 19.0 cm of gain ber. This should eliminate questions about
the eciency of the gain stage. Third, a pulse driver board needs to be built and
optimized for pulsing of the seed laser. The Lumics seed laser can be operated with
currents up to 2 A, with a 30 µs period and 300 ns pulse widths. This current needs
to be taken advantage of to increase the power of the system. Accomplishing these
three tasks should eliminate any questions about the gain of the amplier and allow
lidar data to be taken. At that point, discussion should begin on whether a second
gain stage is needed for completing the science goals in Antarctica.
Conclusion
Design, simulation, construction, and testing of a highly-doped Ytterbium-based
ber optic laser amplier for measuring Antarctic polar stratospheric clouds is well
underway at Goddard Space Flight Center. A prototype ber amplier has been built
and is currently undergoing testing. Initial testing showed that gain was achieved
producing an output power of about 160 mW, which was, however, much lower than
the 283 mW output power predicted through simulations. Several problems have been
suggested that may be reducing the gain, and solutions are being investigated. It is
hoped that after further testing and optimization of this amplier, a prototype of the
Antarctic PSC lidar can be deployed in Alaska for eld testing before being installed
in Antarctica.
168
APPENDIX B
MAXIMUM PERMISSABLE EXPOSURE (MPE) LIMITS AND FEDERAL
AVIATION ADMINISTRATION (FAA)
APPROVALS
169
Introduction
Proper laser safety requires that all personnel involved with the operation of a
laser be aware of the power output of that laser, and whether it exceeds the maximum
permissable exposure (MPE) level for the human body. The MPE is the irradiance
[W/cm2] or energy density [J/cm2] threshold that has been determined to not damage
certain body tissues. The MPE depends on many factors, including the type of laser
(pulsed or cw), the laser wavelength, pulse energy and pulse width, and the type
of tissue upon which the laser is incident. For example, the MPE level for skin
tissue is typically higher than the MPE for laser light that reaches the eye, since
light that is not absorbed by the cornea will be focused onto the retina by the eye's
lens. The calculations for determining MPE are typically somewhat nonintuitive.
The wavelength plays a crucial role in the calculations since the human eye has a
natural blink response for visible wavelengths, but none for wavelengths in the near
infrared. Near IR calculations therefore must take this factor into account. Further
into the IR, near telecommunications wavelengths of 1550 nm, the MPE for the human
eye changes dramatically since these longer wavelengths can no longer penetrate the
cornea and be focused onto the retina. Calculating the MPE for pulsed lasers tends
to be much more dicult than cw lasers. Fortunately, detailed descriptions of the
restrictions and calculations are given in the ANSI National Standard for Safe Use of
Lasers (ANSI Z136.1-2000). All information and calculations described herein were
taken from this source.
ANSI Z136.1-2000 applies to lasers with wavelengths between 180 nm and 1 mm
and is full of useful safety information for anyone involved with lasers. It denes
the ve classes of lasers. Class 1 lasers present no hazards to people and require no
controls. Class 2 lasers are visible lasers (400 nm < λ < 700 nm) that emit < 1 mW
170
of radiant power. Eye protection for this class comes from the natural blink reex and
aversion response. Class 3 lasers are hazards under direct or specular reection, but
not diuse reection, and are split into two categories. Class 3a lasers have outputs
between 1 and 5 times the Class 1 Accessible Emission Limit (AEL, the maximum
emission permitted) for λ < 400 nm or λ > 700 nm, or less than 5 times the AEL in
visible wavelengths. Class 3b lasers exceed the ultraviolet (λ < 400 nm) or infrared
(λ > 700 nm) or visible output of Class 3a lasers but not above an average power of
0.5 W for ≥ 0.25 seconds or energy of 0.125 J in ≤ 0.25 seconds. Because the output
of the amplied diode laser transmitter used in the DIAL instrument is greater than
1 mW but less than 0.5 W, it is classied as a Class 3b laser. Class 4 lasers are
dangerous under direct, specular, and diuse reection and are re hazards. This
classication includes any laser above Class 3b. Other useful denitions are included
as well. Continuous-wave (cw) lasers are dened as lasers that have a continuous
output for a period ≥ 0.25 seconds, since the eye does not have a blink response for
shorter pulses. Beam radius is measured between the point with peak power per unit
area and the point where the power per unit area drops to 1/e of this maximum value.
