Objective Soil erosion in China’s Loess Tableland region has intensified under the combined effects of natural conditions and human activities, posing a serious threat to the ecological security of the Yellow River Basin. Conventional erosion models often fail to adequately quantify the interaction between natural and anthropogenic factors. This study aims to develop an integrated evaluation system that incorporates both natural and anthropogenic factors to support the precise identification of erosion hotspots and enable differentiated zonal management strategies.

Methods Dongshaoliang (E 110°15′–110°18′, N 35°12′–35°15′), a high-erosion-risk zone in the Weibei Tableland, was selected as the study area. Multiple data sources were integrated, including high-resolution (1 m) DEM, multi-temporal Sentinel-2A imagery, UAV-based photogrammetry, and field survey data covering 217 sediment traps and 58 vegetation quadrats. A total of seven factors, including slope gradient, vegetation coverage, cut-slope height, road density, land use intensity, soil erodibility, and surface runoff potential, were selected and weighted using the analytic hierarchy process (AHP) with a consistency ratio (CR) of 0.024. A soil loss sensitivity index (SLSI) was established within a two-tier analytical framework combining slope units and 10 × 10 m grid cells. Furthermore, an optimized comprehensive ecological stabilization (CES) technology was proposed.

Results 1) Slope gradient (weight = 0.20) and vegetation coverage (weight = 0.25) were identified as the primary factors, jointly explaining 78% of the variance in erosion intensity (R² = 0.78, P < 0.01). Human activities such as cut slopes and road construction amplified sediment yield by 32%–41%. 2) Areas with severe erosion (SLSI > 0.75) accounted for 23.8% of the total area, mostly distributed on slopes > 65° with vegetation coverage below 45%. These regions exhibited sediment yields 4.3 times higher than those in stable areas. 3) The optimized CES technology—involving 10% straw fiber incorporation, non-woven fabric coverage (15 g/m²), and smart drip irrigation—was applied in demonstration zones, significantly increasing vegetation coverage to 64.8 ± 3.7% within 24 months (P < 0.01) (an improvement of > 22% compared to pre-treatment levels), while reducing sediment by 42.3%.

Conclusions 1) The interaction between slope gradient and vegetation coverage is the dominant coupling mechanism driving soil erosion on loess tableland slopes, with anthropogenic factors significantly amplifying erosion processes in steep slope areas. 2) The developed AHP-GIS-SLSI coupled model effectively quantifies the synergistic erosion effects of natural and anthropogenic factors. 3) The optimized CES technology is suitable for high and steep loess slopes and can significantly improve vegetation coverage. The proposed zonal management strategy and optimized CES technology parameters can provide technical support for precise soil erosion control and ecological restoration in the loess tableland region, contributing to the Ecological Conservation and High-Quality Development of the Yellow River Basin strategy.