Impact of Watershed Management on Vegetation Cover and Soil Moisture Using Remote Sensing, in Magera and Wutame Micro Watershed, Omo Gibe Basin

No Thumbnail Available



Journal Title

Journal ISSN

Volume Title


Addis Ababa University


Despite the importance of watershed management as an approach to curve land degradation, to date, there has been little research on their impact on hydrological variables. This study evaluated the potential of remote sensing in quantifying and detecting vegetation cover, land cover and soil moisture dynamics as caused by the watershed management in a micro watershed in Ethiopia. To address these, multi-temporal data of Landsat imageries were used to retrieve NDVI for detecting vegetation cover change and to produce the LULC map for assessing the land cover changes from 2010 to 2019 which incompass the period before and after the watershed intervention. Mann-Kendal trend test was used to examine long trends in the monthly NDVI area of vegetation cover. In addition, multiple change-point analyses were carried out using Pettitt’s, Buishand’s and SNHT tests to detect the change point (year), if any, and to find its possible relation with watershed intervention. Long-term station based monthly rainfall data from 2010 was used to check the possible influences of rainfall in the increased the vegetation density. The accuracy of maps was also assessed using the error matrix. Furthermore, a remote sensing-based soil moisture index (SMI) model and ground measurement from 40 sampling scheme was used for soil moisture estimation and validation. The model is validated using adj-R2, root mean squared error (RMSE), the absolute average difference (AAD), and the precision model. The threshold NDVI classification analysis revealed three vegetation cover classes, including no plants or bare land, weak plants or shrub and grassland and healthy plants or forests which were designated in increasing order of vegetation densities. Significant increasing and decreasing trends in vegetation cover classes have been detected from the Mann-Kendal test. The area coverage of healthy plants dramatically increased from 1.5% to 33%; in contrast, the area under bare land decreased drastically from 40.9% to 0.6% post-intervention. The year 2015 was detected as a change point (year) for continues conversion of three vegetation cover classes. The weak and decreasing correlation was shown in between monthly NDVI area and rainfall, which further verified the increment in vegetation cover is not only from the rainfall influences. The fundamental LULC changes in the watershed were, increased in both the forest land and agricultural land area and decreased in bare, shrub and grassland areas. According to SMI data, the value near 0 showed low vegetation cover and little soil moisture, whereas the value near to 1 showed high vegetation cover and high moisture content. The SMI model accurately estimated the ground soil moisture based on its high R2 (0.768) and low RMSE (0.033cm3/cm3). Because of the vegetation cover increased after intervention, the watereshed also experienced an increase in soil moisture over the study period. The study shows that the watershed management intervention has an overall positive impact on the wateshed. Based on the findings of this research, remote sensing approaches have the potential to evaluate watershed intervention and also have to quantify and detect the vegetation cover, land-use changes and soil moisture changes.



Land Degradation, Watershed Management, Landsat, NDVI, SMI, LULC, Soil Moisture