Lasers being operated in the open atmosphere where civilians unaware of laser
safety or the laser itself may be exposed, such as in lidar transmitters, require special
considerations to ensure that no one is harmed. Lidars in general usually need to
be approved for operation by the Federal Aviation Administration (FAA) to ensure
safety to pilots and aircraft passengers. A number of bureaucratic steps and mathe-
matical calculations are needed to obtain approval. Outdoor lasers are now governed
by FAA Order 7400.2F, Procedures for Handling Airspace Matters, Part 6, Chapter
29. As of February 2007, this document could be found by doing a search on the
FAA's website, www.faa.gov, or more specically at http://www.faa.gov/
airports_airtrac/air_trac/publications/atpubs/AIR/index.htm.
171
Obtaining FAA Approval
The steps for getting a laser approved for ring outdoors are: 1) Contact the FAA,
2) Calculate the minimum eye-safe altitude, 3) Obtain independent conrmation of
calculations, and, 4) Notify the FAA of the calculation results, the independent re-
viewer, and all necessary information on location and date and time of laser operation.
This appendix explains the steps and calculations taken from the ANSI Standards
and performed to nd the MPE levels and originally approve the water vapor DIAL
instrument for outdoor use.
Contacting the FAA
The rst step towards approval for vertical operation is to contact the FAA and
notify them of the initial characteristics of the laser. As of August 2004, Bozeman, MT
was governed by the Northwest region of the FAA, whose headquarters are in Seattle.
The contact I worked with for these calculations was Kathie Curran. In conversations
I had with her at that time, she indicated that approvals for the Bozeman area
would soon be handled through Los Angeles International Airport, so her contact
information is likely no longer valid. She also indicated that the person who would
be replacing her for these types of permissions is Gordy Burnet and that the ocial
at FAA headquarters in charge of lasers is Steve Rohring. The person that we are
working with to obtain permissions as of February, 2007, is Rick Roberts, a FAA
system support specialist in Renton, Washington. Contact information for these four
people is given in tables B.4 and B.5.
172
Contact: Kathie Curran Gordy Burnet Steve Rohring
Email: Kathie.Curran@faa.gov gordy.burnet@faa.gov stephen.rohring@faa.gov
Phone: (425) 227-2558 (425) 227-2535 (202) 267-9231
Fax: 425-227-1530 425-227-1530
Table B.4: Contact information for the FAA that I used in 2004.
Contact: Rick Roberts
Email: Richard.Roberts@faa.gov
Phone: (425) 917-6728
Fax: 425-917-6476
Table B.5: Contact information for the FAA, current as of February, 2007.
To begin the approval process, the FAA wants to know at least the laser's location
(latitude and longitude), the eye-safe altitude, and the date and time of operation. For
MSU, the coordinates were given for the top of the roofport room, at the Southeast
corner of the roof of Cobleigh Hall (found using a hand-held GPS unit):
45 degrees, 39.984' North, 111 degrees, 2.744' West, 5003 +/- 18.3 feet.
Ms. Curran indicated that decimals should not be used in the coordinates, all
distances should be expressed in feet instead of meters, and the ground-level altitude
according to her calculations should be 4924 feet. Our updated coordinates for FAA
purposes therefore are:
45 degrees 39' 59 Latitude, 111 degrees 02' 44 Longitude, 4924 feet elevation.
Using the latitude and longitude coordinates of the laser, the FAA will determine
its proximity to airports and designate which of four ight zones the lidar occupies.
