Examples of spatio-temporal data that can be represented as sets of points (called dot patterns) are pervasive in many applications, for example when tracking herds of migrating animals, ships in busy shipping channels and crowds of people in everyday life. The use of this type of data extends beyond the standard remit of Geographic Information Science (GISc), as classification and optimisation problems can often be visualised in the same manner. A common task within these fields is the assignment of a region (called a footprint) that is representative of the underlying pattern. The ways in which this footprint can be generated has been the subject of much research with many algorithms having been produced. Much of this research has focused on the dot patterns and footprints as static entities, however for many of the applications the data is prone to change. This thesis proposes that the footprint need not necessarily be updated each time the dot pattern changes; that the footprint can remain an appropriate representation of the pattern if the amount of change is slight. To ascertain the appropriate times at which to update the footprint, and when to leave it as it is, this thesis introduces the concept of change identifiers as simple measures of change between two dot patterns. Underlying the change identifiers is an in-depth examination of the data inherent in the dot pattern and the creation of descriptors that represent this data. The experimentation performed by this thesis shows that change identifiers are able to distinguish between different types of change across dot patterns from different sources. In doing so the change identifiers reduce the number of updates of the footprint while maintaining a measurably good representation of the dot pattern.

The environmental risk of growing genetically modified (GM) crops and particularly the spreading of GM genes to related non-GM crops is currently a concern in European agriculture. Because the risks of contamination are linked to the spatial and temporal arrangements of crops within the landscape, scenarios of crop arrangement are required to investigate the risks and potential coexistence measures. However, until recently, only manual methods were available to create scenarios. This thesis aims to provide a flexible referenced tool to create such scenarios. The model, called LandSFACTS, is a scientific research tool which allocates crops into fields, to meet user-defined crop spatio-temporal arrangements, using an empirical and statistical approach. The control of the crop arrangements is divided into two main sections: (i) the temporal arrangement of crops: encompassing crop rotations as transition matrices (specifically-developed methodology), temporal constraints (return period of crops, forbidden crop sequences), initial crops in fields regulated by temporal patterns (specifically-developed statistical analyses) and yearly crop proportions; and (ii) the spatial arrangements of crops: encompassing possible crops in fields, crop rotation in fields regulated by spatial patterns (specifically-developed statistical analyses), and spatial constraints (separation distances between crops). The limitations imposed by the model include the size of the smallest spatial and temporal unit: only one crop is allocated per field and per year. The model has been designed to be used by researchers with agronomic knowledge of the landscape. An assessment of the model did not lead to the detection of any significant flaws and therefore the model is considered valid for the stated specifications. Following this evaluation, the model is being used to fill incomplete datasets, build up and compare scenarios of crop allocations. Within the GM coexistence context, the model could provide useful support to investigate the impact of crop arrangement and potential coexistence measures on the risk of GM contamination of crops. More informed advice could therefore be provided to decision makers on the feasibility and efficiency of coexistence measures for GM cultivation.

Doctor of Philosophy / Department of Statistics / Juan Du / It is common to assume the spatial or spatio-temporal data are realizations of underlying random elds or stochastic processes. E ective approaches to modelling of the underlying autocorrelation structure of the same random eld and the association among multiple processes are of great demand in many areas including atmospheric sciences, meteorology and agriculture. To this end, this dissertation studies methods and application of the spatial modeling of large-scale dependence structure and spatio-temporal regression modelling. First, variogram and variogram matrix functions play important roles in modeling dependence structure among processes at di erent locations in spatial statistics. With more and more data collected on a global scale in environmental science, geophysics, and related elds, we focus on the characterizations of the variogram models on spheres of all dimensions for both stationary and intrinsic stationary, univariate and multivariate random elds. Some e cient approaches are proposed to construct a variety of variograms including simple polynomial structures. In particular, the series representation and spherical behavior of intrinsic stationary random elds are explored in both theoretical and simulation study. The applications of the proposed model and related theoretical results are demonstrated using simulation and real data analysis. Second, knowledge of the influential factors on the number of days suitable for fieldwork (DSFW) has important implications on timing of agricultural eld operations, machinery decision, and risk management. To assess how some global climate phenomena such as El Nino Southern Oscillation (ENSO) a ects DSFW and capture their complex associations in space and time, we propose various spatio-temporal dynamic models under hierarchical Bayesian framework. The Integrated Nested Laplace Approximation (INLA) is used and adapted to reduce the computational burden experienced when a large number of geo-locations and time points is considered in the data set. A comparison study between dynamics models with INLA viewing spatial domain as discrete and continuous is conducted and their pros and cons are evaluated based on multiple criteria. Finally a model with time- varying coefficients is shown to reflect the dynamic nature of the impact and lagged effect of ENSO on DSFW in US with spatio-temporal correlations accounted.

