Locally adaptive super-resolution waterline mapping with MODIS imagery

Abstract

ABSTRACTMapping inland water bodies is important for relevant research fields and water resource management. Satellite remote sensing is a routine approach, and various remotely sensed images have been applied to map waterlines. MODerate-resolution Imaging Spectroradiometer (MODIS) images have advantages for waterline mapping in large areas, thanks to its wide scan width and high frequent revisiting period; however, the spatial resolution of MODIS images is too coarse to map waterlines accurately. In this article, a super-resolution mapping (SRM) model was proposed for fine spatial resolution waterline mapping with MODIS images. The proposed SRM model was directly applied in the MODIS band 2 images, which have a fine spatial resolution and a high spectral separability between water and land. In order to further take account of the spatial heterogeneity of endmembers, the reflectance values of water and land were locally calculated for each coarse resolution pixel. The proposed SRM model was assessed in two study areas located in the Tibetan Plateau, China and Wisconsin, United States, including water bodies with different areas and boundary shapes. The results showed that the accuracy values of the proposed SRM model using local endmembers were higher than those of hard classification and the SRM model using average endmembers, showing the effectiveness of the proposed model in fine spatial resolution waterline mapping with MODIS imagery.