This in turn denes the laser irradiance level that must be adhered to. The ight zones
are dened in FAA Order 7400.2E and are listed with their dening characteristics
in table B.6. It was determined with mapping software using the FAA latitude and
longitude coordinates for Gallatin Field airport that the roofport room in Cobleigh
173
Zone Distance to Airport * Max. Irradiance Level
Laser Free Zone 2 NM radial from runway centerline 50nW/cm2
2500 feet each side of runway for 3 NM
Critical Flight Zone 10 NM radial from airport reference point 5µW/cm2
Sensitive Flight Zone Outside Critical Flight Zone having 100µW/cm2
special restrictions
Normal Flight Zone Airspace not dened by other zones MPE
Table B.6: Specications of the FAA's four ight zones. (*Note that minimum altitude
requirements for these irradiance levels also apply. NM=Nautical Mile)
Hall, where most vertically-viewing lidar work occurs, is located a distance of 9.25
miles (14.89 km, 8.04 nautical miles) from the airport, placing it in the Critical Flight
Zone.
Minimum Eye-safe Altitude Calculations
The goal of the next step is to use the ANSI Standards to nd the most conser-
vative MPE level for the laser, and then calculate the minimum altitude where beam
divergence causes the laser beam to fall under this irradiance level, making it eye-safe.
The eye-safety calculation that must be done depends on whether the laser that is
being operated is cw or pulsed.
CW Lasers For cw lasers, the average power and beam area must be known so
that an irradiance can be computed by taking their ratio. This is then compared to
an MPE level that is calculated with help from the ANSI Standards, which takes into
account cumulative damage to tissue based on exposure time. For visible wavelengths,
a maximum occular exposure time of 0.25 seconds can be used due to the eye's natural
aversion response. For UV or IR wavelengths, a maximum exposure time of 10 seconds
is assumed based on natural movements of the eye, which limit the exposure time.
174
Since all MSU lidars to date rely on pulsed lasers for operation, cw MPE calculations
will not be discussed.
Repetitive Pulse (RP) Lasers MPE calculations for pulsed lasers are signicantly
more complicated than for cw lasers. Three rules must be followed. First, the expo-
sure from any single pulse cannot exceed the MPE level. This rule protects against
thermal damage due to any single pulse having greater than average energy. The
specications for the water vapor DIAL transmitter are given in table B.7. The pre-
Specications Preliminary Actual
λ 830.0nm 828.187nm
Pulse Length 1.0µs 500.0ns
Repetition Rate 10.0 kHz 20.0 kHz
Average Power 500.0mW 200.0mW
Pulse Energy 500.0nJ 100.0nJ
Beam Divergence (full angle) 1.43mrad 150µrad
Table B.7: Values used for the maximum permissable exposure calculations.
liminary specications were used in the calculations for the original approval of this
system. The actual specications of the nal operational system as of February, 2007,
are given for comparison. From table 5a of the ANSI Standards, the MPE for a laser
of these specications (specied in J/cm2) is
MPE = 1.8× 10−3 · CA · t0.75, (B.1)
where t [seconds] is the exposure time and CA [unitless] is a correction factor dened
in table 6 of the ANSI Standards as
CA = 102(λ−0.7). (B.2)
175
With CA = 1.82 for the wavelength of 830 nm and a single-pulse exposure time of
t = 1.0µs the single-pulse MPE is MPESP = 1.04× 10−7 J/cm2.
