University of Twente Student Theses


A Quickbird’s-eye view on marmots

Velasco, Marcela (2009) A Quickbird’s-eye view on marmots.

[img] PDF
Abstract:The population of Marmota siberica in Mongolia decreased by 75% in the 1990s . In respond, in 2006 the Mongolian government banned all marmot hunting, but the response of the population this ban is not known. Because demographic research is time-consuming, there is a need for alternative methods, such as remote sensing, to monitor the marmot population. The aim of this study was to determine the applicability of Quickbird imagery to map the mounds of marmot burrows in Hustai National Park, Mongolia, and to assess whether the density of mounds is related to the levels of conservation in the different management zones of the park: a core zone dedicated to conservation, a tourist zone for tourism, and a buffer zone where people live and herd animals. In all bands of the Quickbird imagery, the radiance of active mounds differed from that of non-active mounds and vegetation. In addition mound size and vegetation cover, as recorded in the field, differs between active and non-active mounds. Based on these results, an object-oriented classification rule was built to detect active mounds in the Quickbird images. The resulting mound distribution map had a user accuracy of 69% and a producer accuracy of 87%. The density of active mounds was 279 per km 2 in the core zone, 212 per km 2 in the tourist and 62 per km2 in the buffer zone, respectively. A logistic regression was used to find out if differences in mound density across the study area were only related to environmental variables (topographical elevation, slope, and southern exposure, and vegetation type) or whether, in addition, they were related to conservation practices. Although in general, the model performed poorly as evaluated by area under the ROC, it performed better after the management zones were added as an additional predictor. This study shows for the first time the potential of object-oriented image analysis for marmot population monitoring, through the detection of active marmot mounds. However, it produces a high number of false positive in desertified areas, dry valleys and on unpaved old roads and tracks, because these objects have a reflectance similar to that of active mounds. The logistic regression model suggests that different management practices in the Hustai National Park affect the marmot population.
Item Type:Essay (Master)
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Programme:Geoinformation Science and Earth Observation MSc (75014)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page