Assess And Improve To One Dimension Variational Retrieval Of AMSU Data Under Cloud And Rain

In this paper, the one dimension variational method and observation operator CRTM were used in retrieving AMSU sounding data of typhoon FANAPI, KOPPU physically to get atmosphere status in weather condition of clear sky, cloudy and rainy. Firstly Cloud and rain check is used in the process of variational retrieval to separate AMSU observation into clear sky, cloudy and rainy. Besides, observation error re-estimate was preformed also in retrieval. FNL analysis data, CloudSat and DOTSRAT data are used as validation data to examine the retrieved atmospheric variables. The Variation retrieval experiment shows that:1,The minus between observation and simulated bright temperature by background of channel 2 in AMSU-B, and the rain probability can be used as cloud and rain check and separate AMSU observation into clear sky, cloudy and rainy reasonably. Analysis to retrieved atmospheric variables under different weather condition in detail is possible by using cloud and rain check.2,Observation error re-estimate can make better estimate to observation error and improve the retrieval result to temperature and humidity.3,To temperature, variation retrieval could generate retrieval temperature result of much better accuracy than background temperature in all levels in clear sky, in condition of cloud and rain, variation retrieval could generate better retrieval temperature also in low and high levels, but could not generate better retrieval result in middle levels of atmosphere (about 500hP) for the CRTM can not simulate well under condition of cloud and rain. The retrieved temperature is close to verification data in horizontal and vertical dimension.4,To humidity, in clear sky variation retrieval could generate humidity result of better accuracy than background humidity in high levels, which means that level above 400hPa, but the retrieved humidity's accuracy is worse than background data below 400hPa.5,To water contents such as ice water path (IWP), variation retrieval could generate results of close to CloudSat's Sounding data。6,The background data used in variation retrieval comes from statistical regression, so from the analysis to retrieved temperature and humidity result above, it is clear that variational retrieval is much useful in retrieving temperature than statistical regression in clear sky and at low and high level in cloudy and rainy area. But to the humidity, variation retrieval is better than statistical regression only in levels above 400hPa.