Dr. John D Morgan

Forest disturbance due to natural events, such as wildfires, represents an increasing global challenge that demands advanced analytical methods for effective detection and mitigation. To this end, integrating satellite imagery with deep learning (DL) has emerged as a powerful approach for forest wildfire detection; however, its practical use remains limited by the scarcity of large, well-labeled satellite imagery datasets. In this study, we address this issue by presenting the California Wildfire GeoImaging Dataset (CWGID), a high-resolution bi-temporal collection of over 100,000 labeled RGB "before-" and and "-after" Sentinel-2 wildfire satellite image pairs. We build and label the dataset programmatically, significantly reducing the time and manual effort usually required to create labeled datasets suitable for DL applications. Our methods include data acquisition from authoritative sources, systematic preprocessing, and an initial analysis using three pre-trained Convolutional Neural Network (CNN) architectures for two classification tasks consisting, respectively, in labeling unitemporal and bitemporal inputs as damaged or not damaged by fire. Our results show that using bi-temporal imagery as input during model training and testing can result in improved model performance, with the Early Fusion (EF) EfficientNet-B0 model achieving the highest wildfire detection accuracy of over 92%. These findings suggest that the CWGID and the streamlined programmatic methodology used to build it may help address the scarcity of labeled data for DL-based forest wildfire detection, while providing a scalable resource that could support other DL applications in environmental monitoring.

This research note provides an exploratory analysis of a remarkable - and broadly available - set of documents from the early twentieth century. This preliminary study examined one month of police call logs and warrant books from Pensacola, Florida, in December 1912. Combining perspectives from history, geography, and police practice, the authors integrated these records with digitised spatial data using Geographic Information System (GIS). The analysis revealed significant racial disparities in arrest patterns, unexpected temporal trends in policing activity, and compelling spatial concentrations of law enforcement interactions. This project suggests the potential of applying modern analytical techniques to historical police datasets in order to yield meaningful insights into the evolution of policing practices and their societal impacts. It highlights the wealth of untapped historical arrest records available for similar studies across the United States, suggesting avenues for future comparative research that bridges historical analysis with contemporary policing concerns.

Even more than asserting that geography matters (in the sense that the unique features of place and spatial context directly shape the outcomes of general ('universal') economic, social and cultural processes) geographers have shown the iterative dynamic of place – not just as a container, but as a process. [...]the apparent neutrality of categorisations which we have used 'objectively' to classify place in the past is now understood to be false: terms like 'inner city' are not descriptively neutral, but carry connotation and meaning which need to be examined with care: because the way in which racialised perspectives become sedimented in our minds is the way in which racism becomes normalised. Teaching geography with racial literacy In summary, our argument is that teachers' racial literacy involves a body of knowledge and the ability to understand how race and racism work in shaping society – and that for geography teachers, this has implications for their curriculum making. In other words, through neglecting the unequal distribution of power, diverse perspectives and the agency of ordinary people, geographical patterns and processes will appear to students as natural, inevitable, logical – and therefore legitimate. [...]a great advance in school geography would be more overtly to recognize the historicity of geographical patterns – the history behind the geography. Principles To conclude, we outline six principles that, we argue, underpin a racially literate geography curriculum: * People and places are dynamic and always changing as a result of the interplay of economic, social, cultural, political and environmental processes; furthermore, these do not operate neutrally and in the same way for all people. * Geographical facts are nearly always contingent: they are selected, prioritized and can frequently be contested; and there is nearly always another way of looking at them (the 'danger of a single story' – Biddulph, 2011). * Human agency is rarely unfettered and frequently involves disagreement, tension and struggle: human processes always involve politics (the process by which limited resources are allocated). * Race is not a biological fact, attribute or phenomenon, but it is a real and 'felt' social construction that both produces and is a product of economic, environmental, political and social processes that

Historically, dense vegetation cover near buildings has caused power disruptions during weather phenomena. These types of severe storms impact the coast of Florida each year. However, challenges exist for obtaining both power outage data and calculating the impact of tree cover. NASA's Nighttime Lights, Black Marble, VNP46 product is utilized to analyze the natural and built environments. One aspect of the built environment that can be mapped with the Black Marble data is the megawatts of electricity used by the electrical power grid based on the magnitude of emitted nighttime light energy. This paper discusses using Black Marble data and other landscape variables within a probabilistic model to examine spatial patterns and map electricity outages with Bayesian networks. The research results indicate a high probability of a significant power outage when dense vegetation is present, but nuances in our natural and built environments like electric substations and land cover type alter the chance of reducing energy emissions.

