Downward transport of ozone rich air & 
implications for atmospheric chemistry in the 
Amazon rainforest
Authors: Tobias Gerken, Dandan Wei, Randy J. Chase, 
Jose D. Fuentes, Courtney Schumacher, Luiz A.T. 
Machado, Rita V. Andreoli, Marcelo Chamecki, Rodrigo A. 
Ferreira de Souza, Livia S. Freire, Angela B. Jardine, 
Antonio O. Manzi, Rosa M. Nascimento dos Santos, Celso 
von Randow, Patrícia dos Santos Costa, Paul C. Stoy, Julio 
Tóta, & Amy M. Trowbridge
This is a postprint of an article that originally appeared in Atmospheric Environment on January 
2016.
Gerken, Tobias, Dandan Wei, Randy J. Chase, Jose D. Fuentes, Courtney Schumacher, Luiz A. 
T. Machado, Rita V. Andreoli, Marcelo Chamecki, Rodrigo A Ferreira de Souza, Livia S Freire, 
Angela B Jardine, Antonio O Manzi, Rosa M Nascimento dos Santos, Celso von Randow, Patricia 
dos Santos Costa, Paul C Stoy, Julio Tota, and Amy M Trowbridge. "Downward transport of 
ozone rich air & implications for atmospheric chemistry in the Amazon rainforest." Atmospheric 
Environment 124, Part A, no. 64 (January 2016): 76. 
DOI#: 10.1016/j.atmosenv.2015.11.014
Made available through Montana State University’s ScholarWorks 
scholarworks.montana.edu 
Accepted Manuscript
Downward transport of ozone rich air and implications for atmospheric chemistry in
the Amazon rainforest
Tobias Gerken, Dandan Wei, Randy J. Chase, Jose D. Fuentes, Courtney
Schumacher, Luiz A.T. Machado, Rita V. Andreoli, Marcelo Chamecki, Rodrigo A.
Ferreira de Souza, Livia S. Freire, Angela B. Jardine, Antonio O. Manzi, Rosa M.
Nascimento dos Santos, Celso von Randow, Patrícia dos Santos Costa, Paul C. Stoy,
Julio Tóta, Amy M. Trowbridge
PII: S1352-2310(15)30522-7
DOI: 10.1016/j.atmosenv.2015.11.014
Reference: AEA 14266
To appear in: Atmospheric Environment
Received Date: 22 July 2015
Revised Date: 4 November 2015
Accepted Date: 5 November 2015
Please cite this article as: Gerken, T., Wei, D., Chase, R.J., Fuentes, J.D., Schumacher, C., Machado,
L.A.T., Andreoli, R.V., Chamecki, M., Ferreira de Souza, R.A., Freire, L.S., Jardine, A.B., Manzi,
A.O., Nascimento dos Santos, R.M., von Randow, C., Costa, P.d.S., Stoy, P.C., Tóta, J., Trowbridge,
A.M., Downward transport of ozone rich air and implications for atmospheric chemistry in the Amazon
rainforest, Atmospheric Environment (2015), doi: 10.1016/j.atmosenv.2015.11.014.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
Downward transport of ozone rich air and implications for
atmospheric chemistry in the Amazon rainforest
Tobias Gerkena,∗, Dandan Weia, Randy J. Chasea, Jose D. Fuentesa, Courtney
Schumacherf, Luiz A. T. Machadod, Rita V. Andreolie, Marcelo Chameckia, Rodrigo A.
Ferreira de Souzae, Livia S. Freirea, Angela B. Jardinec, Antonio O. Manzic, Rosa M.
Nascimento dos Santose, Celso von Randowd, Patr´ıcia dos Santos Costae,c, Paul C. Stoyg,
Julio To´tah, Amy M. Trowbridgeg,i
aDepartment of Meteorology, The Pennsylvania State University, University Park, PA, USA
bDepartment of Earth Sciences, The College at Brockport State University of New York, Brockport, NY,
USA
cClimate and Environment Department, Instituto Nacional de Pesquisas da Amazoˆnia (INPA), Manaus,
AM, Brasil
dInstituto Nacional de Pesquisas Espaciais (INPE), Sao˜ Jose´ dos Campos, SP, Brazil
eUniversidade do Estado do Amazonas, Manaus, AM, Brazil
fDepartment of Atmospheric Sciences, Texas A&M University, College Station, TX, USA
gDepartment of Land Resources & Environmental Sciences, Montana State University, Bozeman, MT, USA
hUniversidade Federal do Oeste do Para´, Santare´m, PA, Brazil
iDepartment of Biology, Indiana University, Bloomington, IN, USA
Abstract
From April 2014 to January 2015, ozone (O3) dynamics were investigated as part of
GoAmazon 2014/5 project in the central Amazon rainforest of Brazil. Just above the forest
canopy, maximum hourly O3 mixing ratios averaged 20 ppbv (parts per billion on a volume
basis) during the June–September dry months and 15 ppbv during the wet months. Ozone
levels occasionally exceeded 75 ppbv in response to influences from biomass burning and re-
gional air pollution. Individual convective storms transported O3-rich air parcels from the
mid-troposphere to the surface and abruptly enhanced the regional atmospheric boundary
layer by as much as 25 ppbv. In contrast to the individual storms, days with multiple con-
vective systems produced successive, cumulative ground-level O3 increases. The magnitude
of O3 enhancements depended on the vertical distribution of O3 within storm downdrafts
and origin of downdrafts in the troposphere. Ozone mixing ratios remained enhanced for
∗Corresponding author
Email address: tobias.gerken@psu.edu (Tobias Gerken)
Preprint submitted to Atmospheric Environment October 23, 2015
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
>2 hours following the passage of storms, which enhanced chemical processing of rainforest-
emitted isoprene and monoterpenes. Reactions of isoprene and monoterpenes with O3 are
modeled to generate maximum hydroxyl radical formation rates of 6× 106 radicals cm−3 s−1.
Therefore, one key conclusion of the present study is that downdrafts of convective storms
are estimated to transport enough O3 to the surface to initiate a series of reactions that
reduce the lifetimes of rainforest-emitted hydrocarbons.
Keywords: Isoprene, monoterpenes, air chemistry, convection, mesoscale
convective storms
1. Introduction1
Ozone (O3) mixing ratios in the atmospheric boundary layer of the remote Amazon region2
remain mostly below 40 parts per billion on a per volume basis (ppbv) during the wet period3
(e.g., Bela et al., 2015; Gregory et al., 1988). Low O3 mixing ratios result from the reduced4
photochemistry associated with copious amounts of cloud cover and precipitation that limit5
actinic irradiance reaching near the surface (Gu et al., 2002). In the absence of biomass6
burning, nitrogen oxides (NOX = nitric oxide (NO) plus nitrogen dioxide (NO2)), which are7
the key O3 precursors, only exist in trace mixing ratios in the atmospheric boundary layer8
(Torres and Buchan, 1988) because local soil emissions are small due to characteristically9
moist conditions. In the troposphere over the Amazon, O3 mixing ratios monotonically10
increase with altitude (Kirchhoff et al., 1990). Ozone profile changes result from convection11
(e.g., Folkins et al., 2002; Folkins and Martin, 2005), enhanced formation in the upper12
troposphere due to the presence of NOX from lightening (Goldenbaum and Dickerson, 1993;13
Thompson et al., 1997; Zhang et al., 2003), and surface deposition (Sigler et al., 2002).14
Deep and moist convection is an important feature governing the distribution of trace15
gases and aerosols in the tropical troposphere (Garstang et al., 1988; Scala et al., 1990).16
Due to the internal dynamics of convective storms that have well defined updraft and down-17
draft regions, trace gases undergo frequent vertical redistribution in the troposphere. Within18
the storm updraft, boundary layer air is swiftly transported to the free troposphere where19
reactions of hydrocarbons, NOX, and free radicals (such as hydroxyl radical, HO) can en-20
2
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
hance oxidant and aerosol formation (Bertram et al., 2007). In contrast, within the storm21
downdraft, mid-tropospheric air parcels are transported towards the ground surface with22
reduced moisture content, lower equivalent potential temperature (θe) (Zipser, 1969, 1977),23
and higher (≥ 30 ppbv) O3 content (Betts et al., 2002). While earlier studies (Garstang24
et al., 1988; Thompson et al., 1997; Pickering et al., 1996) examined the transport of trace25
gases from the surface into the free troposphere by mesoscale convective systems (MCS),26
more recent investigations (e.g., Betts et al., 2002; Grant et al., 2008; Hu et al., 2010; Sahu27
and Lal, 2006) indicated that convective-type storms also transport O3-rich air from the28
mid-troposphere to the Earth’s surface. In the areas impacted by the downdraft region of29
convective storms, driven by evaporative cooling from falling precipitation, ground-level O330
can abruptly increase due to the rapidly moving air parcels from layers aloft where O3 levels31
are much higher than at the surface. Ground-level O3 can increase by up to 30 ppbv and32
the magnitude of these O3 enhancements depends on the propagation velocity of MCS and33
origin of downdrafts within storms (Grant et al., 2008) and the tropospheric O3 distribution,34
which can exhibit considerable day-to-day vertical variation in the Amazon (Kirchhoff et al.,35
1990).36
In the Amazon region, maximum cloud cover and precipitation occur in the afternoon and37
early evening (Machado et al., 2004). Storms, with rainfall rates that can exceed 50 mm h−1,38
vary in their organization from locally occurring systems to basin scale convection and squall39
lines (Greco et al., 1990). As a consequence, storm downdraft dynamics depend on the overall40
state and type of the storm systems. In turn, the ground-level O3 enhancements associated41
with storms depend on the type of mesoscale convective systems. To date, these dynamics42
