The summer weather in the East Asia is largely affected by location and activity of the Northern Pacific High (NPH). Consequently, the proper satellite data in numerical weather prediction (NWP) analysis over the western Pacific Ocean is necessary to improve prediction of location of the NPH. However, it is very important to get an accurate observation error making the analysis be optimal in using a satellite data. Traditionally, the satellite observation errors are fixed and do not depend on the atmospheric condition. Moreover, the error caused by radiative transfer model is not considered in general.

In this study, the global 3DVAR system of Korea Meteorological Administration (KMA) is used to assimilate ATOVS data. The satellite observation error depending on the innovation and background error in the radiance space is calculated in order to make up the defect of the observation error treatment. As the ATOVS data is used only over the ocean, background error is sampled over the ocean and converted into radiance space reference to the TIGR profile. In the study, the 3DVAR result with direct assimilation of ATOVS information is compared with the result without ATOVS data, and the calculated observation error in the 3DVAR system are investigated. The experiment is conducted for the summer 2003. The impact of the ATOVS data to the location of the NPH is primarily focused on. As well, the impact to rain band location formed by the frontal system and Typhoon track is also investigated