Susceptibility to gully erosion: applying random forest (RF) and frequency ratio (FR) approaches for a small catchment in Ethiopia

Authors: AMARE, SELAMAWIT - Wageningen University And Research Center; Langendoen, Eddy; KEESSTRA, SASKIA - Wageningen University And Research Center; VAN DER PLOEG, MARTINE - Wageningen University And Research Center; GELAGAY, HABTAMU - Ethiopian Geospatial Information Institute; LEMMA, HANIBAL - Bahir Dar University; VAN DER ZEE, SJOERD - Wageningen University And Research Center

Submitted to: Water
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/14/2021
Publication Date: 1/18/2021
Citation: Amare, S.A., Langendoen, E.J., Keesstra, S.D., Van Der Ploeg, M.J., Gelagay, H.S., Lemma, H., Van Der Zee, S.E. 2021. Susceptibility to gully erosion: applying random forest (RF) and frequency ratio (FR) approaches for a small catchment in Ethiopia. Water. 13(2):216. https://doi.org/10.3390/w13020216.
DOI: https://doi.org/10.3390/w13020216

Interpretive Summary: To effectively manage soil and water resources in the subhumid, northern Ethiopian highlands, it is critical to determine the location of gullies that are the primary cause of soil loss and groundwater table drawdown. Researchers from USDA, ARS, Oxford, MS in collaboration with scientists from Wageningen University and Research, The Netherlands, and Bahir Dar University, Ethiopia, developed a gully erosion susceptibility map for the Minzir catchment, Ethiopia, using Frequency Ratio (FR) and Random Forest (RF) algorithms based on 16 controlling factors. The five-most important factors for gully formation were drainage density, elevation, land cover, groundwater table and topographic wetness index. Results showed that the RF model with a prediction accuracy of 96% outperforms the FR model with a 67% prediction accuracy. The gully erosion susceptibility map developed in this study can be used to plan informed gully erosion rehabilitation and prevention measures in the Minzir catchment. In addition, the approach is recommended in different agro-ecology and geomorphic settings within the degraded subhumid, northern Ethiopian highlands.

Technical Abstract: Soil erosion by gullies in Ethiopia is causing environmental and socio-economic problems. A sound soil and water management plan requires to accurately predict gully erosion hotspot areas. Hence, this study develops a gully erosion susceptibility map (GESM) using Frequency Ratio (FR) and Random Forest (RF) algorithms. A total of 56 gullies were surveyed and their extents were derived by digitizing Google Earth imagery. Literature review and multi-collinearity test resulted in 16 environmental variables for final analysis. Model prediction potential was evaluated using the area under the curve (AUC) method. Results showed that the RF model out-performs the FR model, with 96% and 67% AUC values, respectively. The five most significant factors that determine gully development in the Minzir catchment were drainage density, elevation, land use, GWT and topographic wetness index. Results from separate variable importance analysis showed land cover for Nitisols and drainage density for Vertisols as leading factors that determine gully locations. Factors such as texture, stream power index, convergence index, slope length, and, plan- and profile-curvature were found to have little significance for gully formation in the studied catchment. The outputs of this research can be used to make informed catchment management interventions in sub-humid Ethiopian highland watersheds.