This dissertation extends research that seeks a better understanding of the complex relationships between humans and the urban environment by focusing on one of the more pervasive topics in recent human-environment studies: the investigation and analysis of the connections between transportation and land-use. Currently, the multitude of environmental, economic, and social-welfare concerns incumbent to a society dependent on the automobile have compounded the need to further understand and develop models of these connections. By conceptualizing the urban environment as the locations of housing and the locations of jobs, or urban structure, this dissertation builds upon previous research that utilizes urban commuting to explore connections between transportation and land-use in US metropolitan regions. Motivated by the prospects of providing new insights into the relationship between commuting, sprawl, sustainability and the urban structure; this dissertation develops a methodology to assess and evaluate changes to the urban structure over time by synthesizing elements from both the planning and geographic literatures. The Model of Urban Structure and Evaluation of Change (MUSEC) presented in chapter 6 proposes that for a given city or region, changes to the urban structure can be modeled using homogenous data to model the urban structure and evaluated using the commuting carrying capacity to assess the changes. To better support those assumptions, two analytical chapters are presented exploring the role of homogenous data in commute studies (chapter 4) and the role of the commuting carrying capacity in urban structure assessment (chapter 5). The ability to assess urban structure changes will help broaden the understanding of the transportation/land-use connection and can provide planners, government officials, and geographers' knowledge to address prevalent urban issues such as sprawl and sustainable development. / A Dissertation submitted to the Department of Geography in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Summer Semester, 2012. / June 19, 2012. / Excess Commuting, Policy, Spatial Modeling, Transportation, Urban Sprawl, Urban Sustainability / Includes bibliographical references. / Mark Horner, Professor Directing Dissertation; Timothy Chapin, University Representative; Victor Mesev, Committee Member; Tingting Zhao, Committee Member.

My study explores the role of geography in the construction of cultural identity for place-based communities of practice. I combine recent advances in cultural geography, which focus on the role of cultural traces to culturally order and geographically border a place with recent research on communities of practice, which are communities characterized by a domain focus, an interpretive tradition, and a day-to-day practice. I demonstrate this by exploring the newly constructed town of Ave Maria in the US state of Florida, a community whose domain focus is the day-to-day practice of conservative Catholicism. The study uses a qualitative research methodology to determine the features of the town's landscape that promote the community's domain focus. It uses a quantitative research methodology to investigate the contributions that the spatial configuration of those features makes to the community's cultural identity. An ontology and knowledge base provide a systematic formalization of my qualitative data for subsequent use in quantitative analysis. My results indicate that the cultural ordering and geographical bordering of the community promote a high degree of homogeneity among residents along the community's domain axis. I also conclude with the finding that the community has developed a cultural district whose spatial configuration and location play important roles in the day-to-day lives of its residents. / A Dissertation submitted to the Department of Geography in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Fall Semester, 2012. / July 10, 2012. / communities of practice, cultural geography, cultural identity, cultural traces, landscape analysis, spatial configuration / Includes bibliographical references. / Jon Anthony Stallins, Professor Co-Directing Dissertation; Victor Mesev, Professor Co-Directing Dissertation; Karen L. Laughlin, University Representative; Mark W. Horner, Committee Member; James B. Elsner, Committee Member.

Outward urban growth, driven by increasing population, economic development, and technological advancement, has become a worldwide phenomenon. Such growth is often viewed as the vitality of a regional economy. But it has brought negative impacts on the environment such as biodiversity loss, soil erosion, hydrological perturbation, water and solid pollution, and global warming. Monitoring and modeling urban spatial growth are important for environmental sustainability and urban planning. This dissertation research has aimed at the investigation of urban growth patterns, urban growth processes, and their relevance through the lens of complexity theory to improve our understanding of the spatial and temporal dynamics of urban growth in a rapidly growing metropolitan area. Central to this research effort is the development of a technological framework that tightly integrates satellite imagery processing, artificial intelligence, and geographic information systems (GIS). Specifically, this project includes two principle components. One is to examine the use of artificial neural networks for improving urban land cover change detection from remote sensor data. Due to their capability of dealing with nonlinear and complex phenomena, integrating artificial neural networks with remote sensing has improved the performance of image classification for the fragmented and heterogeneous landscape in an urban environment. The other component is to characterize urban spatial growth at the metropolitan, functional zone, and cell levels by using three approaches: urban land change mapping, landscape metrics analysis, and moving windows analysis. This part of the research has provided insights into urban growth dynamics in urban societies that are not comparable to either industrial or post-industrial cities in the United States through measuring the spatial and temporal variations of urban patterns and processes at different scales. These societies have unique urban forms and development trajectories due to technological robustness and contemporary international and domestic socio-economic conditions. / A Dissertation submitted to the Department of Geography in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Spring Semester, 2012. / March 14, 2012. / artificial neural networks, GIS, landscape metrics, remote sensing, urban growth pattern / Includes bibliographical references. / Xiaojun Yang, Professor Directing Dissertation; Timothy S. Chapin, University Representative; James B. Elsner, Committee Member; Jon A. Stallins, Committee Member; Trajco V. Mesev, Committee Member.

