Search

Chile's temperate forest is a global biodiversity hotspot due to its levels of endemism and rate of habitat loss. Deforestation and biological invasions are two major threats faced by this ecosystem. Both are produced by the expansion of productive forestry, which fragments the native forest and introduces alien species (from the Pinaceae family and Eucalyptus genus). Despite these threats, broad scale ground-based forest and alien species monitoring is not feasible due to logistical and budgetary limitations. Surveying and mapping alien species are key steps to understand biological invasions. This project uses remote sensing to understand biological invasions by alien trees from productive forestry in Chile's temperate forest at different spatial scales. Sentinel-2 is used to assess alien tree occupancy and model its relationship with fragmentation and landscape configuration at a broad scale. A temporal assessment using Landsat historical archive imagery is used to investigate the dynamics of alien tree spread over the last 40 years. A local scale assessment using very high resolution WorldView-3 imagery and object-based image analysis is used to investigate early invasion and its relationship with land cover configuration. The results demonstrate remote sensing methods are useful for alien tree monitoring at different scales and hold potential for management. The expansion of alien trees at a broad scale is dominated by socioeconomic factors such as cost effectiveness and government subsidies. Alien Pinaceae tend to invade the naturally open Araucaria araucana forests, threatening the long term regeneration of this endemic and culturally relevant tree. Early Pinaceae invasion is linked to fragmentation, distance to seed sources, topography and human intervention. Incidentally, this study found that native forest degradation is occurring, increasing its vulnerability to invasion in the future. Conservation and forest management need to adopt sustainable, evidence-based practices to achieve compatibility between productive forestry and native forest conservation.

The main theme of this research is the development of a new hybrid method for change detection of land use and land cover (LULC). LULC change detection is one of most widely used applications of remote sensing. This study used data from two different optical sensors, Landsat-8 images and Sentinel-2A images. Given the newly developed capabilities of these remote sensing satellites, it was necessary to devise appropriate techniques to realise the benefits that they offer. Therefore, three effective change detection methods have been tested, comprehensively analysed, and used to inform the design and development of a new hybrid method of change detection. The studied change detection methods were change vector analysis (CVA), multi-index integrated change analysis (MIICA), and the comprehensive change detection method (CCDM). Case studies were conducted in two regions, Bristol (United Kingdom) and Hail (Saudi Arabia), to provide sufficient variety of inputs to enable the response of more LULC varieties to be recorded. Finally, the Coordination of Information on the Environment (Corine) land cover scheme was used to identify land cover types and LULC changes. In the study area of Bristol, the new hybrid change detection method achieved an overall accuracy of 90% and 0.81 kappa, while the results for the study area of Hail were 74% overall accuracy and 0.40 kappa. The change detection results obtained by the new hybrid method constitute a significant improvement over the implementation of the existing CVA, MIICA and CCDM methods at the two study areas while using Landsat-8 and Sentinel-2A images.

Efforts are increasing globally to harness the potential of forests to alter catchment water runoff and storage dynamics as a 'natural flood management' (NFM) strategy, particularly given a projected rise in the frequency and severity of floods with climate change. Despite decades of research on forest hydrology, knowledge of how forests and land use control catchment runoff is still limited, especially in relation to important, though less investigated, subsurface runoff processes. This PhD research aimed to examine how forest cover interacts with soils and geology to influence runoff pathways at different spatial and temporal scales, focusing on the 67 km2 Eddleston Water NFM pilot site in the Scottish Borders. At the catchment scale, isotopic (2H and 18O)and geochemical tracers (Acid Neutralising Capacity (ANC)), conductivity and pH) were used to investigate whether forest cover is a significant control on water storage and mixing over seasonal and storm event timescales. At the hillslope scale, dense subsurface monitoring (soil moisture, groundwater and time-lapse electrical resistivity tomography (ERT)) compared improved grassland to an across-slope forest strip, similar to those promoted in NFM schemes to control runoff, to reveal water storage potential in soil underneath the forest and the downslope extent of any impacts on subsurface hydrological dynamics. The results revealed complex interactions between land cover and runoff processes at different scales. At the catchment scale, soil type and superficial geology were found to be more dominant controls on catchment storage over seasonal timescales, with land cover playing a secondary role. Dynamic storage estimates for headwater catchments underlain predominantly by glacial till were low, ranging from ~16 mm to 46 mm, and were correlated with low mean transit times, ranging from ~130 to ~210 days. There were no differences in these estimates, within the bounds of error, between catchments with up to 90% forest cover and those with much lower cover (<50%). However, there were significant differences compared to steeper catchments with low glacial till cover. In these catchments dynamic storage estimates ranged from ~160 mm to~200 mm, and were correlated with high mean transit times, ranging from ~320 to~370 days. At the storm event timescale, and comparing two adjacent catchments with similar superficial geology and soils but differences in land cover, forest cover reduced the event water runoff fraction for four high flow events. The fraction of event water runoff at peak discharge during the largest event monitored was 0.37 for the forested catchment but 0.54 for the adjacent partially forested catchment. A third catchment, with minimal glacial till and low forest cover, demonstrated very different dynamics, with much lower runoff ratios for all events, higher groundwater fractions (0.21-0.55 at peak), and 'double-peak' hydrographs, illustrating the impacts of geology on runoff processes. Similar relative differences in runoff fractions were found between catchments across the three winter events, with differences between storms greater than differences between catchments. These findings suggest that while catchment characteristics mediate event responses, the characteristics of the event(rainfall depth, intensity and antecedent conditions) may dominate responses, though it was not possible to disaggregate the different event characteristics with this dataset. The hillslope scale work identified significant differences in subsurface moisture dynamics underneath the forest strip over seasonal timescales: drying of the forest soils was greater, and extended deeper and for longer into the autumn compared to the adjacent grassland soils. Water table levels were also persistently lower in the forest and the forest soils responded less frequently to storm events. Downslope of the forest, soil moisture dynamics were similar to those in other grassland areas and no significant differences were observed beyond 15 m downslope, suggesting minimal impact of the forest at shallow depths downslope. The depth to the water table was greater downslope of the forest compared to other grassland areas, but during the wettest conditions there was evidence of upslope-downslope water table connectivity beneath the forest. The results indicate that forest strips provide only limited additional subsurface storage of rainfall inputs in flood events after dry conditions in this temperate catchment setting. In summary, the research results show that while forests have some seasonal impacts on subsurface moisture dynamics, soil type and underlying superficial geology are primary controls on catchment storage and mixing in temperate upland environments, suggesting limited impacts of changing land use. At storm event timescales increased forest cover has some impact on reducing the amount of event water runoff, but event characteristics are a more dominant control, so forest cover alone is unlikely to lead to significant reductions in peak flows during large flood events. Strategically placed forest cover, such as field boundary planting on hillslopes has some impacts on subsurface moisture dynamics but the effects are spatially limited and not present in winter periods. The processes leading to these findings appear to be similar at the catchment and hillslope scales. From an NFM policy perspective the findings suggest that while tree planting is not a flood management panacea, it may have benefits in certain situations, as well as significant co-benefits. This implies a need for a change in emphasis within flood risk management policy, which 'mainstreams' tree planting as a flood risk strategy into wider policy processes to create multifunctional landscapes. There are still many unknowns about the impacts of land cover on hydrological processes, particularly in the subsurface, and there is a need for enhanced research on these processes. This will also help to reduce some of the large uncertainties surrounding the impacts of NFM, which remain one of the key barriers to its wider implementation.

Ecosystem services are the benefits people obtain from ecosystems. Quantifying these benefits and understanding how they may change under multiple future pressures, such as climate change or land use change, is a highly uncertain exercise. In managing ecosystems to be resilient to future changes, natural resource managers need the most accurate information available, but also need to be informed of when and where they can be confident, or not, in projections of change. In this thesis, I address many of the key aspects of uncertainty in projections of change in ecosystem services, with a particular focus on challenges in West Africa. I show where and for what variables climate models may be reliably used in ecological studies, providing important advice for interpreting the impacts of climate change on biodiversity. Furthermore, I show that uncertainty in climate observations can also have a significant impact on climate change adaptation decisions at both the species level and in terms of protected area management. I also address how vegetation in West Africa may respond to future climate change. I found that even after uncertainties in climate and land use were considered, carbon storage in West African tropical forests was projected to increase where forest degradation remained low or reduced; vegetation productivity was projected to increase in all parts of West Africa, with the exception of locations in the West Sahel where the largest reductions in precipitation were projected; and, importantly for protected areas, ecosystems were projected to shift northwards despite uncertainty in precipitation projections. I also show the sensitivity of 3 major land surface models to uncertainty in vegetation mapping, thereby providing guidance to the remote sensing community on priorities to improve land cover mapping and to the earth system modelling community on bounds of uncertainty in carbon, moisture and energy budgets due to vegetation mapping uncertainties. Lastly, in using the latest land-atmosphere coupled convection-resolving model, I show that it is possible to simulate the observed interaction between landscape heterogeneity and local and regional scale precipitation in West Africa. This provides a timely and relevant tool that will allow scientists and natural resource managers to more accurately assess the impact of changes in land use on the regional climate of West Africa.