Second, the exposure from any group of pulses delivered in a time t cannot exceed
the MPE level for that time. This rule protects against cumulative injury from
photochemical damage mechanisms and also against thermal damage caused by heat
buildup from the laser average power. The maximum exposure limit for an IR laser,
as discussed in the CW Lasers section above, is 10 seconds. With equation B.1, the
MPE for this exposure time is MPE : Hgroup = 0.0184 J/cm2. Averaging over the
number of pulses in this exposure time, the average power limit becomes
MPE : Hgroup/pulse =
(0.0184 J/cm2)
(104 pulses/sec) · (10 sec) = 1.84× 10
−7 J/cm2. (B.3)
Finally, a third rule protects against cumulative injury from thermal buildup below
the allowed threshold levels, and is dened as the single-pulse MPE from rule 1
multiplied by a multiple-pulse correction factor. With the total number of pulses
being the same as calculated in rule 2, 100,000 over 10 seconds, the nal MPE level
is
MPE/pulse = n−0.25 ·MPESP = 5.85× 10−9J/cm2. (B.4)
The MPE calculated from rule 3 gives the most conservative answer, and therefore
must be used to determine the eye-safe altitude. The beam radius where the laser
reaches the eye-safe MPE is calculated by solving the equation
MPE = (PulseEnergy) · (RepetitionRate) · (ExposureDuration)pi(BeamRadius)2 , (B.5)
to nd a value of BeamRadius = 1650 cm. The eye-safe altitude, assuming a cir-
cular beam, is where the divergence of the laser causes the beam radius to grow to
176
this size, or EyeSafeAltitude = (BeamRadius)/tangent(half -angle divergence) =
23.1 km = 75, 787 feet. Rounding the calculations upwards adds an extra safety fac-
tor to the calculations. This is the altitude that must be reported to the FAA, since
each individual laser pulse will exceed the MPE level at any lower altitude.
It is important to note that even for a relatively low-power infrared laser the FAA-
dened eye-safe altitude is over 14 miles above ground level! Due to its much smaller
divergence angle, the actual system would have an even higher calculated eye-safety
altitude, although the actual laser divergence will be larger than designed due to
atmospheric turbulence. This altitude is far dierent than if the MPE calculations
were performed simply by using the average power and divergence of the laser beam,
where the maximum irradiance level of 5µW/cm2 for the Critical Flight Zone would
be reached at an altitude of 2495m = 8187 feet. This enormous dierence is primarily
due to the assumption that any person looking into the beam is assumed to not avert
their eyes for 10 seconds. Visible beams of the same specications come to an eye-
safe level much quicker because of the natural blink response. This is a problem for
infrared lidar systems because the FAA typically will not grant permission to lasers
that are not eye-safe below 20,000 feet since doing so would require high-altitude
ight paths to be altered to avoid the beam. Permission to re the MSU lidar system
at 830 nm was only given because the divergence angle was expanded so that the
eye-safe altitude was calculated to be under the 20,000 feet limit.
One way to possibly circumvent this problem in the future is to make the argu-
ment that the 10-second exposure limit is highly unlikely, since the beam is pointed
vertically and the beam width is not very wide even at high altitudes, making the
chances of a pilot nding the beam and staring into it for 10 seconds extremely im-
probable. For example, at the calculated eye-safe altitude the beam diameter is 33
meters across, meaning an aircraft would need to be ying at 3.3m/s or less to remain
177
in the beam for 10 seconds or more, assuming it was ying directly across the center
of the beam. With the average cruise speed of a small aircraft such as a Cessna 172
Skyhawk listed as ∼ 60m/s, it is highly unlikely that a small plane or jet would be
in the beam for more than a fraction of a second. If the MPE is recalculated using
a more realistic, but still very conservative, exposure time of 1 second, the eye-safe
altitude drops to a much more reasonable value of 5500m = 18, 045 feet. Further
arguing that atmospheric turbulence limits the divergence angle to 1 mrad or more
would make the eye-safe altitude even lower still.
Independent Conrmation of Calculations
After the calculations for minimum eye-safe distance have been performed, the
FAA used to require that they be veried by an independent source. Calculations are
now sent directly to the FAA contact in Washington for verication. If conrmation
of MPE calculations is still needed, it can be done through a number of sources,
including businesses who charge for the service, but the agency recommended by the
FAA in 2004 was the Food and Drug Administration's Center for Devices and Radi-
ological Health (CDRH), who are responsible for enforcing the Federal government's
regulations concerning lasers. The person who was directly in charge of laser safety is
a laser physicist in the CDRH Compliance Oce, Dale Smith. He was very familiar
with lasers and their safety calculations and veried the calculations for the MSU
water vapor DIAL system. His contact information is given in table B.8.