The increase in wind energy production has been relatively rapid and is expected to continue at a global scale. However, numbers of bat carcasses found at wind turbines in North America in the early 21st century raised concern about the plight of this taxon with the growth in wind-energy generation. This led to carcass searches for bats becoming commonplace at wind farms globally. However, few large scale systematic studies have assessed the effects of wind turbines on bats, especially for species considered potentially at higher risk in Europe. In this thesis the number and species of bats killed from wind farms were estimated across Britain, and the important predictors (i.e. activity, turbine characteristics and habitat) of fatality were determined. Insect abundance, biomass and bat activity was also quantified at turbine and control locations, to assess if insects and hence bats were attracted to turbines. In addition, assessments were made of the effects of increasing temporal and spatial replication of acoustic monitoring on estimates of species composition and bat activity. This was assessed for activity monitored at ground and at the centre of the rotor sweep area (the nacelle). Carcass searches were conducted using trained search dogs and concurrently bats were surveyed acoustically at three randomly selected turbines at ground and from the nacelle at 48 wind farms throughout Britain. Bats were also monitored acoustically at paired controls (with a randomly selected turbine) at 20 of the wind farms sites. In addition, nocturnal Diptera were sampled at 18 of the sites using a paired turbine and control design. Across 139 wind turbines, 188,335 bat passes were recorded and 2,973 carcass searches performed. Edge and open aerial foraging species, in particular Pipistrellus pipistrellus and P. pygmaeus were most at risk of fatality 4 at wind farms in Britain. The number of Pipistrellus pipistrellus killed annually in Britain between mid-July and mid-October was estimated at 2,373 95% CI 513 to 4,233 and the number of P. pygmaeus at 3,082 95% CI 1,270 to 4,894. When compared to population estimates, the number of Pipistrellus pygmaeus killed was 57% higher than the number of P. pipistrellus killed (0.19% of the population versus 0.43%, respectively). This may be due to Pipistrellus pygmaeus flying more often within the rotor sweep area compared to P. pipistrellus. Activity measured at the nacelle, which is generally assumed to be a better predictor of fatalities, was not a significant predictor of the probability of a fatality for all species combined, Pipistrellus pipistrellus, or P. pygmaeus. Pipistrellus pipistrellus activity and P. pygmaeus activity, measured at ground level were not good predictors of their respective fatalities. Whilst there was some evidence that Pipistrellus pipistrellus and P. pygmaeus activity monitored at ground level, was a significant predictor of the probability of their respective fatalities occurring, across wide ranging turbine types, fatality estimates were large. This is presumably due to the importance of turbine characterises (the wind speed that turbines become operational (cut-in speeds) turbine and the distance between the ground and blade tip at the bottom of the rotor sweep area) both being important negative predictors of fatalities for Pipistrellus pipistrellus. Predicting from models, if the cut-in speed is increased from 3.5 to 5 m s-1 the number of Pipistrellus pipistrellus fatalities would be reduced by 76% (0.23 fatalities per turbine per month to 0.06). These findings have important implications for guidance, since activity is the ubiquitous measure used to assess fatality risk for all species. Since, Pipistrellus pipistrellus and P. pygmaeus were detected at 98% and 92% of sites respectively; it could be 5 assumed that these species would be detected at the majority of wind farms within their range. Therefore, in a British context, curtailing wind turbines below 5 m s-1 could be an effective mitigation strategy without the costly requirement to monitor activity. Pipistrellus pipistrellus and P. pygmaeus activity was 46% (6.3 ± 1.3 SE mean passes per night c.f. 3.4 ± 1.3 SE) and 34% (4.0 ± 1.4 SE c.f. 2.7 ± 1.4 SE) higher at turbines compared to controls, respectively. Given that habitat and elevation were consistent between paired turbines and controls and monitoring was conducted on the same nights, higher activity at turbines compared to controls provides evidence that these two species are attracted to wind turbines. Furthermore, since the biomass of nocturnal Diptera, the main insect prey for Pipistrellus spp., was higher at controls compared to turbines, and bat foraging at turbines was not predicted by insect abundance or biomass, attraction is unlikely to be due to insects. Evidence presented here shows that bats are attracted to turbines, and therefore measuring activity at pre-construction sites for environmental impact assessments is unlikely to be effective. In conclusion, these results provide further evidence that common species are killed but generally in relatively low numbers, they also support the view that monitoring activity for assessing fatality risk at wind farms is ineffective. It is imperative that wind energy is developed using an evidence based approach. However, it also important that wind energy continues to contribute to an increasing renewable energy sector. In conclusion, results presented here, support that wind turbines are likely to be having a small impact on bat populations in Britain.

