Magmatism is a dominant process on Earth and Mars that has significantly modified and evolved the lithospheres of each planet by delivering magma to shallow depths and to the surface. Two common modes of volcanism are present on both Earth and Mars: central-vent dominated volcanism that creates large edifices from concentrating magma in chambers before eruptions and distributed volcanism that creates many smaller edifices on the surface through the independent ascent of individual magmatic dikes. In regions of distributed volcanism, clusters of volcanoes develop over thousands to millions of years. This dissertation explores the geology of distributed volcanism on Earth and Mars from shallow depths (~1 km) to the surface. On long time scales, distributed volcanism emplaces magmatic sills below the surface and feeds volcanoes at the surface. The change in spatial distribution and formation rate of volcanoes over time is used to infer the evolution of the source region of magma generation. At short time scales, the emplacement of lava flows in these fields present an urgent hazard for nearby people and infrastructure. I present software that can be used to simulate lava flow inundation and show that individual computer codes can be validated using real-world flows. On Mars, distributed volcanism occurs in the Tharsis Volcanic Province, sometimes associated with larger, central-vent shield volcanoes. Two volcanic fields in this province are mapped here. The Syria Planum field is composed three major volcanic units, two of which are clusters of 10s to >100 shield volcanoes. This area had volcanic activity that spanned 900 million years, from 3.5-2.6 Ga. The Arsia Mons Caldera field is associated with a large shield volcano. Using crater age-dating and mapping stratigraphy between lava flows, activity in this field peaked at ~150 Ma and monotonically waned until 10-90 Ma, when volcanism likely ceased.

The Florida manatee (Trichechus manatus latirostris) is a protected species that is vulnerable to both anthropogenic and natural causes of mortality. The ability of wildlife managers to oversee regulation of this species is based on available abundance estimates and mortality data. Using existing manatee mortality data collected by Florida Fish and Wildlife Conservation Commission (FWC) from 1974-2012, this study focuses on identifying significant spatial clusters of high values or “hotspots” of manatee mortality and the temporal patterns of these hotspots using the novel “emerging hotspot analysis” ArcGIS tool. The categories of manatee mortality included in this analysis were watercraft-related, perinatal, cold-stress, and other natural (which includes red tide) and were classified into five hotspot pattern categories. Of interest were the locations where consecutive or new hotspot patterns were identified among the four categories of manatee mortality included in this analysis. Consecutive hotspot clusters were found near Tampa Bay (which includes parts of Pinellas, Hillsborough, and Manatee Counties) and in the counties of Hernando/Pasco, Monroe, Palm Beach/Broward/Miami-Dade, St. Johns/Flagler, and Citrus. New hotspot clusters were found in Tampa Bay (which includes parts of Pinellas, Hillsborough, and Manatee Counties) and in the counties of Nassau, Wakulla, Charlotte/Lee, St. Lucie/Martin, Levy, Duval, Dixie, Volusia/Seminole, and Citrus. These mortality hotspots frequently overlapped areas of higher manatee and human population densities. These hotspot clusters indicate emerging patterns that highlight areas to focus future research by wildlife managers; specifically, on the relationship between human population density and concentration of watercraft activities in coastal areas, as well as the influence coastal development has on the vital resources utilized by manatees.

abstract: Geographically Weighted Regression (GWR) has been broadly used in various fields to model spatially non-stationary relationships. Classic GWR is considered as a single-scale model that is based on one bandwidth parameter which controls the amount of distance-decay in weighting neighboring data around each location. The single bandwidth in GWR assumes that processes (relationships between the response variable and the predictor variables) all operate at the same scale. However, this posits a limitation in modeling potentially multi-scale processes which are more often seen in the real world. For example, the measured ambient temperature of a location is affected by the built environment, regional weather and global warming, all of which operate at different scales. A recent advancement to GWR termed Multiscale GWR (MGWR) removes the single bandwidth assumption and allows the bandwidths for each covariate to vary. This results in each parameter surface being allowed to have a different degree of spatial variation, reflecting variation across covariate-specific processes. In this way, MGWR has the capability to differentiate local, regional and global processes by using varying bandwidths for covariates. Additionally, bandwidths in MGWR become explicit indicators of the scale at various processes operate. The proposed dissertation covers three perspectives centering on MGWR: Computation; Inference; and Application. The first component focuses on addressing computational issues in MGWR to allow MGWR models to be calibrated more efficiently and to be applied on large datasets. The second component aims to statistically differentiate the spatial scales at which different processes operate by quantifying the uncertainty associated with each bandwidth obtained from MGWR. In the third component, an empirical study will be conducted to model the changing relationships between county-level socio-economic factors and voter preferences in the 2008-2016 United States presidential elections using MGWR. / Dissertation/Thesis / Doctoral Dissertation Geography 2020