The role and scale of human impact on the global environment is a question of special importance to the scientific community and the world as a whole. This impact has dramatically increased since the beginning of industrialisation, yet its understanding remains patchy. The sub-Arctic plays a central role in forming the global environment due to the vast territory of boreal forest and tundra. Severe climatic conditions make its ecosystems highly sensitive to any natural and human disturbances. In this context, the dynamics of boreal vegetation, and of the forest/tundra interface (the treeline), is the most representative indicator of environmental changes in the sub-Arctic. For some time now, monitoring land cover and vegetation changes using remote sensing techniques have been a powerful method for studying human impact on environment from landscape to global scales. It is particularly efficient when applied to the sub-Arctic ecosystems. Remote sensing gives access to accurate and specific information about distant and hard-to-reach areas across forest and tundra. Despite all the e orts, there is a lack of uniformity in studying human impact, a shortage of mapping of impact over large territories and a lack of understanding of the relation between human activity and environmental response. This dissertation develops a systematic approach to monitoring land cover and vegetation changes under human impact over northern Fennoscandia. The study area extends north and south of the treeline and covers around 400,000km2 reaching from Finnmark in Norway, through Norrbotten in Sweden, Lapland in Finland up to the Murmansk region in Russia. This is the most populated and industrially developed region of the whole sub-Arctic and, therefore, suffering most from human impact. This dissertation identifies industrial atmospheric pollution, reindeer herding, forest logging, forest fires and infrastructure development as the primary types of human impact close to the treeline. For each type characteristic hotspots are identified and human impact is analysed in the context of physical environment as well as cultural, economical and political development of the area. This dissertation presents an automated workflow enabling large-scale land cover mapping in northern Fennoscandia with high throughput. It starts with automated image pre-processing using image metadata and ends with automated mapping of classification results. A single classifier for multispectral Landsat data is trained on extensive field data collected across the whole region. Open source tools are used extensively to set up the processing scripts enabling rapid and reproducible analysis. Using the developed advanced remote sensing methodology land cover maps are constructed for all identified hotspots and types of human impact. Changes in vegetation are analysed using three or four historical land cover maps for each hotspot. More than 35 Landsat TM and ETM+ images covering the period from the 1980s until 2011 are processed in an automated manner. A strong correlation between the level of impact and the scale of vegetation change is confirmed and analysed. The structure and dynamics of the local treeline and the quality of environment are analysed and assessed in the context of changing levels of impact at each hotspot and regionally.

Deep anthropogenic topsoils – those augmented through long-term additions of mineral bulk among fertilising agents – retain in both their physical and chemical make-up significant indicators for cultural activity. This project researched the geographical distribution and historical context of deep anthropogenic topsoils in Scotland and the Isles, and used this information to investigate the impact of current land cover upon the cultural information they retain. In so doing, the project investigated the potential for conservation of this significant cultural resource. A review of the historical information available on agricultural and manuring practices for Scotland identified several factors likely to affect deep topsoil distribution and frequency. These were: the availability of bulk manures to Scottish farmers, the significance of the seaweed resource in determining fertiliser strategies in coastal areas, and the influence of urban settlement and associated patterns of domestic and industrial waste disposal on the location of deep topsoils. Evidence for widespread deep topsoil development was limited. The primary data source used – the First Statistical Account of Scotland – was manipulated into a spatial database in ArcView GIS, to which geographical data from the Soil Survey of Scotland and national archaeological survey databases were added. This was used to devise a survey programme aiming both to investigate the potential factors affecting soil development listed above, and to locate deep topsoil sites for analysis. Three sites were identified with deep topsoils under different cover types (woodland, arable and pasture). The urban-influenced context of two of these highlighted the significance of urban settlement to the location of Scottish deep topsoils. Analysis of pH, organic matter, and total phosphorus content showed a correlation between raised organic matter and a corresponding increase in phosphorus content in soils under permanent vegetation. By contrast, soils under arable cultivation showed no such rise. This was attributed to the action of cropping in removing modern organic inputs prior to down-profile cycling. The potential for pasture and woodland cover to affect relict soil signatures was therefore observed. Thin section analysis aimed to both provide micromorphological characterisation of the three deep topsoil sites and investigate the effect of modern land cover on micromorphological indicators. Distinctive differences in micromorphological character were observed between the rural and urban deep topsoils, with the latter showing a strong focus on carbonised fuel residues and industrial wastes. All sites showed a highly individual micromorphological character, reflective of localised fertilising systems. There was no correlation between land cover type and survival of material indictors for anthropogenic activity, with soil cultural indicators surviving well, particularly those characteristic of urban-influenced topsoils. Suggestions for preservation strategies for this potentially rare and highly localised cultural resource included the incorporation of deep anthropogenic topsoil conservation into current government policy relating to care of the rural historic environment, and the improvement of data on the resource through ongoing survey and excavation.

Surface mining can impact land cover, forests, and water quality. Current reclamation strategies include revegetation with herbaceous species due to the ease and speed of cover establishment. Herbaceous and woody biomass crops have been used in various studies to reclaim surface mines and act as an alternative to nonrenewable energy sources. The objectives of this study were to quantify the state of vegetation growth on former surface mines in West Virginia over a 9-year period and identify suitable acreage for bioenergy production. During 2011 to 2020, we found that over 40,000 acres had been converted to forest and 40,000 acres of low vegetation was lost during the time frame. We found that land cover change for forest, low vegetation, and barren changed significantly during the period of 2011 to 2020. Further, the results indicated that approximately 23,000 acres of surface mines in the state were low vegetation cover with 15% or less slope for the study. Of the 23,000 acres, 45% were identified as contiguous areas of 10 acres or more. Therefore, reclamation and/or restoration of these lands with bioenergy crops could provide significant feedstocks for bioenergy development projects in the region.

Urbanization alters the quality of the natural environment, with more land areas that were previously covered with forest are now being converted to built-up areas. As a result, watersheds are typically negatively impacted by urbanization. This study analyzed multiple water-quality variables (i.e., temperature, turbidity, pH, dissolved oxygen, and conductance) in 24 to 30 watersheds in the Atlanta, GA region during 2014 – 2021. To understand the impact of urbanization on water quality, monthly means, correlations, and linear regression were used to investigate the relationship between water quality variables and watersheds characteristics. Landcover characteristics such as developed land, imperviousness, population and housing densities, increased stream temperature, and conductance, while decreased DO levels and turbidity. High temperatures and low dissolved oxygen can affect water quality, rendering it unsuitable for various uses. It is recommended that this approach for water quality assessment be used in other metropolitan areas.

Surface urban heat islands (SUHIs) are land surfaces with high concentrations of impervious surfaces like roofs, roads, sidewalks and other infrastructures that trap, absorb, and re-emit heat throughout the day/night and typically present higher temperatures than their surrounding rural areas. In this study, I evaluate how presence of and distance to SUHIs are associated with water temperature in the lower Chesapeake Bay watershed for the summer of 2019. When heavy precipitation events occur, flooding creates stormwater runoff, which is exposed to the hotter temperatures in urban areas. This introduces thermal pollution to nearby rivers and streams disrupting aquatic ecosystems. The hypothesis for this research is that the closer water temperature points are to SUHIs, the warmer the temperature measured will be. To assess this, I processed Landsat 8 and 9 scenes in order to derive land surface temperature (LST), normalized difference vegetation index (NDVI), and normalized difference build-up index (NDBI). I also processed land cover from the national land cover dataset (NLCD) and a digital elevation model (DEM) from which I derived flow direction (FD) and flow accumulation (FA). I used water temperatures measured by water quality stations from 25 sources as well. If areas with a surface temperature were half a standard deviation above the agricultural land cover LST average, they were defined as SUHIs, following Kaplan et al. (2018). The other datasets were used to extract other factors that can impact temperature or the relationship between distance to SUHIs and temperature. In addition, I also processed local climate zones (LCZs) to validate the identified SUHIs. To extract distance from SUHI areas and the water temperature datapoints, I used ArcGIS’s Euclidean Distance and Direction Distance tools. These were calculated for various cases, including; no-distance (contained within SUHI), omni-directional distance, and upstream/downstream distance. Some of these methodological attempts were more successful 2 than others. Overall results do not show a strong relationship between warmer water temperatures and proximity to SUHIs; therefore, in general terms, the hypothesis is not supported. However, there are some noteworthy findings; a) there are warmer water temperatures near urban centers where most of the SUHIs are located; b) elevation has the strongest influence and highest significance on water temperature with the trends of the variables explored (i.e., at higher elevations, the water temperature is cooler while at lower elevations, the water temperature is warmer); and, c) Euclidean distance to SUHIs and NDVI are other significant factors. With more time and resources, I would include more data on environmental confounding factors and use improved methods to calculate various distance measures, which would likely help tease out more specific relationships between water temperature and SUHIs as well as to interpret their correlations.

Fish richness and diversity serve as important indicators of a healthy stream ecosystem, which are influenced by a complex web of ecological factors, including regional climate, watershed characteristics, riparian zone quality, and water quality. Investigating how these factors interconnect and impact fish community is crucial for developing effective management strategies to safeguard freshwater ecosystems. In this study, we used partial least squares regression to develop a causal understanding of how watershed development and climate factors affect fish richness and diversity by altering water temperature, pH, conductivity, total nitrogen, and total phosphorous (TP) in 277 watersheds in the Great Lakesurrounding areas. After identifying TP as a potential threat to fish biodiversity, we examined how watershed land use, slope, and soil interacted to drive changes in TP concentration with multiple linear regression. Results suggested that moderate watershed development (average 5% developed percentage in the study site) can enhance fish diversity by increasing pH, temperature, and conductivity. However, watershed development and riparian degradation diminished fish diversity by increasing nutrient concentrations. We also found that landcover-slope interaction was an important factor affecting TP concentration, while landcover-soil interaction was not significant. Future ecosystem management in the study area should therefore emphasize a dual focus on watershed management approaches and riparian zone preservation to improve fish diversity and stream ecosystem health.

In order to ensure clean drinking water for all, it is crucial to understand potential upland stressors that compromise the quality of source waters treated by local community water systems (CWSs). Contamination associated with specific types of land cover can result in downstream water quality degradation, which may reduce the effectiveness of treatment by CWSs. Surface mining has been hypothesized as a source of drinking water degradation within the Central Appalachian region, which may result in adverse exposures and health disparities. The purpose of this study was to identify potential correlations between land cover and adverse birth outcomes (ABOs) through the application of watershed epidemiology, an emerging environmental health paradigm. Birth records for the Central Appalachian region were acquired from their respective state health departments from 2001 to 2015: each record contained the mother's street address, outcome variables, and covariates. Records were included in later analyses if they fell within an approximated CWS service area. Contributing land cover to each CWS was determined via previously delineated watersheds that relied on CWS intake points. A binomial generalized linear model was used to compare low birth weight (LBW), term low birth rate (tLBW), and preterm birth (PTB) incidence to CWS source watershed land cover, Safe Drinking Water Act (SDWA) violations, CWS size, and covariates related to the birth records. Source watershed mining and SDWA health based (HB) violations were significantly associated with greater risks for preterm birth (PTB) and low birth weight (LBW). Future work should be conducted to explore upstream flow impacts, address missing data in the birth records, and to more accurately represent CWS service areas to better characterize exposure.