and their impact on atmospheric chemistry near the surface have rarely been explored.43
The regional enhancement of O3 in the atmospheric boundary layer can influence several44
atmospheric chemical cycles. For example, during the daytime O3 can undergo photodisso-45
ciation to produce excited oxygen atoms (O(1D)) that readily react with water vapor (H2O)46
to form HO (see reactions 1 and 2). The HO initiates the oxidation of biogenic volatile47
organic compounds (BVOCs), methane (CH4), and carbon monoxide (CO) (reactions 3 and48
3
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
4). Combined with the enhanced sink due to surface deposition, chemical reactions involv-49
ing the O3 transported from the mid-troposphere reduce the greenhouse gas loading in the50
boundary layer (Hu et al., 2010). In addition, elevated O3 mixing ratios can accelerate the51
oxidation rates of plant-emitted isoprene (C5H8), monoterpenes (C10H16), and sesquiterpenes52
(C15H24). Reactions of BVOCs with O3 generate free radicals, acids, carbon dioxide, and HO53
(reactions 5–7). These photo-oxidation processes remain poorly understood but are of great54
importance in the Amazon basin because of the unusually strong source of BVOCs whose55
reaction products can lead to the formation of secondary organic aerosols (SOA) (Martin56
et al., 2010b) which can activate into cloud condensation nuclei (CCN) and thus influence57
cloud formation processes.58
O3 + hν→O2 + O(1D) (1)
59
O(1D) + H2O→HO + HO (2)
60
CH4 + HO→CH3 + H2O (3)
61
CO + HO→H + CO2 (4)
62
C5H8 + O3→Products + HO (5)
63
C10H16 + O3→Products + HO (6)
64
C15H24 + O3→Products + HO (7)
Thus, the present work determines the magnitude and the frequency of ground-level O3 en-65
hancements associated with convective systems in the central Amazon region of Brazil during66
April 2014 to January 2015. An additional objective is to investigate the storm attributes67
associated with downdraft events that enhanced ground-level O3. We also investigate the du-68
ration and the influences of ground-level O3 enhancements on atmospheric chemistry above69
the rainforest. Yields of HO from reactions entailing BVOCs and O3 are estimated to verify70
whether convective storms can measurably impact BVOC oxidation.71
4
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
2. Methods72
2.1. Study site characteristics73
The data used in this work were gathered during a field campaign at the Cuieiras Bio-74
logical Reserve (S 2◦36.11’, W 60◦12.56’; managed by the Brazilian Institute of for Amazon75
Research – INPA) during April 2014 to January 2015 (Fuentes et al., in review). The study76
was part of the GoAmazon 2014/5 (Observations and Modeling of the Green Ocean Amazon)77
project, which designated the study site as T0k (also commonly referred to as ZF2). The78
reserve is located approximately 60 km north-north west of the City of Manaus, Amazonas,79
Brazil (Fig. 1). Due to prevailing easterly winds, air masses arriving at the site mostly come80
from areas with little human influence and show low aerosol concentrations (Martin et al.,81
2010a; Artaxo et al., 2013). However, the study site occasionally experiences the influences82
of Manaus’ urban plume and biomass burning events that occur frequently during the dry83
season and are caused by both natural (e.g., lightning) and anthropogenic processes (e.g.,84
deforestation). The vegetation surrounding the site is a closed primary forest with maximum85
tree heights of 30–40 m. Leaf area index is typically estimated between 6.1 and 7.3 m2 m−286
(Marques Filho et al., 2005; To´ta et al., 2012) depending on measurement technique and87
location. The forest covers valleys and plateaus with altitude differences of circa 60 m. Soils88
in the valleys are more sandy, and vegetation in the valleys tends to be less dense and tall.89
A 50-m tower (known as K34) was used as a platform to deploy measurement systems. The90
tower is located on the top of a plateau where canopy height was approximately 36 m.91
2.2. Ozone and meteorological measurements92
Nine sonic anemometers (model CSAT3, Campbell Scientific Inc, Logan, UT) were de-93
ployed on the tower to measure the three components of the wind speed (u, v, and w) and94
the sonic virtual temperature. The sonic anemometer located at the canopy top (35 m) was95
used to determine wind speed and direction. Measurement frequency was 20 Hz and data96
were processed to generate 30-minute averages. Relative humidity and temperature (model97
HMP-155, Vaisala, Vantaa, Finland) were measured at 32 m above ground. Rainfall mea-98
5
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
Figure 1: Map of South America and the central Amazon indicating the location of the study site (ZF2) in
relationship to the cities of Manaus and Manacapuru. The black lines mark roads.
surements (model RIMB500, McVan Instruments, Melbourne, Australia) from a site located99
3.11 km away from the flux tower are used in this analysis.100
Ambient O3 levels were measured (model 49i, Thermo Fisher Scientific Inc., Waltham,101
MA) at a frequency of 1 Hz. A Proton Transfer Reaction Mass Spectrometer (PTR-MS,102
Ionicon Analytik, Innsbruck, Austria) was used to measure isoprene and monoterpene mixing103
ratios. The air sampling intake for both instruments was placed at 40 m above ground. The104
inlet was placed below a rain shield and had a filter holder with a 1µm pore size Teflon105
membrane to exclude dust and pollen. The membrane was changed weekly. The O3 analyzer106
was housed in an environmentally controlled shed located at the ground level, approximately107
5 m from the tower. A pump moved the air at 12 liters per minute from the inlet to the108
analyzer via a 3/8-inch (9.8 mm) outer diameter Teflon tube. The O3 analyzer was calibrated109
before deployment and regularly zeroed to ensure high data quality.110
Ozone data were averaged to produce 5- and 30-minute quantities. Air temperature (T ),111
water vapor mixing ratio (r), and barometric pressure (p) were used to compute values of112
6
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
equivalent potential temperature (θe)113
θe ≈
(
T +
Lv
cpd
r
)(
p0
p
) Rd
cpd
, (8)
with the latent heat of vaporization (Lv), the dry heat capacity of air (cpd), the gas constant114
of dry air (Rd) and a reference pressure (p0) of 1000 hPa, to identify the periods when storm115
downdrafts occurred at the study site.116
An automated algorithm detected O3 enhancement events that had to satisfy the criteria117
of an increase in 5-minute averaged O3 mixing ratio of at least 2.5 ppbv and a simultaneous118
decrease in θe of 2.5 K within a one-hour time window. Local minimum and maximum values119
of O3 within the time window were used to determine start and end times of the enhancement120
events. After this automatic method, false detections were manually identified and removed.121
The average duration of O3 enhancement events was calculated to determine the periods122
when the atmospheric boundary layer was regionally enhanced with O3.123
The GOES-13 (Geostationary Operational Environmental Satellites) satellite data were124
acquired through the Brazilian National Institute for Space Research (INPE) and used to125
investigate large-scale convection dynamics and extent. In addition, the Sistema de Protec¸a˜o126
da Amazoˆnia (SIPAM) operates a network of S-band (10-cm wavelength), Doppler radars127
with 500 m gate spacing and a 1.8◦ beam width across the Amazon. One of SIPAM’s radars is128
located at the Manaus airport (03◦08.98’ S, 59◦59.48’ W, elevation 102.4 m). The operational129
scan strategy had 17 elevation angles every 12 minutes and data were recorded for scans130
extending 250 km from the radar site. The radar data were analyzed to develop the three-131
dimensional structure of storms impacting the ZF2 site.132
3. Results and discussion133
3.1. Seasonal ozone dynamics134
Ozone mixing ratios at the rainforest study site differed between the wet and dry months135
(Fig. 2). After the end of the wet months (precipitation> 5 mm d−1) in June, O3 mixing136
ratios started to increase as revealed by median and maximum mixing ratios. Maximum137
7
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
O3 mixing ratios occurred in August and coincided with the maximum period of biomass138
burning in the Amazon (e.g., Kuhn et al., 2010; Artaxo et al., 2013). From the start of the139
wet months in October, median O3 mixing ratios declined while their maxima continued140
to rise until November. During December and January, O3 mixing ratios approached the141
values observed at the beginning of the field experiment. While monthly maximum O3 levels142
exceeded 35 ppbv during June to December, the overall frequency distribution indicated that143
these occurrences of elevated O3 were infrequent and represented less than 2.5 % of the time144
and less than 5 % during dry months. During the wet months, the median O3 mixing ratio145
remained close to 10 ppbv with the exception of the months of October and November.146
A M J J A S O N D J
0
20
40
60
80  a)
 O
zo
ne
 [
pp
bv
]
0 10 20 30 40 50 60 70 80 90
0
0.02
0.04
0.06
0.08
0.1