Cette thèse porte sur l’estimation spatiale de la teneur superficielle en carbone organiquedu sol ou teneur en SOC (pour ’Soil Organic Carbon content’), à l’échelle d’une petite région agricolefrancilienne. La variabilité de la teneur en SOC a été identifiée comme étant l’une des principales sourcesd’incertitude de la prédiction des stocks de SOC, dont l’accroissement favorise la fertilité des sols etl’atténuation des émissions de gaz à effet de serre. Nous utilisons des données provenant de sourceshétérogènes décrites selon différentes résolutions spatiales (prélèvements de sol, carte pédologique, imagessatellitaires multispectrales, etc) dans le but de produire d’une part une information spatiale exhaustive,et d’autre part des estimations précises de la teneur en SOC sur la région d’étude ainsi qu’une uneévaluation des incertitudes associées. Plusieurs modèles originaux, dont certains tiennent compte duchangement du support, sont construits et plusieurs approches et méthodes de prédiction sont considérées.Parmi elles, on retrouve des méthodes bayésiennes récentes et performantes permettant non seulementd’inférer des modèles sophistiqués intégrant conjointement des données de résolution spatiale différentemais aussi de traiter des données en grande dimension. Afin d’optimiser la qualité de la prédictiondes modélisations multi-sources, nous proposons également une approche efficace et rapide permettantd’accroître l’influence d’un type de données importantes mais sous-représentées dans l’ensemble de toutesles données initialement intégrées. / In this thesis, we are interested in the spatial estimation of the topsoil organic carbon(SOC) content over a small agricultural area located West of Paris. The variability of the SOC contenthas been identified as one of the main sources of prediction uncertainty of SOC stocks, whose increasepromotes soil fertility and mitigates greenhouse gas emissions. We use data issued from heterogeneoussources defined at different spatial resolutions (soil samples, soil map, multispectral satellite images, etc)with the aim of providing on the one hand an exhaustive spatial information, and on the other accurateestimates of the SOC content in the study region and an assessment of the related uncertainties. Severaloriginal models, some of which incorporate the change of support, are built and several approaches andprediction methods are considered. These include recent and powerful Bayesian methods enabling notonly the inference of sophisticated models integrating jointly data of different spatial resolutions butalso the exploitation of large data sets. In order to optimize the quality of prediction of the multi-sourcedata modellings, we also propose an efficient and fast approach : it allows to increase the influence of animportant but under-represented type of data, in the set of all initially integrated data.

Soil salinization is of great concern in the irrigated arid and semi-arid western United States due to its threat to sustainable agricultural productivity and thus is closely monitored. A widely accepted and traditional standard method for estimating soil salinity is the electrical conductivity of the saturated paste extracts (ECe). However, this method underestimates salinity due to ion pair formation in high ionic strength solution. Numerous studies have recommended the use of an electromagnetic induction (EMI) sensing technique to monitor field-scale soil salinity due to rapidness and non-destructiveness of the sampling. However, because the EMI measurement (ECa) is related to a host of soil properties, calibrating ECa to salinity in a non-homogeneous setting is particularly challenging. The main objective of this study is to understand the dynamics and interacting factors controlling soil salinity using an EMI sensor. Specifically, a correction is made for the underestimation of soil salinity from saturated paste extracts, and a calibration model is developed that is capable of predicting salinity directly from ECa despite the non-homogeneity of potential perturbing factors. A comparison is made of salinity measurement methods based on soil saturated pastes with respect to specific soil management goals. Results show that ion pairing exists even in low ionic strength solution and by diluting the saturated paste extracts to conductivities ≤ 0.03 dS m -1 (ECed), ion pairing is minimized. An improved salinity estimate is obtained by computing total dissolved solids (TDS, in mM) from the ECed values, and then multiplying the TDS by the dilution factor. We also developed a calibration model using quantile regression, which makes no assumption about the distribution of the errors, and which is capable of predicting low range soil salinity (such as that in calcareous soils) from ECa depth-weighted measurements (ECH25ECe). A comparison of ECe, ECed, ECH25ECe, and direct measurement of EC in soil pastes (“ Bureau of Soils Cup ” method, ECcup) across six depths, three texture groups, and the combinations of EC method and depth or texture groups, supports the use of the ECH25ECe method to rapidly and reliably monitor salinity in calcareous soils of arid and semiarid regions.

