Browsing by Subject "Bayesian analysis"

Now showing 1 - 4 of 4
  • Thumbnail Image
    Item
    Effects of Forest Cover Change on Streamflow in Low Relief Glaciated Catchments
    (2020-08) McEachran, Zachary
    Forest cover disturbance, climate change, and their interaction can alter how catchments store and process water, which has ramifications for all aspects of the hydrologic cycle, including flood risk, channel geomorphology, and water quality. Catchments in the boreal-temperate transition zone may be especially vulnerable to these factors. While streams in this glaciated region have low-topographic relief and may originate from expansive wetlands, much of the past research on forest disturbance-streamflow relationships comes from regions where landscape characteristics and subsequent hydrological function is substantially different, e.g. mountainous regions with bedrock close to the soil surface. Further, most work investigating the forest-streamflow relationship occurs at small spatial scales (< 10 km2). I seek to fill a knowledge gap by 1) creating a new conceptual model for how forest cover change affects sediment yield in managed temperate forested catchments that accounts for how sediment yield responds to altered catchment hydrology, 2) developing a new approach to peak-flow analysis using paired catchment experiments at the Marcell Experimental Forest (MEF) in north-central Minnesota, and 3) investigating how forest cover change and climate affect peak flows and water yield in large (> 10 km2) catchments in Minnesota. My results indicate that in low-relief glaciated regions, glacial geology controls sediment yield response to forest harvesting; forest harvesting may affect large peak flows by altering the occurrence probabilities of large peaks at the small catchment scale; and streamflow in larger catchments is largely controlled by climate variation, with land cover a minor yet discernable driver of peak flows and water yield. These results are framed within a new forest harvesting/water quality framework that holistically accounts for all sources of increased sediment yield after forest harvesting in diverse landscapes. Please note that multiple of these chapters are under peer review in scientific journals as of August 2020, and those versions will supersede this dissertation for purposes of citations.
  • Thumbnail Image
    Item
    Estimating annual harvest of American Woodcock (Scolopax minor) in the United States during 1964-2016: data, model code, and supplemental estimates
    (2019-07-09) Arnold, Todd W.; Farr, Matthew T.; Wright, Alexander D.; Saunders, Sarah P.; arnol065@umn.edu; Arnold, Todd W.; Department of Integrative Biology, Michigan State University (MTF, ADW, SPS); Department of Fisheries, Wildlife and Conservation Biology, University of Minnesota (TWA)
    Harvest of American woodcock (Scolopax minor) in the United States has been estimated using two different hunter surveys: 1) the Duck Stamp Survey (DSS, 1964-2001), which estimated harvest by hunters who also hunted ducks or geese and purchased a Federal Migratory Bird Hunting (“Duck”) Stamp, but failed to survey hunters who did not purchase Duck Stamps (which were not required for hunting woodcock); and 2) the Harvest Information Program (HIP, 1999-2019), which was initiated in 1999 and designed to survey nearly all hunters who targeted woodcock. The two surveys overlapped during only 3 years (1999-2001), and in most states, the HIP survey estimated much higher woodcock harvest based on its more complete sampling frame of woodcock hunters. We developed Bayesian hierarchical models to use combined data streams to estimate total harvest during 1964-2016 (Arnold 2019) or 1964-2013 (Saunders et al. 2019) in the Eastern and Central Management Units by estimating unobserved harvest by hunters who never or only occasionally hunted waterfowl. Both approaches used annual Duck Stamp sales as a covariate to assess annual participation by hunters who occasionally hunted waterfowl, and also used the 3 overlap years (1999-2001) to estimate harvest by hunters who never hunted waterfowl. However, our approaches differed in how we assessed participation by occasional waterfowl hunters: 1) as residuals from splines fit to long-term duck stamp sales (Arnold 2019), which posited a smooth change in total waterfowl hunters through time, with residuals reflecting short-term participation or non-participation in waterfowl hunting, or 2) relative to maximum annual duck stamp sales (Saunders et al. 2019), which posited a constant number of potential waterfowl hunters in each state (max stamp sales), but with annual changes in relative participation in waterfowl hunting corresponding to yearly stamp sales. Our estimates were remarkably similar for combined harvest in the Central Management Unit, but diverged substantially for the Eastern Management Unit. We have no way of assessing which set of assumptions is closer to the truth, and present both models here in hopes that future researchers will continue to refine our methods to produce even more robust estimates of historical woodcock harvest.
  • Thumbnail Image
    Item
    On some computational, modeling and design issues in Bayesian analysis of spatial data
    (2012-10) Ren, Qian
    My research on Bayesian spatial analysis can be divided into three challenges: computing, methodology (modeling) and experimental design. My first exploration in research is to find an alternative to Markov chain Monte Carlo (MCMC) for the Bayesian hierarchical model. Variational Bayesian (VB) method would be a choice to tackle the massive computational burden for large spatial data analysis. We discuss applying VB to spatial analysis, especially to the multivariate spatial cases. Different VB algorithms are developed and applied to simulated and real examples. When the number of the locations and the dimension of the outcome variables are large, models with feature of dimension reduction are essential in the real applications. Low-rank spatial processes and factor analysis models are merged together to capture the associations among the variables as well as the strength of spatial correlation for each variable. We also develop stochastic selection of the latent factors by utilizing certain identifiability characterizations for the spatial factor model. A MCMC algorithm is developed for estimation, which also deals with the spatial misalignment problem. In many of the spatial applications (environmental epidemiology, for instance), parameter estimation is the most important objective in the study. Even with carefully constructed models and computing technique, it is always a challenge to handle the large spatial data set. Bayesian experimental design may help us to get the desired information from a spatial survey study with a sample size that can be analyzed by most available software. The problem of finding the optimum experimental design for the purpose of performing one or more hypothesis tests is considered in the context of spatial analysis. The Bayesian decision theoretic approach is used to arrive at several new optimality criteria for this purpose. Different approaches to achieving this goal are explored, including additive weighted loss and convex approximation. Simulated annealing algorithm (SAA) is applied to real examples to find the optimum design based on our objective function.
  • Thumbnail Image
    Item
    R code and output supporting: Time series sightability modeling of animal populations
    (2017-03-29) ArchMiller, Althea A; Fieberg, John R; Dorazio, Robert M; St. Clair, Katherine; althea.archmiller@gmail.com; ArchMiller, Althea A
    The goal of our study was to expand a previously developed model-based approach to include random effects and a temporal spline for time series modeling of multiple years of operational survey data. We developed a Bayesian hierarchical model as our framework to build and compare fixed-effects and temporal model-based sightability models applied to 12 years of MN moose operational survey data. Here, we share the Program R code and data necessary to replicate the manuscript results that demonstrate how our time series sightability modeling approach can increase the precision of population estimators and predict population dynamics with smoother (and thus more realistic trends) through time.