Optimal land Use/Cover Classification Using Remote Sensing Imagery for Hydrological Modeling in a Himalayan Watershed
Saran, S.; Sterk, G.
Erosion and Soil & Water Conservation Group, Wageningen University

Land use/cover is an important watershed surface characteristic that affects surface runoff and erosion. Many of the hydrological models divide the watershed into Hydrological Response Units (HRU), which are spatial units with expected similar hydrological behaviour. The division into HRUís requires good-quality data on land use/cover and its spatial pattern. This paper presents a two-step approach to attain an optimal land use/cover map based on remote sensing imagery for a Himalayan watershed in northern India. The first step in the approach was a digital classification using a decision tree classifier to extract knowledge from the spatial data in the form of classification rules. The results obtained from the decision tree were better compared to a maximum likelihood classification. The obtained land use/cover map was not sufficient for the delineation of HRUs. Moreover, for accurate hydrological modelling, it was necessary to consider the hydrological responses for different crop cover within agricultural class during the monsoon season. Therefore we adopted a visual classification approach using optical data fused with ENVISAT ASAR data with HH polarisation and wide incidence angle (I6). This second step with level II classification system resulted into better classification accuracy within the Ďagricultural landí class which was further combined with topography and soil type to derive HRUís for the physically-based hydrological modelling.


Legal Disclaimer - Glossary - FAQ - Site Map - Contact us Last updated: 09-Feb-2010