Assimilation Application For FY-3C Microwave Humidity Sounder Ⅱ In GRAPES-GFS Model

Data assimilation has been one of the most effective ways to improve the quality of numerical weather prediction.Since 1970s,the implement of satellite data assimilation led to further development of numberical weather forecast.Among numerous satellite observations,the study on the assimilation of satellite data containing water vapor information is particularly critical.It is proved that uncertainties in the mid-level moisture has become one of the main reasons for the insufficient of the forecasts ability,especially for short term forecasts.Due to the low spatial and temporal resolution of vapor information obtained by conventional observations,satellite microwave humidity sounders observations,has become an indispensable information source of water vapor in atmosphere,assimilation of microwave humidity sounders plays a significant role in improving the accuracy of initial field of upper troposphere water vapor.This paper focused on the improvement of the assimilation of the MicroWave Humidity Sounder(MWHS-2)onboard Fengyun-3C(FY-3C)in GRAPES-GFS.First of all,the method of principal component analysis(PCA)is used to eliminate the sacn angle noise in MWHS-2 observations.A five-point smoother is then applied to the first principal component(PC),which effectively removes scan angle noise in the MWHS-2 data.The scan-angle-dependent bias from the reconstructed MWHS data becomes more stable.In the process of quality control for MWHS-2,by analyzing the characteristics of the impacts of cloud parameters on brightness temperatures of different MWHS-2 channels in a single field of view using CRTM,a new index for cloud detection over land regions is developed.This new cloud detection method can remove most of the radiances contaminated by clouds,it has better capability for identifying supercooled water clouds,opaque ice clouds and overlapping than cirrus and water clouds with low cloud heights.The detectable rates of supercooled water clouds,opaque ice clouds and overlapping are up to 80%.The probability distribution of O-B is more consistent with Gaussian distribution for data on clear sky.This algorithm can detect cloudy radiances using MWHS-2 itself and has a good application prospect.Secondly,a new bias correction scheme is proposed in this study according to the scheme of Harris and Kelly.This new scheme is based on the features of MWHS-2 observations and a stepwise regression method is used in air-mass correction to select the best predictors for MWHS-2.Compared with the original scheme,surface temperature and total-column precipitable water are introduced,which makes the probability distribution of O-Bs more consistent with normal distribution after adopting new bias correction scheme.At last,the new cloud detection and bias correction schemes are applied to GRAPES-GFS to explore the influence of MWHS-2 observations on data assimilation and numerical weather prediction.The results show that MWHS-2 channel 11 and 12 observations have positive effects on the analysis and prediction of humidity and temperature field.The combination of new cloud detection and bias correction scheme can further improve the accuracy of initial field and prediction of 500hPa potential height,humidity and temperature.