Indiana University-Purdue University Indianapolis (IUPUI) / Spatio-temporal data is becoming very popular in the recent times, as there are large number of datasets that collect both location and temporal information in the real time. The main challenge is that extracting useful insights from such large data set is extremely complex and laborious. In this thesis, we have proposed a novel 2D technique to visualize the spatio-temporal big data. The visualization of the combined interaction between the spatial and temporal data is of high importance to uncover the insights and identify the trends within the data. Maps have been a successful way to represent the spatial information. Addition- ally, in this work, colors are used to represent the temporal data. Every data point has the time information which is converted into relevant color, based on the HSV color model. The variation in the time is represented by transition from one color to another and hence provide smooth interpolation. The proposed solution will help the user to quickly understand the data and gain insights.

Knowledge discovery on spatio-temporal datasets has attracted<br />growing interests. Recent advances on remote sensing technology mean<br />that massive amounts of spatio-temporal data are being collected,<br />and its volume keeps increasing at an ever faster pace. It becomes<br />critical to design efficient algorithms for identifying novel and<br />meaningful patterns from massive spatio-temporal datasets. Different<br />from the other data sources, this data exhibits significant<br />space-time statistical dependence, and the assumption of i.i.d. is<br />no longer valid. The exact modeling of space-time dependence will<br />render the exponential growth of model complexity as the data size<br />increases. This research focuses on the construction of efficient<br />and effective approaches using approximate inference techniques for<br />three main mining tasks, including spatial outlier detection, robust<br />spatio-temporal prediction, and novel applications to real world<br />problems.<br /><br />Spatial novelty patterns, or spatial outliers, are those data points<br />whose characteristics are markedly different from their spatial<br />neighbors. There are two major branches of spatial outlier detection<br />methodologies, which can be either global Kriging based or local<br />Laplacian smoothing based. The former approach requires the exact<br />modeling of spatial dependence, which is time extensive; and the<br />latter approach requires the i.i.d. assumption of the smoothed<br />observations, which is not statistically solid. These two approaches<br />are constrained to numerical data, but in real world applications we<br />are often faced with a variety of non-numerical data types, such as<br />count, binary, nominal, and ordinal. To summarize, the main research<br />challenges are: 1) how much spatial dependence can be eliminated via<br />Laplace smoothing; 2) how to effectively and efficiently detect<br />outliers for large numerical spatial datasets; 3) how to generalize<br />numerical detection methods and develop a unified outlier detection<br />framework suitable for large non-numerical datasets; 4) how to<br />achieve accurate spatial prediction even when the training data has<br />been contaminated by outliers; 5) how to deal with spatio-temporal<br />data for the preceding problems.<br /><br />To address the first and second challenges, we mathematically<br />validated the effectiveness of Laplacian smoothing on the<br />elimination of spatial autocorrelations. This work provides<br />fundamental support for existing Laplacian smoothing based methods.<br />We also discovered a nontrivial side-effect of Laplacian smoothing,<br />which ingests additional spatial variations to the data due to<br />convolution effects. To capture this extra variability, we proposed<br />a generalized local statistical model, and designed two fast forward<br />and backward outlier detection methods that achieve a better balance<br />between computational efficiency and accuracy than most existing<br />methods, and are well suited to large numerical spatial datasets.