Abstract I examined how abrupt land cover change can be detected, measured, and evaluated via accessible platforms using varied methods applicable to the non remote sensing (RS) specialized analyst. Analysis with free imagery and widely accessible platforms democratizes access to information and research capabilities (Nagaraj, Shears, & de Vaan, 2020). Enabling open access to remotely sensed data can aid in understanding and fostering the sustainability of socioecological systems. As socioecological natural disasters such as abrupt land cover change increase in frequency and intensity, the need for such analysis heightens. Thus, the focus of this thesis is how abrupt land cover change can be measured via readily accessible platforms. As proof of concept for the methods framework, I quantitatively analyzed land cover change with appropriate indices, classification methods, and algorithms. I selected a study area in Zambia’s Luangwa Valley and addressed how land cover changed during the COVID-19 pandemic from prepandemic 2019 to mid-2021. The results supported that the framework of methods I designed could quantify the magnitude of land cover change in km2 and determine whether the rate of change was greater than 10%. Using the random forest algorithm (RF) the magnitude of change in the study area was greater than 10%, with approximately 82% accuracy. With an estimated accuracy of 85%, I concluded that the land cover change extended to an area of 1250 km2. The result was based on the spectral angle (SA) algorithm, which performed with greater accuracy than the minimum distance (MD) and maximum likelihood (MLC) algorithms, respectively. Moreover, analysis via spectral indices and statistics enabled the determination of approximate locations and directions of change. Overall, the devised framework proved worthwhile for conducting a broad land cover change analysis. However, the ability to tailor the methods introduced to a given problem set, such as indices or algorithms, renders this framework worthwhile in studying other socio-ecological problems. The framework of methods introduced has broad applicability.

Quantifying actual evapotranspiration (AET) in urbanized watersheds helps water managers develop accurate water budgets and aids in predicting future water budgets, especially in the face of land-cover and climate change. This paper presents multiple methods, with focus on Penman-Monteith approach, to estimate 20 years of daily, monthly, and annual AET totals in an urbanized watershed, the South River watershed (SRW), in Atlanta, GA. Land cover analysis confirmed NLCD definitions and revealed developed classes 21-24 contain 48.18%, 24.26%, 13.79%, and 4.43% forest cover, respectively. Additionally, it was found that the annual AET of the SRW is approximately 800 mm/yr, with about 10% coming solely from impervious land cover. A decreasing annual AET trend was observed for the SRW, resulting from land cover changes and decreasing incoming solar radiation. Finally, TerraClimate reference AET datasets overestimated urban AET, while statistical models typically underestimated AET when compared to reference water balance AET estimates.

The water quality of the historically industrial Mahoning River continues to be impacted by non-point pollution sources from the watershed. We evaluated the current surface water quality of four important Mahoning River sub-watersheds: Meander Creek, Crab Creek, Lower Mosquito Creek, and Yellow Creek using the National Sanitation Foundation water quality index (NSF-WQI). Water quality parameters measured include, temperature, dissolved oxygen, pH, ammonia, soluble reactive phosphates, total phosphorus, nitrate-nitrogen, total suspended solids, volatile solids, total dissolved solids, total coliform, Escherichia coli and biochemical oxygen demand. All of the samples were taken during the summer at low flow conditions. NSF-WQI of four sub-watersheds are compared along with their land cover types identified as residential, commercial, industrial, forested, and agricultural areas. The NSF-WQI for Lower Mosquito Creek was found to be the highest at 70 and Crab Creek having the lowest at 64.5, respectively being in good and medium according to NSF-WQI range (0-100). Also, the Post-hoc Tukey HSD test indicates Crab Creek is statistically significantly lower than Lower Mosquito Creek. Land cover type of residential in Lower Mosquito Creek had the highest NSF-WQI at 72 and agriculture land cover in Crab Creek had the lowest at 53, respectively being in good and medium NSF-WQI range. When comparing NSF-WQI of overall land cover types, residential had the highest at 70 and agriculture had the lowest at 64, while forest and industrial were same at 67, respectively being in good and rest in medium NSF-WQI range. Tukey HSD test indicates the NSF-WQI for residential land cover is statistically significantly higher than agriculture land cover. The principal component analysis depicts a weak correlation with water variables and NSF-WQI of the sub-watersheds.

Population growth, urbanization, and infrastructure development activities have resulted in the land conversion of forests and farmlands to residential and commercial zones. Such land conversion causes changes in the land cover, as experienced in the Ayung Watershed, in the island of Bali, Indonesia. Here, the land cover undergoes rapid changes due to the growing tourism sector, affecting the runoff coefficient. This study evaluated the changing land cover patterns and surface runoff in the Ayung Watershed between 2012 and 2019. An increase in the surface runoff during the high rainfall events may lead to flooding in the area. The identification of land change patterns in the Ayung Watershed was carried out by a manual digitizing process on Google Earth maps. The runoff coefficient was calculated by Cook’s method using the four physical characteristics of the watershed: land cover, infiltration rate, land slope and drainage density; showing significant changes in the land cover in the study area. Farmlands and forests were reduced by 647.8 ha and 553.1 ha respectively, converted into fast growing grasslands or unproductive land. Such land cover changes have a negative impact by increasing the runoff coefficient in the area. During the study period, the runoff coefficient was consistently found to be more than 0.6 (high-risk category). Several sections in the city of Denpasar experienced an increase in the runoff coefficient by more than 5%. Consequently, there was a high-risk of flooding in the area because of the increasing surface runoff.

Stormwater runoff response in a complex coastal-urban basin varies with the changes in climatic and land cover drivers. This thesis aims to assess the runoff responses to variation in climatic and land cover variables and perform a sensitivity analysis in complex coastal-urban basins. The Buffalo San Jacinto Basin (BUSJ) of southeast Texas, USA was considered a pilot study area for this research. BUSJ is a heavily urbanized coastal basin, draining an area of 3132 km2 through the City of Houston, Texas into the northern Gulf of Mexico, USA. A process-based hydrological model was developed with fine spatio-temporal resolutions for the basin by using U.S. Environmental Protection Agency (EPA)’s Storm Water Management Model 5.1. The model was calibrated and validated with historically measured streamflow data from January 1 to December 31, 2020 (termed 2020). Rainfall and evapotranspiration were selected to represent the key climatic drivers, whereas surface imperviousness represented the major land cover drivers. These stormwater drivers were perturbed by -30 to +30% (with 5 percentage points increments) from their reference 2020 baseline values. The results indicated predominant and varying seasonal sensitivities of stormwater runoff to the changes in rainfall. Evapotranspiration changes led to consequential, but much fewer changes in the runoff than that of rainfall. However, runoff had stronger sensitivities to the changes in imperviousness, compared to that due to evapotranspiration in this heavily urbanized coastal basin. The quantified runoff sensitivities can provide important guidance for stormwater infrastructure management, improvement and development to achieve resilience and sustainability in the City of Houston and other areas along the northern Gulf of Mexico, USA and similar coastal-urban regions around the world.

Changes in climatic and land use drivers typically drive stormwater runoff quantity and quality. This study focuses on determining the stormwater runoff sensitivities to variation in climatic and land cover changes in a coastal natural-urban basin. A mechanistic hydrologic model was developed for the Spring San Jacinto Basin (SSJB) of Texas, USA considering the basin as a pilot study area by using the U.S. Environmental Protection Agency (EPA)’s Storm Water Management Model 5.2. The SSJB drains an area of 1949 km2 ultimately into the northern Gulf of Mexico, USA through peripheral neighborhoods of the Greater Houston area. The model represented locally relevant (e.g., neighborhood-scale) fine spatio-temporal resolutions. The model was calibrated and validated with historically observed streamflow data for 2020. Climatic and land cover variables such as rainfall, evapotranspiration, and imperviousness were perturbed by −30% to +30% (with an increment of 10%) from their average baseline values during 2020s. Basin stormwater runoff showed overall strong and differential seasonal sensitivities to the changes in rainfall. In contrast, runoff had a low sensitivity to the similar perturbations in evapotranspiration. Further, moderately strong sensitivities of runoff were noted towards the changes in the key land cover of surface imperiousness. The quantified sensitivities would guide stormwater management in coastal-natural environments under a changing climate and land use/cover in the U.S. and beyond.

Despite considerable progress made over recent decades in scientific understanding of the structure, evolution, and dynamics of supercell thunderstorms and tornadoes, there are still aspects of their evolution, including tornadogenesis and decay, that have yet to be understood. One area in which scientific understanding is particularly incomplete is how land surface heterogeneity or complex topography interacts with and/or affects tornadoes. Past investigations embarked on the endeavor to quantify the relationship between tornado events with land cover or elevation, with inconclusive or conflicting results. The purpose of this study is to identify the genesis and decay points of tornadoes over an eighteen-year timeframe of 2000-2017 within the state of Arkansas and a ten-year timeframe of 2008-2017 within the State of Oklahoma to determine whether elevation and/or surface roughness of the nearby land cover relative to the broader landscape might impact these events. This study was completed utilizing Storm Prediction Center tornado reports, National Land Cover Data, and USGS digital elevation models (DEMs), with ArcGIS Pro and Microsoft Excel as the modes of data visualization and analysis. A combination of buffer analyses, Mann - Whitney U tests and Getis-Ord Gi* hotspot analyses were incorporated to assess the relationship between genesis/dissipation locations and surface roughness and complex terrain. The qualitative and quantitative results herein suggest a local increase in surface roughness heterogeneity favors genesis events, while regions defined by complex terrain may act to dissipate tornado events. Furthermore, hotspot analyses suggest regions with complex terrain and increased roughness values favor tornado activity.