 b)  Dry months
 F
re
qu
en
cy
0
0.2
0.4
0.6
0.8
1
0 10 20 30 40 50 60 70 80 90
0
0.02
0.04
0.06
0.08
0.1
 c)  Wet months
 F
re
qu
en
cy
 Ozone [ppbv]
0
0.2
0.4
0.6
0.8
1
Figure 2: Ozone mixing ratios during April 2014 to January 2015: (a) Monthly ozone mixing ratios in
ppbv during April 2014 to January 2015. Gray circles indicate minimum and maximum ozone mixing ratios.
Black diamonds denote the mean, the boxplot shows lower-quartile, median and upper-quartile ozone mixing
ratios. The whiskers indicate 10th and 90th percentiles. Frequency distributions of 30-minute averaged ozone
mixing ratios during (b) dry months (Jun–Sept) and (c) wet months. The red lines (right axis) indicate the
cumulative frequency distribution.
Ozone levels showed a clear diurnal variation in response to source and sink processes147
(Fig. 3). Ozone mixing ratios decreased during the night as a result of surface deposition and148
8
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
chemical reactions with plant-emitted hydrocarbons that were mostly confined in the shallow149
and stable nocturnal boundary layer (Jardine et al., 2011; Bouvier-Brown et al., 2009). After150
sunrise, O3 mixing ratios increased due to photochemical formation and entrainment of O3-151
rich air into the boundary layer and to the forest. During the dry months, which included152
June to September, mean O3 mixing ratios were higher than during the wet months (20153
versus 15 ppbv). Air masses arriving at the study site most often originated from clean areas154
where NOx remained in limited levels to contribute to O3 formation. The influence of clean155
air masses was evident in both averaged wet and dry month values that were much lower than156
O3 mixing ratios often recorded near populated areas. The lower O3 mixing ratios during the157
wet months (i.e., 10–15 ppbv) resulted in response to increased cloud cover that attenuated158
the actinic irradiance to drive photochemical processes. Similarly, during the wet months159
maximum diurnal O3 mixing ratios occurred earlier (plateauing 11:00–15:00 local time –160
LT) than during the dry months (peak at 16:00 LT). During the dry months forest fires and161
biomass burning increased NOx concentrations and consequently contributed to enhanced162
O3 formation. Despite its remote location, the study site also experienced some episodes163
of high O3 levels during the wet months. Overall, such episodes were not uncommon and164
provided evidence of the increasingly important influences of human activities on the remote165
rainforest of the Amazon.166
To investigate the potential influence of pollution plumes originating from Manaus and167
biomass burning associated with human settlements between Manaus and the study site on168
near-surface O3, daytime (06:00 – 18:00 LT) O3 mixing ratios were segregated as a function169
of wind direction. During the dry months, elevated O3 mixing ratios were mostly associated170
with air masses coming from south-easterly sectors. For such wind directions, the mean171
O3 mixing ratio was 20 ppbv. Compared to the wet months, there were also more events172
during which O3 levels exceeded one standard deviation from the mean. Similarly, these173
events occurred mostly during winds from southerly to easterly directions and reached O3174
levels exceeding 80 ppbv during one event (Fig. 3 c + d). These high levels of O3 indicate175
the influence of pollutants originating from the City of Manaus to the south-southeast.176
9
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
Increased human settlements along the highway from Manaus to the site, which is located to177
the east, could also contribute to the northward extension of forest fire plumes and additional178
pollution from human activities. In contrast, during the wet months mean O3 mixing ratio179
only reached 10–15 ppbv for all sectors.180
00:00 04:00 08:00 12:00 16:00 20:00 00:00
0
10
20
30
40
  Dry months a)
 O
zo
ne
 [
pp
bv
]
 Time of Day [LT]
0
250
500
750
1000
00:00 04:00 08:00 12:00 16:00 20:00 00:00
0
10
20
30
40
  Wet months b)
 Time of Day [LT]
0
250
500
750
1000
So
la
r 
ir
ra
di
an
ce
 [
W
 m
−
2 ]
5
10
30
50
100
 N
 S
 W  E
 c)
5
10
30
50
100
 N
 S
 W  E
 d)
Figure 3: Average diel cycle of 30-minute mean ozone mixing ratios (red lines) as well as solar irradiance
(black dashed lines) for (a) dry and (b) wet months. Standard deviation is indicated by bars (ozone) and
thin dashed lines (solar irradiance). Subplots (c) and (d) display the mean daytime (06:00 LT – 18:00 LT)
ozone mixing ratio based on wind direction. The standard deviation of daytime O3 mixing ratios for each
sector is indicated with the blue diamonds and red crosses show values exceeding one standard deviation.
3.2. Case studies of ozone enhancements181
Due to its tropical location, the Amazon region experiences frequent occurrences of deep182
and moist convection. In addition to precipitation, storms are associated with downdrafts183
capable of generating substantial mass fluxes from the mid-troposphere to the surface. As184
air masses descend, O3 in the atmospheric boundary layer can abruptly increase (e.g., Betts185
10
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
et al., 2002; Grant et al., 2008) which can be detected by ground-level measurements. Two186
cases are provided to illustrate the unique signatures of O3 enhancements associated with187
convective storms of varying degrees of organization.188
3.2.1. Single storm event189
On 19 June 2014, an isolated MCS with a leading convective line and trailing stratiform190
precipitation passed through the region around 9:30 LT (Fig. 4 a). The fast moving system191
produced a precipitation peak of approximately 6 mm within a 30-minute interval, during192
which the convective line moved over the measurement site with weaker stratiform rain193
persisting for a few hours thereafter (Fig. 5). During the passage of the convective line,194
the surface O3 mixing ratio increased from 5 to more than 20 ppbv with a corresponding195
decrease in θe of approximately 9 K. Equivalent potential temperature is a conserved quantity196
in updrafts and downdrafts, indicating that the increase in O3 was caused by the downward197
transport of air from higher levels in the troposphere. This convective system occurred in198
the late morning hours and regionally enhanced the atmospheric boundary layer with O3199
mixing ratios of 10 ppbv for the rest of the day.200
3.2.2. Multi-storm day201
On 9 May 2014, convective activity was widespread across the region, but much less202
organized than on the 19 June case (Fig. 4 b–d). The site was subject to a sequence of pre-203
cipitation events of increasing strength that were separated by short periods of dry conditions204
(Fig. 6). Each successive rain event (8:00, 11:30, and 13:30 LT) corresponded to an increase205
in the O3 mixing ratio and a decrease in θe. Downdrafts associated with the first rain event206
at 8:00 LT caused an increase of O3 levels from 2 ppbv to more than 10 ppbv as well as a207
decrease in θe of approximately 4 K. However, the first event produced less than 5 mm of208
rain and radar reflectivity was weak (Fig. 4). After the event, O3 mixing ratio decreased to209
5 ppbv and θe rose in response to strong solar irradiance and photochemical activity. Also,210
the atmospheric boundary layer depth likely increased. The second precipitation event at211
11:30 LT increased surface O3 levels by 5 ppbv and a larger drop in θe than the first event.212
Rain rates were similar to the first rain event, but the duration of rainfall was longer and213
11
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
Figure 4: SIPAM S-band (10-cm wavelength) radar images at 0.9◦ elevation angle for 19 June 2014 and
9 May 2014 at the time of the O3 enhancement events. The black squares indicate the location of the
measurement site and the location of Manaus is indicated by the black circles. The storms travel from
north-east to south-west.
the convective system appeared stronger according the radar reflectivity. The third event214
at 13:30 LT produced the largest rain rates, increased O3 levels further, and decreased θe215
approximately 10 K below the maximum. This convective system was also the strongest and216
most widespread of the three events. After 15:00 LT, stratiform rain set in and O3 mixing217
ratio decreased towards characteristic nighttime levels.218
While each of the individual convective events on 9 May 2014 produced O3 increases219
in the order of 10 ppbv, in aggregate, their impacts on surface O3 were close to the single220
large increase observed on 19 June 2014. With the passing of each convective storm, verti-221
cal gradients and absolute O3 levels were reduced in the lower troposphere due to vertical222
redistribution. Therefore, the magnitude of ground-level O3 enhancements depended on223
the altitude where air parcels originated in the troposphere. The third and final event on224
9 May 2014 (Fig. 6) provided an example of a stronger downdraft emanating from a higher225
altitude to further increase ground-level O3.226
12
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT



