abstract: In the field of Geographic Information Science (GIScience), we have witnessed the unprecedented data deluge brought about by the rapid advancement of high-resolution data observing technologies. For example, with the advancement of Earth Observation (EO) technologies, a massive amount of EO data including remote sensing data and other sensor observation data about earthquake, climate, ocean, hydrology, volcano, glacier, etc., are being collected on a daily basis by a wide range of organizations. In addition to the observation data, human-generated data including microblogs, photos, consumption records, evaluations, unstructured webpages and other Volunteered Geographical Information (VGI) are incessantly generated and shared on the Internet. Meanwhile, the emerging cyberinfrastructure rapidly increases our capacity for handling such massive data with regard to data collection and management, data integration and interoperability, data transmission and visualization, high-performance computing, etc. Cyberinfrastructure (CI) consists of computing systems, data storage systems, advanced instruments and data repositories, visualization environments, and people, all linked together by software and high-performance networks to improve research productivity and enable breakthroughs that are not otherwise possible. The Geospatial CI (GCI, or CyberGIS), as the synthesis of CI and GIScience has inherent advantages in enabling computationally intensive spatial analysis and modeling (SAM) and collaborative geospatial problem solving and decision making. This dissertation is dedicated to addressing several critical issues and improving the performance of existing methodologies and systems in the field of CyberGIS. My dissertation will include three parts: The first part is focused on developing methodologies to help public researchers find appropriate open geo-spatial datasets from millions of records provided by thousands of organizations scattered around the world efficiently and effectively. Machine learning and semantic search methods will be utilized in this research. The second part develops an interoperable and replicable geoprocessing service by synthesizing the high-performance computing (HPC) environment, the core spatial statistic/analysis algorithms from the widely adopted open source python package – Python Spatial Analysis Library (PySAL), and rich datasets acquired from the first research. The third part is dedicated to studying optimization strategies for feature data transmission and visualization. This study is intended for solving the performance issue in large feature data transmission through the Internet and visualization on the client (browser) side. Taken together, the three parts constitute an endeavor towards the methodological improvement and implementation practice of the data-driven, high-performance and intelligent CI to advance spatial sciences. / Dissertation/Thesis / Doctoral Dissertation Geography 2018

Contact with nature confers numerous benefits to mental and physical well-being, including enhanced recovery from stress and mental fatigue. Given that stress and mental fatigue likely interfere with performance in school, vegetation in student environments, an element of nature, could support academic achievement and greater educational attainment, a critical component of social mobility and therefore well-being. However, such benefits may be reduced for urban populations, as living in cities often entails reduced access to environments with vegetation, access which may be even lower for minority urban populations. The distribution of vegetation in urban areas therefore has serious implications for well-being and justice in the distribution of environmental amenities, particularly given the increasing majority of humans in cities. Several studies have identified positive relationships between vegetation and various indicators of school-level achievement. However, findings are inconsistent across studies, and in some cases suggest that vegetation may be detrimental to student progress. This research aimed to assess relationships between vegetation and school-level academic achievement and how they vary with vegetation type and social and environmental context, and in so doing, addressed inconsistencies in the literature. This dissertation consists of three studies. The first study identified positive relationships between tree canopy coverage and third-grade reading proficiency at a metropolitan extent. Spatial autoregressive models were applied to account for residual spatial autocorrelation when necessary. The second study examined relationships between vegetation and high school graduation and reading and mathematics rates using a sample of high schools from across the continental US. However, negative binomial mixed-effects models were unable to provide sufficient evidence to suggest academic benefits of vegetation at a national extent, although negative relationships between tree cover and graduation rate were observed for schools serving low-socioeconomic status (SES) attendance areas with greater proportions of African-American students. The third study applied a K-means statistical learning algorithm and mixed-effects beta regression models to a dataset covering the same sample of high schools from the second study to investigate the implications of social and environmental context for academic benefits of vegetation in more detail. Positive relationships between tree cover and agricultural vegetation and graduation rate were identified. The positive association with tree cover was stronger for high-SES populations and low-SES Latino/a populations. A negative association between non-forest vegetation and graduation rate was observed for low-SES, African-American populations. These findings suggest academic benefits of vegetation in student environments at both a metropolitan and national extent, and that those benefits vary according to social and environmental context. Thus, landscape design and management approaches that incorporate vegetation as a resource to support academic achievement should carefully consider the social characteristics of the populations such approaches intend to benefit, as well as the environmental contexts within which those populations live. This work has revealed several opportunities for future research. Those opportunities include investigating which age-groups might benefit most from urban greening with the intent of supporting academic achievement, and investigating the mechanisms behind social and environmental variation in the academic benefits of vegetation.

