Abstract
The present study introduces a method with modifying the gradient-based edge detection method for mapping sandbars automatically using satellite-based information. The target area is a part of the mega-sized Jamuna River in which several different sizes of sandbars exist and often change positions. The authors employ threshold and gradient based methods to detect sandbars. The threshold based method (TBM) detects sandbars by computing the Modified Land Soil Water Index (MLSWI) using MODIS 8day composite data. The proposed modified gradient based method (MGBM) produces better results without any threshold using MODIS daily surface reflectance. To verify the results of the proposed method, the authors conducted field surveys traveling along sandbars with GPS and compare with the Landsat images. The results suggest the sandbars' sizes and shapes identified by the proposed method coincide closely with the field investigations and Landsat images.
