Author(s): Maukthik Srivathsan Jeganathan; Ravindra Vitthal Kale; Bhabagrahi Sahoo
Linked Author(s):
Keywords: LULC; Machine learning; GEE; Temporal Analysis; Chord diagram
Abstract: The dynamics of land use and land cover (LULC) change offer crucial insights of the changing hydrological cycle causing a disturbance in the water supply and demand chain. A Python-based machine learning tool for LULC mapping through an user interface developed by Chen et al., (2023) lacks the option to visualize the temporal variation of land cover. Based on Google Earth Engine (GEE), it utilizes Random Forest algorithm for classifying the satellite imageries. This study aims to enhance this tool by incorporating temporal analysis and change detection through a chord diagram, thereby enabling policymakers and academicians to easily analyze land cover changes. The modified user interface tool was validated with the Kangsabati river basin present in India, which showed classification accuracy of 0.74 but with an erroneous chord diagram. The chord diagram highlights the misclassification, which would have gone unnoticed otherwise. Post-classification approaches should be applied further to fine-tune this tool.
Year: 2024