Geography Theses and Dissertations
Permanent URI for this collectionhttp://hdl.handle.net/1903/2773
Browse
36 results
Search Results
Item TOWARDS OBJECT-BASED EVALUATION OF INDIVIDUAL FIRES AT GLOBAL SCALES(2019) Humber, Michael Laurence; Justice, Christopher O; Geography; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Fire is a complex biophysical variable that has shaped the land surface for over 400 million years and continues to play important roles in landscape management, atmospheric emissions, and ecology. Our understanding of global fire patterns has improved dramatically in recent decades, coincident with the rise of systematic acquisition and development of global thematic products based on satellite remote sensing. Currently, there are several operational algorithms which map burned area, relying on coarse spatial resolution sensors with high temporal frequencies to identify fire-affected surfaces. While wildfires have been analyzed over large areas at the pixel level, object-based methods can provide more detailed attributes about individual fires such as fire size, severity, and spread rate. This dissertation evaluates burned area products using object-based methods to quantify errors in burn shapes and to extract individual fires from existing datasets. First, a wall-to-wall intercomparison of four publicly available burned area products highlights differences in the spatial and temporal patterns of burning identified by each product. The results of the intercomparison show that the MODIS Collection 6 MCD64A1 Burned Area product mapped the most burned area out of the four products, and all products except the Copernicus Burnt Area product showed agreement with regard to temporal burning patterns. In order to determine the fitness of the MCD64A1 product for mapping fire shapes, a framework for evaluating the shape accuracy of individual fires was developed using existing object-based metrics and a novel metric, the “edge error”. The object-based accuracy assessment demonstrated that MCD64A1 preserves the fire shape well compared to medium resolution data. Based on this result, an algorithm for extracting individual fires from MCD64A1 data was developed which improves upon existing algorithms through its use of an uncertainty-based approach rather than empirically driven approaches. The individual fires extracted by this algorithm were validated against medium resolution data in Canada and Alaska using object-based metrics, and the results indicate the algorithm provides an improvement over similar datasets. Overall, this dissertation demonstrates the capability of coarse resolution burned area products to accurately identify individual fire shapes and sizes. Recommendations for future work include improving the quality assessment of burned area products and continuing research into identifying spatiotemporal patterns in fire size distributions over large areas.Item Characterizing tree species diversity in the tropics using full-waveform lidar data(2019) Marselis, Suzanne; Dubayah, Ralph; Geography; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Tree species diversity is of paramount value to maintain forest health and to ensure that forests are able to provide all vital functions, such as creating oxygen, that are needed for mankind to survive. Most of the world’s tree species grow in the tropical region, but many of them are threatened with extinction due to increasing natural and human-induced pressures on the environment. Mapping tree species diversity in the tropics is of high importance to enable effective conservation management of these highly diverse forests. This dissertation explores a new approach to mapping tree species diversity by using information on the vertical canopy structure derived from full-waveform lidar data. This approach is of particular interest in light of the recently launched Global Ecosystem Dynamics Investigation (GEDI), a full-waveform spaceborne lidar. First, successful derivation of vertical canopy structure metrics is ensured by comparing canopy profiles from airborne lidar data to those from terrestrial lidar. Then, the airborne canopy profiles were used to map five successional vegetation types in Lopé National Park in Gabon, Africa. Second, the relationship between vertical canopy structure and tree species richness was evaluated across four study sites in Gabon, which enabled mapping of tree species richness using canopy structure information from full-waveform lidar. Third, the relationship between canopy structure and tree species richness across the tropics was established using field and lidar data collected in 16 study sites across the tropics. Finally, it was evaluated how the methods and applications developed here could be adapted and used for mapping pan-tropical tree species diversity using future GEDI lidar data products.Item Parameterized and Machine Learning Methods for Estimating Evapotranspiration from Satellite Data(2019) Carter, Corinne Minette; Liang, Shunlin; Geography; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)The studies in this dissertation present evaluation of and improvement to parametric and machine learning regression methods for estimating evapotranspiration from remote sensing. It includes three main parts. The first part is an assessment