多源卫星观测数据在全球四维变分同化系统中的应用

Application of Multi-source Satellite Obervations in the Global 4D-VAR System

首先描述高分辨率全球四维变分资料同化系统的基本软件框架,并对多源卫星资料同化的关键技术作阐述;其次通过统计分析,说明新型卫星观测数据的引入不但能够增加同化系统中的信息量,而且能够提高其他类型观测数据的利用率;然后通过一个月的统计检验结果表明:无论是从距平相关还是均方根误差而言,在有效同化无线电掩星和卫星风资料后,高分辨率同化预报系统预报技巧的提高是十分明显的;最后,通过一个强降水个例的分析结果表明:基于新的初始场全球模式降水预报准确性较高,就强降水中心区域的预报而言,模式预报和观测实况较为一致,优于国外模式降水预报。

At first,the infrastructure software frame of a high-resolution global four-dimensional variational data assimilation system( YH4DVAR) is described,and the key technologies for assimilation of multi-source satellite data are also introduced. Secondly,it is shown from statistics analyses that the new types of satellite observations not only increase the amount of information in the meteorological data assimilation system,but also improve the rate of utilization for other kinds of observations. Then,according to abnormal correlation and RMS error from a complete month's statistical verification,it can be concluded that the introduction and effective assimilation of new satellite data( GPS RO and AMV) can greatly improve the skill score of high-resolution numerical weather prediction( NWP) system. At last,the global spectral forecast model( YHGSM) can predict the strength and location of a heavy rain weather process more accurately with the new initial field.The rain accumulation from YHGSM forecast agrees very well with atmospheric observations and is superior to rain forecast from NWP models abroad.