Impact of physics parameterizations on high-resolution weather prediction over two Chinese megacities

The 1 km Institute of Urban Meteorology (IUM) operational model has a high-temperature bias, especially at night, and a high wind speed bias in urbanized areas, limiting the ability of IUM to provide accurate, high-resolution prediction of thermal stress and air quality for the densely populated Beijing-Tianjin metro region. This study provides an assessment of the IUM WRF-based operational model setups and performs a diagnostic analysis to isolate the contributions of model physics parameterization schemes to operational forecast bias over complex urban regions. Results show that non-turbulent kinetic energy (TKE) planetary boundary layers (PBL) schemes perform better than their counterpart TKE-based schemes at night, reducing the warm bias by about 1°C in nonurban areas. However, the best performing urban PBL scheme still produces ~2°C warm bias. Considering aerosol effects in the solar radiation scheme improves downward solar radiation and surface energy budgets but has negligible effect on the simulated temperature. Urban canopy models and the specification of various urban model parameters have comparable or even more significant contributions to forecast biases in temperature and wind speed than PBL schemes. The predicted PBL height using an optimized urban parameter table is lower by about 100–200 m, which is about 50–100% of the interurban scheme effect on the PBL height. Overall, the Building Effect Parameterization urban scheme with the default parameter table, or a parameter table with less urban heat storage, is recommended for the best results in urban areas and shows that most of the urban areas of Beijing and Tianjin have a greater than 4°C improvement in absolute temperature bias and more than 1 m s−1 improvement in absolute wind speed bias.