of parametric regression methods for obtaining evapotranspiration from vegetation index and other variables. It was found that including more variables tends to improve results, but the form of the regression formula does not make a large difference. Algorithm performance is not as good for wetland and agricultural sites as for other land cover types. Re-training of algorithms for those surface type results in some improvement. The second part consists of an evaluation of ten machine learning techniques for retrieval of evapotranspiration from surface radiation and several other variables. It is found that the best results are obtainable using all available input variables to train the bootstrap aggregation tree, random kernel, and two- and three- hidden layer neural network algorithms. Performance is again found to be weaker for wetland and agricultural surface types than for other surface types. However, separate training of the machine learning algorithms with data from those surface types does not significantly improve performance. The third part consists of further refinement to the machine learning algorithms and application of the bootstrap aggregation tree method to generate evapotranspiration maps of the continental United States for 2012. It is found that separating snow and non-snow data points improves performance. Performance for all tested algorithms was similar against the validation data set, but best for the bootstrap aggregation tree using an independent test data set. Monthly mean maps of the continental United States are generated for the drought year 2012 using the bootstrap aggregation tree. Evapotranspiration levels are lower than those shown in comparison data sets for the growing season in the eastern United States, resulting from a low bias at high evapotranspiration values. Retraining with the training data set weighted towards higher evapotranspiration values reduces this discrepancy but does not eliminate it. It is clear that machine learning evapotranspiration algorithm results have a significant dependence on training data set composition.Item CHARACTERIZING HYDROLOGICAL PROCESSES WITHIN THE DATA-SCARCE ENVIRONMENT OF THE CONGO BASIN(2019) Munzimi, Yolande; Hansen, Matthew C; Geography; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)The Congo Basin in Africa is the world’s second largest river basin. Centrally located and with the greatest water resources in Africa, the basin is a vital resource for water and energy supply for a continent with increasing needs for safe water and energy. The Congo Basin’s streams and rivers could be impacted by human activities in the region, notably by land cover and land use change (LCLUC) considering the strong interactions between hydrology and ecosystem processes in the humid tropics. It could impact flow discharge downstream Congo River and hydropower potential at the Inga hydroelectric site, the largest such installation in Africa, located 150km upstream from the river’s mouth. The seasonal rainfall regime, to which the Congo River owes its regular flow regime, play an important role in mediating freshwater resources. An improvement to our baseline information on the Congo’s rainfall and streamflow dynamics allows for a greater quantitative understanding of the basin’s hydrology, necessary for the current and future management of Congo Basin water resources. The hydrometeorological observation network in the Congo Basin is very limited, and this environment of scarce ground data necessitates the use of remotely-sensed data for hydrological modeling. This dissertation reports the use of hydrological modeling supported by remotely-sensed data to 1) characterize precipitation and climate in the Congo Basin, 2) characterize daily streamflow across the basin, 3) assess the hydrological response to LCLUC, including the additional response caused by climatic feedbacks following LCLUC. The study uses rainfall gauge data within the Democratic Republic of Congo (DRC) to re-calibrate a TRMM science product. It then describes a physically-based parameterization of a semi-distributed hydrological model, augmented with a spatially-distributed calibration that enables the model to simulate hydrologic processes in the Congo Basin, including the slowing effect of the basin’s central wetlands, the Cuvette Centrale. Model simulations included scenarios of 25% to 100% conversion of the Basins forest cover to agricultural mosaic and compared simulated flows to those of the current baseline conditions. The dissertation also reports on the estimated impacts of the hydrological response to LCLUC on the river’s hydropower potential. Re-calibration of TRMM improved rainfall accuracy at the gauges by 15% and correctly captured important rainfall patterns such as the ones representative of the highland climate. Model calibration of daily streamflow resulted in a model with high predictive power (Nash–Sutcliffe coefficient of efficiency of 0.70) when compared to Kinshasa gauge downstream Congo River, near its outlet. Model shows realistic seasonal and spatial patterns that can be explained by the ITCZ-driven rainfall patterns in the Congo Basin. Models of the direct effects alone of 25% to 100% forest conversion produce increases in peak flows of 7% to 8%, respectively, relative to the