A Real-Time Urban Microscale Forecasting System Driven by CMA-MESO: Performance and Sensitivity

Urban microscale meteorological forecasting is critical for the safety of low-altitude activities, environmental assessment, and disaster prevention and mitigation. However, effectively integrating mesoscale-to-microscale coupling frameworks into national operational forecasting systems and quantitatively evaluating their added value for near-surface meteorological variables in continuous real-time forecasts remain major challenges. This study develops a real-time forecasting system that couples the Chinese operational mesoscale model (CMA-MESO) with the Parallelized Large-Eddy Simulation Model (PALM) and conducts a comprehensive one month (June 2025) performance evaluation and mechanism diagnosis. The results show that: (1) The system is operationally feasible, completing a 12-h forecast in about 29 min on 1024 computing cores, thereby synchronizing with the CMA-MESO operational cycle. (2) Compared with the driving CMA-MESO (3-km resolution), the control experiment at 54-m resolution significantly improves forecast accuracy, increasing the correlation coefficient from 0.51 to 0.61. (3) Sensitivity experiments reveal that the primary physical mechanism for forecast improvement stems from the explicit representation of neighborhood-scale urban morphology (building-drag effects) rather than from grid refinement alone. This study demonstrates the potential of the CMA-MESO/PALM coupled system for operational urban microscale forecasting and provides quantitative insights into the relative roles of grid resolution and urban surface representation in enhancing forecast skill.