Abstract: The warm-sector torrential rain in South China often occurs in the warm sector 200-300 km ahead of a cold front or without a cold front. Because of the lack of distinct synoptic systems, the predictability of this type of rain is at a very low level. To discuss the impact of initial moisture condition on warm-sector convection triggering and development, two numerical experiments (the control run CTRL and the cloud analysis run CLD) are designed to study a case occurred over the western coast of Guangdong province on 16 May 2015. These experiments are conducted using 3-km resolution mesoscale GRAPES (Global/Regional Assimilation and PrEdiction System) model and ADAS (ARPS Data Analysis System) complex cloud analysis system. Results show that:(1) the initial condition of cloud analysis run CLD is characterized by higher moisture content, more instable stratification, larger convective available potential energy, increased K index and precipitable water, and lower lifting condensation level and convective inhibition, and smaller LI index. Thereby the vertical velocity is enhanced and the convection is triggered more quickly; (2) there exist some differences between the mechanisms for the warm-sector convection triggering and development, i.e., the triggering is enhanced by latent heat released from water vapor condensation, which contributes to the positive buoyancy; during the convection development period, convergence is caused by the outbreak of the surface cold pool and the warm advection from the sea. Sensitivity experiments further indicate that when reducing the water vapor content to 70%, the initial convergence is weakened, and the convection triggering is subsequently weakened, while the dissipation is also delayed. These results can increase our understanding of the predictability of warm-sector torrential rain in South China. Specifically, results of the present study can be used as a reference for future development of operational cycling assimilation system.