-XQD
2
]R
QH
>S
SE
Y@



E
θ
H>
.
@
     




F
5
DL
QI
DO
O>
P
P


K
í
@
7LPHRI'D\>/7@





6
RO
DU
LU
UD
GL
DQ
FH
>:
P
 @
Figure 5: Case study of a single storm ozone enhancement on 19 June 2014: (a) Ozone mixing ratio [ppbv],
(b) equivalent potential temperature (θe, [K]), and (c) 30-minute accumulated precipitation [mm] and solar
irradiance [Wm−2] (red line). The dashed line indicates the times of the radar images in Figure 4.
13
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT





0D\D
2
]R
QH
>S
SE
Y@



E
θ
H>
.
@
     




F
5
DL
QI
DO
O>
P
P


K
í
@
7LPHRI'D\>/7@





6
RO
DU
LU
UD
GL
DQ
FH
>:
P
 @
Figure 6: Case study of a multi-storm day on 9 May 2014.: (a) Ozone mixing ratio [ppbv], (b) equiva-
lent potential temperature (θe, [K]), and (c) 30-minute accumulated precipitation [mm] and solar irradiance
[Wm−2] (red line). The dashed lines indicate the times of the radar images in Figure 4.
14
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
3.2.3. Dynamics of ozone enhancement events227
The magnitude of O3 enhancement depends on the strength of the downdraft and the228
origin of O3-enriched air parcels transported down to the surface. For the case of 19 June
Figure 7: Contoured frequency with altitude diagram (CFAD) of radar reflectivity observed by the SIPAM S-
band radar located in Manaus for 22 km×22 km box centered above ZF2 during a 12-min period, on 19 June
2014. The contours represent the reflectivity occurrence at each height and a minimum count is applied so
that the distribution at data sparse heights is not plotted.
229
2014, contoured-frequency-by-altitude diagrams (CFADs, see Yuter and Houze (1995)), show230
the storm’s evolution between 9:24 and 10:00 LT, i.e., the period when the leading convective231
line passed over the site (Fig. 7). As the leading line passed over the site (09:24 LT), echo232
tops reached 12 km but radar reflectivity close to the surface was generally less than 30 dBz.233
Twelve minutes later (09:36 LT), echo tops were greater than 14 km and an arm of strong234
reflectivity starting around 7 km descended to the surface with a large portion of reflectivity235
values greater than 40 dBz. This feature persisted into the 09:48 LT volume scan and then236
disappeared by 10:00 LT as the leading line of the storm passed over the site and a more237
stratiform distribution of reflectivity (i.e., more tightly grouped contours at lower heights and238
reflectivity values) was subsequently observed. The descending arm of strong reflectivity was239
consistent with hydrometeors growing too large to be maintained by the convective updraft,240
creating a strong downdraft and heavy rain as they fall. The descending arm also coincided241
with the timing of the O3 enhancement and θe drop (Fig. 5).242
15
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
Figure 8: Contoured frequency with altitude diagram (CFAD) of radar reflectivity observed by the SIPAM
S-band radar located in Manaus for 22 km×22 km box centered above ZF2 during a 12-min period, observing
the storm system on 9 May 2014. The contours represent the reflectivity occurrence at each height and a
minimum count is applied so that the distribution at data sparse heights is not plotted.
16
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
Figure 8 shows the evolution of the three convective systems that passed over the site on243
9 May 2014. The first event during 7:36 LT shows echo tops reaching only 8 km and surface244
reflectivity never exceeding 40 dBz, consistent with the low reflectivity values in Fig. 4 and245
low rainfall rate in Fig. 6. Similarly, no descending arm was evident, although there appeared246
to be a broadening of the reflectivity values below below 5 km, the nominal 0 ◦C level in247
the tropics, which may be indicative of re-evaporation of precipitation and thus downdrafts.248
While this event appeared to be much weaker than the 19 June 2014 case, it still produced a249
strong O3 enhancement end θe reduction. Such an increase in O3 still appears plausible, based250
on the O3 mixing ratio profile from the AIRS-Instrument (Atmospheric InfraRed Sounder)251
on the Aqua satellite of 25 ppbv at approximately 2000 m above ground level. The second252
event during 11:00 LT had higher echo tops (note that upper level radar data is missing253
during the 11:48 LT scan) and stronger surface reflectivity than during 8:00 LT. Once again,254
a weak and more shallow descending arm occurred as the convective system passed over the255
site. The third and strongest event of 9 May 2014 had consistently high echo tops and the256
most robust descending arm of the case studies that appeared to be linked to processes above257
the 0 ◦C level.258
3.2.4. Contrasting local and basin-scale convection259
The local effects from the passage of convective systems are associated with the large-260
scale storm dynamics (Fig. 9). On 19 June 2014 (Fig. 9 a) a small, single MCS propagated261
quickly through the basin and over the site. Conversely, a large basin scale system was262
present on 9 May 2014 (Fig. 9 b), so that O3 enhancements likely affected a much larger area263
of the Amazon basin. However, despite the small spatial extent of individual storms, once264
occurring simultaneously throughout the basin, these storm types can also substantially and265
regionally increase ambient O3 levels in the atmospheric boundary layer.266
In addition to the regional extent, large-scale dynamics also impact the photoxidation267
process discussed in section 3.4. Basin scale convection, produces persistent cloud cover,268
lowering the solar irradiance for extended periods of time, which can reduce actinic fluxes269
and the photoxidation of the BVOC. In contrast, local convection events are often followed270
17
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
by clear conditions shortly after the enhancement of O3 allowing for photoxidation to occur271
at high rates. Consequently, the type of convection as well as its spatial distribution within272
the Amazon basin, influence not only atmospheric chemistry within the atmospheric bound-273
ary layer, but can also constitute a mechanism of O3 downward transport and subsequent274
destruction on the basin scale.275
Figure 9: GOES-13 channel 4 images illustrating large-scale convection in the Amazon basin. (a)
19 June 2014 1400 UTC (10:00 LT), (b) 9 May 2014 1400 UTC (10:00 LT). Overlaid colors show cloud top
brightness temperatures below -30 ◦C. The cross indicates the location of the research site.
3.3. Ozone enhancement event characterization276
In the tropical troposphere, O3 mixing ratios increase with altitude while θe decreases.277
During the wet season near Manaus, Kirchhoff et al. (1990) found an average O3 gradient of278
6 ppbv km−1 within 3–5 km of the troposphere. Such O3 vertical distribution resulted in a279
positive correlation between the strength of O3 enhancements and the magnitude of change280
in θe (Fig. 10). Between 5 and 10 km, O3 remained almost invariant with height (Kirchhoff281
et al., 1990). However, upper tropospheric O3 mixing ratio for the same altitude exhibited282
large temporal variability (sometimes exceeding 25 ppbv) due to interactions between con-283
vection and vertical O3 distribution (e.g., Folkins et al., 2002; Folkins and Martin, 2005) and284
increased O3 formation from lightning-produced NOx (Goldenbaum and Dickerson, 1993;285
Thompson et al., 1997; Zhang et al., 2003). Those variations in O3 profiles and θe gradients286
18
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
were likely the reasons for the observed scatter in the relationship of O3 and θe changes.287
Furthermore, they may also help to explain the different magnitude of O3 enhancements288
associated with single (e.g., 19 June) versus multiple storm events (e.g., 9 May).289
The magnitude of O3 enhancements depended on the time of day when storms occurred.290
During 2014, more daytime than nighttime storm events occurred (Fig. 10). This observation291
is in agreement with the preferential timing of convective precipitation during late morning292
and early afternoon hours. The nighttime response of O3 to changes in θe appeared to be293
stronger than during the daytime, which could be in response to temporal O3 variability294
in the atmospheric boundary layer. After sunset, O3 in the stable boundary layer becomes295
mostly depleted due to dry deposition and chemical reactions while O3 above the boundary296
layer does not exhibit much diurnal variability (Kirchhoff et al., 1990). As a result, large O3297
gradients likely developed between air in the stable boundary layer and the mid-troposphere,298
where the downdrafts originated. Conversely, during the daytime the net effect of O3-rich299
downdrafts was less pronounced as photochemical processes increased O3 levels in the bound-300
ary layer, thereby likely yielding reduced O3 gradients between the convective boundary layer301
and the mid-troposphere. The convective boundary layer, whose depths typically reach be-302
tween 800 and 1200 m, remained well mixed with respect to O3. While there seemed to303
be little difference in the strength of O3 enhancements between days with the passage of304
single or multiple convective systems, strong enhancements (∆O3 > 15 ppb) for days with305
multiple events were rare. This finding seems to be counter-intuitive, as more organized and306
basin scale convection was associated with much stronger precipitation as well as downdraft307
strengths. However, the repeated passage of convective systems caused increased vertical308
mixing of mid-tropospheric air towards and into the boundary layer such that mixing ratio309
gradients between near surface O3 and upper levels were reduced. It thus diminished the310
overall magnitude of individual events associated with the more organized and basin scale311
convection. Storm downdrafts increased O3 between 5 and 25 ppbv (Fig. 10 b).312
The length of time when the atmospheric boundary layer remains enriched with O3 is313
needed to investigate whether O3 enhancements significantly influence regional air chem-314
19
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
    