Hurricanes threaten the United States every year. It is important to quantify the risk of these events for emergency managers. Extreme value statistics are used to model hurricane characteristics at different locations. Using wind speeds over a specified threshold, the risk of extreme winds are estimated in twelve Florida cities. The risk estimates are provided as statistical return periods, or the expected frequency of specific hurricane magnitudes. Results show that the city of Miami can expect to see hurricane winds blowing at 50 ms-1 (45.5-54.5) [90% CI] or stronger, on average, once every 12 years. In comparison, the city of Pensacola can expect to see hurricane winds of 50 ms-1 (46.9-53.1) [90% CI] or stronger once every 24 years. Hurricanes in the vicinity of Florida are found to be increasing in intensity over time as a product of higher offshore intensification rates. The risk of hurricane strikes can be further understood by including an additional variable, storm surge. Using observational data, the joint probability distribution of hurricane wind speeds and storm surge is found at three Air Force Bases along the U. S. Gulf coast. The quartile pointwise uncertainty is quantified using a Monte Carlo procedure. Eglin Air Force Base can expect wind speeds blowing at 50 ms-1 and surge heights of 3 m, on average, once every 28 years (23-36). MacDill Air Force Base can expect wind speeds blowing at 50 ms-1 and surge heights of 3 m, on average, once every 27 years (22-34). Keesler Air Force Base can expect wind speeds blowing at 50 ms-1 and surge heights of 3 m, on average, once every 15 years (13-18). Utilizing a spatial tessellation across the North Atlantic and Gulf of Mexico provides additional insight into the risk of hurricane strikes. Parameters from the extreme value model are mapped across space to visualize patterns. Sea surface temperature is included as a covariate in a geographically weighted regression model with each parameter. It is found that as sea surface temperatures increase, the expected hurricane wind speed for a given return period also increases. / A Dissertation submitted to the Department of Geography in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Spring Semester, 2012. / March 15, 2012. / Extreme Value, Hurricane, Risk / Includes bibliographical references. / James Elsner, Professor Directing Dissertation; Joseph Donoghue, University Representative; Victor Mesev, Committee Member; Tingting Zhao, Committee Member.

Urban cloud-to-ground (CG) lightning enhancement has been well documented for Atlanta, Georgia. This study builds on those investigations using modeling techniques. Numerous styles of analyses and regressions were conducted to establish patterns of CG lightning over the North Georgia region. CG lightning demonstrated clustering for all years of data: 1995-2008. However, the first strike of each day with lightning was randomly distributed according to a Poisson distribution, demonstrating the clustering is not due to permanent features. Attempts were unsuccessful to model CG lightning clusters as either a Matern or Thomas Poisson point process. Regressions of CG lightning with built environment covariates- FAA aviation obstacle locations and heights, population density, road length density, distance to the center of Atlanta, PM10 emissions data, distance to highways, and coal plant locations- as well as natural variables such as projected coordinate easting, northing, and NWS severe thunderstorm status were executed at resolutions of 1km, 2km, 4km, and 8km. Analyses demonstrated significantly higher flash frequency near FAA aviation obstacles. With an R2 value of 0.22, taller obstacles are struck more frequently than shorter obstacles. Regressions with road length density revealed little explanatory power (maximum R2=0.19), but demonstrated a positive correlation independent of scale. A multi-level visualization technique demonstrates the road length density correlation loses accuracy within dense urban corridors. Distance from Atlanta shows a negative correlation, but only at larger scales. Subsetting both regressions by direction reveals a significant difference on the Eastern and Western sides of Atlanta. Subsetting both regressions only to Gwinnett County, Georgia illustrates road length density has no correlation with flash frequency, and distance to Atlanta is still a scale dependent process. PM10 emissions analysis suggests that CG amplification is most likely to occur between 15 and 33µg/m3, but the data has significant flash frequency variability even within these PM10 levels. Immediate proximity to highways proves not to be a significant variable in modeling flash frequency at any resolution. Coal plant proximity has the potential to enhance CG flash frequency, but the inherent variability in frequency precludes a strong p-value compared to randomly generated areas. However, if the data is subset by flashes/day over 500, then the areas close to coal plants have significantly more lightning (p=2.19e-5). Using a subset of the twenty-five highest frequency flash days in three equal area study areas (Haralson and Polk Counties representing rural, Cobb County representing developed and upwind of Atlanta, and Gwinnett County representing downwind and developed) Mann-Whitney tests are completed to determine if NWS severe thunderstorm storms are significantly different in CG flash frequency. In each study area, the NWS severe thunderstorms are not significantly different in flash frequency than non-severe storms. This analysis also suggests spatial tendencies of high frequency storms in each area. This multi-scale analysis also suggests that when examining CG lightning, more than one scale of examination should be used. Some processes of lightning amplification appear to occur at very local scales (500m), whereas others are coarser (up to 8km). There appears to be no goldilocks scale of analysis for CG lightning. However, if only one resolution is to be used, 2km is recommended. / A Thesis submitted to the Department of Geography in partial fulfillment of the requirements for the degree of Master of Science. / Summer Semester, 2012. / June 20, 2012. / Atlanta, CG Lightning, ggplot, MAUP, NLDN, Urban Lightning / Includes bibliographical references. / James B. Elsner, Professor Directing Thesis; Victor Mesev, Committee Member; Tetsuo Kobayashi, Committee Member.