This study explores various relationships of streamflow and stream temperature over the Portland Metropolitan area in two urbanizing watersheds. Four stream temperature and discharge metrics were derived from USGS stream gauges in the Tualatin River and Johnson Creek watersheds and were analyzed for monotonic trends. Additionally, this study explored the sensitivity of stream temperature to air temperature and streamflow to assess where locations throughout the watershed may be more sensitive to these changes. Relationships among stream temperature, air temperature, and streamflow were assessed using linear and nonlinear bivariate regression for yearly values and summer months. Additionally, this study seeks to explain the spatial variations of thermal sensitivity throughout the Johnson Creek watershed using predictors derived using different weights at the contributing watershed scale and the buffer scale. Results indicate significant increasing trends in stream temperature metrics at various locations throughout the study area. Decreasing baseflow does not appear to coincide with increasing temperature metrics. Significant increasing trends in October and November are present in runoff ratio and TQmean. In both watersheds, air temperature appears to have a greater influence than streamflow on stream temperature, though the addition of discharge generally improves model fit. Increasing thermal sensitivity in Johnson Creek is related to increasing and decreasing standard deviation of slope, increasing mean slope, increasing open water and wetlands, less forest area, increasing standard deviation of NDVI, decreasing restoration area, increasing gray infrastructure density, and increasing upstream flow length. At most, ordinary least squares explained 30% of the variance in thermal sensitivity when only including stream temperature monitoring locations in the mainstem of the creek. Modelling tributary only stream temperature monitoring locations used a variety of watershed, buffer-scale, areal average and inverse distance weighted variables. The findings of this study highlight the importance of temporal scale and complex hydro-climatic influences along an urban-rural gradient in assessing patterns of discharge and temperature. These results have important implications for watershed managers, local agencies, and stakeholders who have worked to restore Johnson Creek and help to guide future water quality planning throughout the watershed.

Precipitation timing and magnitude is essential to human, ecological, and economic systems. Climate change may be altering the character of precipitation locally to globally, thus it is vital that resource managers, practitioners, and decision makers understand the nature of this change. This thesis was conducted in partnership with the City of Portland Bureau of Environmental Services (BES), and the Portland Water Bureau (PWB) in order to support resiliency planning around precipitation and precipitation extremes. This work has two primary phases, which are discussed in chapter 2 and 3 of this thesis. The first phase of this research entails characterization of the large-scale meteorological patterns (LSMPs) associated with high hourly intensity and heavy daily accumulation of precipitation over Portland, OR. Heavy precipitation is associated with a multitude of impacts on urban environments, thus it is important to understand the meteorological drivers behind these events. This phase of work describes the range of meteorological patterns associated with heavy precipitation totals and high intensity precipitation days over the city of Portland, Oregon. The range of large-scale meteorological patterns (LSMPs) associated with high intensity precipitation days are clustered using the self-organizing map (SOM) approach and are defined using sea level pressure, 500 hPa geopotential height, and 250 hPa wind. Results show that an array of LSMPs are associated with heavy precipitation days, the majority of which occur in fall and winter, usually driven by extratropical cyclones and associated atmospheric rivers. Spring and summer heavy and high intensity precipitation days, while less common than in fall and winter, are typically related to upper level disturbances. Examination of two case studies, one occurring in summer and one in winter, supports the ability of the SOMs approach to realistically capture key observed storm types. Methods developed here may be extensible to other locations and results build an observational foundation for validating the ability of climate models to simulate the LSMPs associated with local extremes. The second phase of this thesis involves evaluation of the fifth phase of the Coupled Model Intercomparison Project (CMIP5) to simulate wet season LSMPs and associated precipitation in the Pacific Northwest of North America. As in the first phase, LSMPs are identified using the self-organizing maps (SOMs) approach, except in this phase all wet season days are included, and defined with sea level pressure, 500 hPa geopotential height, and 250 hPa wind speed. Using SOMs, the range of LSMPs over the region is constructed with reanalysis, providing the target for the multi-model evaluation. Overall, the CMIP5 models are able to reproduce reference LSMPs with reasonable fidelity, though the low pressure LSMPs are generally captured better than the ridging patterns. Furthermore, there is a hierarchy in model ability to capture key LSMPs, with some models exhibiting overall higher fidelity than others. To further evaluate model fidelity, precipitation associated with the LSMPs is evaluated. In general, the observations, reanalysis, and CMIP5 models agree on the LSMPs associated with wet and dry days, but wet patterns are captured somewhat better than dry patterns. The LSMPs associated with the driest and wettest conditions in the PNW are generally overrepresented, while the LSMPs associated with light average daily precipitation across the pacific northwest are underrepresented in the models. Results provide a mechanistic perspective on model fidelity in capturing synoptic climatology and associated precipitation characteristics across the PNW. This research focuses on Portland and the Pacific Northwest, but has helped to develop methodology that is extensible to any location. The first phase gives us target LSMPs to understand future extreme precipitation over Portland, and the second phase of work lays the groundwork for developing projections of future changes to precipitation and precipitation extremes.

