Geography Theses and Dissertations

Permanent URI for this collectionhttp://hdl.handle.net/1903/2773

Browse

End date: 2009Subject: search.filters.subject.Biology, Ecology

Search Results

Now showing 1 - 6 of 6
  • Thumbnail Image
    Item
    Changes in Amazon Forest Structure from Land-Use Fires: Integrating Satellite Remote Sensing and Ecosystem Modeling
    (2008-11-17) Morton, Douglas; DeFries, Ruth S; Geography; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
    Fire is the dominant method of deforestation and agricultural maintenance in Amazonia, and these land-use fires frequently escape their intended boundaries and burn into adjacent forests. Initial understory fires may increase forest flammability, thereby creating a positive fire feedback and the potential for long-term changes in Amazon forest structure. The four studies in this dissertation describe the development and integration of satellite remote sensing and ecosystem modeling approaches to characterize land-use fires and their consequences in southern Amazon forests. The dissertation contributes three new methods: use of the local frequency of satellite-based active fire detections to distinguish between deforestation and maintenance fires, use of satellite data time series to identify canopy damage from understory fires, and development of a height-structured fire sub-model in Ecosystem Demography, an advanced ecosystem model, to evaluate the impacts of a positive fire feedback on forest structure and composition. Conclusions from the dissertation demonstrate that the expansion of mechanized agricultural production in southern Amazonia increased the frequency and duration of fire use compared to less intensive methods of deforestation for pasture. Based on this increase in the frequency of land-use fires, fire emissions from current deforestation may be higher than estimated for previous decades. Canopy damage from understory fires was widespread in both dry and wet years, suggesting that drought conditions may not be necessary to burn extensive areas of southern Amazon forests. Understory fires were five times more common in previously-burned than unburned forest, providing satellite-based evidence for a positive fire feedback in southern Amazonia. The impact of this positive fire feedback on forest structure and composition was assessed using the Ecosystem Demography model. Scenarios of continued understory fires under current climate conditions show the potential to trap forests in a fire-prone structure dominated by early-successional trees, similar to secondary forests, reducing net carbon storage by 20-46% within 100 years. In summary, satellite and model-based results from the dissertation demonstrate that fire-damaged forests are an extensive and long-term component of the frontier landscape in southern Amazonia and suggest that a positive fire feedback could maintain long-term changes in forest structure and composition in the region.
  • Thumbnail Image
    Item
    ENDANGERED DRY DECIDUOUS FORESTS OF UPPER MYANMAR (BURMA): A MULTI-SCALE APPROACH FOR RESEARCH AND CONSERVATION
    (2006-09-11) Songer, Melissa A.; DeFries, Ruth S.; Geography; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
    Tropical dry forests are critically endangered and largely unprotected ecosystem. I used a multi-scale research approach to study Upper Myanmar's dry deciduous forests. At the broad scale I assessed how well existing land cover data can be used to map and monitor dry forests, and estimated the extent, distribution, and level of protection of these forests. At the landscape level I assessed spatial and temporal dynamics of deforestation in and around a dry forest protected area, Chatthin Wildlife Sanctuary (CWS), investigated land use pressures driving these changes, and evaluated effectiveness of protection efforts within the sanctuary. At the local scale I studied the degree to which people rely on dry forests for subsistence and the socioeconomic variables correlated with dependence on forest products. Using MODerate Resolution Imaging Spectroradiometer (MODIS) Normalized Difference Vegetation Index (NDVI) data to delineate remaining dry deciduous forests, I found that only 24,000 km2 of this forest type remain in Upper Myanmar--only 4% inside protected areas. At 81% accuracy, this map scored higher than existing global and regional land cover classifications for predicting dry forest. Employing satellite images covering the landscape in and around CWS (Landsat MSS, TM, ETM+ and ASTER) between the years 1973-2005 , I found that 62% of forest was lost (1.93% annual rate) primarily from agricultural conversion and hydroelectric development. Sanctuary protection has been effective in slowing decline: loss rates inside CWS were 0.49% annually (16% total). However, forest inside the sanctuary is still declining at a rate above the global average and shows evidence of impact from forest product extraction around the boundaries. Based on interviews with 784 people living in 28 subsistence-based agricultural communities located in and around CWS, I found virtually all survey respondents depended on CWS for food, medicine, housing materials, and, above all, fuelwood. Poverty and socioeconomic limitations drive extractive activities. While CWS has been effective in slowing deforestation rates, alternative use strategies that benefit people will improve prospects for long-term conservation in the area. My results demonstrate that a multi-scaled research approach is essential for understanding the drivers impacting the rapidly-declining dry forests of Upper Myanmar.