baseline, and decreases in low flow of 1% and 6%, for 75% and 100% forest conversion respectively, relative to the baseline. However, 25% and 50% forest conversion produce increases in low flows of 3% and 1% respectively indicating a possible sensitivity of the hydrological response to the spatial variability of forest conversion. Models of the combined direct and indirect effects of 25% to 100% conversion produce decreases in peak flows of 7% to 5% respectively and decreases in low flow of 8% to 11% respectively. Model estimates of the impacts on hydropower potential range from 11% decrease during dry season to 10% increase during rainy season, with greater impacts (year-round decrease) for increasing LCLUC models including indirect effect. The modeled loss in hydropower potential during dry season reaches -5,797 MW corresponding to the hydropower potential of countries such as Zambia or Angola and of grand projects such as the Grand Ethiopian Renaissance Dam. The dissertation has showed the adequacy of TRMM precipitation products for Congo Basin rainfall regime representation and daily flow estimation particularly in capturing the timing and the seasonality of the flow. The results of these modeling efforts can be useful in research and decision-making contexts and validate the application of satellite-based hydrologic models driven for large, data-scarce river systems such as the Congo Basin by producing reliable baseline information. We recommend a prioritization of further data collection and more gauges installation required to enable further satellite-derived data calibration and models simulations. Likewise, the results from LCLUC analysis support the need for field campaigns to better understand sub-watersheds responses and to improve the calibration of currently used simulation models.Item Quantifying the Spatial and Temporal Variation of Land Surface Warming Using in situ and Satellite Data(2019) Rao, Yuhan; Liang, Shunlin; Geography; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)The global mean surface air temperature (SAT) has demonstrated the “unequivocal warming”. To understand the impact of the global warming, it is very important to quantify the spatial and temporal patterns of the surface air temperature change. Currently, most observational studies rely on in situ temperature measurements over the land and ocean. But the uneven and sparse nature of these temperature measurements may cause large uncertainty for the climate analysis especially at local and regional scales. With the rapid development of satellite data, it is possible to estimate spatial complete surface air temperature from satellite data using advanced statistical models. The satellite data-based estimation can serve as a better data source for local and regional climate analysis to reduce analysis uncertainty. In this dissertation, I firstly examined the uncertainty of four mainstream gridded SAT datasets over the global land area (i.e., BEST-LAND, CRU-TEM4v, NASA-GISS, NOAA-NCEI). The comprehensive assessment of these datasets concludes that different data coverage may cause remarkable differences (i.e., -0.4 ~ 0.6°C) of calculated large scale (i.e., global, hemispheric) average SAT anomaly using different datasets. Moreover, these datasets show even larger differences at regional and local scale (5°×5°). The local and regional data differences can lead to statistically significant differences on linear trends of SAT estimated using different datasets. The correlation analysis shows strong relationship between the uncertainty of estimated SAT trends and the density of in situ measurements across different regions. To reduce the uncertainty of surface air temperature data, I developed a statistical modelling framework which can estimate daily surface air temperature using remote sensing land surface temperature and radiation products. The framework uses machine learning models (i.e., rule-based Cubist regression model and multivariate adaptive regression spline) to characterize the physical difference between land surface temperature and surface air temperature by including radiation products at both surface and the top of the atmosphere. The model was firstly developed for the Tibetan Plateau using Cubist model trained with Chinese Meteorological Administration station measurements. Comprehensive evaluation show that the Cubist model can estimate the surface air temperature with nearly zero degree Celsius bias and small RMSEs between 1.6 °C ~ 2.1 °C. The estimated SAT over the entire Plateau for 2000-2015 show that the warming of the western part of the Plateau has been more prominent than the rest of the region. This result show the potential underestimation of conventional station measurements based studies because there are no station measurements to represent the rapid warming region. The machine learning model is then extended to the northern high latitudes with necessary modification to account for the regional difference of the diurnal temperature cycle as well as the large data volume of the northern high latitudes. The MARS model trained using data over the northern high latitudes from the Global Historical Climatology Network daily data archive show a reasonable model performance with the bias of around -0.2 °C and the RMSE ranging