Urban forests have positive impacts on human and ecosystem health, reduce stress on aging stormwater infrastructure, increase property values, and reduce energy consumption. The scale of these benefits ranges from the hyper-local to the global. While the benefits of urban forests can extend well beyond the boundaries of cities, they often do not reach all residents of the city equally. Urban forest policies do not adequately address environmental equity or employ planting strategies with knowledge of the social and political factors that determine the spatial variations of tree canopy extent in cities. Chapter I analyzes the determinants of current canopy extent in Portland, OR using spatial regression analysis. Chapter II uses current landcover datasets to identify potential planting opportunities. Results of spatial regression show that income and education level are significantly positively linked to tree canopy, while sewer pipe density, an indicator of development, is negatively associated with canopy. The majority of tree canopy and potential in the city occurs on private, residential lands. Distribution of canopy potential is not even, with greater amounts in north and outer east side areas. Findings presented here will inform efforts to expand tree canopy in Portland in a manner that is spatially explicit and based on Portland's unique demographics, land use assemblage, and development policies.

Frequent intense precipitation events can mobilize and carry sediment and pollutants into rivers, degrading water quality. However, how seasonal rainfall and land cover affect the complex relationship between discharge and turbidity in urban watersheds is still under investigation. Using hourly discharge, rainfall, and turbidity data collected from six stations in three adjacent watersheds between 2008 and 2017, we examined the temporal variability of the discharge-turbidity relationship along an urban-rural gradient. We quantified hysteresis between normalized discharge and turbidity by a Hysteresis Index (HI) and classified hysteresis loops during 377 storm events in early, mid, and late wet season. Hysteresis loop index and direction varied by site land cover type and season. Turbidity values peaked quicker in the watersheds with higher degrees of urban development than a less urbanized watershed. The positive relation between discharge and turbidity was highest in two downstream stations in the mid wet season, while it was highest in two upstream stations in the early wet season. Correlation and regression analysis showed that maximum turbidity was best explained by discharge range, and the sensitivity of turbidity to discharge change was higher in the larger downstream watershed than in the small upstream watersheds. A flashiness index was negatively associated with the slope of turbidity versus discharge, suggesting that turbidity is difficult to predict solely based on discharge in flashy urban streams. This paper contributes to a deeper understanding of the spatial and temporal variation of discharge-concentration relationship in urbanizing watersheds, which can help water managers increase the resiliency of water-related ecosystem services to impacts of climate change.

Within mid-to-high elevation conifer forests in the Cascade Range, wildfire extent, severity, and frequency are expected to rise due to increasingly drier forest fuels under climate change. Considering dominant species composition, existing forests may be poorly adapted to absorb stress and recover following altered wildfire patterns. We tested the hypothesis that increased fire activity may disrupt the recovery of upper-montane and subalpine forest types by quantifying post-fire forest structure and conifer regeneration after spatially large, severe, and rapidly repeated wildfires in the Central Cascade Range. A stratified random sampling design was used to select field plots (n=122) and drivers of conifer regeneration were modeled using logistic and negative binomial regression models. Median conifer regeneration was very poor among sample plots that experienced either a single high-severity fire (49 seedlings/ha) or rapid reburn (14-28 seedlings/ha). Distance to seed source primarily drove seedling abundance, with shade-tolerant species abundance being most sensitive to increasing seed source distances and dry, exposed, post-fire environmental conditions. Rapidly repeated fire increased the size of high-severity patches by killing live seed source trees spared during an initial fire, with chronological sequence of burn severity promoting regeneration of all conifer species or primarily fast growing, fire-adapted pines. Low-seedling densities, a general lack of seed source, and future warming trends suggest these forests affected by expansive high-severity and/or repeated wildfire will transition into a patchy, low-density forest state. This early-seral state ecosystem will be composed of fire-adapted pines farther from live seed source and incorporate a patchwork of shrubby grassland that in turn, may be more resilient to frequent wildfire than prior forests. If future wildfire patterns manifest as expected in the Cascade Range, recovering mid-to-high elevation forests may begin resembling their drier, lower-elevation mixed-conifer counterparts in structure and composition.

