MODIS: An Alternative for Updating Land Use and Land Cover in Large River Basin

Abstract

Land use and land cover information is essential for effective land resource management such as water resource. However, mapping of land use and land cover at regional or large river basin scale based on high resolution satellite image is usually expensive and time consuming. This research attempts to map out the status of land use and land cover in 2010 based on existing land use and land cover in 2003 that was generated from Landsat. The mapping was undertaken in Tonle Sap basin, located in the northwest part of Cambodia that covers an area of about 80,000 km², where a conventional maximum likelihood classification was performed upon Moderate-resolution Imaging Spectroradiometer (MODIS) Imagery. The MODIS land cover products were evaluated using existing Landsat TM based land cover maps in 2003 as a major referenced data. Results showed that the MODIS classification was more successful at mapping the forested area and less successful at mapping the shrub land and upland crop. Whereas the smaller habitats (e.g. settlement area) in many cases occupy in small community areas that are smaller than the MODIS pixels were not able to identify. Accuracy assessment by comparison between classified LULC 2003 based on MODIS and the existing Landsat based LULC 2003 shows that overall accuracy was more than 80% and Kappa index obtained was 0.81 which is considered as an acceptable level. This study showed the benefits of using MODIS satellite imagery that provide new options for regional land use and land cover mapping.

Keywords: MODIS, Land Use and Land Cover, Remote Sensing, Image Processing