APPLYING MACHINE LEARNING TO LAND USE TRANSITION INVENTORY USING SENTINEL-1 AND SENTINEL-2 DATA (CLASSIFYING LAND TRANSITION PURPOSES)

Abstract

This study aims to address the challenge of land use classification and monitoring changes over time to support sustainable land resource management in Thuan Hoa commune, Soc Trang province. Utilizing the Random Forest model and satellite imagery data from Sentinel-1 and Sentinel-2, accessed through the Open Data Cube platform under the EASI CSIRO Asia project, combined with ground-truth data, the study analyzes and compares classification results with the 2022 land inventory map. The methodology involves overlaying RF-generated classification maps with reference data in shapefile format to evaluate land use changes. Results demonstrate the RF model's effectiveness in accurately classifying various land use types. These changes are visualized clearly through GIS maps, providing valuable insights into land use trends within the region. This research highlights the potential of integrating machine learning with remote sensing technology for land use classification and monitoring. The findings contribute not only to advancing scientific understanding but also to practical applications in land resource management, supporting strategic planning for sustainable development, particularly in the context of climate change and growing urbanization pressures.