Flood Inundation Mapping of a River Stretch Using Machine Learning Algorithms in the Google Earth Engine Environment

Abstract

Floods are among the most common natural disasters in India, causing significant socio-economic and environmental impacts. This study focuses on a frequently flooded stretch of the Godavari River in Telangana, India, to analyze the flood event that occurred between 14th July 2022 and 20th July 2022. Sentinel-1 SAR data from 6th July 2022 to 20th July 2022 were used to perform flood inundation mapping. Various machine learning algorithms, including Support Vector Machine (SVM), Random Forest (RF), Gradient Boosting Tree (GBT), and Classification and Regression Tree (CART), were employed. The analysis revealed that out of the total study area of 1,556,544 ha, SVM classified 59,823 ha, RF classified 60,088 ha, GBT classified 57,497 ha, and CART classified 58,374 ha as flooded areas. In contrast, Otsu's Thresholding technique identified a significantly larger flooded area of 359,253 ha. For validation, 70 flooded and 30 non-flooded points were randomly selected from the flood map provided by the National Remote Sensing Center (NRSC). The RF algorithm achieved the best performance, correctly classifying 58 flooded points and 26 non-flooded points, resulting in an overall accuracy of 84%. The findings highlight the effectiveness of machine learning algorithms, particularly Random Forest, in flood inundation mapping.