Remote sensing and machine learning techniques to monitor fluvial corridor evolution: The Aras River between Iran and Azerbaijan

Abstract Nowadays, remote sensing and machine learning techniques provide an unprecedented potential for the monitoring of fluvial corridors. To exemplify their usefulness, the evolution of an international river was assessed to detect spatio-temporal changes and human artificialization of the international border between Iran and Azerbaijan along the period 1984–2020. by supervised classification and geomorphological indexes calculation. Results demonstrated that the active channel has been narrowed dramatically (− 88.9%) with a narrowing ratio of 16.6 m/year. The RNCI shows that deposition was the dominant process and the channel moved profusely over the fluvial area. Channel displacement shifted toward the non-Iranian part (~ 133 m on average). Vegetation cover, water body, and No-farming cover have been reduced more than a half. By combining different approaches, we note the progressive impoverishment of the land transitions with artificialization being the most frequent change.