This dissertation presents and evaluates the use of geographic techniques for understanding the role of environmental variables as they relate to the odds of settlements being targeted for attack within a civil conflict. Geospatial technologies are demonstrated for their utility in deriving environmental variables that are entered into statistical models in order to explore relationships between the environmental landscape variables and conflict incidence. The methodologies presented offer novel approaches for the disaggregation of spatial as well as longitudinal variation in statistical relationships when applied to the sub-national analysis of conflict, for which most hypotheses do not account for the timing of when factors will be relevant or for describing geographical variation in the strength of these relationships. This dissertation contributes to research on the sub-national analysis of conflict by demonstrating techniques that foster deeper consideration of how spatial and temporal variation may impact otherwise static theory. / A Dissertation submitted to the Department of Geography in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Fall Semester, 2011. / September 22, 2011. / Conflict, Counterinsurgency, Darfur, GIS, Remote Sensing, Subnational / Includes bibliographical references. / Tingting Zhao, Professor Directing Dissertation; William H. Moore, University Representative; Victor Mesev, Committee Member; James Elsner, Committee Member.

The official tornado report record is limited by a significant bias. Termed the "population bias", it communicates misleading climatological trend information by displaying an increasing number of reports from 1950 through the early 2000s. This increase in reports is most likely due to increasing technology and awareness regarding tornadoes. This led to more reports being made and less being missed, as opposed to an increase of tornado events. We propose a method for quantifying this population bias in four regions, the Central Plains, Northern Plains, Southeast, and Midwest using bounding boxes spanning multiple states in each region. A spatial model is developed, which calculates the distance from the nearest city on a 128 X 128 grid within each region. The city locations are based on the 1990 census. The tornado report density is also computed at each grid point, and we explore the relationship between report density and distance from the nearest city. We then use that relationship to quantify the population bias as the ratio of the maximum report density to minimum report density in the vicinity of a city. We find that this population bias has a general decrease throughout the record, and the most significant decrease is after the induction of radar in the early 1990s. The bias has diminished for all regions within the most recent years of the record. We find that the changing nature of this population bias has a regional dependency, which must be addressed when considering a model with a population bias term in future studies. We also speculate that we are now reporting all tornadoes in all regions. / A Thesis submitted to the Department of Geography in partial fulfillment of the requirements for the degree of Master of Science. / Summer Semester, 2013. / June 28, 2013. / geography, population bias, spatial model, tornado / Includes bibliographical references. / James Elsner, Professor Directing Thesis; Jay Baker, Committee Member; Chris Uejio, Committee Member.

