Multi model forecast biases of the diurnal variations of intense rainfall in the Beijing-Tianjin-Hebei region

Forecasts of the intense rainfall events are important for the disaster prevention and reduction in the Beijing-Tianjin-Hebei region (BTHR). What are the common biases in the forecasts of intense rainfall in the current operational numerical models? What are the possible causes of model bias? In this study, intense rainfall events in the BTHR were categorized into two types: those mainly due to strong synoptic forcings (SSF) and those with weak synoptic forcings (WSF). The results showed that, the numerical forecasts tend to overestimate the frequency of intense rainfall events but underestimate the rainfall intensity. Of these, the overestimation of precipitation frequency mainly appeared in the mountainous areas in the afternoon. Compared with global models, high-resolution mesoscale models showed a notable improvement in forecasting the afternoon intense rainfall, while they all have an obvious bias in forecasting the nighttime rainfall. For the WSF type, both global model and mesoscale model have a low forecast skill, with large biases in subdaily propagation feature. The possible causes are related to a poor performance of the model in reproducing the local thermodynamical circulations and the dynamical processes in the planetary boundary layer. So, the biases in forecasting the WSF type intense rainfall showed notable features of nonlinearity, which made it really challenging to understand their physical processes and to improve the associated forecasts.