Theses and Dissertations at Montana State University (MSU)
Permanent URI for this communityhttps://scholarworks.montana.edu/handle/1/732
Browse
9 results
Search Results
Item Studies in alternative theories of gravity and advanced data analysis(Montana State University - Bozeman, College of Letters & Science, 2024) Gupta, Toral; Chairperson, Graduate Committee: Neil J. Cornish; This is a manuscript style paper that includes co-authored chapters.The field of gravitational wave astronomy is generating groundbreaking findings, yielding unique insights on some of the most extraordinary phenomena in the universe and providing invaluable information on testing the principles of general relativity. All gravitational wave signals detected so far appear to come from compact binaries - black holes and neutron stars. We use information from these sources to probe strong fields of gravity and to constrain modified theories of gravity. However, solely relying on template- based searches for known astrophysical sources biases our gravitational wave signal search towards well-modeled systems, potentially overlooking unpredicted sources with limited theoretical models, hindering the extraction of new physics. Further work in this thesis focuses on building improved signal and noise models to enhance our capability of detecting gravitational signals of all within and beyond the constraints of theoretical predictions. This includes introduction of new basis functions with added modifications to develop a signal-agnostic waveform reconstruction model using Bayesian inference. Additionally, this study discusses improvements in the speed and performance of the BayesWave trans-dimensional Bayesian spectral estimation algorithm, which includes implementing a low-latency analysis and various enhancements to the algorithm itself. In essence, this study is centered on developing a comprehensive understanding, both theoretical and observational, of astrophysical objects along with the spacetime that governs their dynamics.Item Forecasting the response of invasive plant-infested communities to management(Montana State University - Bozeman, College of Agriculture, 2003) Rinella, Matthew James; Chairperson, Graduate Committee: Roger L. Sheley.Item Simultaneous estimation of risk in several 2 x 2 contingency tables : an empirical Bayes approach(Montana State University - Bozeman, College of Letters & Science, 1981) Bergum, James StanleyItem Development of a total systems approach to multi-pest management decision analysis(Montana State University - Bozeman, College of Letters & Science, 2012) Keren, Ilai Naftaly; Co-chairpersons, Graduate Committee: James Robison-Cox and Fabian D. MenalledThe concentration of wheat production in the Northern Great Plains has resulted in the development of specialized multitrophic agricultural pest complexes whose members interact in both positive and negative ways. In this context, management recommendations based on the traditional single-species pest control paradigm may lead to undesirable outcomes. Our goal was to develop a modeling framework to make multi-pest management decisions that take into account the existence of direct and indirect interactions among pests. We a Bayesian decision theory approach in combination with a analysis where observed intermediate nodes were replaced with error terms. This model holds several advantages current decision models, in particular it allows intuitive of coefficient estimates as the total, direct and through pest interactions, impact of management on the. We evaluated interactions between Bromus tectorum), Fusarium crown rot, and Cephus cinctus (wheat stem sawfly) as well as assessed the joint response of these pests to wheat seeding rates, cultivar competitiveness, and cultivar stem sawfly tolerance. Results indicate that yield differences be more readily explained as a result of the effects of on pests and multi-pest interactions, rather than just by direct effect of any particular management scheme on yield. Our provides a framework for finding the balance between simplicity and the complexity of the process being modeled, also bridges the gap between making inferences from experimental observational studies and ecological management decisions.Item Conceptual hydrologic modeling : insights into Bayesian analysis, model development, and predictions in ungauged basins(Montana State University - Bozeman, College of Agriculture, 2012) Smith, Tyler Jon; Chairperson, Graduate Committee: Lucy Marshall.; Ashish Sharma, Lucy Marshall, Raj Mehrotra and Scott Sisson were co-authors of the article, 'Development of a formal likelihood function for improved Bayesian inference of ephemeral catchments' in the journal 'Water resources research' which is contained within this thesis.; Lucy Marshall and Ashish Sharma were co-authors of the article, 'A Bayesian likelihood function specification methodology for conceptual hydrologic modeling' in the journal 'Water resources research' which is contained within this thesis.; Lucy Marshall and Brian McGlynn were co-authors of the article, 'Improving hydrologic model calibration using a flow-corrected time transformation' in the journal 'Water resources research' which is contained within this thesis.; Lucy Marshall, Brian McGlynn and Kelsey Jencso were co-authors of the article, 'Using field data to inform and evaluate a new model of catchment hydrologic connectivity' in the journal 'Water resources research' which is contained with this thesis.; Lucy Marshall, Brian McGlynn and Kelsey Jencso were co-authors of the article, 'A cross-catchment comparison and sensitivity analysis of the catchment connectivity model' in the journal 'Water resources research' which is contained within this thesis.; Lucy Marshall and Ashish Sharma were co-authors of the article, 'Predicting hydrologic response through a pooled catchment knowledgebase' in the journal 'Water resources research' which is contained within this thesis.Water is a fundamental resource, essential for human activities and underpinning economic and environmental sustainability. As such, management of water resources is a critical task, even more so in this time of climate uncertainty and change. Though hydrologic models have long been used to address