Machine learning approach to detect Land Use Land Cover (LULC) change in Chure region of Sarlahi district, Nepal

Abstract

Land Use and Land Cover (LULC) changes are major global environmental issues, affecting ecological systems, climate change, and biodiversity. The Chure region of the Sarlahi district in Nepal is a critical ecological zone that has experienced significant LULC changes in recent years. In this study, our aim was to apply a machine learning approach to detect LULC changes in the Chure region using Google Earth Engine (GEE) and the Random Forest classifier. We utilized Landsat imagery of 2007 and 2022 to generate land cover maps for each year, which were then compared to identify changes over time. The major findings of this study indicate that the forest cover in the region has increased by approximately 16% over the past 15 years, while the agriculture and built-up areas have also shown a significant increase. Conversely, the barren land and water areas have decreased. The classifier obtained an overall accuracy of 85.7% and a kappa coefficient of 81.2% for the year 2022, and an overall accuracy of 82.2% and a kappa coefficient of 76.8% for the year 2007, which demonstrates the high accuracy of the proposed approach. The use of GEE and random forest classifiers provided a cost-effective and efficient method for analysing large datasets and producing accurate LULC maps. Our findings can inform policymakers and conservationists about the need for sustainable land management practices to preserve the ecological integrity of the Chure region. The approach can be applied to other regions to monitor and manage LULC changes and support effective conservation efforts.