Contact: Dale Smith
Email: lds@cdrh.fda.gov
Phone: (301) 594-4654 x147
Table B.8: Contact information for the CDRH.
178
Approval
Once the FAA has been notied of the calculation results, the independent re-
viewer, and the requested location, date, and time of laser operation, they will either
disapprove or approve the request. Approval comes in the form of a Letter of Non-
Objection which states the the allowed dates and times of operation and any safety
stipulations required by the FAA. Common examples of stipulations include: hav-
ing a phone line where operators can immediately be contacted by the FAA, having
spotters watching for planes ying near the beam, and notifying whomever is on duty
at the Air Trac Control Tower (ATCT) at Gallatin Field Airport in Bozeman and
the Air Route Control Center (ARTCC) at Salt Lake International Airport in Salt
Lake City 30 minutes before commencing and upon shutting down. The Letter of
Non-Objection lists the numbers to be used for contacting air trac control (ATC).
For MSU, the numbers are given in table B.9. Any changes in the lidar design or
laser requires new approval before operation. The Letter of Non-Objection takes
about 3 weeks to get all of the necessary signatures. A copy is typically faxed upon
completion, with a hard copy mailed simultaneously.
Gallatin Field Airport (BZN) ATCT (406) 388-9082
Salt Lake City International Airport (ZLC) ARTCC (801) 320-2561
Table B.9: Contact information for airports in Bozeman and Salt Lake City, Utah.
Conclusion
Obtaining and maintaining FAA approval for ring lidar systems is obviously a
time-consuming, tedious, and somewhat confusing process, although it becomes easier
once a particular lidar system is initially approved. The ideal situation would be to
179
ensure laser safety while circumventing the need for continual FAA approval, which
can at times stand in the way of scientic objectives when interesting atmospheric
conditions call for using a lidar that does not have current FAA approval. As of
February, 2007, representatives of the FAA are insisting that human spotters on
the roof of Cobleigh Hall are still necessary even with the installation of an X-band
radar capable of detecting airborne vehicles near the MSU campus, although they
are trying to get around this requirement. Until a solution for allowing lidar systems
to be operated safely at any time is found that is suitable to both MSU and the
FAA, lidars at MSU will continue to be subject to the FAA's approval and reapproval
processes and restrictions.
180
APPENDIX C
DIAL OPERATING GUIDE
181
This appendix consists of a user's guide for operating the DIAL. The step-by-step
directions for operation of the system are:
1. Make sure the APD power cord is disconnected.
2. Apply power to the power strip on instrument cart containing the wavemeter,
power meters, MCS, AWG, AOM power supply, and fan power supply.
3. Turn on both power meters, wavemeter, the power supply to the AOM driver,
the power supply to fan, and the AWG used to provide the pulsing waveform
to the AOM driver. The wavemeter will take several up to 10 minutes to warm
up. The 1830c power meter should be set to fast averaging by pushing the
AVG button once, and should be set for use with the attached attenuator by
pushing the ATTN button once. The wavemeter wavelength should be set to
Vacuum. The voltage to the AOM driver should 28 V. The fan that cools the
second tapered amplier should be supplied with about 20 V.
4. Verify that all wires are correctly connected, especially the signal BNC cable
from the APD to the MCS, the power cables to the AOM driver, and the output
BNC cable from the AWG to the AOM driver, as these tend to be disconnected
or loosened more than other cables.
5. Turn on the two laser-tuning AWG's, and the piezo driver for the ECDL.
(a) Set channel 1 of the ECDL bias AWG (located on the left) to a sinusoidal
waveform, with 0.1 V amplitude and 150 kHz frequency.
(b) Set channel 2 of the ECDL bias AWG (located on the left) to a dc bias
waveform with an amplitude of -4 V.
182
(c) Set channel 1 of the ECDL tuning AWG (located on the right) to a dc
bias waveform and activate it. The amplitude of this waveform will set
the wavelength to which the ECDL is tuned, as described in Chapter 5.
