Spatiotemporal Analysis of Agricultural Variability in Eastern Cape Villages: Employing Google Earth Engine for Climate Change Assessment

Abstract

Satellite-derived vegetation indices and climate data from 2018 to 2024 were analysed to quantify smallholder agricultural responses to climate variability in two rural villages in the Eastern Cape, South Africa. Using Google Earth Engine, R programming 4.4.0, and ArcGIS Pro 3.6, the study assessed spatiotemporal trends in vegetation condition in relation to bioclimatic variables and plot-scale land ownership. The results showed an overall accuracy of 96%, with producer and user accuracies at 79% and 85%, respectively, and a kappa coefficient of 0.95. Time-series analysis revealed a trend of decreasing rainfall and increasing temperatures across the study area, accompanied by elevated Plant Senescence Reflectance Index (PSRI > 0.294) values indicative of advanced vegetation stress. Spatial analysis showed that valley areas exhibited higher moisture accumulation potential and aligned with drainage networks, reflecting enhanced soil moisture retention relative to surrounding terrain. These findings demonstrate the strong influence of topography-mediated water availability on vegetation health in rain-fed smallholder systems. In accordance with the Sustainable Development Goals, the study stresses the importance of gender equity in combating climate change and achieving food security, highlighting the value of integrating multi-scale remote sensing and climate data to identify localised agricultural vulnerability, and underscores the importance of gender-responsive, climate-aware land management strategies to support food security under changing environmental conditions. By situating smallholder agriculture within a land system science framework, the study advances understanding of how topography-mediated soil moisture retention, climate variability, and gendered land governance jointly shape land system trajectories in communal tenure settings.