D
&KDQJHLQθ
H
>.@
&
KD
QJ
H
LQ
R
]R
QH
>S
SE
Y@


'D\6LQJOH
1LJKW6LQJOH
'D\0XOWLSOH
1LJKW0XOWLSOH
      









E (YHQWV 
'D\ 
1LJKW 
)
UH
TX
HQ
F\
2]RQH>SSEY@
Figure 10: Aggregated changes in ozone during enhancement events from April 2014 to January 2015:
(a) Change in ozone with respect to change in equivalent potential temperature for daytime (yellow) and
nighttime (blue) cases as well as single (circle) and multiple events (triangle); (b) Frequency distribution of
ozone events.
istry. Prior to the event onset, daytime O3 mixing ratios decreased while θe did not change315
significantly (as illustrated by an extreme case in Fig. 11 a+b) in response to atmospheric316
boundary layer air swiftly moving with storm updrafts. Stronger winds and updrafts trans-317
ported warm, moist, and O3-depleted in-canopy air to sensor level above the canopy, thus318
increasing θe and decreasing O3 for short periods of time. During the nighttime, there was319
no corresponding increases in θe due to reduced temperature. This can be explained by the320
difference of in-canopy gradients of O3 mixing ratio and temperature as well as moisture,321
which dominate the observed changes in θe.322
The median nighttime change in θe was smaller than the daytime change (approximately323
-3 and -5K), as nighttime values of θe are generally smaller and have less temporal vari-324
ation. Shortly before the start of the event, O3 mixing ratios decreased during daytime325
conditions, while θe remained virtually unchanged. This was due to the dynamics of the326
passing storms. While chemical reactions and dry deposition cause noticeable O3-gradients327
within the canopy, the in-canopy air was much better mixed with respect to temperature328
and humidity. Subsequently, O3 mixing ratios and θe started to change due to storm down-329
20
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
drafts. As storms propagated, downdrafts dominated the transport process. On average,330
nighttime median changes of O3 were greater than daytime changes with approximately 7.5331
and 4 ppbv, respectively. Maximum O3 levels were reached after circa 1 h and 50minutes332
for nighttime and daytime events followed by a gradual decline of O3 mixing ratios during333
daytime conditions, which was likely due to chemical reactions with BVOCs. For nighttime334
events, increases in O3 persisted for two hours.335
The increase of the median O3 levels for all events exceeded 5 ppbv, while 25% of all336
events exceeded 9 ppbv. Given the range of O3 changes between 1 and 10 ppbv and typical337
ambient O3 mixing ratios between 5 and 25 ppb, the impact of these events on air chemical338
processes cannot be neglected, especially as their influence on O3 levels was still detectable339
for at least two hours after the onset of the event and thus had the potential to affect chemical340
reactions for extended periods of time. Similarly, Sigler et al. (2002) have shown that there341
is a difference in the diurnal cycle of O3 mixing ratios for days with and without the passage342
of convective systems.343
í í     
í