Typical voltage values are 0.7 V for the on-line wavelength and 3.0 V for
the o-line wavelength, although these are variable.
6. While waiting for the wavemeter to warm up, plug in the power to the laptop
LidarLap1.
7. Connect the laptop to the USB-GPIB connector that is attached to the laser
tuning and bias AWG's.
8. Activate power to the laptop, and login to the lidarlap1 user.
9. Start LabVIEW and open software dialv10.vi.
10. Create a new folder for writing data in E:\Vertical_WV_DIAL\Data_runs
with the current date in the format of MMDDYY, i.e. 010107 for January 1st,
2007.
11. When the wavemeter is ready to read the ECDL wavelength, make sure the
optical ber monitoring the ECDL is attached to the wavemeter and correctly
aligned, switch the ECDL bias electronics box to o and activate power to
the ECDL power supply. The power indicator on the wavemeter (provided the
optical ber monitoring the ECDL laser is correctly aligned and attached to the
wavemeter) will jump from zero power to about 25%.
12. Activate channel 2 of the ECDL bias AWG, supplying full current to the ECDL.
The power indicator on the wavemeter should now to about 75%. Note that
the ECDL wavelength may jump by ∼3 nm, but should settle to between 828
183
nm and 833 nm if it is operating correctly. If the power drops to zero or is
intermittent, it is most likely a grounding problem within the ECDL electronics
box.
13. Activate channel 1 of the ECDL bias AWG, supplying a sinusoidal modulation
to the ECDL for tuning purposes. Note that this step has been known to cause
instabilities in the ECDL wavelength, running the wavelength quickly to 800
nm or 850 nm uncontrollably. If this is the case, deactivate the sinusoidal modu-
lation and wait several minutes until the ECDL is more temperature-stabilized.
The instabilities only seem to manifest themselves if applied to the ECDL be-
fore it has sucient time to equilibrate with the environment. Generally, the
longer the ECDL can be active without the sinusoidal modulation, the better
it will behave.
14. Remove the tapered amplier covers and activate the tuning reference power
meter.
15. Turn on the lock-in amplier. When the lock-in amplier has found the tuning
reference power meter signal, it will lock, at which point the output BNC cable
can be attached.
16. Turn o the room lights.
17. Plug in the power cord for the APD at least 15 minutes before it is to be used.
As with other instruments, the more warmup time the APD has, the better it
will respond. It is good practice to leave the room lights o even though the
APD should be protected at this point within its sealed box.
18. Open the roofport at least 15 minutes or more before use to stabilize the tem-
perature and pressure within the room. This is especially critical when a large
184
temperature gradient exists between the inside and the outside of the building.
19. Verify that the 1930c power meter detector head is placed after the second
amplier-alignment iris, right before the second tapered amplier, and activate
the rst tapered amplier. The drive current should be set to roughly 2100mA
to provide about 20-22 mW of seed power to the 1930c power meter.
20. Wait for the rst TA to stabilize at about 20-22 mW seed power into the 1930c
power meter. Using the two alignment mirrors immediately before the rst TA,
peak up its power using the 1930c power meter.
21. Apply the cw setting to the AOM driver AWG. This should be a dc oset of
1.5 V amplitude, and was saved in the AWG memory as memory slot 2, called
cw.
22. Remove the 1930c power meter detector head and activate the second TA.
23. Verify that the second tapered amplier cooling fan is being supplied with about
20 V, and increase the second TA drive current to about 1000 mA.
24. Using the two alignment mirrors immediately before the second TA, the z-axis
stage for the second TA, and the focusing lens attached to a translation stage
between the two alignment mirrors, peak up the output power of the second
TA using the 1930c power meter detector head placed after the second TA.
Adjusting the alignment of the second TA tends to be easier below full drive
currents, which is why the second TA should be set to about 1000 mA at this
time.
25. Use the 1930c power meter detector head to again check the seeding power
incident on the second TA. Adjust the drive current of the rst TA accordingly
185
so that the seed power is 20-22 mW. Note that the limit for seeding either
tapered amplier is 20 mW according to the manufacturer's specications, but
that not all of the power incident on the power meter actually is seeded into
the gain region of the amplier.
