Abstract:
Using radial wind data from 18 CINRAD (China New Generation Weather Radar)/S and CINRAD/C radars in southwest China, S–S, C–C, and C–S dual-radar wind fields were retrieved, and a mosaic was created using a three-dimensional dual-radar wind retrieving technique, which accounted for standard atmospheric refraction in a dynamic earth coordinate system. To study the impact of three-dimensional variational assimilation of dual-radar retrieved wind fields with different horizontal resolutions on a heavy rainfall forecast, eight cloud-resolving numerical experiments were performed. These experiments used the WRF4.2 model and the WRFDA assimilation system to simulate a rainstorm event in southwest China on 6–7 July 2017. The results are as follows. (1) The three-dimensional dual-radar wind retrieval technique, which accounts for standard atmospheric refraction in a dynamic earth coordinate system, is highly effective for retrieving wind data from different band radars in southwest China. (2) For the heavy rain band from northeast Sichuan to northern Chongqing, eight retrieval-wind assimilation experiments with different horizontal resolutions all significantly improved the forecast. These enhancements included better predictions of the direction, precipitation area, range, location, and intensity of the heavy rain band, making the results much closer to reality compared to the control experiment. The spatial correlation coefficient and TS score for 24-hour accumulated precipitation (valid at 6–30 hours) showed significant improvement, while both missed and false forecasting rates decreased. The greatest improvements were observed for rainfall amounts equal to or greater than 25 mm and 50 mm. The best results came from the wind-field assimilation experiment with a horizontal resolution of 0.25°. At this resolution, an appropriate analysis increment field was generated relative to the background wind field through the assimilation of well-resolved wind-field data. This led to an analysis wind field that closely aligned with actual conditions. Additionally, the model integration continuously corrected the wind-field structure from the control experiment, producing simulated features such as vortices, shear lines, high-altitude troughs, wind direction, and wind speed being closer to the observed conditions. (3) From the heavy rain band from northern Yunnan and southern Sichuan to northwestern Guizhou, the assimilation of retrieved winds with a resolution of 0.15° or higher had a significant negative impact on the prediction results. The negative impact became more significant as the resolution of retrieval wind increased. This issue stemmed from data correlation caused by the assimilation of overly dense retrieval winds. This resulted in excessively high and inaccurate analysis increment fields, which ultimately disrupted the simulated wind-field structure and misaligned the position of the rain band. (4) The assimilation of retrieval winds with different horizontal resolutions significantly affects numerical prediction results. Overall, this study identified a horizontal resolution of 0.2° as the critical threshold for the thinning scheme of retrieved winds under the condition of using a triply nested domain with resolutions of 18 km /6 km /2 km and conducting data assimilation across all domains. When the horizontal resolution of the retrieval wind does not exceed this threshold, it generally enhances prediction accuracy. Among these, a horizontal resolution of 0.25° proved to be the optimal thinning scheme.