2
]R
QH
>S
SE
Y@ F 1LJKW
í í     
í
í


G 1LJKW
θ
H>
.
@
í í     
í




H 'D\
7LPHVLQFHHYHQWVWDUW>K@
í í     
í
í


I 'D\
7LPHVLQFHHYHQWVWDUW>K@
í í     
í





D
í í     
í
í


E
Figure 11: Changes in ozone and equivalent potential temperature (θe) during ozone enhancement events:
1-minute averages for (a) ozone and (b) θe on 22 Aug 2014, 20:55 LT; (c+d) median change for nighttime
events; and (e+ f) daytime events from 5-minute averaged data. The colored patches give the interquartile
range of events.
21
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
3.4. Influences on atmospheric chemistry344
The rainforest emits numerous hydrocarbon species including isoprene, monoterpenes345
(e.g., α-pinene), sesquiterpenes (e.g., caryophyllene), and oxygenated compounds (e.g., methanol).346
At the study site, the most abundant and important hydrocarbons measured above the rain-347
forest included isoprene and monoterpenes whose daily maximum mixing ratios reached 15348
and 2 ppbv, respectively (Fuentes et al., in review). Isoprene and monoterpenes react with349
O3 and the reactions provide an important source of HO (Paulson and Orlando, 1996), which350
can subsequently enhance the oxidation of rainforest-emitted hydrocarbons. To determine351
the magnitude of HO formation from reactions of O3 with BVOCs, the HO formation rates352
were computed following equation 9.353
∂[OH]
∂t
=
N∑
i
[BVOC]iKi [O3] Yieldi (9)
N denotes the total number of identified hydrocarbons above the rainforest, [BVOC]i repre-354
sents the concentration (molecules cm−3) of speciated hydrocarbon, i (i.e., isoprene, α-pinene,355
etc.), Ki is the reaction rate coefficient (cm
3 molecules−1 s−1) of [BVOC]i with O3, [O3] is the356
O3 concentration (molecules cm
−3), and Yieldi is the HO yield (unitless) from the reaction357
of [BVOC]i with O3. The speciated monoterpenes measured above the forest (Jardine et al.,358
2015), associated O3 reactivities, and HO yields are included in Table 1. Following storms,359
isoprene and monoterpene mixing ratios exhibited an interquartile ranges of 1.5 to 3.0 and360
0.3 to 0.7 ppbv.361
Ozone enhancements due to storms resulted in substantial HO production rates. For362
example, an O3 enhancement from 5 to 30 ppbv would increase HO production rates from363
0.15–0.29×105 free radicals cm−3 s−1 to 0.88–1.75×105 free radicals cm−3 s−1 for the range364
of 1.5 to 3.0 pbbv of isoprene (Fig. 12 a). Likewise, HO formation rates increase from 0.18–365
0.42×106 free radicals cm−3 s−1 to 1.08–2.53×105 free radicals cm−3 s−1, for a monoterpene366
range of 0.3 to 0.7 ppbv (Fig. 12 b).367
These HO formation rates closely compare with estimates obtained for regions support-368
ing intense summertime photochemical activity dominated by enhanced NOX and BVOC369
22
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
levels (Fuentes et al., 2007). Ozone enhancements also substantially reduced the lifetimes of370
isoprene and monoterpenes, defined as τIsop = (KIsop [O3])
−1 and τMono =
∑N
i (Ki [O3])
−1.371
For instance, an O3 enhancements from 10 to 40 ppbv caused τIsop to decline from 90 to 20372
hours whereas τMono decreased from 40 to 8 hours (Fig. 12 c). The reductions in isoprene373
and monoterpene lifetimes (Fig. 12) do not include the sinks resulting from reactions with374
HO. Nonetheless, in environments such as the rainforest, where BVOCs exist in elevated375
concentrations (e.g., Jardine et al., 2015), O3 initiates a chain of chemical reactions involv-376
ing BVOCs that in the process generate not only abundant amounts of HO but also reaction377
products which serve as precursors of secondary organic aerosols.378
23
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
0 10 20 30 40 50
0
1
2
3
4
5
6
7
x 10
6
 H
O
 f
or
m
at
io
n 
[r
ad
ic
al
s 
cm
−
3  
s−
1 ]
 
 
 b)
[Monpterpene] = 0.25 ppbv
[Monpterpene] = 0.50 ppbv
[Monpterpene] = 0.75 ppbv
[Monpterpene] = 1.00 ppbv
0 10 20 30 40 50
0
0.5
1
1.5
2
2.5
3
x 10
5
 H
O
 f
or
m
at
io
n 
[r
ad
ic
al
s 
cm
−
3  
s−
1 ]
 
 
 a)
[Isoprene] = 1.5 ppbv
[Isoprene] = 2.0 ppbv
[Isoprene] = 2.5 ppbv
[Isoprene] = 3.0 ppbv
0 10 20 30 40 50
   1
  10
 100
 Ozone [ppbv]
 L
if
et
im
e 
[h
]
 
