Regreening Kenya: Assessing the effectiveness of moisture retention techniques using GIS-based temporal analysis

Abstract

This study evaluates the effectiveness of semi-circular bunds (demi-lunes) in enhancing vegetation health and stability in the semi-arid Loitokitok subcounty of southern Kenya over a multi-year period (2016–2024). Using high-resolution Planet Labs satellite imagery, the research quantifies vegetation dynamics through Soil-Adjusted Vegetation Index (SAVI) merged composites, comparing bund plots to reference areas matched for geographical, topographical, and soil characteristics. Key methodological parameters included testing indices (NDVI vs SAVI), composite intervals and merging methods (mean, median, maximum) to optimize data accuracy and mitigate raster defects such as cloud cover and glare. Results indicate that semi-circular bunds improved vegetation resilience, with bund plots outperforming reference areas in peak SAVI (up to +86.2%), greening rates (+37.1% in Amboseli plots), and resistance to wilting (+18.4%). Older bund plots (Amboseli group, initiated in 2016) demonstrated more pronounced benefits compared to newer plots (Chyulu group), suggesting time-dependent efficacy. Bi-weekly composites with median merging emerged as optimal, balancing data density (13.7% loss) and accuracy, while SAVI proved superior to NDVI in detecting sparse vegetation, aligning with semi-arid conditions. Overall, bund plots exhibited a 5.6% average increase in mean SAVI compared to reference areas. Limitations included sparse pre-2018 satellite data and processing constraints that restricted advanced algorithms. The study underscores the potential of semi-circular bunds as a scalable tool for combating desertification, supported by remote sensing methodologies. Future research should integrate field validation, historical pre-intervention data, and machine learning to refine temporal analysis and account for climatic variability.