Abstract:Based on the conventional observation data and the Guangzhou Doppler weather radar data, this article analyzed the synoptic background, mesoscale systems and predictability of the torrential rain process which occurred in Guangzhou on 7 May 2017. By contrasting the members of ECMWF ensemble forecast which successfully forecasted the local heavy precipitation on 7 May 2017 with the failed member, key triggering factor affected this process was investigated. The results show that the ambient condition and dynamic forcing were weaker on 7 May 2017. In the context of weak wind field, a convergence line formed between the warm flow from the sea and the cold downslope wind in the front of mountain, combined with the intense temperature gradient, initially triggering the generation of convective storms. Subsequently, the outflow of the preexisting thunderstorms impacted the warm, moist flow at boundary layer, inducing new convergence lines continually. Therefore the weakened storm-cells enhanced again and aroused the second stage of heavy rainfall. The first stages of the local torrential rain on 7 May 2017 was induced by short-lived severe local storm with a meso-vortex evolved from the steady and strong block-shaped echo. The second stage of this torrential rain was caused by long-lived heavy precipitation (HP-type) supercell storm. However, the band echo merging from a mass of new-born convective cells and moving eastward with the short-wave trough was responsible for the third stage of this torrential rain. The radar echo had the low-quality core vertical structure and warm cloud precipitation character during all the three stages. It has turned out that more enhanced temperature gradient and surface wind convergence might be the important triggering factors for local severe precipitation. This finding is based on the member comparison of ECMWF ensemble forecast that successfully forecasted the local heavy and weak precipitation on 7 May 2017. Now, it is still difficult for numerical models to forecast the extremely heavy rains in short lead-time forecast under the condition of warm sector and, especially, the weak wind field. The main method for this is to enhance rainstorm disaster monitoring and early warning.