实时GPS可降水量资料的变分同化个例研究

A case study of variational assimilation of true real-time GPS precipitable water

用实时GPS可降水量(Precipitable Water,PW)资料GPSPW,针对2007年7月2—3日发生在日本九州南部地区的一次梅雨锋降水过程,设计控制试验和4个同化试验,利用WRF和WRF-Var模式进行三维变分同化研究,并用ETS评分客观评价各方案的预报效果。结果表明:实时GPSPW同化能有效地改进模式初始时刻水汽场以及预报时间内(0～6h)的PW。背景误差水平尺度调节系数的大小对水汽场的调整范围和PW的改进程度有一定影响。GPSPW的实时性能提高降水预报业务的实时性。

Most GPS precipitable water（PW） data used in variational assimilation are semi-real or near real because of the delay of zenith total delay （ZTD） estimates after its observation. The data used in this paper are from Japan＇s GPS Earth Observation Network （GEONET） , which has the ability of real- time data transmission with several seconds latency. So we define the GPSPW as true real data. A heavy rain associated with Meiyu Front occurred from July 2 to July 3, 2007 in the south of Kyushu Japan and is discussed in the paper. We utilize WRF-Var as the variational assimilation system, design a control and four variational experiments and evaluate the results using Equitable Threat Score （ETS）. The results show that true real-time GPSPW data are able to be assimilated into WRF and the initial water vapor fields and PW predictions within forecast period are adjusted after assimilation. Different horizontal scaling of background errors leads to different modified areas of water vapor fields and PW predictions. True real-time GPSPW will contribute to more effective operational application.