This research investigates particulate matter (PM) deposition to rock pigeons (Columba livia) in urban environments within Denton County, Texas. Pigeon habitat was characterized within a 2-km radius at eight locations using the 2016 National Land Cover Database (NLCD). In summer 2020, feathers were sampled from 10 rock pigeons at two locations (n = 20) differing in degree of urban development. Birds were captured using walk-in funnel traps baited with bird seed. Based on molt pattern and appearance, four old flight feathers were identified and sampled from each bird. New primary feathers were obtained from each population as reference samples. Feathers were washed three times with double deionized water and acetone, and the solution vacuum filtered through a glass microfiber filter to collect all particles >1.5 µm in diameter. Particulate matter mass was determined by gravimetric analysis and calculated per unit feather surface area. Relative PM accumulation rates were significantly different between the populations. Characteristics of urban land cover, proximity to and types of emissions sources, wind exposure, and building density were drivers of variability in PM deposition to feather surfaces. The results from this study should be useful for subsequent research to help identify best practices for using feathers collected from pigeons or other urban restricted bird species to monitor PM levels across multiple spatial scales.

Previous research has indicated that agricultural land-use often reduces water quality in streams. This includes: 1) an increase in total suspended solids (TSS) due to elevated soil erosion and 2) changes in dissolved organic carbon (DOC) concentrations. The goal of this study is to determine if a forested state park can mitigate these potentially negative impacts. This study examines spatial and temporal dynamics of TSS and DOC in four streams, located in an agriculturally impacted watershed in SW Ohio, as they flow from agricultural land-cover through a naturally forested state park. Eight surface water sites were sampled bimonthly over a one-year period from December 2019 to December 2020. Each water sample was processed to quantify concentrations of TSS and DOC. In addition to examining the impact of land use, this study also examined the effect of discharge and the normalized difference vegetation index (NDVI) on changes to TSS and DOC concentrations. A significant increase in TSS concentrations was found within the state park. The changes in TSS concentrations were not significantly related to stream discharge nor NDVI. A significant decrease in DOC concentrations was found in two streams within the state park. Discharge was negatively correlated with changes in DOC concentrations in three of four streams. The results suggest that small streams in the U.S. Midwest are able to reduce fluvial DOC concentrations in agricultural watersheds within naturally forested areas. Further research is needed to determine why TSS concentrations increased within the forested state park.

Coastal areas are more sensitive to climate change and are more vulnerable to increase in temperatures. For the last decades, residents of New Orleans have been experiencing hotter summer temperatures conditions and extreme weather conditions. Meanwhile, with economic development and the impact of human activities, the land cover of the region has experienced substantial changes. Since Hurricane Katrina in 2005, Parks and Parkway has undertaken a re-greening effort to replace the many thousands of trees lost. A specific benefit of tree cover is its ability to reduce the heat storage capacity of urban surfaces that creates urban heat islands. This study (1) creates an up-to-date, high spatial resolution land cover map and examine how the components of Land cover, particularly trees vary across post-Katrina New Orleans; (2) quantifies the magnitude and seasonality of LST change between 2006 to 2020 within the same areas of Land cover maps; and (3) contextualizes Land cover and LST patterns with census‐based demographic data and to draw inferences about the heat vulnerability implications.

Modern hydrologic modeling and data systems are growing in scale and resolution to answer urgent, data intensive questions related to water resources. The focus of this dissertation is how real-world entities can be represented as flexible and efficient computational elements that capture processes and ease the exchange of information. Part 1 focuses on evaluating the state-of-the-art National Water Model (NWM) and Height Above Nearest Drainage flood mapping techniques to establish a community benchmark for improvement. We propose a outline for what an intelligent emergency response system might look like with these systems in place but note challenges related to (1) data usability (2) assumptions of spatial homogeneity in explicitly heterogeneous phenomena, and (3) the application of historical conventions that are inefficient in a federated system.Part 2 focuses on (1) improving the data format of NWM streamflow output (2) parameterizing Manning’s roughness across the national river network for improved river depth estimates, and (3) the digital representation and estimation of hydraulic relations at the national scale for increased interoperability. Part 3 utilizes these techniques to reimagine a flood forecasting system that circumvents the need to produce flood maps and instead computes feature-level flood forecasts through a relational data system that allows building-level and demographic flood assessments to be extracted from a national inventory in a matter of seconds for any area of interest.Lastly, part 4 explores the role of land cover as the mediating layer of human, water, and energy processes. The questions explored include how land cover data is best represented at coarser scales and what are the hydrologic consequences of variable spatial allocation and static land cover data. In approaching these, we develop a new raster resampling technique, evaluate the hydrologic influence of fire over California in the last two decades, and quantify the sensitivity of hydrologic models to different land cover representations. The dissertation concludes by sharing a simulation of California land use change out to 2100 and a discussion of the potential impacts faced.

This work aims to develop a deep learning model for land cover quantification through hyperspectral unmixing using an unsupervised autoencoder. Land cover identification and classification is instrumental in urban planning, environmental monitoring and land management. With the technological advancements in remote sensing, hyperspectral imagery which captures high resolution images of the earth's surface across hundreds of wavelength bands, is becoming increasingly popular. The high spectral information in these images can be analyzed to identify the various target materials present in the image scene based on their unique reflectance patterns. An autoencoder is a deep learning model that can perform spectral unmixing by decomposing the complex image spectra into its constituent materials and estimating their abundance compositions. The advantage of using this technique for land cover quantification is that it is completely unsupervised and eliminates the need for labelled data which generally requires years of field survey and formulation of detailed maps. We evaluate the performance of the autoencoder on various synthetic and real hyperspectral images consisting of different land covers using similarity metrics and abundance maps. The scalability of the technique with respect to landscapes is assessed by evaluating its performance on hyperspectral images spanning across 100m x 100m, 200m x 200m, 1000m x 1000m, 4000m x 4000m and 5000m x 5000m regions. Finally, we analyze the performance of this technique by comparing it to several supervised learning methods like Support Vector Machine (SVM), Random Forest (RF) and multilayer perceptron using F1-score, Precision and Recall metrics and other unsupervised techniques like K-Means, N-Findr, and VCA using cosine similarity, mean square error and estimated abundances. The land cover classification obtained using this technique is compared to the existing United States National Land Cover Database (NLCD) classification standard.

Thousands of communities across the nation are exposed to health risks from contaminated drinking water. Upstream anthropogenic land covers have been linked with the degradation of source drinking water quality and likely pose a threat to a community water system's (CWS's) ability to provide safe drinking water. The goal of this study was to predict the differences in compliance with the Safe Drinking Water Act (SDWA) between CWSs based on their upstream land cover, economic situation, and system characteristics. In Central Appalachia, from 2001 to 2016, proportions of land cover in each target CWS's upstream source water watershed were weighted based on their distance to a CWS's source water intake. Violations to the SDWA at respective CWSs over the same period were modeled with their distance weighted land cover proportions, economic status of the county served, and system characteristics as covariates. The major findings were that increases in low intensity development increased the likelihood of a health-based violation, larger CWSs were less likely than smaller CWSs to obtain a monitoring and reporting violation, and CWSs that distributed purchased water were the least likely to incur either violation type. These results suggest that communities that have CWSs that are repeatedly failing to remain in compliance with the SDWA may be able to reduce public health risks associated with drinking water by purchasing from a larger CWS. Further to protect public health, community managers should consider source water protection and/or upgrading a CWS's treatment capacity prior to developing a previously undeveloped area.

Over the last 20 years, Lyme disease has grown to become the most common vector-borne disease affecting Americans. Spread in the eastern U.S. primarily by the bite of Ixodes scapularis, the black-legged tick, the disease affects an estimated 329,000 Americans per year. Originally confined to New England, it has since spread across much of the east coast and has become endemic in Virginia. Since 2010 the state has averaged 1200 cases per year, with 200 annually in the New River Health District (NRHD), the location of our study. Efforts to geographically model Lyme disease primarily focus on landscape and climatic variables. The disease depends highly on the survival of the tick vector, and white-footed mouse, the primary reservoir. Both depend on the existence of forest-herbaceous edge-habitats, as well as warm summer temperatures, mild winter lows, and summer wetness. While many studies have investigated the effect of forest fragmentation on Lyme, none have made use of high-resolution land cover data to do so at the peridomestic level. To fill this knowledge gap, we made use of the Virginia Geographic Information Network’s 1-meter land cover dataset and identified forest-herbaceous edge-habitats for the NRHD. We then calculated the density of these edge-habitats at 100, 200 and 300-meter radii, representing the peridomestic environment. We also calculated the density of

This study is focused on the population of federally-endangered American burying beetles in south-central Nebraska. It is focused on changes in land cover over time and at several levels of spatial scale, and how management efforts are impacting both the beetle and a changing landscape. Our findings are applicable to a large portion of the Great Plains, which is undergoing the same shift from grassland to woodland, and to areas where the beetle is still found.

