Evaluation of the performance of the non-hydrostatic RegCM4 (RegCM4-NH) over Southeastern China

This study evaluates for the first time the performance of the latest version of the non-hydrostatic RegCM4 (RegCM4-NH) customized over two vast urban agglomerations in China (i.e., the Pearl River Delta, PRD, and the Yangtze River Delta, YRD). A one-way double nesting configuration is used, with a mother domain (20 km grid spacing) driven by ERA-Interim data (0.75°) forcing and two nested domains (4 km grid spacing). The analysis focuses on how the dynamical core (hydrostatic versus non-hydrostatic) employed in the driving mother domain simulation can affect the regional characteristics of temperature and precipitation patterns in the PRD and YRD regions simulated by the 4 km resolution nested RegCM4-NH. In addition, we assess the sensitivity of the 4 km model results to the use of a convective parameterization scheme (CPS), since the 4 km grid size can be considered as a grey-zone resolution at which deep convection is partially resolved and may still need to be parameterized. For mean temperature, a reasonably good performance is shown by all simulations, with the summer season mean bias being mostly less than ± 1 °C when averaged over the PRD and YRD. However, the simulated daily temperature distribution is excessively peaked around the median value, indicating a large probability concentrated on a small temperature range. Although the higher resolution slightly ameliorates this deficiency, the effect of the dynamical core and CPS tends to be marginal. Conversely, precipitation behaves quite differently across simulations. The forcing from the non-hydrostatic mother domain simulation helps to reduce a severe dry bias seen over the PRD due to a reduction in convection inhibition. Also, the use of the Emanuel CPS tends to intensify localized precipitation events over mountainous regions in connection with stronger ascending motions over topographical features. The higher resolution also improves the phase of the diurnal cycle of precipitation, both with and without the use of the CPS. In general, the performance of RegCM4-NH over the PRD and YRD is found to be the best when driven by a non-hydrostatic mother domain simulation and when turning on the Emanuel CPS. With the growing demand for high-resolution climate information, our evaluation study of RegCM4-NH will be a valuable reference to facilitate a wider use of this non-hydrostatic version of the model.