Post-war LULC Change Trajectories Indicate Rising Human Activities in the Okavango Basin

Land use/cover (LULC) change is one of the most important contemporary environmental problems threatening water availability and distribution in transboundary basins. This creates the need for regular monitoring of LULC to inform management. The objective of this study was to evaluate trajectories of LULC change between (1996 and 2020) and identify priority management areas (PMAs) for LULC change in the Okavango basin based on Remote Sensing and Spatial modelling. The Deep Neural Network (DNN) was used to classify Landsat 7 and 8 images into eight classes namely bare land, built-up land, bushland, forest/woodland, grassland, cultivated land, water and wetland. Post-feature selection based on Recursive Feature elimination with a wrapper and climate-based regionalization were used to enhance the performance of the classification algorithm. Change in the extent of LULC classes was mapped and assessed in the years 1996, 2004, 2013 and 2020 to establish trajectories of LULC change. Overlay analysis was used to detect patches of LULC-changed sites. Centroids of changed sites were determined and categorized based on the initial class and final class into Category A-transitions (from natural classes to artificial classes); Category B-transitions (from artificial classes to natural classes); and Category C-transitions (from natural classes to another natural class). To assess PMAs for LULC, an ensemble of machine learning algorithms was firstly used to evaluate the susceptibility of the landscape to LULC change. Thresholding of susceptibility maps was done to identify the most susceptible areas which were regarded as PMAs. Results show that anthropogenic classes (croplands, builtup areas and barelands) generally increased at the expense of natural classes (wetlands, woodlands, water and shrubland) during the period between 1996 and 2020. The extent of PMAs for anthropogenic driven transitions (Category A) was larger than that of naturally driven transitions (Categories B and C). Our work demonstrates the dominance of anthropogenic activities in driving LULC change in transboundary basins. Our findings suggest that management actions should focus on strategies to promote sustainable anthropogenic practices to ensure reliable availability and distribution of water in transboundary basins.