University of Twente Student Theses

Login

Effect of spatial resolution on erosion assessment in Namchun watershed, Thailand

Polous, Khatereh (2010) Effect of spatial resolution on erosion assessment in Namchun watershed, Thailand.

[img] PDF
2MB
Abstract:Water erosion is one of the most important environmental problems in Thailand. To control water erosion, assessing soil erosion at regional scale is important. Satellite imagery data and Digital Elevation Models are being used increasingly to assess erosion at different scales, but the main restrictions for these assessments are availability and quality of data. Therefore, it is necessary to analyse the reliability of erosion assessments at different spatial resolutions. To analyse the effect of spatial resolution of satellite images on erosion assessment independent of other sensor characteristics, several images with different spatial resolutions were simulated. Effects of spatial resolution was simulated by aggregating an ASTER image dataset (3 VNIR bands in 15m and 6 SWIR bands in 30m resolution) to three coarser spatial resolutions (30m, 90m, and 250m) through averaging function. In addition to analyse the effect of DEM resolution on erosion assessment, several DEMs in different resolutions (5m, 15m, 30m, 90m, and 250m) were created from digitized contour line map. In order to isolate the effect of spatial resolution, all simulated satellite images and DEMs in different resolutions were disaggregated to the finest resolution (5m). In this study, the RMMF model was applied to assess the effect of resolution on erosion. As spatial resolution of satellite images became coarser, both overall accuracy and kappa coefficient of final erosion maps were significantly decreased, since the erosion classes were progressively converted to each other. Most conversions appeared between the classes that had completely different erosion rates. However, the general trend of conversions was from classes with lower erosion rate to the classes with higher erosion rate. The main reason behind these changes was the conversion of land use/cover classes with different erosion rates to each other. Firstly, as about 36% of the whole watershed was covered by agriculture as dominant class, the averaging caused a very high conversion of the other classes to agriculture in coarser resolutions. Secondly, by decreasing the spatial resolution, spectral details were progressively combined and the variation within satellite image reduced. Subsequently, classification results in coarser resolutions showed that bare and forest (with minimum and maximum spectral values in the near-IR band) disappeared and converted to the other classes. By decreasing DEM resolution, both overall accuracy and kappa coefficient of final erosion maps were moderately decreased, since the erosion classes were converted to each other. Most conversions occurred between the classes that had almost the same erosion rates with general trend from classes with higher erosion rate to the classes with lower erosion rate, which proved the underestimation of erosion in coarser resolutions. Indeed, by changing DEM resolution, the slopes and the distribution of slopes have changed within watershed; the average slope, standard deviation, and maximum slope values reduced as DEM resolution became coarser, in other words the topographic features of the watershed were smoothed; which in turn affected the soil erosion prediction. - ii - The results of this study showed, although by using satellite images in coarser resolutions soil erosion was slightly overestimated, but the RMMF model is not sensitive to the land use/cover factor; the spatial pattern of erosion maps in coarser resolutions even in 250m resolution approximately coincided with 5m resolution. Therefore, for stakeholders who want to assess erosion at regional scale, MODIS images at 250m resolution still can give reasonable results. Likewise, although more than 60% of the study area had steep slope (>20°), but the spatial pattern of erosion maps in coarser resolutions even in 90m resolution conformed very well to the 5m resolution; thus SRTM could be an appropriate choice to assess erosion at regional scale with acceptable results. This implies that it may not be necessary to use costly, fine resolution remote sensing and DEMs data for the application of the erosion models at regional scale. Keywords: Satellite imagery data resolution; DEMs resolution; SAM classification, RMMF model
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:https://purl.utwente.nl/essays/90724
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page