In the United States, the Centers for Disease Control and Prevention (CDC) is the national agency that conducts and supports public health research and practice. Among the CDC’s many achievements is the development of a social vulnerability index (SVI) to aid planners and emergency responders when identifying vulnerable segments of the population, especially during natural hazard events. The index includes an overall social vulnerability ranking as well as four individual themes: socioeconomic, household composition & disability, ethnicity & language, and housing & transportation. This makes the SVI dataset multivariate, but it is typically viewed via maps that show one theme at a time. This paper explores a suite of cartographic techniques that can represent the SVI beyond the univariate view. Specifically, we recommend three techniques: (1) bivariate mapping to illustrate overall vulnerability and population density, (2) multivariate mapping using cartographic glyphs to disaggregate levels of the four vulnerability themes, and (3) visual analytics using Euler diagrams to depict overlap between the vulnerability themes. The CDC’s SVI, and by extension, vulnerability indices in other countries, can be viewed in a variety of cartographic forms that illustrate the location of vulnerable groups of society. Viewing data from various perspectives can facilitate the understanding and analysis of the growing amount and complexity of data.

Passive design and landscape variables (e.g., rooftop albedo and shading vegetation) are frequently proposed as important green building techniques. However, there is a paucity of literature demonstrating their large-scale effects with empirically measured building stocks and observed residential energy consumption. This paper uses a spatial Durbin error model (SDEM) approach to test the effects of passive building performance indicators, such as orientation, albedo, and NDVI, on city-wide summertime household billed energy data in Gainesville, FL. Our findings suggest that vegetation and albedo reduce energy consumption, but our model did not produce similar significant results for building orientation and footprint compactness. These results provide evidence to suggest high albedo roofing and purposeful shading are important energy conservation strategies for energy-efficient residential neighborhoods.
•This study uses a spatial Durbin error model (SDEM) to test the effects of passive design elements on energy consumption patterns in Gainsville, FL.•NDVI, as a proxy for near home vegetation and shade, has significant direct and indirect (spillover) effects on reducing summertime energy consumption.•Rooftop albedo (reflectivity) has significant direct effects on reducing summertime energy consumption, even in a context where there are few “high albedo” roofs.•Roof-top albedo is emperically correlated with reductions in energy consumption in our study sample.•Building codes may discourage maximizing energy reductions from passive design by setting minimum standards which may lead to offsetting.

Trumbo’s (1981) ideas on bivariate choropleth design have been underexplored and underutilized. He noted that effective map design (including color selection) is directly informed by the intended goal or use of the map (i.e., what questions might the map answer), and he identified three common spatial relationships that can be displayed by a bivariate choropleth: inverse relationships, a range of one variable within another, and direct relationships. Each is best suited to answering different map readers’ questions. Trumbo also suggested sample color palettes to focus the map reader’s attention on pertinent data. In consultation with Trumbo, we extended his ideas, first by creating focal models that illustrate his three spatial relationships. We then constructed sample maps to examine each of the focal models, and finally compared each model by mapping the same two data sets (of obesity and inactivity). We investigated the visual differences in each of the resulting maps, and asked spatial questions regarding the relationships between obesity and inactivity. Our work validates Trumbo’s ideas on bivariate choropleth map design, and we hope our focal models guide cartographers towards making color choices by linking their map purpose to the appropriate focal model.

With the increasing availability of web mapping application programming interfaces, those tasked with the job of developing and designing web-based geographic information systems (Web GIS) are faced with the question of how to (re)develop new and existing capabilities. While Web GIS often reproduces features currently available within desktop GIS, usage requirements often demand it be both easier to use and more specific in its purpose. As GIS has become more integrated with evolving web technologies, many semantic ways in which GIS functions depend on specific user interactions (UX) where underlying functions are abstracted to user interfaces (UI). Our study explores the utilization of concept maps to facilitate learning about Web GIS within the context of an online graduate GIS course. Specifically, our project involves a case study that illustrates the following: 1) a design and implementation of having students create concept maps as part of the learning process around building custom Web GIS; 2) a discussion of how an instructor can approach the assessment of student-built concept maps; 3) an illustration of how student-build concept maps can encourage student interaction specific to course content.

Purpose The southeast of the United States (USA) is a high stroke prevalence area otherwise known as the BStroke Belt™ in order to characterize the increased stroke morbidity and mortality rates found there. The purpose of this study was to characterize the
relationship between socio-cultural factors and dietary habits related to stroke occurrence within the affected states.
Methods The 2015 Behavioral Risk Factor Surveillance System (BRFSS) data was analyzed for both bivariate and multivariate models aimed at studying the interaction between socio-cultural factors and the prevalence pattern of stroke in the southeastern area of the USA.
Results Overall, 4% of individuals who lived in the US southeastern states of Alabama, Florida, Georgia, Louisiana, and Mississippi had stroke. Of these, Mississippi had the most stroke victims, followed by Alabama, Louisiana, Georgia, and Florida, with 5.1%, 4.7%, 4.5%, 4.3%, and 3.4% respectively (χ2 = 18.68 and p value < 0.01). The logistic regression showed that individuals who consumed vegetables every day were less likely to be stroke victims than those who did not consume vegetables on a daily basis, with aOR = 0.74; CI = 0.59–0.91; p value < 0.01. Surprisingly, individuals who drank alcohol were less likely to report stroke than those who did not drink alcohol (aOR = 0.63; CI = 0.51–0.79; p value < 0.001).
Conclusion Characterization of factors associated with stroke prevalence in a region of the USA known for its adverse stroke rates is essential for offsetting the burden of this public health issue and for promoting health.