Diagnosing WRF Forcing Biases in PALM-4U Urban Heat Stress Modelling for Climate Adaptation in a Humid Continental Climate

Abstract

PALM-4U is a popular large eddy simulation (LES) based CFD model for urban climate research, enabling assessment of the effects of urban layout on outdoor thermal comfort at fine scales through advanced urban parameterizations. However, the accuracy of PALM-4U simulations strongly depends on the quality of boundary conditions (BCs). While LES can partially attenuate errors introduced at the boundaries, it cannot systematically correct biases. In most applications, PALM-4U BCs are preprocessed using the Weather Research and Forecasting (WRF) model, which provides mesoscale atmospheric forcing that is dynamically downscaled to the microscale within PALM-4U. To examine the performance of PALM-4U in outdoor urban thermal comfort indices across seasons, field campaigns were conducted at 21 sites across University of Guelph, Ontario, Canada. Biometeorological data were collected at four time points per day during July and February using mobile measurement platform. PALM-4U simulations were evaluated with in situ observations using multiple performance metrics. Results show that PALM-4U represents mean radiant temperature (MRT) more accurately than the Universal Thermal Climate Index (UTCI) and performs better in winter. The strong performance in simulating MRT indicates that radiative geometry and its spatial distribution are well captured, while remaining biases are more likely associated with uncertainties in dynamic forcing. To further explore sources of uncertainty, radiation, surface temperature, and wind speed simulated by WRF were compared with in situ observations. The results indicate that biases in mesoscale forcing contribute to uncertainty in PALM-4U simulations, likely arising from the combined effects of multiple WRF parameters. This work aids identification of boundary uncertainty sources and provides insight for screening PALM-4U dynamic forcing data, thereby reducing error propagation and improving the reliability of urban thermal comfort simulations.