  • Thumbnail Image
    Item
    Correlates of Terrestrial Vertebrate Species Richness: an Evaluation of Environmental Hypotheses over the Western Continental USA
    (2006-04-24) Slayback, Daniel Andrew; Prince, Stephen D; Geography; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
    An explanation for the unequal distribution of life forms across the Earth's surface has been a persistent and problematic question in modern ecology ever since these patterns were first noted, over 100 years ago. Most empirical research supports one of three environmental hypotheses to explain these patterns: environmental energy (ambient environmental energy or ecosystem productivity); climatic variability; or habitat heterogeneity. This research examines these hypotheses using better datasets than those commonly considered, and using a consistent methodology that addresses often neglected statistical and analytic details. The environmental datasets used in this study are derived from time series of satellite and ground station data, including the Daymet climate data, and net primary productivity data from the GLOPEM model. Species richness is derived from the individually modeled vertebrate distributions provided by the individual state Gap Analysis Projects for the western US states of California, Oregon, Washington, Idaho, Montana, Wyoming, Utah, and Colorado, which define the spatial extent of this study. The study methodology relies upon the summary of results from many model variants for each hypothesis. These variants are constructed by creating regression models at each of four different spatial scales (8, 16, 32, and 64 km grid cells), for each class of vertebrates (amphibians, birds, mammals, reptiles, and all), and over each of the eight states considered. Preliminary studies found that ordinary least squares would be a sufficient model form, although conditional autoregressive models were extensively considered. Other preliminary work examined issues of spatial autocorrelation and variable selection. The results indicate that the energy/productivity hypothesis consistently outperforms all other hypotheses in explaining species richness, across almost all spatial scales, geographic regions, and vertebrate classes. The performance of the climatic variability and habitat heterogeneity hypotheses varies for particular states or vertebrate classes. Vertebrate data quality was important; results for Colorado and Washington were frequently unusual, suggesting an incompatibility between their modeled vertebrate distributions and those of other states. Models of reptile richness also often showed substantially different characteristics than those for other vertebrates. Overall the results provide additional support to the energy/productivity hypothesis, from a more comprehensive methodological basis.
  • Thumbnail Image
    Item
    MEASURING AND MAPPING FOREST WILDLIFE HABITAT CHARACTERISTICS USING LIDAR REMOTE SENSING AND MULTI-SENSOR FUSION
    (2005-12-05) Hyde, Peter; Dubayah, Ralph O.; Geography; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
    Managing forests for multiple, often competing uses is challenging; managing Sierra National Forest's fire regime and California spotted owl habitat is difficult and compounded by lack of information about habitat quality. Consistent and accurate measurements of forest structure will reduce uncertainties regarding the amount of habitat reduction or alteration that spotted owls can tolerate. Current methods of measuring spotted owl habitat are mostly field-based and emphasize the important of canopy cover. However, this is more because of convenience than because canopy cover is a definitive predictor of owl presence or fecundity. Canopy cover is consistently and accurately measured in the field using a moosehorn densitometer; comparable measurements can be made using airphoto interpretation or from examining satellite imagery, but the results are not consistent. LiDAR remote sensing can produce consistent and accurate measurements of canopy cover, as well as other aspects of forest structure (such as canopy height and biomass) that are known or thought to be at least as predictive as canopy cover. Moreover, LiDAR can be used to produce maps of forest structure rather than the point samples available from field measurements. However, LiDAR data sets are expensive and not available everywhere. Combining LiDAR with other, remote sensing data sets with less expensive, wall-to-wall coverage will result in broader scale maps of forest structure than have heretofore been possible; these maps can then be used to analyze spotted owl habitat. My work consists of three parts: comparison of LiDAR estimates of forest structure with field measurements, statistical fusion of LiDAR and other remote sensing data sets to produce broad scale maps of forest structure, and analysis of California spotted owl presence and fecundity as a function of LiDAR-derived canopy structure. I found that LiDAR was able to replicate field measurements accurately. Additionally, I was able to statistically combine LiDAR with passive optical and RaDAR (SAR backscatter and InSAR range) data to produce broad scale maps of forest structure that are consistent and accurate relative to field data and LiDAR data alone. Finally, I was able to demonstrate that these forest structural attributes predict spotted owl presence and absence as well as productivity.
