Development of a land cover reclassification scheme for Malwatu Oya basin in Sri Lanka

Abstract

Land cover (LC) data is a valuable resource in a wide array of remote sensing applications to monitor dynamic changes on Earth's surface. A series of accurate land cover maps play a crucial role in capturing extreme hazard events. There are multiple remotely sensed LC products available today, yet the absence of a common classification system hinders their ability to use for hydrological applications. In this study, a common classification system was developed for WaPOR and MD12Q1 data sets by using LANDSAT 8 satellite images. The classification scheme consisting of five land cover classes was then employed by logical sequence to extract land cover information of the Malwatu Oya River Basin. Among several indices tested, the Normalized Difference Water Index (NDWI) was selected as a suitable index for identifying water areas, which achieved an accuracy of 85%. Additionally, a combination of the Normalized Difference Vegetation Index (NDVI) and the slope found water areas with an acceptable accuracy of over 75%. The NDVI proved to be the most effective index for capturing forest areas and cropland, with an accuracy exceeding 80%. Furthermore, the NDVI difference method was adopted to identify cropland areas. The Modified Normalized Difference Water Index (MNDWI) was identified as a moderately suitable index for delineating built-up areas. The developed land cover map provided a datum for assessing temporal changes in watershed LC and the proposed methodology can be used for integrating remote sensing technology for water resource management in other river basins.