This dissertation investigated the potential impacts of changing climate, land cover, and sea level on the future stormwater budget and extreme runoff in coastal urban-natural environments. Two large-scale basins, namely Florida Southeast Coasts Basin (7117 km2) and Saint Johns River Basin (24928 km2), were selected for this study. The basins represented gradients in climate, land cover, and hydrology across the Atlantic coasts of Florida and the southeast U.S. Two mechanistic hydrologic models were developed for the basins using U.S. Environmental Protection Agency (U.S. EPA)’s Storm Water Management Model (SWMM) 5.1. The models were calibrated and validated with observed historical streamflow of the 2010s (2004-2013), computing the corresponding runoff volume as a historical reference. Runoffs for 2050s (2044- 2053) and 2080s (2076-2085) were quantified by incorporating climatic projections from 20 General Circulation Models (GCMs) and land cover projections from EPA under the Representative Concentration Pathways (RCP) 4.5 and 8.5 scenarios. Results suggested a predominant climatic control on potential runoff changes and high vulnerability in the coastal urban-natural environments. The relative increases in runoff were higher during the dry season and transitional months (October-May) than the wet season (June-September). Based on the basin-scale annual averages, runoff sensitivity to changes in rainfall was substantially stronger than that to changes in watershed imperviousness and evapotranspiration. Further, the concurrent changes in climate and land cover led to synergistic (stronger) nonlinear responses of runoff, compared to the linear summation of their individual effects. The dissertation also evaluated potential future changes in extreme events (rainfall and sea level) and associated runoff extremes across the two basins. Significance and direction of trends in 50-year annual maximum rainfall and sea level of 1-7 day durations for the 2000s (1964-2013) were evaluated by non-parametric Mann-Kendall test and Theil-Sen slope estimator. At the 5% level of significance, statistically insignificant mixed (increasing and decreasing) trends were noted in historical rainfall and significant increasing trends were observed in sea level. Accordingly, historical and future stationary design rainfalls and nonstationary design sea levels for 1-7 day durations with varying return periods (2, 5, 10, 25, 50, and 100 year) were estimated using Gumbel probability distribution, representing the family of generalized extreme value distributions. Based on the analyses, the extreme rainfalls and sea levels over various durations would substantially increase in the 2050s (2025-2074) and 2080s (2050-2099). Higher (than other locations) runoff increases were noted at and around the urban centers across the two basins. The projected extreme stormwater runoff under the changing regimes of extreme rainfall, sea level, and land cover suggested intensified risks of pluvial flooding and water pollution across the Atlantic coasts of Florida. The findings would guide stormwater management and ecosystem protection for coastal built and natural environments in the southeast U.S. and around the globe.

Little research has been done on the land cover change of unconventional oil and gas (UOG) wells across the Appalachian basin. In the last decade, UOG development has increased drastically across the region. Permitted well data from 2007 to 2017 were assembled into a geospatial database and analyzed with current aerial imagery to determine the activity of permitted wells in Ohio, Pennsylvania, and West Virginia. To capture all disturbance related to UOG development, a 25-hectare buffer was placed around each active well as the study extent. A supervised classification approach was used to determine the pre-development and post-development landcover across all 4,212 well pad buffers in the region. Overall, the average forested and grass land cover percentage decreased by 4.3 and 0.8 percent across the Appalachian basin, while the impervious surface percentage gained an average of 5.2 percent respectively. The largest land cover change category was forest to grass, with an average of 3.8 hectares per well pad buffer across the region. Forest and grass land cover changed to impervious land cover by an average of 0.8 and 0.9 hectares per well buffer. The trend suggests that well pads are currently being placed in areas with higher grass landcover. Thus as development continues in the region, results indicate that impacts could increase in more forested cover types.

Lyme disease has been a growing problem in the United States over the last few decades, and is currently the most common vector-borne disease in the country. This research evaluates the land cover within specified counties of northern Virginia to determine if a correlation exists between forest fragmentation, suburbanization, and cases of human Lyme disease as has been demonstrated in other Lyme endemic regions in the United States. Few studies have focused specifically on northern Virginia when considering the impacts of land cover change on Lyme disease. Discovered through the use of geospatial and statistical analysis, the cluster of Lyme disease cases in northern Virginia are associated with forest fragmentation within the study region, which creates an ideal habitat for black-legged ticks and the white-footed mouse, allowing for an increase in Lyme disease transfer from vector to humans. The goal is for the research findings to be applicable to other regions with similar land cover types. Regions with similar characteristics would then be able to recognize the potential risk of human Lyme disease and implement ways to reduce the Lyme disease risk associated with suburban development. The purpose of this study is to answer the following research questions: 1) How has the spatial distribution of Lyme disease in Northern Virginia changed since 2005 with respect to land cover? 2) Which suburban communities are more at risk for Lyme disease when considering their land cover types and the increasing spatial distribution of Lyme disease?

Indigenous community land tenure systems in many locations worldwide are shifting toward individually parcelized and privatized systems. Among the drivers of this shifting land tenure, are distant political-economic forces and commodity markets, from local to global. Accompanying the observed land tenure changes are shifts in livelihoods, away from subsistence-based, and toward market-oriented activities. These changes can ultimately impact land use and land cover patterns. I investigated a global-to-local casual pathway, from agriculture, livestock and forestry production for distant markets, extending through shifting land tenure and livelihoods, to impacts on land use and land cover within ejidos (a type of community landholding) across Yucatán, México where Maya people are the primary land stewards. My research questions are: 1) what underlying and broad-scale social processes influence changes to ejido land tenure and livelihoods, and 2) in what way do the changes to ejido land tenure and livelihoods influence land use and land cover? To investigate these questions, first I use descriptive statistics to initially assess the shift in ejido land tenure, from community to individually parcelized systems, and the shift in a principal subsistence livelihood and land use activity, from maize cultivation to cattle rearing. Next, I use ordinary least squares regression, mapped variables, and variographic analyses to assess spatial patterns and correlations. I further explored relationships among variables using spatially-explicit simultaneous autoregressive models. Finally, I use remotely sensed satellite images to map and analyze changes to land use and land cover patterns across the state, and particularly among ejidos (pre- and post- parcelization). I show that commodity production for distant markets is strongly related to parcelized ejido lands. Moreover, I show that changes to land use and land cover patterns among ejidos are associated with changes to ejido land tenure. Specifically, parcelized ejido lands are often deforested. Conversely, I show that community-managed ejidos comprise a larger percentage of conservation lands or lands undergoing forest regeneration, and therefore are much more likely to be densely forested. In all, I conclude that land privatization can threaten the conservation potential of community-managed forests and landscapes across the state of Yucatán.

This research is composed of three parts: 1) Adaptation of Soil Judging to Libya, 2) Predicting the classes and distribution of salt-affected soils in Northwest Libya, and 3) Geospatial analysis of land use change in the Savannah River Basin using Google Earth. Soil judging (Evaluation) plays an important role in soil science education. Libya has six soil orders according to the U.S. Soil Taxonomy (Entisols, Aridisols, Alfisols, Inceptisols, Vertisols, and Mollisols) and the most common soil orders are Entisols and Aridisols. A Soil judging (Evaluation) scorecard was tested at two different universities in Libya: The University of Tripoli and University of Zawia. Eighty-two percent of Libyan users were not aware of Soil Judging prior to this study. After completing Soil Judging trials in various locations in Libya, ninety-five percent of those surveyed indicated that Soil Judging is useful to the natural science education. Libya is mostly a dry and arid country, where sodicity and salinity problems are often accelerated by the prevailing climatic condition and geographical setting of the area. A framework was identified for classifying and mapping salt-affected soils in northwest Libya using field measurements (ECe, soil pH, and SAR) and Geographic Information Systems (GIS). The majority of soils in this region of Libya are normal (slight degree of limitation). Twenty percent of the topsoil is saline-sodic (extreme degree of limitation). Land use change and the loss of wildlife habitats are serious issues facing the Southeastern United States. Across the Savannah River basin, the major change of land use was deforestation and reforestation during the entire study period with most of the changes located near lakes and water tributaries.

Excessive suspended sediment and nutrient concentrations often result in negative environmental impacts within rivers and on downstream water bodies. Land cover, stream discharge, and effluent from waste water treatment plants (WWTPs) can all influence sediment and nutrient dynamics in rivers, yet often these factors are not considered together. The purpose of this study was to determine the impacts of land cover, stream discharge, and WWTP discharge on sediment and nutrient concentrations, across a large land cover gradient (4-71% agricultural; 7-92% urban), in Southwest Ohio streams. Weekly baseflow samples were collected from eight streams over a one year period. Total suspended sediment, total nitrogen, nitrate, total phosphorus, and phosphate concentrations, along with total nitrogen:total phosphorus ratios were measured. Results show that WWTPs were primarily responsible for increasing nutrient concentrations in the study area. Agricultural land cover also significantly increased total nitrogen and nitrate concentrations and marginally increased phosphate concentrations, however, was a weaker predictor of water quality relative to WWTP discharge. Stream discharge significantly increased total suspended sediment and all nutrient concentrations, except phosphate, for which it had a marginal impact. Results suggest that to improve water quality in the study area, efforts should focus on improving WWTP effluent treatment and agricultural land management.