This dissertation starts by evaluating the applicability of using a commercially available, cost-effective, sidescan sonar system to detect benthic habitats, in particular hardbottom habitats, in the nearshore northeastern Gulf of Mexico. Hardbottom habitats are likely to function as essential fish habitat for several important fisheries species. Identifying and mapping these habitats is a crucial first step to learn more about their roles in sustaining the associated fisheries species and their ecological importance in harboring biodiversity. The locations of nearshore hardbottom habitats in the Gulf of Mexico are largely unknown in part because mapping the seafloor is generally an expensive and time-consuming process. To illustrate the capability of low-cost devices in mapping benthic habitats, I tested the Humminbird 997c SI unit marketed to fishermen at a cost of approximately $2,000. Methodological approaches to effectively capture and process the Humminbird sidescan imagery were developed. Humminbird sidescan data from three sites were compared to overlapping sidescan imagery acquired by the National Marine Fisheries Service using a standard, much more expensive (~$20,000) Marine Sonic system. This analysis verified that the classification results of sand and hardbottom habitats based on data collected using the Humminbird sidescan system were similar to those produced using the traditional and more expensive Marine Sonic sidescan equipment. The similarity of the hardbottom and sand classes extracted from both datasets was large for all three sites with overall accuracies of 86.2%, 78.5%, and 81.5%. The results indicate that the Humminbird system can be used independently to create valid benthic imagery of nearshore marine habitats. Thirty-three sites in total were then mapped with the Humminbird system and sampled using dive surveys. Seascape pattern metrics were calculated from the classified Humminbird sidescan maps. The dive survey data included measurements of the geomorphology, physical attributes of the water column (e.g. temperature, depth, and visibility), and coverage and heights of the benthic biota. The coverage and heights of the biota were compared to the geomorphology, seascape, and water column variables to identify patterns in the distribution and community composition of the sessile organisms. Sponges and red algae were dominant at the western sites while hard corals and brown algae dominated the eastern sites. A cluster analysis revealed four communities, each with unique species indicators: group 1 - bivalves and bryozoans, group 2 - sea urchins and sponges, group 3 - red algae, group 4 - brown algae. Groups 1 and 4 in the east were higher in diversity than groups 2 and 3 in the west. The cluster analysis results also showed the longitudinal pattern of sponge and red algae communities (groups 2 and 3) to the west and brown algae (group 4) to the east. Within the study area, visibility was found to vary with longitude. Sites in the east showed higher visibility than sites in the west and this may be driving the community patterns that were identified. Relationships were identified between the four most abundant taxa (sponges, hard corals, brown algae, and red algae) and the geomorphology, physical, and seascape variables. However, the relationships were often complicated and the biota did not strictly follow gradients or boundaries in substrate or geoform (physical feature or landform), even though these features are often used to classify habitats and biotopes. The percent cover of rock was a significant geomorphology variable for red algae and hard coral coverage while geoforms were related to the heights of sponges and brown algae. Seascape metrics also had significant effects on the sessile biota particularly related to patch edges, heterogeneity, core areas, nearest neighbor distances, and the percent cover of hardbottom. Despite the fact that sessile organisms do not move much, if at all following their planktonic larval stage, the surrounding seascape contributes to the patterns we see in their distribution, coverage, and heights. The third chapter focuses on applying a new classification standard to the benthic habitats in the nearshore northeastern Gulf of Mexico. The United States Geological Survey (USGS) has a standardized system for classifying terrestrial and aquatic habitats found across the U.S. which has been in place for almost 40 years. This classification standard does not include marine and most coastal habitats. Therefore, marine researchers developed a number of classification systems for coastal and marine habitats relevant to their local or regional studies in U.S. waters. A national standardized method for classifying marine and coastal habitats was not adopted until recently. The Coastal and Marine Ecological Classification Standard (CMECS) developed by the Federal Geographic Data Committee was approved last year and is intended to fill the gap in U.S. marine habitat classification standards. Since the classification standard is in its infancy, it has not been applied in many geographic areas. My third chapter is the first study to apply the CMECS to the benthic habitats in the nearshore northeastern Gulf of Mexico off the coast of northwest Florida. Hardbottom and sand habitats are characteristic of this area. In the previous chapter, the underwater surveys revealed that the dominant taxa at the sites within the study area were hard corals, sponges, and macroalgae. I used CMECS to broadly classify the sites where the surveys were completed. I found that habitat heterogeneity and a wide variety of environmental characteristics influenced the distribution of taxa at the local scale. This made applying CMECS at scales finer than the composite study area unfeasible without major modifications. CMECS worked well for classifying the broad scale in this region but was not appropriate for classifying complex fine-scale biotopes. / A Dissertation submitted to the Department of Geography in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Summer Semester, 2013. / June 14, 2013. / classification systems, Gulf of Mexico, hardbottom, sessile biota, sidescan sonar, spatial analysis / Includes bibliographical references. / Tingting Zhao, Professor Directing Dissertation; Markus Huettel, University Representative; Xiaojun Yang, Committee Member; Christopher Uejio, Committee Member.

