Identifying morphological hotspots in large rivers by optimizing image enhancement

Limited access to observe in-situ sediment changes requires viable means for quantifying sediment transport in large rivers for effective management of changes in river channels. This study developed a remote sensing-based framework to identify erosion hotspots by magnifying sediment concentration from Sentinel-2 and Landsat-8/9 multispectral images of the Brahmaputra River and the Indus River. First, uncorrelated independent bands were produced to boost the spectral information using the Principal Component Analysis (PCA). The optimal band composite was then identified by applying the Optimum Index Factor (OIF) on the Principal Components (PCs). This approach determined a 3-PCs composite having the highest variance with the least correlation to highlight active morphological changes during flood times. The results of the study reaffirm the significance of the minor PCs (PC4, PC5 and PC6) to characterize the small variation in the data, whereas the main PCs depict the majority of the brightness values around means. The approach was applied to Sentinel-2 imagery acquired on September 2018 in the Brahmaputra River, and Landsat-8/9 images of 2015 and 2022 in the Indus River during flood time to enhance and identify active riverbank erosion hotspots. Precise and timely monitoring of erosion-prone areas can support the control of riverbank erosion and improve soil conservation practices.