Analysis of Uncertainties in Convection-Permitting Ensemble Simulations of Land Breeze and Nocturnal Coastal Rainfall in South China

Through daily convection-permitting ensemble simulations conducted over a 3-month period, the forecast uncertainty for the land breeze and associated coastal rainfall during early-summer rainy season over South China is investigated. The ensemble includes 12 sets of physics parameterization schemes for boundary layer, radiation, surface layer, and land surface processes. Observations from air–sea buoys at sea, coastal weather stations, and radiosondes are employed to evaluate the diurnal variations and vertical structures of the simulated land breezes. Results suggest that the forecast uncertainty of land breeze circulations is closely associated with the model’s representation of the nocturnal near-surface air temperature on land sides. A systematic underestimation of nocturnal air temperature is recognized in most ensemble members, while the diverse errors of daytime air temperature on land can be diminished through the ensemble mean. The cold bias tends to create stronger land breezes, resulting in prolonged and widespread coastal rainfall through more intensive coastal convergence. By comparing the relative contributions of multiple parameterization schemes, it is found that the systematic underestimation for nocturnal air temperature primarily results from the surface layer and land surface parameterization schemes. To improve the nighttime temperature forecast over this rainfall hotspot, it is essential to implement an advanced land surface model that incorporates complex thermodynamic processes tailored to this climate regime. Additionally, improved parameterization schemes for the planetary boundary layer and surface layer are necessary to enhance the nocturnal turbulent intensity under near-neutral conditions.