water resource management, there is a growing desire to develop more reliable, transferable, and physically meaningful hydrologic models in light of data scarcity and uncertainty. In this dissertation, the three primary topics of (1) statistical implementation of models under uncertainty, (2) hydrologic model development, and (3) model applicability/transferability under data scarcity are considered. As a result of this research, we have: (1) developed methods for improved statistical implementation of hydrologic models through the development of a formal likelihood function for ephemeral catchments, the creation of a framework for identifying adequate likelihood functions across catchment conditions, and the introduction of a time domain transformation that was not previously used in hydrologic modeling; (2) developed a new model of catchment hydrologic connectivity based on extensive empirical observations that was shown to be consistent with both external and internal hydrologic variables, as well as transferable; and (3) introduced a new hierarchical Bayesian approach that was shown to accurately quantify uncertainty at an ungauged catchment based on pooled information from similar gauged catchments. This dissertation is divided into six key chapters/manuscripts that each address various aspects involved in improving hydrologic model performance, whose cumulative contribution results in a better understanding of how to construct, implement, and apply hydrologic models to lead to improved water resource management and planning problems of societal importance.Item Spatial and seasonal variability of watershed response to anthropogenic nitrogen loading in a mountainous watershed(Montana State University - Bozeman, College of Agriculture, 2010) Gardner, Kristin Kiara; Chairperson, Graduate Committee: Brian L. McGlynn.; Brian L. McGlynn was a co-author of the article, 'Seasonality in spatial variability and influence of land use/land cover and watershed characteristics on streamwater nitrogen export in a developing watershed in the Rocky Mountain West ' in the journal 'Water resources research' which is contained within this thesis.; Brian L. McGlynn was a co-author of the article, 'A multi-analysis approach to assess the spatio-temporal patterns of watershed response to localized inputs of nitrogen' in the journal 'Water resources research' which is contained within this thesis.; Brian L. McGlynn, and Lucy A. Marshall were co-authors of the article, 'Quantifying watershed sensitivity to spatially variable nitrogen loading and the relative importance of nitrogen retention mechanisms' in the journal 'Water resources research' which is contained within this thesis.Anthropogenic activity has greatly increased watershed export of bioavailable nitrogen. Escalating levels of bioavailable nitrogen can deteriorate aquatic ecosystems by promoting nuisance algae growth, depleting dissolved oxygen levels, altering biotic communities, and expediting eutrophication. Despite these potential detrimental impacts, there is notable lack of understanding of the linkages between anthropogenic nitrogen inputs and the spatial and seasonal heterogeneity of stream network concentrations and watershed nitrogen export. This dissertation research seeks to more accurately define these linkages by investigating the roles of landscape position and spatial distribution of anthropogenic nitrogen inputs on the magnitude and speciation of watershed nitrogen export and retention and how these roles vary seasonally across contrasting landscapes in a 212 km ² mountainous watershed in southwest Montana. Results indicate localized inputs of anthropogenic nitrogen occurring in watershed areas with quick transport times to streams had disproportionate effects on watershed nitrogen export compared to spatially distributed or localized inputs of nitrogen to areas with longer transport times. In lower elevation alluvial streams, these effects varied seasonally and were most evident during the dormant winter season by amplified nitrate peaks, elevated dissolved organic nitrogen:dissolved organic nitrogen (DIN:DON) ratios and lower dissolved organic carbon (DOC):total dissolved nitrogen (DOC:TDN). During the summer growing season, biologic uptake of nitrogen masked anthropogenic influences on watershed nitrogen export; however, endmember mixing analysis of nitrate isotopes revealed significant anthropogenic influence during the growing season, despite low nitrate concentrations and DIN:DON ratios. In contrast, streams draining alpine environments consisting of poorly developed, shallow soils and small riparian areas exhibited yearlong elevated nitrate concentrations compared to other sites, suggesting these areas were highly nitrogen enriched. Watershed modeling revealed the majority of watershed nitrogen retention occurred in the upland environment, most likely from biological uptake or lack of hydrologic connectivity. This work has critical implications for watershed management, which include: 1) developing flexible strategies that address varying landscape characteristics and nitrogen loading patterns across a watershed, 2) avoiding clustering nitrogen loading in areas with quick travel times to surface waters, 3) seasonal monitoring to accurately gauge watershed nitrogen saturation status, and 4) incorporating spatial relationships into streamwater nitrogen models.Item Bayesian optimization and uncertainty analysis of complex environmental models, with applications in watershed management(Montana State University - Bozeman, College of Engineering, 2010) Mashamba, Able; Chairperson, Graduate Committee: Edward L. Mooney; Lucy Marshall (co-chair); Lucy Marshall was a co-author of the article, 'A new approach to watershed management practices assessment using the soil and water assessment tool, SWAT' in the journal 'Journal of soil and water conservation' which is contained within this thesis.; Lucy Marshall was a co-author of the article, 'A case study examining factors affecting the performance of response surface modeling during Bayesian optimization and uncertainty analysis of hydrologic models' in the journal 'Journal of environmental modeling and software' which is contained within this thesis.; Lucy Marshall was a