between 2.1 – 2.6 °C. Further evaluation shows that the model performs worse over permanent snow and ice surface due to the insufficient training data to represent this specific surface conditions. Overall, this research demonstrated that leveraging advanced statistical methods and satellite products can help generating high quality surface air temperature data which can provide much needed spatial details to reduce the uncertainty of local and regional climate analysis. The model developed in this research is generic and can be further extended to other regions with proper modification and training using high quality local data.Item Outgoing Longwave Radiation at the Top of Atmosphere: Algorithm Development, Comprehensive Evaluation, and Case Studies(2019) Zhou, Yuan; Liang, Shunlin; Geography; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Outgoing longwave radiation (OLR) at the top of the atmosphere (TOA) represents the total outgoing radiative flux emitted from the Earth’s surface and atmosphere in the thermal-infrared wavelength range. It plays a role as a powerful diagnostic of Earth’s climate system response to absorbed incoming solar radiation (ASR). Long-term measurements of OLR are essential for quantitatively understanding the climate system and its variability. However, inconsistencies and uncertainties have been always existing in OLR estimation among different datasets and algorithms. The objective of this dissertation is to carry out a comprehensive investigation on OLR with three specific questions: 1) How large are the discrepancies in estimates from various OLR products and what are their spatial and temporal patterns? 2) How to generate more accurate and more useful OLR estimates from multi-spectral satellite observations? 3) How does OLR respond to extreme climate and geological events such as El Niño/Southern Oscillation (ENSO) and giant earthquakes, and does the newly developed OLR products have any advantage to predict such events? To address those questions, this dissertation 1) conducts comprehensive evaluations on multiple OLR datasets by performing inter-comparisons among different satellite retrieved OLR products and different reanalysis OLR datasets, respectively; 2) develops an algorithm framework for estimating OLR from multi-spectral satellite observations based on radiative transfer simulations and statistical approaches; 3) investigates the correlation between OLR anomalies and historical ENSO events and a typical giant earthquake, and makes an attempt to predict ENSO and earthquake through OLR variations. Results indicate that 1) obvious discrepancies exist among different OLR datasets, with the two Japanese Meteorological Agency’s (JMA) Japanese Reanalysis project (JRA) OLRs displays the largest differences with others. However, all OLR products and datasets have comparable magnitude of inter-annual variability and monthly/seasonally anomaly, resulting in similar capability to capture the tropical expansion and ENSO events; 2) the developed OLR algorithm framework can generate reliable OLR estimates from multi-spectral remotely sensed data including Moderate Resolution Imaging Spectroradiometer (MODIS) and Advanced Very High Resolution Radiometer (AVHRR); 3) OLR has a potential to predict ENSO events through traditional statistical approach and machine learning methods, and it has slight advantage over the sea-surface-temperature (SST) as a metric for this purpose. The developed high resolution AVHRR OLR performs better than High-Resolution Infrared Radiation Sounder (HIRS) and NOAA interpolated AVHRR OLR in predicting ENSO. In addition, the singularities in OLR spatial anomalies around the giant earthquake epicenter starting three days prior to the earthquake days also suggests the OLR as an effective precursor of such an event, and the developed AVHRR OLR showed much stronger sensitivity to the coming earthquake than the existing NOAA interpolated AVHRR OLR, suggesting that the former one as a better indicator for the earthquake prediction. In this dissertation, the in-depth inter-comparisons among various OLR datasets will contribute as a reference for peers in the climate community who use OLR as one of inputs in their climate models or other diagnostic purpose. The developed OLR algorithm framework could be utilized to estimate OLR from future multi-spectral satellite data. This study also demonstrates that OLR is a promising indicator to predict ENSO and testifies that it is a precursor of giant earthquakes, which has implications for decision making aimed at alleviating the impacts on life and property from these extreme climate variations through some preventive measures such as releasing weather alert and conducting evacuations.Item CHARACTERIZING RICE RESIDUE BURNING AND ASSOCIATED EMISSIONS IN VIETNAM USING A REMOTE SENSING AND FIELD-BASED APPROACH(2018) Lasko, Kristofer; Justice, Christopher O; Geography; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Agricultural residue burning, practiced in croplands throughout the world, adversely impacts public health and regional air quality. Monitoring and quantifying agricultural residue burning with remote sensing alone is difficult due to lack of field data, hazy conditions obstructing satellite remote sensing imagery, small field sizes, and