In this study, we extend the spatial weight matrix defined by Getis and Aldstadt (2004) to a more general case. The modified spatial weight matrix performs better than the original spatial weight matrix since the modified spatial weight matrix adjusts weights of observations based on the distance from other observations. Both the simulation study and the application to the ecological process of invasion of non-native invasive plants (NNIPs) provide evidences for the better performance of the modified spatial weight matrix. We also develop procedures that can be used to quantify the invasion stages of NNIPs. The resultant map of invasion stage on county-level provides a useful and meaningful tool for policy makers; especially, it can be used to optimize allocation of management resources. The result of simultaneous autoregressive model shows that not only the biotic and abiotic factors but also human activities play an important role in the establishment and spread of multiflora rose in the Upper Midwest. It also shows the tendency of the establishment and spread of multiflora rose (Rosa Multiflora, Thunb. ex Murr.) in the Upper Midwest.

Mining processes generate waste rock, tailings, and slag that can increase heavy metal concentrations in soils. Un-reclaimed, abandoned mine sites are particularly prone to leaching these contaminants, which may accumulate and pose significant environmental and public health concerns. The characterization and spatial delineation of heavy metals of such soils is vital for risk assessment and soil reclamation. Bumpus Cove, once one of the richest mineralized districts of eastern TN, is home to at least 47 abandoned, un-reclaimed mines that were all permanently closed by the 1950s. This study evaluated 52 soil samples collected within a 0.67 km2 study area containing 6 known abandoned Pb, Zn, and Mn mines at the headwaters of Bumpus Cove Creek for heavy metal concentrations. Soil samples were analyzed for Zn, Mn, Pb, Cu, and Cd by means of microwave-assisted acid digestion and flame atomic absorption spectrometry (FAAS). Using the measured values and digital elevation model (DEM) derived from lidar data, ordinary kriging and cokriging interpolation techniques were used to predict the trend of heavy metal concentrations throughout the study area. Concentrations for Zn, Mn, and Pb show significant variability between sample sites (ranges of 12 – 1,354 mg/kg Zn, 6 – 2,574 mg/kg Mn, 33 – 2,271 mg/kg Pb). Cu and Cd were much less variable, with ranges of 1 - 65 mg/kg and 7 – 40 mg/kg, respectively. Of the measured heavy metals, only Zn and Pb exceed permissible limits in soils. Results show that ordinary kriging interpolation methods produced improved results over ordinary cokriging with and without lognormal transformations for all metals. Mn and Pb were found to transport further downhill following the natural drainage, whereas Zn, Cu and Cd concentrations exhibit localized variability without a clear transportation path. This study can provide a reference for state and local entities responsible for heavy metal monitoring in Bumpus Cove, TN.

No description available.

Nature-based solutions (NbS) are becoming increasingly popular in cities around the world; however, such efforts have not been widely incorporated into analyses of urban flood vulnerability nor the total population and property loss of flooding to date, except for a few studies that examined the effectiveness of green infrastructure or only wetlands in flood regulation. The proposed research sought to understand if the existing pattern and composition of NbS can mitigate flood vulnerability and loss of flooding in one of the fastest urbanizing regions in the United States, the Philadelphia Metropolitan Area. This research made key contributions to our understanding of how urban areas can grow without exacerbating flooding and inequity.First, a systematic mapping was conducted to reveal the most common spatial metrics of NbS that mitigate urban flooding in countries around the world. These findings identified important research areas for urban geographers, policymakers, planners, and civil engineers. This review indicated that the effectiveness of NbS varies spatially based on land use/land cover, climatic, and other contextual factors. The results indicated that the location, distribution, and arrangement of NbS may have different impacts on runoff mitigation and flood loss. Also, flood hydrology was the most common topic addressed, and the spatial configuration of NbS, especially connectivity was consistently identified as an important factor in flood regulation. Second, the potential of NbS as a flood loss mitigation tool in one of the fastest-growing and flood-prone counties of Pennsylvania, Montgomery County, using the Generalized Linear Model (GLR) and Geographically Weighted Regression (GWR) techniques was examined. The findings partially contradicted previous research by revealing an unexpected relationship between NbS quantity in floodplains and expected annual loss. Findings also demonstrated that lower-sized and disconnected patches of NbS in floodplains in some dense urban areas effectively reduce total losses from flood events. Third, the spatial coincidence between the density of NbS and flood vulnerability within eight neighboring urbanizing regions situated in Montgomery County was analyzed by using the Local Indicator of Spatial Association (LISA). The results of LISA identified regions of concern characterized by elevated flood vulnerability scores and reduced concentrations of two tree canopy types as well as shrubs and grasses. Taken together, these results emphasize the significance of strategically integrating and improving NbS, especially in areas grappling with distinct flood-related issues. It also emphasized the potential for significant enhancements in flood resilience and mitigation policies thoughtful urban planning and the adoption of NbS. / Geography

