Role of convection representation across the gray zone on forecasting an extreme heavy rainfall event in Shanghai, China

The representation of convection has been a major source of model biases and uncertainties for precipitation forecasts in current weather /climate models. However, the role of convection and its representation across the gray zone at grid spacing around 1-10 km remains an open question in modeling community. This study investigated the role of convection representation across the gray zone using a multi-grid nesting approach based on the Weather Research and Forecasting (WRF) model. An extreme heavy precipitation event occurred in Shanghai on 25 May 2018 was used to examine the precipitation forecast sensitivity using three convective treatments, i.e., the traditional Kain-Fristch (KF) and scale-aware Grell-Freitas (GF) cumulus parameterization, and explicit convection (EC) solution and different nesting configurations across grid spacings from 27, 15, 9, 5, 3 to 1 km. Results showed at the outer coarser grids (> 10 km), GF produced wet biases with delayed rainfall peaks, while KF produced dry biases with realistic peak timing. At both 9-km and 5-km grids, KF outperformed GF and EC in forecasting the diurnal variation by effectively regulating convective precipitation. For the inner 3-km or 1-km grid simulations, using KF in the outer nesting grid showed advantages over using EC and GF in predicting the extreme precipitation. Double nesting with a large grid ratio (15:1 or 9:1) showed advantages over conventional triple nesting with an intermediate grid of a smaller ratio.