active field management. This dissertation highlights the uncertainties, discrepancies, and underestimation of agricultural residue burning emissions in a small-holder agriculturalist region, while also developing methods for improved bottom-up quantification of residue burning and associated emissions impacts, by employing a field and remote sensing-based approach. The underestimation in biomass burning emissions from rice residue, the fibrous plant material left in the field after harvest and subjected to burning, represents the starting point for this research, which is conducted in a small-holder agricultural landscape of Vietnam. This dissertation quantifies improved bottom-up air pollution emissions estimates through refinements to each component of the fine-particulate matter emissions equation, including the use of synthetic aperture radar timeseries to explore rice land area variation between different datasets and for date of burn estimates, development of a new field method to estimate both rice straw and stubble biomass, and also improvements to emissions quantification through the use of burning practice specific emission factors and combustion factors. Moreover, the relative contribution of residue burning emissions to combustion sources was quantified, demonstrating emissions are higher than previously estimated, increasing the importance for mitigation. The dissertation further explored air pollution impacts from rice residue burning in Hanoi, Vietnam through trajectory modelling and synoptic meteorology patterns, as well as timeseries of satellite air pollution and reanalysis datasets. The results highlight the inherent difficulty to capture air pollution impacts in the region, especially attributed to cloud cover obstructing optical satellite observations of episodic biomass burning. Overall, this dissertation found that a prominent satellite-based emissions dataset vastly underestimates emissions from rice residue burning. Recommendations for future work highlight the importance for these datasets to account for crop and burning practice specific emission factors for improved emissions estimates, which are useful to more accurately highlight the importance of reducing emissions from residue burning to alleviate air quality issues.Item TOWARDS FINE SCALE CHARACTERIZATION OF GLOBAL URBAN EXTENT, CHANGE AND STRUCTURE(2017) Wang, Panshi; Huang, Chengquan; Liang, Shunlin; Geography; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Urbanization is a global phenomenon with far-reaching environmental impacts. Monitoring, understanding, and modeling its trends and impacts require accurate, spatially detailed and updatable information on urban extent, change, and structure. In this dissertation, new methods have been developed to map urban extent, sub-pixel impervious surface change (ISC), and vertical structure at national to global scales. First, an innovative multi-level object-based texture classification approach was adopted to overcome spectral confusion between urban and nonurban land cover types. It was designed to be robust and computationally affordable. This method was applied to the 2010 Global Land Survey Landsat data archive to produce a global urban extent map. An initial assessment of this product yielded over 90% overall accuracy and good agreement with other global urban products for the European continent. Second, for sub-pixel ISC mapping, the uncertainty caused by seasonal and phenological variations is one of the greatest challenges. To solve this issue, I developed an iterative training and prediction (ITP) approach and used it to map the ISC of entire India between 2000 and 2010. At 95% confidence, the total ISC for India between 2000 and 2010 was estimated to be 2274.62±7.84 km2. Finally, using an object-based feature extraction approach and the synergy of Landsat and freely available elevation datasets, I produced 30m building height and volume maps for England, which for the first time characterized urban vertical structure at the scale of a country. Overall, the height RMSE was only ±1.61 m for average building height at 30m resolution. And the building volume RMSE was ±1142.3 m3. In summary, based on innovative data processing and information extraction methods, this dissertation seeks to fill in the knowledge gaps in urban science by advancing the fine scale characterization of global urban extent, change, and structure. The methods developed in this dissertation have great potentials for automated monitoring of global urbanization and have broad implications for assessing the environmental impact, disaster vulnerability, and long-term sustainability of urbanization.Item Towards Better Understanding of Agricultural Drought: A Comprehensive Analysis of Agricultural Drought Risk, Impact and Monitoring from Earth Observations(2017) Zhang, Jie; Justice, Christopher; Geography; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)With a changing climate, the frequent occurrence of drought has resulted in unprecedented grain prices and severe market instability, threatening global food security. Earth observation, especially satellite-based observation, has proven its potential for near real-time drought monitoring and early warning. This dissertation undertakes a comprehensive analysis of