<br /><br />We addressed the third challenge by mapping non-numerical variables<br />to latent numerical variables via a link function, such as logit<br />function used in logistic regression, and then utilizing<br />error-buffer artificial variables, which follow a Student-t<br />distribution, to capture the large valuations caused by outliers. We<br />proposed a unified statistical framework, which integrates the<br />advantages of spatial generalized linear mixed model, robust spatial<br />linear model, reduced-rank dimension reduction, and Bayesian<br />hierarchical model. A linear-time approximate inference algorithm<br />was designed to infer the posterior distribution of the error-buffer<br />artificial variables conditioned on observations. We demonstrated<br />that traditional numerical outlier detection methods can be directly<br />applied to the estimated artificial variables for outliers<br />detection. To the best of our knowledge, this is the first<br />linear-time outlier detection algorithm that supports a variety of<br />spatial attribute types, such as binary, count, ordinal, and<br />nominal.<br /><br />To address the fourth and fifth challenges, we proposed a robust<br />version of the Spatio-Temporal Random Effects (STRE) model, namely<br />the Robust STRE (R-STRE) model. The regular STRE model is a recently<br />proposed statistical model for large spatio-temporal data that has a<br />linear order time complexity, but is not best suited for<br />non-Gaussian and contaminated datasets. This deficiency can be<br />systemically addressed by increasing the robustness of the model<br />using heavy-tailed distributions, such as the Huber, Laplace, or<br />Student-t distribution to model the measurement error, instead of<br />the traditional Gaussian. However, the resulting R-STRE model<br />becomes analytical intractable, and direct application of<br />approximate inferences techniques still has a cubic order time<br />complexity. To address the computational challenge, we reformulated<br />the prediction problem as a maximum a posterior (MAP) problem with a<br />non-smooth objection function, transformed it to a equivalent<br />quadratic programming problem, and developed an efficient<br />interior-point numerical algorithm with a near linear order<br />complexity. This work presents the first near linear time robust<br />prediction approach for large spatio-temporal datasets in both<br />offline and online cases. / Ph. D.

No description available.

The statistical modeling of spatial extreme events, augmented by graphical models, provides a comprehensive framework for the development of techniques and models to describe natural phenomena in a variety of environmental, geoscience, and climate science applications. In a changing climate, the impact of natural hazards, such as wildfires, is believed to have evolved in frequency, size, and spatial extent, although regional responses may vary. The aforementioned impacts are of great significance due to their association with air pollution, irreversible harm to the environment and atmosphere, and the fact that they put human lives at risk. The prediction of wildfires holds significant importance within the realm of wildfire management due to its influence on the allocation of resources, the mitigation of detrimental consequences, and the subsequent recovery endeavors. Therefore, the development of robust statistical methodologies that can accurately forecast extreme wildfire occurrences across spatial and temporal dimensions is of great significance. In this thesis, we develop new spatial statistical models, combined with popular machine learning techniques, as well as novel extreme-value methods to enhance the prediction of wildfire risk using graphical models. First, in order to jointly efficiently model high-dimensional wildfire counts and burnt areas over the whole continguous United States, we propose a four-stage zero-inflated bivariate spatiotemporal model combining low-rank spatial models and random forests. Second, to model high values of the McArthur Forest Fire Danger Index over Australia, we develop a novel spatial extreme-value model based on mixtures of tree-based multivariate Pareto distributions. Our new methodology combines theoretically justified spatial extreme models with a computationally convenient graphical model framework to spatial problems in high dimensions efficiently. Third, we exploit recent advancements in deep learning and build a parametric regression model using graphic convolutional neural networks and the extended Generalized Pareto distribution, allow us to jointly model moderate and extreme wildfires observed on irregular spatial grid. We work with a novel dataset of Australian wildfires from 1999 to 2019, and analyse monthly spread over areas correspond to Statistical Area Level 1 regions. We highlight the efficacy of our newly proposed model and perform risk assessment for Australia and dense communities.

