Research On An Ensemble Kalman Filter Data Assimilation System With A Global Spectral Model T106L19 At A Medium-Range Resolution

There have been many significant results and greater progress both in theory and in practical applications since Ensemble Kalman Filter (EnKF) was applied to atmospheric data assimilation in the late 1990's. There has been more than two decades that research on Numerical Weather Prediction at a medium-range resolution in our country, while data assimilation at a medium-range resolution is being researching at the same time. At present, the data assimilation scheme is 3D-Var (3-Dimentional Variational). However, there is in the developing stage that EnKF scheme is used at a medium-range resolution. In this paper, the applicability and feasibility of EnKF in real atmospheric model-the global medium range spectral model T106L19 are studied by assimilating real conventional observational data and real ATOVS (Advanced Television and Infrared Observation Satellite Operational Vertical Sounder) satellite data. And a reasonable EnKF data assimilation system with a global spectral model at a medium-range resolution is set up. The main results are as follows:1. An EnKF data assimilation system is set up, which assimilates conventional observational data and ATOVS satellite data with a global spectral model at a medium-range resolution.2. For conventional data assimilation, EnKF makes the analysis field close to observations, while errors still exist, which are mainly in the near of trough and ridge and more obvious in the southern hemisphere due to few conventional observations. Compared the effect of T106L19-EnKF assimilation/forecast with that of T106L19-OI (There is original in the model T106L19), the results show as following: (1) the errors of analysis field of T106L19-EnKF are smaller than those of T106L19-OI in the middle and high latitude region of northern hemisphere (NH) and tropical and subtropical region (TR), while they are almost same in the middle and high latitude region of southern hemisphere (SH). (2) The errors of forecast field increase with the time of the forecast growth, and the errors increase more rapidly in region of SH and NH than in region of TR. (3) In region of TR and NH, the results of T106L19-EnKF forecast are better than those of T106L19-OI, however, such superiority is not founded in region of SH.3. Based on the characteristic of ATOVS radiance data and the existing bias correction methods of TOVS and ATOVS, a bias correction system for model T106L19 is set up for bias correction and quality control of ATOVS data, which are done before satellite data assimilation. The distribution of observation residual is more similar to Gaussian distribution than that without bias correction and its peak is shifted to zero after bias correction.4. There is a significant improvement for the analysis field, not only in the southern hemisphere that conventional data is sparse but also in the northern hemisphere that conventional data is relative abundance after assimilating ATOVS with bias correction and quality control based on conventional data assimilation. The results of assimilation show that T106L19-EnKF are better than T106L19-OI in majority of the global. For results of forecast, T106L19-EnKF are better than T106L19-OI in region of TR and NH, however, there is no superiority in the first three days in region of SH.5. To investigate the effect of T106L19-EnKF assimilation/forecast further, T106L19-EnKF assimilation is used to the Typhoon "Prapiroon" track prediction. For the Typhoon "Prapiroon" track prediction, four test are done, including T106L19-EnKF or T106L19-OI assimilating conventional data only, and T106L19-OI or T106L19-EnKF assimilating conventional data and ATOVS satellite data. The results show that the Typhoon track is predicted by the analysis field of T106L19-EnKF that assimilate conventional data and ATOVS data as the initial fields matches best with the observed best track and its circulation at 500 hPa matches better with "the real field".