 
 c)
Isoprene
Monoterpenes
Figure 12: Hydroxyl radical formation rates from the oxidization of (a) isoprene with ozone, (b) monoterpenes
with ozone and (c) isoprene and monoterpene lifetimes as a function of ambient ozone mixing ratios.
24
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
09:00 10:00 11:00
0
5
10
15
20
25
O
zo
ne
 [
pp
bv
]
 a)
09:00 10:00 11:00
0
0.5
1
1.5
2
2.5
Is
op
 &
 M
on
ot
 (
× 
3)
 [
pp
bv
]
 b)
09:00 10:00 11:00
0
1
2
3
4
5
6
7
x 10
4
H
O
 (
Is
op
) 
[r
ad
ic
al
s 
cm
−
3  
s−
1 ]
 c)
Time of Day [LT]
09:00 10:00 11:00
0
2
4
6
8
10
12
x 10
5
H
O
 (
M
on
ot
) 
[r
ad
ic
al
s 
cm
−
3  
s−
1 ]
 d)
Time of Day [LT]
Figure 13: Hydroxyl radical formation rates [radicals cm−3 s−1] during the single storm day (19 June 2014) in
response to (a) ozone, (b) isoprene (red) and monoterpene (yellow) mixing ratios for oxidation of (c) isoprene
and (d) monoterpenes. Monoterpene mixing ratios were multiplied by a factor of 3 for scale reasons.
25
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
The passage of the storm on 19 June 2014 provides one example of the impact of O3379
enhancement events on HO formation rates (Fig. 13). The increase in O3 of this event380
compensated for the decrease in isoprene and monoterpene levels, leading to increased HO381
formation rates during and after the storm. The increase HO formation rates coincided with382
a period of low actinic irradiance (as deduced from incoming solar irradiance).383
Table 1: Ozone reactivities and HO yields for the speciated hydrocarbons found over the rainforest at the
Cuieiras Biological Reserve during the wet months of 2014.
Chemical species Ozone reactivities HO yield Reference
[cm3 molecules−1 s−1]ψ [Unitless]
Isoprene 1.30 x 10−17 0.25 (Paulson et al., 1998)
α-pinene 2.35 x 10−17 0.83 (Rickard et al., 1999)
β-pinene 3.79 x 10−17 0.35 (Atkinson et al., 1992)
3-Carene 2.35 x 10−17 0.86 (Aschmann et al., 2002)
Camphene 9.00 x 10−19 0.18 (Atkinson et al., 1992)
Limonene 2.50 x 10−16 0.67 (Aschmann et al., 2002)
Sabinene 2.50 x 10−16 0.33 (Aschmann et al., 2002)
β-Myrcene 4.70 x 10−16 0.63 (Aschmann et al., 2002)
α-Ocimeme 5.40 x 10−16 0.55 (Aschmann et al., 2002)
Terpinolene 1.90 x 10−15 0.74 (Aschmann et al., 2002)
γ-Terpinene 1.40 x 10−16 0.81 (Aschmann et al., 2002)
ψ: Kinetics data are from (Atkinson et al., 2006).
4. Summary and conclusions384
Convective storms are ubiquitous features that redistribute O3 in the tropical tropo-385
sphere. In the Amazon region, convective downdrafts frequently transport air parcels to386
the surface that provide distinct signatures of O3 enhancements. Isolated storms generate387
26
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
single and abrupt O3 enhancements whereas convectively active days produce multiple and388
successive O3 increases. Depending on the strength of storm downdrafts and height from389
which air parcels emanate, O3 enhancements can reach as much as 30 ppbv. Such O3 en-390
hancements are associated with rapid declines in θe, ranging from 3 to 20 K. Radar and θe391
data indicate that air parcels, which transport O3-rich air to the surface originate 2–7 km392
above the ground. While there are comparatively few O3 enhancement events during the393
nighttime, these events typically lead to greater O3 enhancements due to the O3 difference394
in the atmospheric surface layer and the free troposphere. Following convective storms,395
the atmospheric boundary layer can be enriched with O3 for periods lasting >2 hours, the396
first hour experiencing the greatest O3 enhancements. Due to their long lifetime and large397
size, basin scale systems increase cloud cover which reduces photochemical processes for ex-398
tended times, whereas local systems are often followed by clear conditions and thus enhance399
photochemistry.400
Convective storms sufficiently enhance the atmospheric boundary layer with O3 to ini-401
tiate a chain of chemical reactions involving rainforest-emitted hydrocarbons. Reactions of402
O3 with hydrocarbons generate substantial amounts of HO which additionally contribute to403
the oxidation of hydrocarbons. In response to reactions of O3 with prevailing monoterpenes404
and isoprene, maximum HO formation rates can reach 6x106 radicals cm−3 s−1. Moreover, O3405
enhancements can substantially reduce the lifetimes of isoprene and monoterpenes. There-406
fore, one key conclusion derived from this investigation is that, in the Amazon rainforest,407
convective systems can transport sufficient O3 to the canopy to drive oxidation of rainforest-408
emitted hydrocarbons. It is hypothesized that O3 enhancement events can thus accelerate409
the production of SOAs which can activate into cloud condensation nuclei, and thereby might410
influence cloud formation.411
5. Acknowledgments412
The U.S. Department of Energy Office of Biological and Environmental Research’s Cli-413
mate and Environmental Sciences Division supported the field studies as part of the GoA-414
mazon 2014/5 project (grant SC0011075). Fundac¸a˜o de Amparo a` Pesquisa do Estado de415
27
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
Sa˜o Paulo (FAPESP) and Fundac¸a˜o de Amparo a` Pesquisa do Estado do Amazonas (FA-416
PEAM) funded the Brazilian component of the field studies. We acknowledge the support417
from the Central Office of the Large Scale Biosphere Atmosphere Experiment in Amazonia418
(LBA), the Instituto Nacional de Pesquisas da Amazonia (INPA), and the Universidade do419
Estado do Amazonia (UEA). The work was conducted under 001030/2012-4 of the Brazilian420
National Council for Scientific and Technological Development (CNPq). We acknowledge421
logistical support from the ARM Climate Research Facility. Juliane Mercer and Brazilian422
students Raoni Aquino Silva de Santana, Debora Tanya, and Francisco assisted with the field423
studies. Satellite images are courtesy of Instituto Nacional de Pesiquisas Espacias (INPA)424
Divisa˜o de Satelites e Siste´mas Ambientais. Radar data are courtesy of SIPAM and Hannah425
Upton and Aaron Funk of Texas A&M University. The SIPAM radar analysis and imagery426
were completed with support from DOE ASR grant DE-SC0008561 and the Texas A&M427
University-CAPES Collaborative Research Grant Program. LATM acknowledges the sup-428
port of FAPESP (Grant 2009/15235-8). PCS acknowledges the support from the Alexander429
von Humboldt Foundation.430
References431
Artaxo, P., Rizzo, L. V., Brito, J. F., Barbosa, H. M. J., Arana, A., Sena, E. T., Cirino, G. G.,432
Bastos, W., Martin, S. T., Andreae, M. O., 2013. Atmospheric aerosols in Amazonia and433
land use change: from natural biogenic to biomass burning conditions. Faraday Discussions434
165 (0), 203–235.435
Aschmann, S. M., Arey, J., Atkinson, R., 2002. OH radical formation from the gas-phase436
reactions of O3 with a series of terpenes. Atmospheric Environment 36 (27), 4347–4355.437
Atkinson, R., Aschmann, S. M., Arey, J., Shorees, B., 1992. Formation of OH radicals in438
the gas phase reactions of O3 with a series of terpenes. Journal of Geophysical Research439
97 (D5), 6065–6073.440
Atkinson, R., Baulch, D. L., Cox, R. A., Crowley, J. N., Hampson, R. F., Hynes, R. G.,441
28
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
Jenkin, M. E., Rossi, M. J., Troe, J., IUPAC Subcommittee, 2006. Evaluated kinetic and442
photochemical data for atmospheric chemistry: Volume II – gas phase reactions of organic443
species. Atmospheric Chemistry and Physics 6 (11), 3625–4055.444
Bela, M. M., Longo, K. M., Freitas, S. R., Moreira, D. S., Beck, V., Wofsy, S. C., Gerbig, C.,445
Wiedemann, K., Andreae, M. O., Artaxo, P., 2015. Ozone production and transport over446
the Amazon Basin during the dry-to-wet and wet-to-dry transition seasons. Atmospheric447
Chemistry and Physics 15, 757–782.448
Bertram, T. H., Perring, A. E., Wooldridge, P. J., Crounse, J. D., Kwan, A. J., Wennberg,449
P. O., Scheuer, E., Dibb, J., Avery, M., Sachse, G., Vay, S. A., rawford, J. H., Mc-450
Naughton, C. S., Clarke, A., Pickering, K. E., Fuelberg, H., Huey, G., Blake, D. R., Singh,451
H. B., Hall, S. R., Shetter, R. E., Fried, A., Heikes, B. G., Cohen, R. C., 2007. Direct452
measurements of the convective recycling of the upper troposphere. Science 315 (5813),453
816–820.454
Betts, A. K., Gatti, L. V., Cordova, A. M., Silva Dias, M. A. F., Fuentes, J. D., 2002.455
Transport of ozone to the surface by convective downdrafts at night. Journal of Geophysical456
Research 107 (D20), 8046.457
Bouvier-Brown, N. C., Goldstein, A. H., Gilman, J. B., Kuster, W. C., de Gouw, J. A., 2009.458
In-situ ambient quantification of monoterpenes, sesquiterpenes, and related oxygenated459
compounds during BEARPEX 2007: Implications for gas- and particle-phase chemistry.460
Atmospheric Chemistry and Physics 9, 5505–5518.461
Folkins, I., Braun, C., Thompson, A. M., Witte, J., 2002. Tropical ozone as an indicator of462
deep convection. Journal of Geophysical Research 107 (D13), 4184.463
Folkins, I., Martin, R. V., 2005. The vertical structure of tropical convection and its impact464
on the budgets of water vapor and ozone. Journal of the Atmospheric Sciences 62 (5),465
1560–1573.466
29
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
Fuentes, J. D., Wang, D., Bowling, D. R., Potosnak, M., Monson, R. K., Goliff, W. S., Stock-467
well, W. R., 2007. Biogenic hydrocarbon chemistry within and above a mixed deciduous468
forest. Journal of Atmospheric Chemistry 56 (2), 165–185.469
Fuentes, J. D., Chamecki, M., Nascimento dos Santos, R. M., von Randow, C., Stoy, P. C.,470
Katul, G. G., Fitzjarrald, D. R., Manzi, A. O., Gerken, T., Trowbridge, A., Freire, L. S.,471
Ruiz-Plancarte, J., Furtunato Maia, J. M., To´ta, J., Dias, N. L., Fisch, G., Schumacher, C.,472
Acevedo, O. C., Mercer, J. R., Linking meteorology, turbulence, and air chemistry in the473
Amazon rainforest during the GoAmazon project. Bulletin of the American Meteorological474
Society, in review.475
Garstang, M., Scala, J., Greco, S., Harriss, R., Beck, S., Browell, E., Sachse, G., Gregory, G.,476
Hill, G., Simpson, J., Tao, W.-K., Torres, A., 1988. Trace gas exchanges and convective477
transports over the Amazonian rain forest. Journal of Geophysical Research 93 (D2),478
1528–1550.479
Goldenbaum, G. C., Dickerson, R. R., 1993. Nitric oxide production by lightning discharges.480
Journal of Geophysical Research 98 (D10), 18333–18338.481
Grant, D. D., Fuentes, J. D., DeLonge, M. S., Chan, S., Joseph, E., Kucera, P., Ndiaye, S. A.,482
Gaye, A. T., 2008. Ozone transport by mesoscale convective storms in western Senegal.483
Atmospheric Environment 42 (30), 7104–7114.484
Greco, S., Swap, R., Garstang, M., Ulanski, S., Shipham, M., Harriss, R. C., Talbot, R.,485
Andreae, M. O., Artaxo, P., 1990. Rainfall and surface kinematic conditions over central486
Amazonia during ABLE 2B. Journal of Geophysical Research 95 (D10), 17001–17014.487
Gregory, G. L., Browell, E. V., Warren, L. S., 1988. Boundary layer ozone: An airborne488
survey above the Amazon Basin. Journal of Geophysical Research 93 (D2), 1452–1468.489
Gu, L., Baldocchi, D., Verma, S. B., Black, T.A., Vesala, T., Falge, E. M., Dowty, P. R.,490
2002. Advantages of diffuse radiation for terrestrial ecosystem productivity. Journal of491
Geophysical Research 107 (D6), 4050.492
30
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
Hu, X.-M., Fuentes, J. D., Zhang, F., 2010. Downward transport and modification of tropo-493
spheric ozone through moist convection. Journal of Atmospheric Chemistry 65 (1), 13–35.494
Jardine, A. B., Jardine, K. J., Fuentes, J. D., Martin, S. T., Martins, G., Durgante, F.,495
Carneiro, V., Higuchi, N., Manzi, A. O., Chambers, J. Q., 2015. Highly reactive light-496
dependent monoterpenes in the Amazon: Amazon light-dependent monoterpenes. Geo-497
physical Research Letters 42.498
Jardine, K., Yae˜z Serrano, A., Arneth, A., Abrell, L., Jardine, A., van Haren, J., Artaxo,499
P., Rizzo, L. V., Ishida, F.Y., Karl, T., Kesselmeier, J., Saleska, S., Huxman, T., 2011.500
Within-canopy sesquiterpene ozonolysis in Amazonia. Journal of Geophysical Research501
116, D19301.502
Kirchhoff, V. W. J. H., da Silva, I. M. O., Browell, E. V., 1990. Ozone measurements in503
Amazonia: Dry season versus wet season. Journal of Geophysical Research 95 (D10),504
16913–16926.505
Kuhn, U., Ganzeveld, L., Thielmann, A., Dindorf, T., Schebeske, G., Welling, M., Sciare, J.,506
Roberts, G., Meixner, F. X., Kesselmeier, J., Lelieveld, J., Kolle, O., Ciccioli, P., Lloyd,507
J., Trentmann, J., Artaxo, P., Andreae, M. O., 2010. Impact of Manaus City on the508
Amazon Green Ocean atmosphere: Ozone production, precursor sensitivity and aerosol509
load. Atmospheric Chemistry and Physics 10 (19), 9251–9282.510
Machado, L. a. T., Laurent, H., Dessay, N., Miranda, I., Apr. 2004. Seasonal and diurnal511
variability of convection over the Amazonia: A comparison of different vegetation types512
and large scale forcing. Theoretical and Applied Climatology 78 (1-3), 61–77.513
Machado, L. a. T., Rossow, W. B., Guedes, R. L., Walker, A. W., Jun. 1998. Life cycle514
variations of mesoscale convective systems over the Americas. Monthly Weather Review515
126 (6), 1630–1654.516
Marques Filho, A. d. O., Dallarosa, R. G., Pacheco, V. B., 2005. Radiac¸a˜o solar e distribuic¸a˜o517
31
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
vertical de a´rea foliar em floresta – Reserva Biolo´gica do Cuieiras – ZF2, Manaus. Acta518
Amazonica 35 (4), 427–436.519
Martin, S. T., Andreae, M. O., Althausen, D., Artaxo, P., Baars, H., Borrmann, S., Chen,520
Q., Farmer, D. K., Guenther, A., Gunthe, S. S., Jimenez, J. L., Karl, T., Longo, K.,521
Manzi, A., Mu¨ller, T., Pauliquevis, T., Petters, M. D., Prenni, A. J., Po¨schl, U., Rizzo,522
L. V., Schneider, J., Smith, J. N., Swietlicki, E., Tota, J., Wang, J., Wiedensohler, A.,523
Zorn, S. R., 2010a. An overview of the Amazonian Aerosol Characterization Experiment524
2008 (AMAZE-08). Atmospheric Chemistry and Physics 10 (23), 11415–11438.525
Martin, S. T., Andreae, M. O., Artaxo, P., Baumgardner, D., Chen, Q., Goldstein, A. H.,526
Guenther, A., Heald, C. L., Mayol-Bracero, O. L., McMurry, P. H., Pauliquevis, T., Po¨schl,527
U., Prather, K. A., Roberts, G. C., Saleska, S. R., Silva Dias, M. A., Spracklen, D. V.,528
Swietlicki, E., Trebs, I., 2010b. Sources and properties of Amazonian aerosol particles.529
Reviews of Geophysics 48 (2), RG2002.530
Paulson, S. E., Chung, M., Sen, A. D., Orzechowska, G., 1998. Measurement of OH radical531
formation from the reaction of ozone with several biogenic alkenes. Journal of Geophysical532
Research 103 (D19), 25533–25539.533
Paulson, S. E., Orlando, J. J., 1996. The reactions of ozone with alkenes: An important534
source of HOx in the boundary layer. Geophysical Research Letters 23 (25), 3727–3730.535
Pickering, K. E., Thompson, A. M., Wang, Y., Tao, W.-K., McNamara, D. P., Kirchhoff, V.536
W. J. H., Heikes, B. G., Sachse, G. W., Bradshaw, J. D., Gregory, G. L., Blake, D. R.,537
1996. Convective transport of biomass burning emissions over Brazil during TRACE A.538
Journal of Geophysical Research 101 (D19), 23993–24012.539
Rickard, A. R., Johnson, D., McGill, C. D., Marston, G., 1999. OH yields in the gas-phase540
reactions of ozone with alkenes. The Journal of Physical Chemistry A 103 (38), 7656–7664.541
Sahu, L. K., Lal, S., 2006. Changes in surface ozone levels due to convective downdrafts over542
the Bay of Bengal. Geophysical Research Letters 33, L10807.543
32
M
AN
US
CR
IP
T
 
