Development of QGIS plugin for flood inundation mapping: applying Otsu’s thresholding technique

Flooding is one of the most significant and frequent phenomena in India, with Bihar and Assam being among the most severely affected states. The consequences of flooding extend to the loss of fertile soil, agricultural productivity, infrastructure, and, more significantly, the life of living beings. Although flood inundation can be seen in aerial imagery, obtaining cloud-free datasets during flood events is often challenging, complicating the mapping process. In response, Synthetic Aperture Radar (SAR) datasets emerge as important tools, given their effectiveness in mapping flood inundation, especially during the monsoon season when clouds prevent flood mapping using aerial imagery. However. previous attempts in flood inundation mapping using SAR data have highlighted the difficulties associated with manual thresholding, demanding significant expertise, rigorous sampling, and detailed field knowledge. In this study, a QGIS plugin based on Otsu’s thresholding was developed to streamline flood inundation mapping across diverse geographical conditions. The results obtained using the developed plugin was compared with flood inundation maps derived from aerial imagery. The discrepancy in flooded area between SAR and aerial imagery ranged from 0.74 to 5.13%, demonstrating a high level of agreement. The accuracy of flood mapping using SAR data was consistently high, exceeding 86% across most datasets, with highest recorded as 96.82% for an image dated October 2016 in Darbhanga district. The highest overall accuracy of Barpeta district was observed as 89.50% and overall accuracy of DSNP was observed as 94.26%. This study underscores the efficacy of the developed plugin in delivering accurate flood inundation maps. The tool offers a valuable resource for decision-makers, enabling rapid flood assessments and supporting the development of effective flood management strategies in vulnerable regions.