GIS-Based Surface Water Changing Analysis in Rajshahi City Corporation Area Using Ensemble Classifier

Abstract

Water is one of the largely used elements of this nature. Water is available in different forms like surface water and groundwater. The amount of surface water in the study area is not constant because the study area is situated to the north of Bangladesh. Some years this region faces flood and some year faces heavy droughts. Many ponds, canals and a large portion of the river are filled by dumping wastages and for construction purposes. Many industries and factories discharge wastewater containing harmful substances directly into the aquatic environment and there is no preliminary purification carried out. In this way, the water resources are becoming unsuitable for subsequent use, especially for drinking water supply. It makes the amount of fresh water scarcer. Heavy rainfall causes flood and river erosion changes the direction of river flow. This change is found by the Geological Information System (GIS). The Landsat images are collected from the United States Geological Survey (USGS). In this research, the Landsat 4–5 Thematic Mapper and Landsat 8 Operational Land Imager are used. ArcGIS is used to extract the features from the images. Finally, ensemble classification, i.e., Random forest algorithm is used to make the prediction. The accuracy measured about 92%. Also, in this study, precision, recall, and F1 scores calculated to justify the prediction accuracy.