A Study Of Assimilating AMSU-A Radiance With Ensemble Square Root Filter

The numerical weather prediction has become one of the important means for weather forecast. Assimilating satellite radiance data with a reasonable method can provide accurate initial states for the numerical weather models and thus improve the accuracy of numerical weather prediction. In this paper, the possibility of improving the mesoscale model initial states is investigated under completely two different weather conditions by assimilating NOAA satellite AMSU-A channel5microwave radiance data into the Weather Research and Forecast (WRF) model which is driven by coarser spatial resolution NCEP reanalysis data with the Ensemble Square Root Filter (EnSRF). Furthermore the impact of the horizontal covariance localization on assimilation is also examined. The main contents are as follows:Firstly, by using an observation system simulation experiment (OSSE), five experiments are designed respectively under the sunny weather condition in jianghuai region, including one experiment without the horizontal covariance localization and four experiments with different localization radius. The results show that: under the sunny weather condition, the mesoscale analysis states including the temperature, humidity and wind are closer to the real states compared with the background states, especially at the levels with the peak energy contribution to the observed radiance for the channel5of AMSU-A; the effect of assimilation is different among meteorological elements, of which the temperature is improved most; among the five experiments, the assimilation results with the horizontal covariance localization are significantly better than the results without it, especially when the localization radius is400km.Secondly, an ideal experiment (i.e., OSSE) and an actual experiment are carried out under the rainy weather condition in the same region. Set up of the ideal experiment is similar to the OSSE in the experiment on the sunny day and the results showed that:the assimilation also has positive contribution to the four states in particular vertical model levels; and there’s evident difference between four model states, overall, the improvement of assimilation on the rainy day is less significant than that on the sunny day. The assimilation effect of horizontal covariance localization can not be generalized, and it can have positive impact on the temperature, humidity, wind only when the radius is reasonable.Finally, the results of actual experiment under rainy weather condition show that: the assimilation of AMSU-A microwave radiance data have a certain influence on model states, it can enrich the initial states information of temperature, humidity, wind relatively; and the6-h and24-h WRF forecasts with the updated model states can describe the intensity and range of the precipitation more accurately to a certain extent compared with the forecasts using original background states, however, the assimilation results are not good enough in some area, indicating that satellite data assimilation under precipitation remains to be further studied.