Journal: Discover Soil
Published on 24th November, 2024
Kenyan drylands have over the years undergone extensive land use and land cover (LULC) changes due to population increase, urbanization, agricultural expansion, industrialization and infrastructural developments. There is however limited information on their historical and future spatio-temporal patterns. This study assessed the spatio-temporal LULC change patterns in Kibwezi West for the period 1990–2021 and predicted the LULC map of 2051. Six LULC classes (Forested land, shrubland, grassland, cropland, water body and other lands) covering 1,040.9 Km2 were examined. Landsat imageries (1990, 2000, 2011 and 2021) were classifed using Random Forest algorithm in R software, while LULC change was analyzed using ERDAS Imagine. The 2051 LULC map was predicted using Artifcial Neural Network and Cellular Automata algorithms. OpenLand software was used for visualization of LULC patterns using Sankey diagrams. Overall classifcation accuracy of 78.04% was obtained with 0.61 kappa coefcient. A net loss in forested land (−112.8 km2 ), shrubland (−54.48 km2 ) and water body (−0.688 km2 ) had occurred, with a net gain in cropland (146.03 km2 ), grassland (20.24 km2 ) and other lands (1.66 km2 ) between 1990–2021. Further, a net loss in shrubland (−110.48 km2 ), forested land (−89.1 km2 ), water body (−0.38 km2 ) and other lands (−0.32 km2 ) was predicted in 2051 while a net gain was predicted in cropland (176.90 km2 ) and grassland (23.39 km2 ). The study pointed out historical and future encroachment into natural ecosystems like forested lands and shrublands. The fndings of this study contribute to the body of knowledge on LULC dynamics in drylands. These results will inform evidence-based decision-making processes for sustainable land use planning, natural resource management and environmental conservation eforts in Kibwezi West and other similar landscapes.
Keyword Drylands; Sankey diagrams; Sustainable land management; Change detection