A heavy rainfall process which occurred over the Taihang Mountains in central and northern Shanxi Province during 4-5 August 2019 was analyzed by using the GRAPES cloud analysis system, with which numerical experiments were also designed. The effects of the introduction of lightning mapping imager event (LMIE) data of FY-4A on the radar reflectivity, cloud microphysical variables, and rainfall prediction of the model calculation were mainly analyzed. The results indicate that the radar echo calculated by adding the LMIE data was closer to the measured radar echo, and the cloud microphysical elements such as cloud water, cloud ice and snow got adjusted, significantly increasing the content of cloud microphysical elements in the area where lightning occurs. The extreme centers of microphysical elements were generally consistent with the distribution of active areas of lightning. Thus, cloud information initialization could effectively improve the accuracy of precipitation forecast within 24 hours and reduce the spin-up phenomenon of numerical model. With the addition of LMIE data, the accuracy of 1-12 h precipitation forecast could be further improved.