Abstract: Objective Topographic factors are key parameters of the Revised Universal Soil Loss Equation (RUSLE), as they represent hillslope runoff convergence and the associated accumulation of erosive energy. The conventional slope length factor (L) is formulated under the assumption of continuous downslope runoff. However, this assumption is frequently violated in karst landscapes, where widespread exposed rock can interrupt surface flow paths and shortens the effective slope length, potentially resulting in biased estimates of soil erosion. Methods To address this limitation, a typical karst watershed in Yunnan Province, Southwest China, was selected as the study area. An improved slope length factor (L_Rc) incorporating the truncation effect of exposed rock was developed using ¹³⁷Cs tracing, runoff plot observations, and the RUSLE framework. Wavelet analysis was applied to identify the dominant spatial scales of erosion-deposition processes on karst slopes and to examine how exposed rock constrains effective slope length. The exposed rock ratio was subsequently incorporated into the slope length calculation to derive L_Rc factor and establish RUSLE-Rc model. Model performance was evaluated against both runoff plot measurements and ¹³⁷Cs -derived soil erosion rates. Results Erosion-deposition process on karst slopes exhibited significant spatial periodicity, with a dominant scale of approximately 70 m, markedly shorter than the 200 m scale identified on non-karst slopes. This contrast indicates that exposed rock not only alters surface cover conditions, but also modifies the spatial scale of erosion–deposition processes by interrupting runoff pathways. Incorporating this truncation effect enabled L_Rc to more realistically represent effective slope length in karst landscapes and improved the agreement between RUSLE-Rc estimates with both ¹³⁷Cs-derived erosion rates and runoff plot observations. Compared with the conventional RUSLE model, RUSLE-Rc consistently improved model performance for both validation datasets, with R² values increasing to 0.61 and 0.69, NSE values increasing to 0.52 and 0.43, and PBIAS values decreasing to 10.31% and 5.91%, respectively. Conclusions These results indicate that slope length in karst areas is better interpreted an effective runoff-contributing scale jointly controlled by exposed rock truncation and surface flow-path connectivity, rather than simply as a geometric distance. Incorporating exposed rock truncation into the L factor therefore provides a process- informed parameterization for soil erosion modelling under complex karst surface conditions.