Significant uncertainties exist in the ability of numerical models to reproduce heavy rainfall events over warm sector during the pre-summer rainy season in southern China. One such event occurred with heavy rainfall along the coastal line of Guangdong Province during the period of 29–30 May 2020, and all operational numerical models failed to predict this event at that time. Consequently, eight experiments were conducted to assess the influence of nudging surface-intensive observations into the numerical simulations. The findings reveal that nudging all surface meteorological elements, in what is termed the control experiment (EXP1), reproduces well the development of linear convection and the spatial and temporal evolution of heavy precipitation along the coastline of Guangdong Province. Sensitivity experiments focusing on nudging different surface elements indicate that nudging surface water vapor plays a pivotal role in convection initiation, primarily attributed to the swift escalation of relative humidity from 80% to near saturation (99%) within 3 hours in the low levels, accompanied by a marked reduction in convective inhibition (CIN), lifting condensation level (LCL), and level of free convection (LFC). Nudging surface temperature accentuates thermal buoyancy by amplifying potential temperature perturbations, thereby influencing the initiation and organization of convection. The occurrence and development of convection are considerably delayed in the absence of the nudging surface temperature, and its organization is less structured. Nudging surface wind helps to correct the near-surface southwesterly wind direction, aligning the convection evolution and the rainfall more closely with the observations. Additional six sensitivity experiments were carried out to further explore the impact of nudging duration on simulations. The results suggest that nudging all surface meteorological elements for 6 hours mirrors the outcomes of the control run, effectively reproducing well the heavy rainfall along the coastline. Despite a swift rise in water vapor within the initial 3 hours, the maintenance of water vapor for a certain time (another 3 hours) promotes rapid convection development. Thus, the numerical prediction performance of heavy rainfall in the warm sector over southern China can be improved to a certain extent by nudging surface intensive observations using the surface-grid nudging technique during the initial several (6) hours of model integration.