<p dir="ltr">The first chapter considers the l_α regularized linear regression, also termed Bridge regression. For α ∈ (0, 1), Bridge regression enjoys several statistical properties of interest such</p><p dir="ltr">as sparsity and near-unbiasedness of the estimates (Fan & Li, 2001). However, the main difficulty lies in the non-convex nature of the penalty for these values of α, which makes an</p><p dir="ltr">optimization procedure challenging and usually it is only possible to find a local optimum. To address this issue, Polson et al. (2013) took a sampling based fully Bayesian approach to this problem, using the correspondence between the Bridge penalty and a power exponential prior on the regression coefficients. However, their sampling procedure relies on Markov chain Monte Carlo (MCMC) techniques, which are inherently sequential and not scalable to large problem dimensions. Cross validation approaches are similarly computation-intensive. To this end, our contribution is a novel non-iterative method to fit a Bridge regression model. The main contribution lies in an explicit formula for Stein’s unbiased risk estimate for the out of sample prediction risk of Bridge regression, which can then be optimized to select the desired tuning parameters, allowing us to completely bypass MCMC as well as computation-intensive cross validation approaches. Our procedure yields results in a fraction of computational times compared to iterative schemes, without any appreciable loss in statistical performance.</p><p><br></p><p dir="ltr">Next, we build upon the classical and influential works of Neal (1996), who proved that the infinite width scaling limit of a Bayesian neural network with one hidden layer is a Gaussian process, when the network weights have bounded prior variance. Neal’s result has been extended to networks with multiple hidden layers and to convolutional neural networks, also with Gaussian process scaling limits. The tractable properties of Gaussian processes then allow straightforward posterior inference and uncertainty quantification, considerably simplifying the study of the limit process compared to a network of finite width. Neural network weights with unbounded variance, however, pose unique challenges. In this case, the classical central limit theorem breaks down and it is well known that the scaling limit is an α-stable process under suitable conditions. However, current literature is primarily limited to forward simulations under these processes and the problem of posterior inference under such a scaling limit remains largely unaddressed, unlike in the Gaussian process case. To this end, our contribution is an interpretable and computationally efficient procedure for posterior inference, using a conditionally Gaussian representation, that then allows full use of the Gaussian process machinery for tractable posterior inference and uncertainty quantification in the non-Gaussian regime.</p><p><br></p><p dir="ltr">Finally, we extend on the previous chapter, by considering a natural extension to deep neural networks through kernel processes. Kernel processes (Aitchison et al., 2021) generalize to deeper networks the notion proved by Neal (1996) by describing the non-linear transformation in each layer as a covariance matrix (kernel) of a Gaussian process. In this way, each succesive layer transforms the covariance matrix in the previous layer by a covariance function. However, the covariance obtained by this process loses any possibility of representation learning since the covariance matrix is deterministic. To address this, Aitchison et al. (2021) proposed deep kernel processes using Wishart and inverse Wishart matrices for each layer in deep neural networks. Nevertheless, the approach they propose requires using a process that does not emerge from the limit of a classic neural network structure. We introduce α-stable kernel processes (α-KP) for learning posterior stochastic covariances in each layer. Our results show that our method is much better than the approach proposed by Aitchison et al. (2021) in both simulated data and the benchmark Boston dataset.</p>