agriculture-oriented drought risk, impact and monitoring using time-series satellite observation combined with ancillary earth observation data, thus providing a better understanding of agricultural drought. Agricultural lands exhibit more severe drought regimes during the agricultural growing season. At the global scale, the U.S. Corn Belt, Spain & Eastern Europe, Central Russia, India, North China and Australia, are shown to be the hotspots of agricultural drought risk. For the last three decades, different agricultural drought risk change patterns are found in different regions with a relatively stable but slight declining drought risk overall for the globe, while Australia exhibits a continuous increase and Brazil exhibits a continuing decrease. Land Surface Temperature (LST) and Evapotranspiration (ET) based indicators show similar capabilities for drought monitoring and have an immediate response after drought; while for Normalized Difference Vegetation Index (NDVI) derived indicators, there shows a lagged and inconsistent drought response. The relationships between NDVI- and LST- derived drought indicators are variable, exhibiting changing functions in both spatial and temporal domains, which provides basis for effectively integrating different data sources for developing a synthetic index. Drought results in varying impacts during the growing season, with generally increasing impacts during the winter wheat main growing season and the most severe drought effects during the grain filling stage around vegetative peaks. As for the Drought Severity Index, better performance is found in rainfed-dominated than irrigation-dominated regions. This dissertation calls for continuing work to develop an improved impact-oriented agricultural drought indicator by integrating the contributions of different data sources, the dynamics of NDVI-LST interactions as well as the varying drought impacts during the growing season. Improved agricultural drought monitoring and impact assessment, together with agricultural risk analysis, can help prototype an enhanced and integrated agricultural drought monitoring system, thus offering reliable and timely information for drought mitigation, preparedness, response and recovery.Item HUMAN-INDUCED VEGETATION DEGRADATION IN A SEMI-ARID RANGELAND(2017) Jackson, Hasan; Prince, Stephen D; Geography; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Current assessments of anthropogenic land degradation and its impact on vegetation at regional scales are prone to large uncertainties due to the lack of an objective, transferable, spatially and temporally explicit measure of land degradation. These uncertainties have resulted in contradictory estimates of degradation extent and severity and the role of human activities. The uncertainties limit the ability to assess the effects on the biophysical environment and effectiveness of past, current, and future policies of land use. The overall objective of the dissertation is to assess degradation in a semi-arid region at a regional scale where the process of anthropogenic land degradation is evident. Net primary productivity (NPP) is used as the primary indicator to measure degradation. It is hypothesized that land degradation resulting from human factors on the landscape irreversibly reduces NPP below the potential set by environmental conditions. It is also hypothesized that resulting reductions in NPP are distinguishable from natural, spatial and temporal, variability in NPP. The specific goals of the dissertation are to (1) identify the extent and severity of degradation using productivity as the primary surrogate, (2) compare the degradation of productivity to other known mechanisms of degradation, and (3) relate the expression of degradation to components of vegetation and varying environmental conditions. This dissertation employed the Local NPP Scaling (LNS) approach to identify patterns of anthropogenic degradation of NPP in the Burdekin Dry Tropics (BDT) region of Queensland (14 million hectares), Australia from 2000 to 2013. The method started with land classification based on the environmental factors presumed to control NPP to group pixels having similar potential NPP. Then, satellite remotely sensing data were used to compare actual NPP with its potential. The difference, in units of mass of carbon fixed in NPP per unit area per monitoring interval and per year, also its percentage of the potential, were the measures of degradation. Degradation was then compared to non-green components of vegetation (e.g. wood, stems, leaf litter, dead biomass) to determine their relationship in space and time. Finally, the symptoms of degradation were compared to land management patterns and the environmental variability (e.g. drought, non-drought conditions). Nearly 20% of the region was identified as degraded and another 7% had significant negative trends. The average annual reduction in NPP due to anthropogenic degradation was -17% of the non-degraded potential, although the severity of degradation varied substantially throughout the region. Non-green vegetation cover was strongly correlated with the inter-annual and intra-annual temporal trends of degradation. The dynamics of degradation in drought and non-drought years provided evidence of multiple stables states of degradation.