Flood risk management in the U.S. has contributed to overdevelopment in at-risk areas, increases in flood losses over time, significant deficits in federal emergency programs, and inequitable outcomes to households and communities. Addressing these issues in a cost-effective and socially equitable manner relies on the ability of policy analysts to identify and understand complex interactions that characterize coupled natural-human systems, and tools for accurate estimates of the risks that arise from these interactions. This dissertation addresses this need by developing and investigating a flood risk analysis system that integrates data on property locations, assessments and transactions, high resolution flood hazard models, and flood risk policy and impacts across the coterminous United States. We focus on the degree to which markets accurately value their exposure to flooding and its impacts, and the accuracy of procedures and tools to estimate flood losses. In the first chapter, we identify heterogeneous valuation of storm risk in the Florida Keys that depends on the presence of structural defense and proximity to damaged homes after Hurricane Irma. This result suggests that stranded assets, properties with increasing vulnerability to storms but unable to rebuild structures and recover wealth, and overvalued assets at risk, which raise disaster costs, can occur simultaneously. This runs counter to the common framing of competing drivers of observed market valuation. In the second, we show that conventional methods employed in flood loss assessments to achieve large spatial scales introduce large aggregation bias by sacrificing spatial resolution in inputs. This investigation adds important context to published risk assessments and highlights opportunities to improve flood loss estimation uncertainty quantification which can support more cost effective and equitable management. In the final chapter, we conduct a nationwide study to contrast the predictive accuracy of predominantly used U.S. agency flood damage prediction models and empirical alternatives using data on 846 K observed flood losses to single-family homes from 446 flood events. We find that U.S. agency models mischaracterize the relationships of losses at the lowest low and high inundation depths, for high-valued structures, and structures with basements. Evaluated alternatives improve mean accuracy on these dimensions. In extrapolation to 72.4 M single-family homes in the U.S., these differences translate into markedly different predictions of U.S.-wide flood damages to single-family homes. The results from this dissertation provide an improved empirical foundation for flood risk management that relies on the valuation and estimation of flood risk from county to continental scales.

Recently, increased attention has been given to the social and environmental context in which crimes occur (Wells & Weisheit 2004). This new interest in the human ecology of crime is largely demographic, both in terms of subject matter and increasingly in terms of the analytic methods used. Building on existing literature on the social ecology of crime, this study introduces a new approach to studying sub-county geographies of reported crime using existing census place and county definitions coupled with spatial demographic methods. Spatially decomposing counties into Census places and what Esselstyn (1953) earlier called “open country,” or non-places, allows for the development of a unique but phenomenological meaningful sub-county geography that substantively holds meaning in conceptualizing rural and urban localities in the demographic analysis of crime. This decomposition allows for the examination of core-periphery relationships at the sub-county level, which are hypothesized to act similarly to those at the national level (Agnew 1993; Lightfoot and Martinez 1995). Using 1990 to 2000 Agency-level UCR data within this approach, I propose to use spatial statistics to describe and explain patterns of crime across differing localities. Potential processes of spatial mobility in regards to the spread of criminal activity from places to non-place localities are also examined. In order to adequately understand these spatial patterns of crime while testing the usability of the new place-level geography, several of the generally accepted theories of crime and a number of explanatory factors and covariates are tested. Furthermore, using this sub-county geography, significant patterns of spatial diffusion and contagion are through the implementation and modification of the spatio-temporal model, which provides the current point of departure and put forth by Cohen and Tita (1999). The results are promising and suggest a meaningful contribution to the ecological analysis of crime and the larger sub-discipline of spatial demography.