The primary goal of this dissertation was to leverage the capabilities of remote sensing technology for capturing detailed spatial information at different spatial resolutions to monitor agricultural crops and generate accurate input datasets for water resources models. This dissertation is divided into three different research studies. In the first study, a remote sensing classification method was developed for classifying irrigated and non-irrigated fields that integrates Vegetation indices with surface energy balance fluxes. The method was applied in the COHYST2010 hydrological model region with wide climate variation and to multiple growing seasons with results that were 92.1% accurate and explained 97% variation in National Agricultural Statistics Service (NASS) county irrigation statistics. In the second study, a new method was developed (referred to as “footprint method”) of re-projecting Moderate Resolution Imaging Spectroradiometer (MODIS) satellite images that preserves the geometric orientation and size of satellite sensor pixels. It properly represents satellite sensor pixel orientations in fields, and eliminates artifacts introduced by conventional processing methods. Statistical results of field comparison in AmeriFlux experimental fields US-Ne1 and US-Ne2 based on Leaf Area Index (LAI) equation of Myneni eat al. showed improvement in LAI estimation when the footprint method was applied with reduced RMSE by 16.05%, ubRMSE by 26.25%, and nRMSE by 16.1% in average. On the contrary, the results of statistical analysis of MODIS Green LAI estimates based on Green LAI equation of Viña et al. does not support this conclusion. A third study explored the potential opportunities and benefits of utilizing gridded precipitation data, which is the combination of remotely sensed and weather stations precipitation data with more detailed spatial variability in water resources models. This study explored differences in spatial patterns between precipitation and recharge maps generated by interpolating data from weather stations and maps generated by gridded method. The percentage difference in annual average precipitation volume over 16 million acres of the Republican River basin area was around 14%. In a sensitivity analysis of precipitation in the watershed model, the effects of same rates of precipitation were found to be different for different types of soils, crops, and irrigation settings. Advisors: David M. Admiraal and Trenton E. Franz

Shrublands and successional ecosystems are of special interest to land managers and conservation groups because to be maintained, they must be disturbed periodically. Shrubland birds rely on variation in the landscape, regular disturbance, and ephemeral patches of shrubs for breeding, foraging, and nesting. Many of these avian species are in decline, so monitoring and mapping shrublands is necessary for future conservation and management strategies (DeGraaf & Yamasaki, 2003; Howard et al., 2015; Litvaitis, 2003). Large proportions of shrubland bird populations rely on shrublands in the northeast states for breeding habitat, and the identification and land use monitoring of these regions are important for their continued survival. Remote sensing approaches were used for habitat modeling in this study because comprehensive field analyses of the status and condition of all the shrubland habitat in the study area (Monroe County, NY) would have been resource and labor intensive. To determine the quantity, distribution, and features of the quality of shrubland habitat in Monroe County, two types of models were created, and their similarities, differences, and accuracies were assessed. The first model used the National Wetland Inventory (NWI) and the National Land Cover Database (NLCD) to define optimal conditions for eight species of passerine songbirds based on habitat preferences found in literature. This classification produced an overall accuracy of 71.1% and a K coefficient of 0.65. The second model type was a supervised classification intended to find important patches of habitat not included in the NLCD classification scheme. The second classification yielded a different set of habitat requirements with an overall accuracy of 71.4% and a K coefficient of 0.66. A preliminary study on the separability of woody shrub species based on their spectral signatures was promising but inconclusive statistically and unable to be applied to high-resolution imagery to remotely determine invasive vegetation cover.

Quality of life in urban areas is strongly linked to land use and land cover, in part because green vegetation mitigates much of the negative consequences of urbanization and population pressures. However, the green vegetation of urban parks, forests, street trees, and landscaping is inequitably distributed in the urban environment. The social and economic processes that give rise to these uneven outcomes are not well-understood, while the rise in the availability of spatially explicit, fine-scale data on neighborhood conditions has created the conditions for an empirically rich investigation into neighborhood socio-ecological change. This dissertation assimilates new observations from different sources with new modes of inquiry to address persistent knowledge gaps: the dependence of socio-ecological relationships on scale and urban or metropolitan context; understanding the duration and significance of neighborhood improvement or decline; and the outstanding need for comparative analyses and novel analytical techniques for comparing neighborhood change between multiple metropolitan areas. Time-series satellite remote sensing of 30 years of vegetation cover is combined with population and housing market data to provide a comprehensive picture of the neighborhood environmental quality, demographic composition, and housing stock conditions. Three different metropolitan areas, Detroit, Los Angeles, and Seattle, are used to elucidate how our common assumptions of socio-ecological relations---and the underlying analytical approaches in which remote sensing plays a pivotal role---often fail to accurately capture the complexities and contradistinctions in the social and economic drivers of neighborhood-level biophysical changes. Results indicate that while population decline confounds conventional explanations for socio-economic differences in environmental quality, neighborhood advantages and disadvantages persist for multiple decades, with wealthier neighborhoods tending to resist cyclical declines in the housing market and accrue yet higher home values while preserving and increasing vegetated cover through irrigation and likely several policy tools. Historical conditions, particularly racial residential segregation, also yield surprising outcomes today, in some places reducing vegetation disparities and exacerbating them in others, depending on metropolitan-level population pressures and the balance of municipal political economies.

Evapotranspiration (ET) is a hydrologically and eco-agronomically important process that can bealtered by soil properties, crop type and mechanisms of photosynthesis (e.g. C3 and C4), cropstatus, agricultural practices (crop rotation and monoculture), and meteorology.In particular, corn monoculture, which is widely used in the U.S, may affect over agriculturalfields differently than soybean and wheat deu to the different (C4) photosynthesis mechanism,and thus can have an impact on local hydrologic cycle and climate. Satellite observations are themost sophisticated technology to monitor different rates of ET at large scale. This study used datafrom two satellites, Landsat 8 and Sentinel 2, to examine the capability of combining those datain ET time series to explore the differences between ET rates for C3 (soybean and winter wheat)and C4 (corn) crops. ET was estimated for a study area located in the Western Lake Erie basinfor 2016 and 2017 using satellite data and the Boreal Ecosystem Productivity simulator (BEPS),a process based ecosystem model, modified for the agricultural ecosystem. Satellite images(from which land cover/land use data, and leaf area index were generated), weather (Gridmetdata), and soil data (SSURGO data) were main inputs to BEPS. In addition, a sensitivity analysiswas conducted to estimate ET for different percent increments of the total area covered by cornto the point of becoming a monoculture using synthetically developed land covers and LAIimages. For both years, corn and soybean reach the maximum ET rate in the mid-growing seasonas expected with the peak being somewhat later in the season for soybean. The ET relationshipbetween two sensors was strong during the mid-season (r = 0.95 for July) when LAI was high,and at the end of the season, when many crops were harvested and soil exposed (r = 0.98 forivOctober). A high correlation was also observed when data were acquired within a short period of time (e.g. 2 days). The correlation was moderate at the beginning of the growing season (r = 0.59), in June, due to the 10-day difference of data acquisition and high heterogeneity of the land cover. The sensitivity analysis showed that an elevated stomatal conductance increased the ET rate; however, for the same stomatal conductance, soybean transpired at a higher rate than corn, suggesting that corn monoculture would not increase the ET rate most likely due to its high water efficiency as a C4 plant. Thus, this study suggests that different percentages of corn cover affect the ET rate of croplands but not considerably. Stomatal conductance of corn has more effect on the ET rate than the increase of the corn cover area. These results are valuable for other studies conducted in the Western Lake Erie basin that are related to the land-use management of the shoreline where crops with a high ET rate are expected to be more beneficial in combating lake eutrophication.

Many parts of the High Plains region are facing declining aquifer levels, which threatens the long-term viability of irrigated agriculture. Furthermore, some areas of the High Plains region, like the Republican River Basin in Nebraska, need to keep groundwater levels high enough in the short-term to ensure that hydrologically connected rivers have enough streamflow to fulfill surface water obligations, such as Nebraska's interstate river compact with Colorado and Kansas. To better manage groundwater, it is important to understand the unintended effects of policies that may not be aimed at groundwater conservation, such as the USDA- Conservation Reserve Program (USDA-CRP). The CRP pays farmers to take cropland out of production and put it into conservation covers, mainly grassland. Environmental benefits include reduced soil erosion, improved surface water quality, and increased wildlife habitat. But, the changes in land cover due to CRP enrollment could also impact the infiltration of precipitation through the soil, thus changing groundwater recharge. The paper estimates the potential effect of CRP on groundwater levels using data from Ogallala Aquifer region of Kansas and the Republican River Basin portion of Nebraska. The analysis relates disaggregated aquifer level data with spatial land cover data, weather, soil, and groundwater extraction data. Grassland land cover is used as a proxy for grassland put in by CRP. Findings suggest that grassland leads to a lower yearly recharge rate than common crop land covers in the Republican River Basin of Nebraska. Recharge in the Ogallala Aquifer region of Kansas seems too small to have a detectable impact though. These results imply that in addition to other environmental benefits, policymakers need to pay attention to the impact of CRP enrollment on regional aquifer conditions in regions where groundwater levels are a concern but can expect recharge to take place. Adviser: Karina Schoengold and Taro Mieno

Lyme disease is the most significant vector-borne disease in the United States. Its southward advance over the last several decades has been quantified, and previous research has examined the potential role of climate change on the disease's expansion, but no research has considered the role of future land cover patterns upon its distribution. This research examines Lyme disease risk in the southeastern United States based on estimated land cover projections under four different Intergovernmental Panel on Climate Change Special Report Emissions Scenarios (IPCC-SRES) A1B, A2, B1, and B2. Results are aggregated to census tracts which are the basic unit of analysis for this study. This study applied previously established relationships between Lyme disease and land cover in Virginia to the projected land cover layers under each scenario. The study area, the southeastern United States, was defined from Level III Ecoregions that are present in Virginia and extend throughout the Southeast. Projected land cover data for each scenario were obtained from the USGS. The projected land cover datasets are compatible with the National Land Cover Dataset (NLCD) categories and had seventeen land cover categories. The raster datasets were reclassified to four broad land cover types: Water, Developed, Forest, and Herbaceous areas and the relationship between certain landscape configurations were analyzed using FRAGSTATS 4.2. Significant variables established in previous research were used to develop a spatial Poisson regression model to project Lyme disease incidence for each decade to the year 2100. Results indicated that potential land cover suitability for Lyme disease transmission will increase under two scenarios (A1B and A2) while potential land cover suitability for Lyme disease transmission was predicted to decrease under the other two scenarios (B1 and B2). Total area under the highest category of potential land cover suitability Lyme disease transmission was calculated for each year under each scenario. The A2 scenario experiences the most rapid acceleration of potential land cover suitability for Lyme disease transmission, with an average increase of 16,163.95 km² per decade, while the A1B scenario was projected to show an average increase of 3,458.47 km² per decade. Conversely, the B1 scenario showed an average decrease of 595.7 km² per decade and the B2 scenario showed the largest decrease of potential land cover suitability for Lyme disease transmission with an average decrease of 2,006.83 km² per decade. This study examined the potential spatial distribution of potential land cover suitability for Lyme disease transmission in the southeastern United States under four different future land cover scenarios. The results indicate geographic regions of the study area that are at greatest risk of potential land cover suitability for Lyme disease transmission under four different predictive scenarios developed by the IPCC. The A1B and A2 land cover projections are predicted to have an overall increase in areas where the Lyme disease transmission cycle will be enhanced by 2100 and the scenarios have a primary focus on economic development. Economic concerns outweigh environmental concerns for the A1B scenario, in addition to a high standard of living. The A2 scenario describes rapid population growth which results in high rates of land cover conversion to developed land; in addition, this scenario describes a reduction of environmental protection. The B1 and B2 land cover projections are predicted to have an overall decrease in areas of high Lyme disease transmission by 2100 and these scenarios have a central focus on environmental sustainability. The B1 scenario is characterized by a high environmental awareness which results in lower demand for forest products. A common theme for the B1 scenario is restoration and forest protection. Finally, the B2 scenario is described as improving local and regional environmental value which results in a high demand for biofuels and repossession of degraded lands, and an overall increase of forest cover. This study was the first to predict potential land cover suitability for Lyme disease risk and geographic distribution using projected land cover in the southeastern United States, and the results of this research can aid in the reduction of Lyme disease as it continues to expand in the south.

