Aiming at the applicability of cloud mode forecasting products in artificial rainfall operations in complex terrain areas of Guizhou, based on the quantitative forecasting products of CMA-CPEFS (Cloud-Precipitation Explicit Forecasting System) mode developed by the Artificial Impact Weather Center of the China Meteorological Administration, the multi-source observation data fusion diagnosis method of ground-based radar, static satellite and pronging profile is adopted to test the mode prediction performance of 30 cases of rainfall of aircraft in Guizhou from 2020 to 2021. The results show that: (1) The accuracy of mode cloud phase discrimination is 93% (28/30), and the matching degree between the operation height layer and the catalyst magnitude configuration is 93% (28/30) and 100% (30/30) respectively, confirming that the mode's characterization ability of the cloud physical parameter field meets the business practical standard. ( 2) There is a systematic positive deviation in the height of the zero-degree layer (average error +288 m, 95% CI: ±150 m, p<0.01), and the cloud speed prediction error is significantly higher than the shift error (the relative error is 23%±5% and 17%±3% respectively). ( 3) The TS score of the rain increase potential area is 0.67, but there is a significant spatial epitaxial deviation (the forecast range is 18% ± 3% larger than that of the actual situation), which is significantly related to the subjective analysis error and excessive mode estimation of cloud water content (deviation +0.15g/m3) (R2=0.82, p<0.001). The results of this study provide key physical constraints for the optimization of artificially affecting weather numerical models in the karst geomorphological area where Guizhou is located.