AC
CE
PT
ED
ACCEPTED MANUSCRIPT
Scala, J. R., Garstang, M., Tao, W.-k., Pickering, K. E., Thompson, A. M., Simpson, J.,544
Kirchhoff, V. W. J. H., Browell, E. V., Sachse, G. W., Torres, A. L., Gregory, G. L.,545
Rasmussen, R. A., Khalil, M. a. K., 1990. Cloud draft structure and trace gas transport.546
Journal of Geophysical Research 95 (D10), 17015–17030.547
Sigler, M., J., Fuentes, J. D., Heitz, C., R., Garstang, M., Fisch, G., 2002. Ozone dynamics548
and deposition processes at a deforested site in the Amazon Basin. AMBIO 31 (1), 21–27.549
Thompson, A. M., Tao, W.-K., Pickering, K. E., Scala, J. R., Simpson, J., 1997. Tropical550
deep convection and ozone formation. Bulletin of the American Meteorological Society551
78 (6), 1043–1054.552
Torres, A. L., Buchan, H., 1988. Tropospheric nitric oxide measurements over the Amazon553
Basin. Journal of Geophysical Research 93 (D2), 1396–1406.554
To´ta, J., Fitzjarrald, D. R., da Silva Dias, M. A. F., 2012. Amazon rainforest exchange555
of carbon and subcanopy air flow: Manaus LBA site–A complex terrain condition. The556
Scientific World Journal 2012, 1–19.557
Yuter, S. E., Houze, R. A., 1995. Three-dimensional kinematic and microphysical evolution558
of Florida cumulonimbus. Part II: Frequency distribution of vertical velocity, reflectivity,559
and differential reflectivity. Monthly Weather Review 123, 1941–1963.560
Zhang, R., Tie, X., Bond, D. W., 2003. Impacts of anthropogenic and natural NOx sources561
over the U.S. on tropospheric chemistry. Proceedings of the National Academy of Sciences562
100 (4), 1505–1509.563
Zipser, E. J., 1969. The role of organized unsaturated convective downdrafts in the struc-564
tureand rapid decay of an equatorial disturbance. Journal of Applied565
Zipser, E. J., 1977. Mesoscale and Convective–Scale downdrafts as distinct components of566
squall-line structure. Monthly Weather Review 105 (12), 1568–1589.567
33