In this dissertation, I study black geographic visions, experiences, and practices of the Marching 100 (M100) band at Florida A&M University (FAMU) and show how the black place-making practices of the Marching 100 (re)produces the black geographies of FAMU, Tallahassee, and M100 rehearsal spaces. This dissertation both draws from and makes conceptual and empirical contributions to the sub-discipline of black geographies. I show throughout this dissertation the usefulness of taking a place-making approach in studying black geographies and focus on how black place-making can be deployed as part of an affirmative celebration of black life. Conceptually, I draw on black feminist scholars to offer scholars interested in affirmative black geographies places of respite as an analytic and ontological object that is produced by (and productive of) visions and practices of black life. These places, I argue, provide relief from the burdens of oppressive articulations of society and space, and their existence amounts to a critique of these oppressive articulations. These places also offer opportunities to resist and heal harms of oppression. I also analyze the use of celebration as an affirmative, transgressive claiming of place within the city. Such celebrations, I argue, are transgressive place-making practices that can transform places and extend, temporarily, the sense of belonging places of respite provide. I also show, however, the precarity of black place-making claims. Together these chapters show the socio-spatial power of black joy/celebration and highlight the importance of black life in the production of black geographies. / A Dissertation submitted to the Department of Geography in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Spring Semester 2019. / March 29, 2019. / Black Geographies, HBCU, Place-making, Places of Respite / Includes bibliographical references. / Mary Lawhon, Professor Co-Directing Dissertation; Tyler McCreary, Professor Co-Directing Dissertation; Michael Nair-Collins, University Representative; Adam Bledsoe, Committee Member; Ronald Doel, Committee Member; Joseph Pierce, Committee Member.

The 2009-2010 H1N1 influenza pandemic was caused by a novel strain, made up of genetic material from human, swine and avian influenza viruses. While the 2009 H1N1 strain originated in Mexico, Southeast Asia, and southern China in particular, remains the putative epicenter of new viral emergence. Using genetic and epidemiological information from 433 H1N1 viral isolates taken during the 2009-2010 pandemic in China, we examined the spatial and temporal characteristics of viruses in concert with their genetic characteristics, identifying the spatial and temporal diffusion patterns. Then we applied Moran's I test to see if the gene distances of the H1N1 virus are spatially autucorrelated. We then explored the suspected factors driving the evolution of H1N1 viruses during the pandemic. Regression methods were applied to test the association of H1N1 virus's spatial and temporal patterns with environmental, social and biological variables. Temperature& humidity, railway transportation, population density, morbidity of H1N1, population's accessibility to tap water, sampled patient's age are some of the variables considered in the regression. We find that during the 2009- 2010 pandemic H1N1 influenza viruses evolved more through time, that further evolved strains have a trend of spreading from Northern China to Southern China, and continued evolving in the Southern China. Among the 8 genes of H1N1 virus, the HA gene and MP gene showed statistically significant positive spatial autocorrelation, showing that the genetic distances of the genes are related to the genetic distances of those genes in nearby isolates. Statistically significant positive spatial autocorrelation is also shown for the total 8 genes' genetic distance. Results of the spatial regression models indicate that the influence of environmental, social and biological variables varies not only across space but also by gene segment under consideration. We find that population, environment and behavior are all playing a role in the evolution of H1N1 viruses in the 2009 pandemic in China. Thus understanding the dynamics of H1N1 incidence and the ecology of H1N1 virus evolution in China can be enlightening in establishing public health policy.

Recent shifts in the hydrologic regime of the Apalachicola River have been attributed to anthropogenic changes in the watershed, particularly those associated with dam and reservoir construction. To assess the impact of these changes on the forests on the river's floodplain, a two-tiered methodology was applied to a 1-ha forest plot. First, repeat-survey data spanning a 27-year interval was subjected to multivariate analysis for identification of major trends in forest composition. While the changes identified were small, the increased representation of upland tree species on the plot was a potential indication of ecological response to declining river stages. To examine patterns in tree growth in more detail, a dendrochronological approach was used, beginning with the collection of core samples from every species present on the plot. Of these, cores from 4 different species were selected as suitable for analysis. The annual growth increment record from these cores was subjected to correlation and multiple regression analysis with various hydrologic and climatic parameters in an attempt to isolate the primary factors influencing growth. The use of the hydrologic analysis software IHA allowed the development of parameters characterizing the site-specific flooding regime, including parameters approximating annual frequency, average duration, and average timing of inundation events. Finally, intervention detection analysis was used to identify major trends, shifts, and unusual features in the growth record. Growth in all four species was found to correlate most strongly with hydrologic parameters, particularly with the site-specific parameters generated by IHA. Regression models developed primarily from hydrologic parameters were successful in accounting for variance in growth in all four species, particularly after removal of disturbance-related outlier years (0.36 / A Thesis submitted to the Department of Geography in partial fulfillment of the requirements for the degree of Master of Science. / Degree Awarded: Spring Semester, 2007. / Date of Defense: March 30, 2007. / Celtis laevigata, Quercus laurifolia Quercus lyrata, Taxodium distichum, Flood regime, Ordination, Tree rings / Includes bibliographical references. / J. Anthony Stallins, Professor Directing Thesis; James B. Elsner, Committee Member; Xiaojun Yang, Committee Member.