Human activities on land have grown significantly changing the entire landscape, while most of the changes have occurred in the tropics. The change has become a serious environmental concern at the local, regional and global scales. The intensity, speed, and degree of land use / land cover (LULC) changes are nowadays quicker compared to the past because of the development of society. Moreover, the rapid increase in population resulted in disturbing a large number of landscapes on the Earth. The main objective of this study was to detect historical (1990– 2020) and predicted (2020–2050) LULC changes in the Welmel River Watershed, which is located in the Genale-Dawa Basin, South Eastern Ethiopia. The dataset of 1990, 2005, and 2020 was generated from Landsat 5, Landsat 7 and Landsat 8 respectively to determine the historical LULC map. The result of this study revealed that agriculture/ settlement increased by 6.85 km ²∙y ^–1, while forestland declined by 9.16 km ²∙y ^–1 over the last 31 years between 1990 and 2020. In the coming 31 years (by 2050), if the existing trend of the LULC change continues, agriculture/settlement land is expected to increase from 290.64 km ² in 2020 to 492.51 km ² in 2050 at the rate of 6.73 km ²∙y ^–1, while forestland is expected to shrink from 690.48 km2 in 2020 to 427.01 km ² in 2050 by a rate of 8.78 km ²∙y ^–1.

This study focuses on the problem of mapping impervious surfaces in urban areas and aims to use remote sensing data and orthophotos to accurately classify and map these surfaces. Impervious surface indices and green space assessments are widely used in land use and urban planning to evaluate the urban environment. Local governments also rely on impervious surface mapping to calculate stormwater fees and effectively manage stormwater runoff. However, accurately determining the size of impervious surfaces is a significant challenge. This study proposes the use of the Support Vector Machines (SVM) method, a pattern recognition approach that is increasingly used in solving engineering problems, to classify impervious surfaces. The research results demonstrate the effectiveness of the SVM method in accurately estimating impervious surfaces, as evidenced by a high overall accuracy of over 90% (indicated by the Cohen’s Kappa coefficient). A case study of the “Parkowo-Lesne” housing estate in Warsaw, which covers an area of 200,000 m², shows the successful application of the method. In practice, the remote sensing imagery and SVM method allowed accurate calculation of the area of the surface classes studied. The permeable surface represented about 67.4% of the total complex and the impervious surface corresponded to the remaining 32.6%. These results have implications for stormwater management, pollutant control, flood control, emergency management, and the establishment of stormwater fees for individual properties. The use of remote sensing data and the SVM method provides a valuable approach for mapping impervious surfaces and improving urban land use management.