  • Thumbnail Image
    Item
    Investigating Uncertainties in Trace Gas Emissions from Boreal Forest Fires Using MOPITT Measurements of Carbon Monoxide and a Global Chemical Transport Model
    (2005-08-02) Hyer, Edward Joseph; Kasischke, Eric S; Geography; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
    Boreal forest fires are a significant contributor to atmospheric composition in the high northern hemisphere, and are highly variable both spatially and temporally. This study uses a new emissions model [Kasischke et al., 2005] to generate input to the University of Maryland Chemical Transport Model [Allen et al., 1996], with the goal of examining and constraining the key uncertainties in current understanding of boreal forest fire behavior. Model outputs are compared with data from the MOPITT instrument as well as in situ measurements of CO. A case study of CO transport during the summer of 2000 is used to examine several key uncertainties in the emissions estimates, describing how current levels of uncertainty affect atmospheric composition and applying atmospheric measurements can be applied to constrain uncertainty. Source magnitudes determined by inverse methods were shown to be highly sensitive to the assumed injection properties. For the boreal forest in 2000, the best agreement with observations was obtained with a pressure-weighted profile of injection throughout the tropospheric column, but detailed examination of the results makes clear that any uniform parameterization of injection will be a significant source of error when applied globally. Comparison of simulated CO distributions from daily, weekly, and monthly aggregate emissions sources demonstrated that while model data sources produced a valid representation of emissions at weekly resolution, the atmospheric distribution outside the source region has very little sensitivity to temporal variability at scales finer than 30 days. Different estimates of burned area produced large differences in simulated patterns of atmospheric CO. The GBA-2000 global product and the data sources used by Kasischke et al. [2005] gave better agreement with atmospheric observations compared to the GLOBSCAR product. Comparison of different estimates of fuel consumption indicated that atmospheric measurements of CO have limited sensitivity to spatial variability in fuels, but that current fuels maps can improve agreement with atmospheric measurements. These results provide a clear indication of how atmospheric measurements can be used to test hypotheses generated by emissions models.
  • Thumbnail Image
    Item
    Improving the Estimation of Leaf Area Index from Multispectral Remotely Sensed Data
    (2003-10-27) Fang, Hongliang; Liang, Shunlin; Prince, Stephen D.; Townshend, John R.; Weismiller, Richard; Geography
    Leaf Area Index (LAI) is an important structural property of surface vegetation. Many algorithms use LAI in regional and global biogeochemical, ecological, and meteorological applications. This dissertation reports several new, improved methods to estimate LAI from remotely sensed data. To improve LAI estimation, a new atmospheric correction algorithm was developed for the Enhanced Thematic Mapper Plus (ETM+) imagery. It can effectively estimate the spatial distribution of atmospheric aerosols and retrieve surface reflectance under general atmospheric and surface conditions. This method was validated using ground measurements at Beltsville, Maryland. Several examples are given to correct AVIRIS (Airborne Visible/Infrared Imaging Spectrometer), MODIS (Moderate Resolution Imaging Spectroradiometer) and SeaWiFS (Sea-viewing Wide Field-of-view Sensor) data using the new algorithm. Next, a genetic algorithm (GA) was incorporated into the optimization process of radiative transfer (RT) model inversion for LAI retrieval. Different ETM+ band combinations and the number of "genes" employed in the GA were examined to evaluate their effectiveness. The LAI estimates from ETM+ using this method were reasonably accurate when compared with field measured LAI. A new hybrid method, which integrates both the RT model simulation and the non-parametric statistical methods, was developed to estimate LAI. Two non-parametric methods were applied, the neural network ((NN) algorithms and the projection pursuit regression (PPR) algorithms. A soil reflectance index (SRI) was proposed to account for variable soil background reflectances. Both atmospherically corrected surface reflectances and raw top-of-atmosphere (TOA) radiances from ETM+ were tested. It was found that the best way to estimate LAI was to use the red and near infrared band combination of surface reflectance. In an application of this hybrid method to MODIS, the PPR and NN methods were compared. MODIS LAI standard products (MOD15) were found to have larger values than my results in the study area.