Comprehensive Analysis of Land Use and Cover Dynamics in Djibouti Using Machine Learning Technique: A Multi-Temporal Assessment from 1990 to 2023

Abstract

Understanding land use and land cover (LULC) dynamics in semi-arid regions is vital for unraveling complex environmental processes and resource management. This study delves into the intricate interplay of land patterns and resource dynamics, offering indispensable insights into the environmental repercussions of these changes. The study aims to quantify land use categories in Djibouti's semi-desert region using remote sensing. It analyzes temporal changes and evaluates Random Forest (RF) algorithms for land use classification. Through meticulous quantification and comprehensive temporal analysis, the research contributes significantly to remote sensing and environmental science by enhancing understanding of land use dynamics and informing sustainable land management practices. Leveraging machine learning supervised classification on the Google Earth Engine (GEE) platform using Landsat data spanning four time periods (1990, 2002, 2012, and 2023), alongside spectral indices and Digital Elevation Model (DEM) data, our study achieves unprecedented insights. Our findings reveal a significant landscape transformation, delineating seven major land cover classes: mangroves, bushes, farmland, built-up areas, water bodies, barren land, and salt plains. With overall accuracy ranging from 89% to 95%, our assessments demonstrate significant changes in land use types over the studied period. Notably, mangroves, bushes, farmland, and salt areas witnessed declines, while built-up areas, water bodies, and barren lands expanded. This research underscores the pivotal role of remote sensing in monitoring long-term land use changes and their ecological impacts. By harnessing technological advancements, our study empowers stakeholders to make informed decisions for sustainable resource management and environmental conservation in semi-arid landscapes.