co-author of the article, 'Bayesian constrained optimization and uncertainty analysis using radial basis random local fitting' in the journal 'Journal of stochastic environmental research and risk assessment' which is contained within this thesis.; Lucy Marshall was a co-author of the article, 'Bayesian uncertainty analysis of the distributed hydrology soil-vegetation model using radial basis functions' in the journal 'Journal of environmental modeling and software' which is contained within this thesis.This dissertation presents results of research in the development, testing and application of an automated calibration and uncertainty analysis framework for distributed environmental models based on Bayesian Markov chain Monte Carlo (MCMC) sampling and response surface methodology (RSM) surrogate models that use a novel random local fitting algorithm. Typical automated search methods for optimization and uncertainty assessment such as evolutionary and Nelder-Mead Simplex algorithms are inefficient and/or infeasible when applied to distributed environmental models, as exemplified by the watershed management scenario analysis case study presented as part of this dissertation. This is because the larger numbers of non-linearly interacting parameters and the more complex structures of distributed environmental models make automated calibration and uncertainty analysis more computationally demanding compared to traditional basin-averaged models. To improve efficiency and feasibility of automated calibration and uncertainty assessment of distributed models, recent research has been focusing on using the response surface methodology (RSM) to approximate objective functions such as sum of squared residuals and Bayesian inference likelihoods. This dissertation presents (i) results on a novel study of factors that affect the performance of RSM approximation during Bayesian calibration and uncertainty analysis, (ii) a new 'random local fitting' (RLF) algorithm that improves RSM approximation for large sampling domains and (iii) application of a developed automated uncertainty analysis framework that uses MCMC sampling and a spline-based radial basis approximation function enhanced by the RLF algorithm to a fully-distributed hydrologic model case study. Using the MCMC sampling and response surface approximation framework for automated parameter and predictive uncertainty assessment of a distributed environmental model is novel. While extended testing of the developed MCMC uncertainty analysis paradigm is necessary, the results presented show that the new framework is robust and efficient for the case studied and similar distributed environmental models. As distributed environmental models continue to find use in climate change studies, flood forecasting, water resource management and land use studies, results of this study will have increasing importance to automated model assessment. Potential future research from this dissertation is the investigation of how model parameter sensitivities and inter-dependencies affect the performance of response surface approximation.Item Applying novel approaches to old datasets : utilizing opportunistic observations and Bayesian estimation to describe spatial use patterns for Steller sea lions(Montana State University - Bozeman, College of Letters & Science, 2010) Himes Boor, Gina Kristine; Chairperson, Graduate Committee: Daniel GoodmanDespite two decades of satellite telemetry studies conducted on Steller sea lions, scientists still lack basic spatially-explicit knowledge about Steller sea lion habitat use. The Platforms of Opportunity data collected by the National Marine Fisheries Service contain Steller sea lion sighting records throughout the species' entire range and have the potential to fill the critical gap in knowledge about what areas Steller sea lions are using. The Platforms of Opportunity data have not previously been used to identify marine mammal habitat because they contain sightings without associated effort records (e.g. time spent surveying or area sampled). In this study a novel approach was used to overcome this issue through development of an effort index that allowed for calculation of effort-corrected Steller sea lion encounter rates. A Bayesian negative binomial model was used to quantify both the encounter rate and the uncertainty surrounding that rate within 15 km² grid cells across the species' entire range. Year-round encounter rate estimates were derived in addition to breeding and non-breeding season encounter rates. Although the results of this analysis confirmed many of the areas known to be important Steller sea lion habitat, several previously unrecognized high-use areas were identified. Current critical habitat designated areas only encompass about 37% of high use areas estimated using this methodology.Item A comprehensive Bayesian approach to gravitational wave astronomy(Montana State University - Bozeman, College of Letters & Science, 2009) Littenberg, Tyson Bailey; Chairperson, Graduate Committee: Neil J. CornishThe challenge of determining whether data from a gravitational wave detector contains signals which are cosmic in origin is the central problem in gravitational wave astronomy. The "detection problem" is particularly challenging for low amplitude signals embedded in "glitchy" instrument noise. It is imperative that we can robustly distinguish between the data being consistent with instrument noise alone, or noise and a weak gravitational wave signal. In response to this challenge we have set out to develop a robust, general purpose approach that can locate and characterize gravitational wave signals, and provided odds that the signal is of cosmic origin. Our approach employs the Markov Chain Monte Carlo family of algorithms to construct a fully Bayesian solution to the challenge - the Parallel Tempered Markov Chain Monte Carlo (PTMCMC) detection algorithm. The PTMCMC detection algorithm establishes which regions of parameter space contain the highest posterior weight, efficiently explores the posterior distribution function of the model parameters, and calculates the marginalized likelihood, or evidence, for the models under consideration. We illustrate our approach using simulated LISA and LIGO-Virgo data.