In the present study, a new local breeding growth mode (LBGM) method is used to conduct convection-allowing ensemble prediction experiments regarding two typical squall line cases.Next, by comparing the results with the traditional breeding growth mode (BGM) method, the actual forecast effect of the LBGM is tested.The simulations of precipitation are performed using the probability-matched mean (PMM) method, then compared with the observational data.Finally, the fraction skill score (FSS) is used to replace the traditional equitable threat score (ETS), so as to reasonably evaluate the precipitation prediction.We obtained the following results:(1) The LBGM is observed to be superior to the BGM in the forecast of the precipitation in the squall line.(2) The FSS better represents the convection-allowing ensemble prediction capability better than ETS, particularly for extremely severe rain assessments.(3) The precipitation scoring results show that the EM is superior to the PMM method for light, moderate and heavy rainfall, whereas the PMM method is better for severe and extremely severe rainfall.