Pathogenic impairments are widespread across the United States and the source of the pollutant is both complicated and often elusive. The boundary of Greenville County, South Carolina crosses four Hydrologic Unit Code 8 watershed designations, and within those, there are over one hundred approved Total Maximum Daily Loads related to fecal indicator bacteria. This region is experiencing rapid growth and development; consequently, it is critical for planners to better understand the nexus between land use choices and water quality impacts, as zoning, ordinance decisions, and comprehensive plans can affect this relationship. The goal of this research was to correlate urbanized land cover from the National Land Cover Database and land uses by examining impervious surface types with E. coli levels in Greenville County and from those results, provide insight for land use planning policy. I hypothesized that there was a relationship between land cover typology and impervious land uses with E. coli, which could generate land management policy adjustments. The null hypothesis was that no correlation could be determined between land cover or land use and E. coli levels in Greenville County. The analysis revealed the following results, rejecting the null hypothesis. First, both developed land cover and land use types lead to increases in E. coli levels. Second, the presence of National Pollutant Discharge Elimination System permits and agricultural fecal spreading permits lead to decreases in E. coli. Third, and most importantly, knowledge of the sampling condition, specifically wet weather, is highly correlated to E. coli levels and leads to significant increases in measured E. coli. This third observation indicates that land use, the presence of sewer overflows, failing septic tanks, the resuspension of bottom sediment living E. coli, or a combination of these appears to elevate E. coli levels. Despite the presence of this relationship, the resulting adjusted coefficients of determination (R2) from both multi linear regression analysis (0.41-0.42) and factor analysis (0.21) suggest that there are sources of E. coli for which this study does not account. Accordingly, it will be essential to conduct additional research that examines the relationship between more accurately-specified land use types and E. coli levels. Also, there is a need to understand the nuances of the relationship between wet weather sampling conditions and E. coli presence and concentration fluctuation. Regardless of the remaining gaps, this research reinforces the imperative involvement of planners in conversations about water quality within their jurisdictions. The consequences of land use policy making are evident in water quality, and planners should be cognizant of water quality consequences as they accommodate and plan for future urban growth.

This study identifies land cover changes associated with a ten-year drought period and discusses the importance of vegetation change in Diné Bikéyah, a semi-arid land located in a remote part of the southwestern United States (US). This study concludes that drought produced slight changes in vegetation within a 540 km2 study area in the Tselani-Cottonwood Chapter (TCC) in Diné Bikéyah. The data for this study consist of three Landsat images for the years 1998, 2002, and 2009. The methods used to analyze these Landsat images included image pre-processing, calculation of normalized difference vegetation index (NDVI) images, and supervised (maximum likelihood) classification of land cover. The classification analysis yielded five land-cover categories; land-cover change was assessed using standard change-detection techniques, followed by accuracy assessment of the results. Land-cover change was minimal over the ten-year study period in the TCC, which suggests that vegetation in the study area is resilient to drought and livestock grazing. Identifying land-cover change provides insight on the impact of a drought in Diné Bikéyah. Given the importance of vegetation in Diné livelihoods, and the limited knowledge of land-cover change associated with drought and other climate events, analysis of land-cover change using remote sensing offers the ability to understand how Diné people and livestock might contribute to land-cover change in addition to developing sustainable livestock management strategies.

Precipitation and land cover in the West African Sahel have changed dramatically over the past 50 years. Region-wide data on land cover change in the Sahel, however, have been sparse or unreliable. I present a new 30 meter 2000-2016 annual resolution land cover dataset for the West African Sahel. The product is built from hand-classified land cover maps using random forest machine learning methods with Landsat, precipitation, and topography features. The resulting maps confirm the widespread extensification of agriculture in the region over this time period. Contrary to the common narrative of desertification, this increase in agriculture has not been accompanied by an increase in bare soil or sandy area. Land cover change volatility is shown to be spatially heterogeneous, both at local and regional scales. In addition to the new land cover dataset, I present spatial and temporal analyses of precipitation during the recent years of increased variability in the West African Sahel. I examine seasonal trends, interannual variability, and differences among datasets representing precipitation in the Sahel. Region-wide spatial organization of precipitation is identified using the self-organizing mapping pattern recognition technique. The number of days spent in the monsoon transition period is strongly negatively correlated with annual precipitation anomaly` indicating a tradeoff with the peak monsoon period, a result that supports the upped-ante hypothesis of precipitation in the Sahel.

Vector-borne pathogens, such as Malaria, Dengue, West Nile and Zika virus, infect hundreds of millions of people each year and lead to widespread human morbidity and mortality, with enormous spatial and temporal variation in disease risk. The recent emergence of West Nile virus (WNV) into the developed world offers a unique opportunity to better understand ecological drivers that contribute to variation in disease risk, as a step toward more effective disease management through targeted interventions. In Chapter 1, we examined correlations between rice cultivation and WNV human disease incidence in rice-growing regions within the United States (US). We found WNV human disease incidence increased with the fraction of each county under rice cultivation in California, but not in the southern US. We show that this is likely due to regional differences in the mosquitoes transmitting WNV. These results illustrate how cultivation of particularly water-demanding agricultural crops can increase mosquito-borne disease risk and how spatial variation in vector ecology can alter the relationship between land cover and disease. In Chapter 2, we examined the effect of irrigation, climate and land cover on mosquito abundance and WNV human disease cases across California. Irrigation made up nearly a third of total water inputs to the region, with irrigation exceeding precipitation in some dry regions. Irrigation reduced seasonal variability in mosquito abundance by more than 40%, and increased abundance by more than an order of magnitude. In addition, irrigation increased human WNV cases and explained 33% of variation in WNV cases among California counties. These results suggest that irrigation can increase and decouple mosquito populations from natural precipitation variability, resulting in sustained and increased disease risk. In Chapter 3, we quantified the risk of WNV transmission to humans from 6 Culex mosquito species by integrating mosquito abundance, infection prevalence, vector competence, and blood feeding patterns, and examined correlations between risk indices and human disease cases. Human WNV cases were strongly correlated with the density of infectious vectors feeding on humans. However, different mosquito vector species contributed to transmission in different land use types and within seasons and across years. Culex tarsalis was more abundant in agricultural areas, whereas Culex pipiens and Culex quinquefasciatus were abundant in developed and agricultural areas, and Culex erythrothorax was abundant in wetland areas. As a result, WNV risk did not change substantially along either agricultural or urbanization land use gradients because the diversity of vectors maintained high disease risk across a range of habitats. These results show how a diversity of vectors can maintain an ecosystem disservice - vector borne disease – by their differential response to environmental disturbance.

Abstract: Thunderstorms are common in boreal forest regions and can cause dangerous hazards such as lightning, forest fires, hail, wind, and flooding. Significant research has been conducted to help predict thunderstorms to mitigate or avoid the hazards and damage. The development of thunderstorms depends on many factors, including land cover variations. In this thesis, the Athabasca oil sands development (an anthropogenic land cover modification), and the Canadian Shield (a natural land cover variation), will be examined. The oil sands development creates a massive almost 1000 km2 land disturbance, changing boreal forest to barren land, tailings ponds, and bitumen upgrading facilities. The Canadian Shield is a drastic land cover change from lusher boreal forest on soil to sparser boreal forest on exposed Precambrian bedrock interspersed with intermittent deep, cold lakes. The effect of land disturbances in the boreal forest is relatively unexplored, and cooler, drier climatic conditions and land cover could result in unexpected effects on thunderstorms. The first results chapter in this thesis compares temperature, humidity, precipitation, and lightning near and away from the oil sands development, searching for temporal trends as the oil sands development increases in size. Comparing how weather near the oil sands development changes over time, with respect to weather away from it, eliminates differences due to local effects and general climate variation. The precipitation and lightning did not change over time, thus, the oil sands development does not appear to affect thunderstorm climatology. However, a strengthening heat island and dry island were detected within the oil sands development. In particular, the overnight temperatures have increased about 1.2°C relative to the surroundings. These phenomena are thought to be caused by clearing the land, causing a higher Bowen ratio, and emissions of waste heat from oil upgrading, which might be able to trigger thunderstorms in rare situations. The second results chapter in this thesis used the Weather Research and Forecasting (WRF) model to perform sensitivity experiments. Factor separation was used to quantify the effect of adding/removing two major environmental factors caused by the oil sands development: the land cover disturbance, and the emissions of waste heat. The effect of the oil sands development on thunderstorm intensity was insignificant on all ten days investigated. However, on two of the case study days, the oil sands development appeared to cause thunderstorms to occur one to two hours earlier. The analysis indicates that the oil sands likely triggered storms earlier in these cases, but they appear to be rare. Aircraft measurements indicated that the oil sands development affects thunderstorms mostly when the vertical totals index is greater than 30°C. However, there are probably not enough of these cases to measurably affect climatology. The third results chapter in this thesis analyzed the variation of detected cloud-to-ground lightning density near the Canadian Shield boundary. The results show that significantly less lightning occurs inside the Canadian Shield than just outside, and a strong cloud-to-ground lightning density gradient exists along the boundary. Various statistical analyses suggest that the lightning density gradient is statistically significantly higher near the Canadian Shield boundary than away from it. However, the signal was not detected in regions with large lakes or more complex topography, which seemed to dominate over the Canadian Shield effect. In one region, the Lightning density decreased from 10 strikes per square km down to 6 over less than 100 km. This effect may be caused by a lower detection efficiency in the Canadian Shield. However, it also could be caused by the sudden change in land cover from more lush, moist, higher transpiring broadleaf forest to sparser, lower transpiring needle-leaf forests on bare bedrock. This thesis shows that two unique land cover variations in the boreal forest can affect thunderstorms. However, the magnitude of the effect was not always what was expected. The Athabasca oil sands were expected to enhance thunderstorms significantly, but their effect was much weaker than expected. The Canadian Shield was not anticipated to have much effect at all; however, it significantly diminished thunderstorms. It seems that because thunderstorms occur less in a shorter season and are generally weaker in the boreal forest, that land cover variations could cause thunderstorms to be more difficult to enhance, and easier to diminish.

