To support the ecological benefit compensation system of key ecological function areas in China, Remote Sensing RMMF (RS-RMMF) soil erosion model is established by selecting the optimized remote sensing factors and RMMF (The revised Morgan, Morgan and Finney model) model. RS-RMMF model eliminates the soil erosion changes caused by the fluctuation of meteorological elements. The function of soil and water conservation is evaluated by constructing three evaluation indexes of unit rainfall interception rate, unit runoff erosion amount and unit runoff transportation volume. Karst functional areas in Guangxi, Guizhou and Yunnan are taken as the study area based on RS-RMMF model. The results show that compared with 2011, the rainfall interception rate PI₀ in 2019 increased by 1.94%, the runoff erosion amount H₀ decreased by 5.96×10^-4 Mg/hm², the runoff transportation volume TC₀ decreased by 6.0×10^-7 Mg/hm², and the comprehensive score increased by 0.83. The overall water and soil conservation function has been slightly improved. Compared with RUSLE, the results of RS-RMMF model are basically consistent in the overall trend from 2011 to 2019. There are significant differences in the spatial distribution of the model results. The spatial distribution of the evaluation score of RS-RMMF is dominated by vegetation, but is less affected by precipitation factors. RS-RMMF model can objectively calculate the soil and water conservation capacity of the ecosystem, and provide scientific support for the application of ecological benefit compensation system.