Research In Online Bias Correction Of Polar Orbit Satellite Observations In GRAPES-3DVar

Bias correction is the key technology of data assimilation for satellite radiance data. Currently, global GRAPES variational assimilation system considers scan bias correction and air-mass bias correction based on static bias correction scheme. But when deviation property changes (such as the degradation of forecast model, instruments aging and data drifting and so on), the correction effect will be degraded, the effect of assimilation analysis will be worse.Therefore, in order to solve the problem, this paper plans to put forward a scheme called online bias correction based on Harris and Kelly’s TOVS radiation error correction scheme and the concept of online bias correction which is put forward by foreigners in NWP system. The scheme considers the character of GRAPES system and instrument characteristics of National Satellite Meteorological Center. It will keep good effect of deviation correction in the process of long time., real-time circulation assimilation analysis. This is the first time that online bias correction scheme are applied to GRAPES. The online bias correction is an adaptive bias correction system, it uses variational method to adjust the coefficients of predictors.To test the effect of the new scheme, some assimilation experiments which last two months are carried out. The results show that, compared with the effect of degraded static bias correction, the online bias correction scheme is effective, it can optimize the degenerate bias equation automatically and quickly, and keep the effect of bias correction stable; compared with the effect of static bias correction (local optimal correction), we can say the effect of online bias correction scheme better than the effect of the local optimal correction for some areas. The results are encouraging,On the whole, the results of two months assimilation tests are satisfactory. Deviation of each channel is largely reduced, and analysis quality has also improved. We can give more accurate prediction of initial value. It is a good foundation for online bias correction scheme applied in GRAPES.