With the continuous improvement of the horizontal resolution of the numerical weather prediction model, cumulus clouds are not distinguishable by the grid scale in the “grey area” (1-10 km) of the convective parameterisation scheme, making the traditional cumulus convective parameterisation scheme difficult to apply to the simulation of precipitation at different resolutions. Therefore, the development of scale-aware cumulus cloud parameterisation schemes is one of the development trends of numerical model refinement. By introducing the scale-aware KFeta (Kain-Fritsch eta) convective parameterisation scheme into the CMA-MESO (China Meteorological Administration Mesoscale model) model, the paper combines three resolutions (1 km, 3 km, and 10 km) and designs three sets of experiments so that a numerical simulation test is carried out on a precipitation process on 11-12 July 2023, in the mountainous area of southwest China, and a 7-day batch test is carried out on the precipitation process on 17-23 June 2023. This is done to test and evaluate the effect of the scale adaptation scheme on precipitation simulation in the southwest mountainous area. The results show that: (1) In the case experiment, the scale adaptation scheme has a good grasp of the precipitation rain band and precipitation intensity, and the simulation of the heavy precipitation centre has certain advantages. With the increase in resolution, the precipitation intensity and falling area of the scale adaptation group are closer to reality, which greatly reduces the false precipitation caused by cumulus clouds and increases the grid precipitation. (2) The combined reflectance factor and vertical velocity simulation results in the strong convection region of the scale adaptation scheme are better. (3) In the batch test, the scale adaptation scheme has a good and stable simulation ability for precipitation in the southwest mountain area, especially the simulation results of the scale adaptation scheme for moderate rain and above, which are more advantageous, and the precipitation TS score of the original scheme improves by 4% overall. In general, the scale adaptation scheme performs well in the forecast of heavy precipitation in the southwest mountainous area, providing a basis for improving the current numerical model of heavy precipitation forecasts.