26. Remove the 1930c power meter detector head from the laser beam and increase
the drive current to the second TA to a full power of about 2800-2900 mA.
Note that the current limit is 3000 mA but that the tuning characteristics of
the ampliers seem to degrade when driven above about 2900 mA.
27. Verify that the reference power being measured by the 1830c is roughly 3-4%
of the total output power of the second amplier after the AOM. It is VERY
IMPORTANT to note here that if the total transmit power is not as high
as expected, it is almost always not a hardware or alignment problem, but
most likely due to obstruction of the beam somewhere in the experiment. The
transmit system stays well aligned over weeks of use, and small misalignments
due to temperature variations or natural vibrations typically do not reduce the
power by a large fraction. Thus, be sure to verify that the beam path is clear
of obstruction before looking for a solution with misalignment or a hardware
malfunction. Also, verify that all power and data cables are correctly connected,
and that the AOM driver AWG is actually activated, all of which are common
mistakes if the system is not transmitting.
28. Replace the top covering for the experiment to ensure that stray transmit laser
light is blocked from the user and the telescope. At this point the system should
be ready to take data using DIAL_v10.vi!
29. When the experiment is ready to be stopped, rst disable the AOM driver AWG,
186
so that the laser is no longer fully transmitting. Reduce the drive current on
the second TA to about 1000 mA or less and disable it. Disable current to the
rst TA. Close the roofport.
30. Unplug the APD so that the room lights can be turned on.
31. Unplug the output BNC cable at the front of the lock-in amplier. Shut down
the lock-in amplier. Disable the voltage to the piezo driver, channel 1 on the
ECDL tuning AWG, and turn o the piezo driver. Deactivate the sinusoidal
waveform to the ECDL on channel 1 of the ECDL bias AWG.
32. Disable the -4 V dc signal to the ECDL, channel 2 on the ECDL bias AWG.
Switch the ECDL bias electronics box to on to deactivate the ECDL, and shut
o the ECDL power supply. Shutdown the wavemeter, both power meters, the
second TA's cooling fan, MCS, and tuning reference power meter. Shut down
the ECDL bias and tuning AWG's.
33. Shut o the MCS box, which should produce an audible tone on the laptop as
it loses GPIB signal to the MCS.
34. Switch o the instrument cart power strip. Close LabVIEW and all programs
on the laptop and shut it down.
35. Carefully replace the tapered amplier covers after the amplier have cooled
for about ten minutes. BE CAREFUL to not slide the covers fully onto the
TA's, as this may put pressure on internal wires and damage the TA's, which
has happened in the past.
36. Replace the top covering of the experiment to keep dust o while it is not in
use.
187
APPENDIX D
LIST OF ACRONYMS
188
AERI Atmospheric Emitted Radiance Interferometer
AIRS Atmospheric Infrared Sounder
APD Avalanche Photodiode
ASE Amplied Spontaneous Emission
AOM Acousto-optic Modulator
AWG Arbitrary Waveform Generator
CDRH Center for Devices and Radiological Health
CW Continuous Wave
DC Direct Current
DIAL Dierential Absorption Lidar
ECDL External Cavity Diode Laser
FAA Federal Aviation Administration
FOV Field of View
GSFC Goddard Space Flight Center
LASER Light Amplication by Stimulated Emission of Radiation
LIDAR Light Detection and Ranging
MAML Multi-Application Montana Lidar
MCS Multi-channel Scalar
MM Multi-Mode
189
MSU Montana State University - Bozeman
NA Numerical Aperture
NASA National Aeronautics and Space Administration
NB Narrow Band
OSA Optical Spectrum Analyzer
PBS Polarizing Beam Splitter
PMT Photomultiplier Tube
PZT Piezo-electric Tuner
SM Single-Mode
SNR Signal-to-Noise Ratio
TA Tapered Amplier
TEC Thermo-electric Cooler