North Atlantic hurricanes constitute a threat to both life and property. The warm seas found in tropical low-latitudes provide a breeding ground for hurricanes, with nearly continuous heat and moisture fluxes into near-surface air. Traditionally, the sun's role in hurricane climate studies is acknowledged as a time-marker for ocean heat content, with calendar date predicting hurricane frequency and intensity. However, a series of investigations into a different type of sun-hurricane relationship has uncovered a link between solar activity and hurricane intensity and frequency. High solar activity at a daily timescale is understood to weaken hurricanes in the southwest Atlantic yet correspond to increased hurricane intensity in the southeast Atlantic. At a seasonal timescale, high solar activity is shown to correspond with fewer U.S.-landfalling hurricanes. A gap in the knowledge exists on how and where solar activity influences seasonal hurricane frequency over and within the North Atlantic basin. This study is quantitative featuring exploratory analysis and inferential modeling, with diagnosis and prediction of the sun-hurricane count relationship over space being the primary contribution to science and society. It is carried out via exploratory data analysis and statistical modeling. Hurricane and climate data are binned in equal-area hexagon regions. Count differences for periods of high solar activity (i.e, high sunspot number) feature fewer hurricanes across the Caribbean, Gulf of Mexico, and along the eastern seaboard of the United States when sunspots are numerous. In contrast, fewer hurricanes are observed in the central North Atlantic when sunspots are few. The sun-hurricane connection is as important as the El Ni\~no Southern Oscillation toward statistically explaining regional hurricane occurrences. Regression results indicate a 30\% reduction in probability of annual hurricane occurrence for southeastern Cuba, the southern Bahama islands, Haiti, and Jamaica when the September sunspot number is 115 sunspots. In contrast, hurricane risk in regions of the southeastern Atlantic is predicted to more than double when the September sunspot number is 160 sunspots. Regions within the southwest Atlantic indicate a negative relationship. A physical explanation for the eastern basin increase in counts and count probability is still unclear. Additional warming of the sea surface in these regions from increased solar activity would lead to increased hurricane frequency. However, the sea-surface temperature response to solar activity appears marginal. Future work will address potential explanations including circulation changes to African weather systems in response to changes in solar activity, and earlier hurricane development leading to more storms leaving the deep tropics and tracking into these regions due to coriolis effects on the storm. The study can be expanded to include storms worldwide. / A Dissertation submitted to the Department of Geography in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Spring Semester, 2013. / March 29, 2013. / hierarchical model, hurricane, relative risk, sunspots / Includes bibliographical references. / James Elsner, Professor Directing Thesis; Robert Hart, University Representative; Victor Mesev, Committee Member; Tetsuo Kobayashi, Committee Member; Thomas Jagger, Committee Member.

Environmental justice studies tend to deemphasize political economic processes, social relations between various social groups, scale and the social constructions of environmental issues which hide environmental injustice. Political ecology, in contrast, emphasizes issues of social relations, scale, political economy, marginalization, and social constructionism, which are useful for identifying the environmental injustice often hidden by both developmentalist and mainstream environmentalist approaches to environmental conflict. Using a political ecology framework, this research uncovers the environmental injustice hidden in conflicts over coal plant installation and expansion on Yeongheung Island, South Korea. Methods include analysis of public reports and statements and qualitative interviews with 20 local residents. Even though the residents live in poverty and suffer the effects of coal-dust, noise from transmission towers, and tidelands destruction, environmental justice has little presence in environmental debates and decision-making. Instead, these derive from dominant social relations among developers, governments and environmentalists, with the local residents excluded, marginalized and dis empowered from the discourses. At the scale of national and global capitalism, localized environmental injustice is a consequential result of global and national scale of political economy. In addition, the environmental injustice tends to be hidden by social constructions of coal energy, nature and environment and scale, which are manipulated by developers, environmentalists and governments. Political ecology provides the theoretical lenses to see how money floats upward and pollution sinks downward as environmental injustice is produced and hidden. / A Dissertation submitted to the Department of Geography in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Fall Semester, 2009. / July 13, 2009. / Energy Issue Environmental Study, Geography, Political Ecology, Environmental Justice / Includes bibliographical references. / Dan Klooster, Professor Directing Dissertation; Tony Stallins, Professor Directing Dissertation; Ivonne Audirac, Outside Committee Member; Barney Warf, Committee Member; Jonathan Leib, Committee Member.