【摘要】 利用国家气象中心的数值天气预报（ＮＷＰ）６ｈ预报场，作为同步物理反演模式初估场，由气象卫星垂直探测（ＴＯＶＳ）资料推导出大气湿度，并给出了试验结果．结果表明，ＴＯＶＳ湿度反演的相对误差（ＲＭＳ）约为２０％．其中，夏季反演结果略优于冬季，而且低层（１０００ｈＰａ）和高层（５００ｈＰａ以上）与初估场相比有较大改进，中层（８５０，７００ｈＰａ）二者大致相当．与统计回归反演结果相比其精度约提高１０％．该研究为在我国建立“卫星反演／数据同化／数值预报”相互影响的循环系统奠定了基础更多还原

【Abstract】 This paper presents a new approach for the satellite retrieval of atmospheric humidity from TIROS N Operational Vertical Sounder (TOVS) observations with 6 hour numerical forecasts from the National Meteorological Center (NMC) as the first guess of the Simultaneous Physical Retrieval Model (SPRM). It is shown that the RMS error of SPRM retrievals is about 20%, and the results for the summer time is slightly better than those for the winter time. As compared to the first guess. the SPRM retrievals have greatly improved especially for the lower level (1000 hPa) and higher levels (above 500 hPa). Also, the SPRM results are much better (about 10%) than those from the statistic regression method. This research will lay the foundation for establishment of a cycle system of interaction among the satellite retrieval, data assimilation and numerical prediction.更多还原