风廓线雷达资料对华南区域模式预报的影响

Impact of Wind Profiler Data on Regional Model Prediction in South China

设计基于GRAPES_Meso的不同试验模拟2014年3月28日-4月8日的广东前汛期降水过程,评估风廓线资料对同化和预报的影响。对资料同化后分析增量的分析表明:相比同化时仅使用自动气象站资料,风廓线雷达资料对1000 hPa到850 hPa纬向风增量均有贡献,在850 hPa,700 hPa高度以上贡献迅速减小。应用3个试验的预报结果计算探空站、风廓线雷达站预报值与观测值的11 d均方根误差发现,同化加入风廓线雷达资料对各预报要素的改善在850 hPa高度最明显,其中风速预报误差显著降低,为0.7 m·s^(-1)。此外,风廓线雷达资料对700 hPa风速预报有一定改善,而在925 hPa高度模拟效果反而降低。通过对2014年3月30日12:00(世界时)的个例分析发现,同化加入风廓线雷达资料的风速预报均方根误差在大雨级别以上的降水落区更大,其原因还有待于进一步研究。

Weather analysis demonstrates that upper-level jet, low-level jet, and wind shear are closely related with rainstorms and severe convections in South China. Wind profiler radar can continuously observe wind, making it the most direct resource of upper wind observation comparing with conventional observations. A network of observation stations with 18 wind profiler radars is built in Guangdong, data of which are assimilated every 3 hours in GRAPES_Meso model in real time, and the influence of wind profiler data is evaluated. A precipitation process in the pre-flood season of South China from 28 March to 9 April in 2014 is simulated through three designed experiments by GRAPES_Meso model. Results of assimilation trials show that wind profile data contribute a lot to analysis increment of zonal wind at levels from 1000 hPa to 850 hPa, especially at 850 hPa, and this effect rapidly diminishes above 700 hPa level. The root mean square error （RMSE） of forecasted variables at radiosonde stations are calculated in terms of sounding observations and the outcome of three experiments. Results show that profiler data mostly improve RMSE at 850 hPa, which announces a 0.7 m · s-1 reduce of forecasted wind speed error, for 700 hPa level there＇s no evident improvement, and for 925 hPa level it becomes even worse. The same RMSE analysis is done at 12 wind profiler stations. The result is in accordance with radiosonde stations, which shows that the RMSE decreases at 850 hPa as well and the improvement is not evident at 925 hPa. The analysis indicates that the quality of wind profiler data is relatively better at 850 hPa. Results of two sensitive experiments for 1200 UTC 30 March 2014 is examined, revealing that the RMSE of the forecasted wind speed is even greater when wind profiler data are used in assimilation at locations of heavy rain grade or a- bove. Besides, it seems that the RMSE of the forecasted wind speed also increases in the downstream direction of this heavy rain location. Causes of these two phenomena still need to be analyzed.