Assessment of Land Cover Changes in Litumbandyosi-Gesimasowa Game Reserve using Remote Sensing and GIS

Abstract

As countries in Sub-Saharan Africa strive to reduce deforestation in Miombo woodlands, it is essential to use the appropriate, reliable, and cost-effective tools for assessing land cover changes. This study employed Remote Sensing and GIS techniques to assess land use and its changes in the Litumbandyosi-Gesimasowa Game Reserve between 1990 and 2020. The tools employed were GEE and Collect Earth. The study employed Sentinel-2 and Landsat-5 TM imagery and also incorporated the Atmospheric Resistant Vegetation Index (ARVI) for improving classification by overcoming the effects of Non-Photosynthetic Vegetation (NPV) and phenology. The study produced highly accurate land cover maps, with an overall accuracy of 99.53% and a kappa coefficient of 98.11% in 1990, 99.84% and a kappa coefficient of 98.69% in 2011, and 92.10% and an 89.62% kappa coefficient in 2020. The findings of the post-classification revealed an alarming change in land cover over the last 30 years, with heavy forestland decreasing by 10.77%, shrubland increasing by 12.19%, and grassland increasing by 13.35%. Furthermore, farmland expanded by 4.58%, barren land grew by 5.82%, and wetlands grew by 0.74%. Significant agents of change have been identified as forest fires, overgrazing, crop farming, and mining.