气象卫星资料同化的科学问题与前景

Scientific issues and perspective of assimilation of meteorological satellite data

从数值天气预报资料同化的角度,分析了气象卫星观测与常规气象观测的不同特点形成了卫星资料同化的特殊科学问题。由于各类星载遥感仪器所观测到的是一定波长的电磁辐射,不能像传统的直接观测资料一样被预报模式直接应用。又由于卫星观测对象是整个大气层,而不是特定高度的大气,因此确定卫星观测影响的高度和它的潜在垂直分辨率成为一个关键问题。卫星观测还可能存在着系统性的偏差,这与直接观测的误差互相独立也有很大区别。资料同化通常建立在模式预报(即背景信息)与观测量的比较的基础上,为了实现同化,需要将模式的基本大气变量转换成星载仪器所获得的特定波长的电磁波特征量,或者将观测到的电磁辐射特征量反算成大气的特征量。前者需要引入复杂的观测算子,后者则将复杂的反演过程交给了前处理阶段。这就形成了直接与间接同化卫星资料的两种不同策略,策略的选择取决于同化系统处理复杂观测资料的能力,对同化效果有决定性的影响。逐个分析了目前用于数值预报的5种卫星观测资料,即星载大气垂直探测器资料、大气运动矢量资料、散射仪海面风资料、卫星观测的云与降水信息资料与GPS掩星观测资料的同化的进展与有待解决的主要问题,概述了中国近年在大气垂直探测器等卫星资料同化中的研究进展及其业务应用的效果,并提出了今后需要予以特别关注的研究方向。

The main features of the observations by meteorological satellites and related scientific issues in terms of data assimilation in numerical weather prediction are analyzed in comparison with the in situ data.The raw observations of the satellite-borne instruments are electromagnetic radiances which may not be digested directly by the NWP（Numerical Weather Prediction） models.The measured radiances are affected by the state of the whole column of the atmosphere rather than an individual level,so that the sensitive height and the equivalent vertical resolution of the satellite observations are crucial.Different from most conventional observations,the measurements from the satellite are not bias-free.In order to assimilate the satellite data,it is fundamental to convert the model predictands to the electromagnetic radiances of specific channels sensed by the satellite-borne instruments or conversely.The former,which is so called direct assimilation,needs complicated observational operators,but in the latter,known as indirect assimilation,the retrieval must be included as one step of the data preprocessing.The choice of direct or indirect strategies depends on the capability of the data assimilation system in dealing with the complicated observations and is crucial for the effects of the satellite data usage in NWP.Following the general discussions,the achievements in assimilation of five main categories of satellite data,i.e.the satellite soundings,satellite derived atmospheric motion vectors,space based scatterometors,precipitation affected radiances and GPS radio occultation data are reviewed with the stress on the discussion of the further improvements.The progresses of satellite data assimilation in China are also reviewed followed by concerns for the near future developments.