This dissertation introduces a novel method for automatic pattern-based regionalization of large raster maps. The method utilizes techniques from several disciplines including data science and machine vision. A detailed analysis of different techniques and ideas leads to the construction of this robust methodology. The methodology consists of four steps: (1) dividing a raster into an arbitrary regular grid of blocks of cells, (2) calculating histograms of color co-occurrence, which are numerical representations of patterns in each block, (3) a region growing segmentation of the grid, (4) Ward's hierarchical clustering of the segments. The method inherits its unsupervised character from the usage of a self-seeding region growing segmentation algorithm, and a clustering algorithm. Because the process is unsupervised, the delineation of regions comes directly from the structure of data, without having any additional input from the analyst. To obtain the best possible regionalization, the final number of clusters of patterns can be found by using either the silhouette technique or the L-method. The performance of the methodology is presented as a study of landscape pattern types (LPTs), spatial units each containing a unique quasi-stationary pattern of land cover categories, over the entire conterminous United States. A hierarchy of nested LPTs in the form of dendrogram is the result. The hierarchy of LPTs is studied by cutting the dendrogram and mapping the resulting clusters. Finally, the L-method finds that there are nine most distinct land cover patterns in the United States.

Research on the impacts of transport infrastructure is limited, and most of it is either focused on rural areas or on developed areas before modern geospatial technologies were available. This study aimed to fill this gap in transport research by providing a holistic look at the regional changes that occurred due to new transport infrastructure construction in Northwest Arkansas between 1980 and 2011. The National Land Cover Database was used to create a time-series of land cover across the region between 1992 and 2011. These data were then used to predict future growth in the region. Additionally, growth patterns of the four largest cities (Bentonville, Fayetteville, Rogers, and Springdale) were studied, and monocentricity values for the region were calculated to study population movement. Data on income, retail trade, gross merchandise trade, and worker flows were used to economically characterize the region for the time period of study. The results show that development is mostly occurring on top of agricultural land, though higher intensity developments do are often replacing lower intensity uses. A third of all development occurs within a half-mile of a US highway, with a third of these developments occurring within a half-mile of I-540/I-49. The main cities of Northwest Arkansas are expanding their borders at a rate that maintains roughly 50% developed land cover, such expansion and land use track could make for an easily quantifiable measure of urban sprawl that could be combined with other factors to better quantify sprawl than current methods. Economically, Madison County, AR and McDonald County, MO both saw up to 80% permanent drops in economic trade that align with major milestones in the construction of I-540. This is further supported by a ten point drop in residents who chose to work outside of Benton or Washington County once the bypass was completed. Meanwhile, Benton County saw a short-lived doubling of trade, and Washington County saw a smaller, lagged effect that was also not permanent. These findings suggest that the I-540/I 49 created severe spatial and economic competition in the region with no obvious winner.

Due to their hazardous nature, most thunderstorm observations today come from remote sensing platforms such as radar, satellite, and lightning detection sensors. Advancements in these sensor networks provide the ability to identify and track thunderstorms at finer spatial and temporal scales than ever before. Thunderstorms, however, are products of interactions between the land and atmosphere with certain land use and land cover (LULC) types augmenting the frequency and intensity of thunderstorms. Yet, these LULC effects may not be directly apparent when examining the remote sensing fields in isolation. This dissertation represents three research endeavors, each containing a multi-year climatology of a unique remote sensing dataset, to examine how the addition of LULC information affects the identification of thunderstorms and their attendant hazards and the interpretation of remote sensing products. First, a 20-year climatology of cloud-to-ground (CG) lightning data at 500 m spatial resolution quantifies an increase in isolated regions of high CG lightning frequencies in concert with the construction of antenna towers to accommodate the expansion of broadcasting and telecommunications technologies across the United States. CG lightning occurrence is correlated with antenna height and 96% of towers examined had a higher lightning density with 1 km of a tower compared to 2 km to 5 km away. Comparing tower strikes in the northern Great Plains reveals that shorter towers are more likely to observe larger CG lightning densities in the meteorological winter/fall months compared to the spring/summer months. Second, a five-year multi-radar/multi-sensor retrospective examining the effects of city size on thunderstorm initiation and longevity reveals an increase in thunderstorm frequency downwind of both cities larger than 1100 km^2 on convective days with ingredients historically shown to be conducive for urban-enhancement (i.e., summer months, afternoon initiation, synoptically weak days, non-supercell modes). As a result, downwind regions experienced a higher frequency of more intense composite reflectivity, vertically integrated liquid, and maximum expected size of hail values compared to equivalent distances upwind. Such effects were not observed in the smaller two cities and were not observed in any of the cities when examining the full five-year dataset. Finally, Landsat 5 Thematic Mapper and Landsat 7 Enhanced Thematic Mapper Plus surface reflectance data were acquired to examine how tornadoes alter the spectral behavior of grassland, forest, and urban land cover. Generally, independent of land cover type, tornadoes tend to increase the surface reflectance in the visible and shortwave-infrared spectral bands and decrease surface reflectance as measured within the near-infrared (NIR) spectral band. This results in a higher tasseled cap brightness, lower tasseled cap greenness and wetness, and a lower Normalized Difference Vegetation Index (NDVI) values. With tornado damaged forests providing an analogous signature to forest clearing, a five-year climatology of Landsat imagery was acquired to compare NDVI, a common damage identification metric derived from the red and NIR spectral bands, to a disturbance index (DI) derived from the tasseled cap indices to examine tornado damage in forests. DI is more resilient to seasonal variability as its coefficients are derived on an image-by-image basis, making it an optimal metric for a pixel-based identification and tracking of damage and subsequent recovery.

Habitat loss and fragmentation have been identified as one of the sources of biodiversity loss. Those issues are predominantly triggered by human activities such as the development of settlements, conversion of forest habitat into agricultural areas, and the development of infrastructure such as transportation and pipelines. Gunung Halimun Salak National Park (GHSNP) as one of the protected areas in Indonesia is deforested each year by approximately 1,473 ha or 1.3% of the total area while the capability of GHSNP to rehabilitate the degraded forest is only 500-800 ha annually because of a limited budget and at the shortage of human resources. The focus of this study was primarily on an ecological assessment of landscape scale habitat impacts on mammals and birds caused by infrastructure construction and land conversion. This was accomplished, by firstly identifying the current land cover and land cover change over a 15-year periods within the study area, GHSNP, Indonesia, as a case study; secondly, by assessing a class-level landscape metrics and their changes to detect forest fragmentation over a 15-year period within GHSNP; and finally, by analysing changes in the habitat network caused by fragmentation, land conversion, and disturbance. The potential effects on wildlife were also discussed, including how to derive and interpret such fragmentation effects and disturbance by applying Geographical Information System (GIS)-based quantitative modelling. Results of this study show that land cover in GHSNP has changed drastically with Forest and Agriculture experiencing the biggest decrease and increase by 35.63% and 463.74% respectively. It was also found that landscape metrics for forest cover in GHSNP show changes with a decrease in class area (CA), patch size (MPS), patch size coefficient of variation (PSCoV), mean shape index (MSI), mean core area (MCA) and mean proximity index (MPI) and an increase in number of patch (NP) and edge density (ED). These suggests that the forest cover in GHSNP has been fragmented for the past fifteen years. Regarding the effects of fragmentation and disturbance on mammals and birds, the responses of the forest-grassland mammals (high area and medium area demands), forest mammals (area below 1500 metres above sea level and area above 1500 metres above sea level and bird profiles (forest area) were similar to each other, with class area (CA) and number of patches (NP) being reduced as a response to both effects, except for small mammals requiring forest area below 1500 metres above sea level, which underwent a relatively huge increase in the number of patch for the disturbed area. Nevertheless, among all Ecological profiles established, mammals having a high area demand are considered to be most impacted from fragmentation and disturbance effects.

Urbanization has been associated with the degradation of streams, and a consequence of forest to urban land transition is a change in streamflow. Therefore, the purpose of this thesis is to examine the impacts of land-cover change in ten different watersheds in the rapidly urbanizing Atlanta, GA USA metropolitan area. Streamflow and precipitation data for a 30-year period (1986-2016) were analyzed in conjunction with land cover data from 1992, 2001, and 2011. Big Creek and Suwanee Creek experienced the most urbanization and increases (20%) in streamflow and runoff, and high flow (>95th percentile of flow) days doubled and increased 85%, respectively. Precipitation-adjusted streamflow for Peachtree Creek and Flint River decreased about 17%. Runoff ratios for South River were the highest among all watersheds, even the Etowah River, which remained moderately forested and had the most precipitation and slope.