Background In the Loess Plateau region, constrained by limited soil and water resources, the potential for vegetation restoration has reached its ecological carrying capacity limit. The vegetation patterns within small watersheds have not yet effectively revealed the coupling relationship between vegetation spatial arrangement and erosion-induced sediment yield.

Methods The Lü'ergou watershed was selected to collect underlying surface condition data. Six typical vegetation patterns were established based on altitude and slope gradients. The sediment reduction effects of these patterns were simulated using the SWAT model and an improved connectivity index (IC_ZQ).

Results 1) The SWAT model demonstrated good performance for runoff simulation (R² > 0.93, NS > 0.89) and acceptable performance for sediment simulation (R² > 0.68, NS > 0.64). 2) The vegetation pattern implemented in high elevation areas (HA) achieved the maximum sediment reduction rate of 41%. 3) At the inter-annual scale, a significant negative exponential relationship (P < 0.01) was observed between sediment yield and the mean IC_ZQ value across all typical vegetation patterns. 4) At the sub-watershed scale, a highly significant relationship (P < 0.001) existed between changes in IC_ZQ and changes in sediment yield modulus for all typical patterns.

Conclusions The SWAT model is suitable for simulating sediment transport under different vegetation patterns. The HA pattern (vegetation in higher altitudes) demonstrates optimal effectiveness in reducing erosion and blocking sediment transport. Optimizing the spatial configuration of vegetation can regulate sediment connectivity (IC_ZQ), thereby enhancing soil and water conservation benefits.