Unveiling gully erosion susceptibility: A semi-quantitative modeling approach integrated with field data in contrasting landscapes and climate regions

Abstract

This study addresses the challenge of mapping gully erosion susceptibility, which is often hindered by limited observed data, the complexity of controlling factors, and the uncertainties associated with characterizing these factors. We utilized a semi-quantitative modeling approach that integrates field-based data and ten controlling factors in the Chemoga watershed of Ethiopia's Upper Blue Nile basin. The resulting gully erosion susceptibility map was compared with a random forest-based approach to assess the methodological applicability. Additionally, an independent dataset from adjacent watersheds was used to validate the approach. The findings revealed that certain landscape positions with specific elevation ranges and slope steepness were more susceptible to gully erosion due to factors such as rainfall, lithological formations, soil characteristics, and agricultural activities. Approximately 10 % of the watershed area was affected by gully erosion, with varying susceptibility levels. The comparison between the semi-quantitative and random forest approaches demonstrates a total agreement of around 58 %, with minimal differences in susceptibility classes. The study also highlights a strong agreement between simulated and observed susceptibility maps, with a 76 % PBIAS (Percent Bias) value for the simulation and a lower 48 % agreement for the random forest approach. Furthermore, in the adjacent watershed, 65 % of the area exhibits no discrepancies between observed and simulated maps. This suggests that the semi-quantitative approach is effective in extrapolating gully erosion susceptibility when detailed data is limited, offering a cost-effective and efficient solution. The study emphasizes the utility of the semi-quantitative modeling approach in mapping gully erosion susceptibility and its potential for practical applications in land management and intervention strategies.