Impacts of Data Assimilation on the K-PPM for Hydrometeorological Application: Based on Analysis of Typhoon Cases Landed on the Korean Peninsula during 2012

Abstract

To assess data assimilation effect for the hydrometeorological application, the operational K-PPM and its 3D-Var assimilation experiments based on typhoon cases landed on the Korean Peninsula during 2012 were carried out. Though the model initial time of all preliminary control experiments set at least 36 hours prior to typhoon landing on the Korean Peninsula, it appeared that the typhoon track and intensity forecast generally showed good performance compared with the RSMC best track except for some track differences of Typhoon Khanun and Sanba cases near the Korean Peninsula. However, it was found that there was no significant effect for the K-PPM 3D-Var assimilation to predict the more accurate typhoon track and intensity. It seemed that data assimilation with global (CV3) and regional (CV5) background error covariances often made negative effects if the results of control run were similar to observation. Especially, in Typhoon Sanba, the typhoon tracks predicted by control run shifted eastward, while typhoon tracks predicted by the application of data assimilation shifted westward compared with the best track. In only Typhoon Tembin case, there was positive effects of data assimilation to improve the typhoon forecast. Though the applications of data assimilation were not enough to improve the typhoon forecast, it showed the result of C1_CV3 which was data assimilation experiment with global background error covariances based on control experiment C1, was relatively stable considering the forecast of typhoon track, central pressure, precipitation, and the analysis of RMSE. Three types of observation data among 10 categories of conventional and non-conventional observation data which were used in C1_CV3 experiment were classified with GTS, SATOB, and BOGUS considering the type and distribution of observation data, and the impact of data assimilation and the contribution of each data were analysed. It showed that the impact of data assimilation by only GTS data on the model initial field were most widely dominant in whole domain. Though the change of each initial field assimilated by only SATOB or BOGUS appeared clearly, the impact of them on the initial field assimilated by all observation data were negligible because of quality control and data filtering in the K-PPM 3D-Var assimilation processes. The data assimilation experiments using only typhoon bogussing data made a largest error of typhoon track, intensity, and RMSE, and in the strong typhoon Bolaven and Sanba cases, the results of typhoon track forecast were very similar to the whole data assimilation. It seems the application of a improper typhoon bogussing data may lead to the poor performance of a whole data assimilation. On the other hands, the data assimilation experiments using only GTS data showed the best performance with typhoon track forecast error within 100 ㎞ until 36 hours forecast compared with the best track and position, and the data assimilation experiments using only SATOB data except typhoon Tembin case also showed a good performance. The initial intensity of Typhoon Bolaven and Sanba which were weakened by application of data assimilation using only typhoon bogussing data was forcibly multiplied, and the impact of the data assimilation according to multiplied bogussing data on the sensitivity of a typhoon forecast was analysed. According to multiplied typhoon bogussing, the tracks of typhoon were biased to the west compared with the best track. It seemed that it is because the more strengthened typhoon tends to keep its moving direction at early stage of typhoon forecast. Generally, it showed that the impact of strengthened initial intensity by bogussing was rapidly disappeared within 24 hours forecast. The distributions and concentration areas of precipitation accompanied with typhoon strongly depended on the typhoon track forecast. The distribution of 48 hours accumulated precipitation of Typhoon Sanba by the strengthened typhoon was more similar to observation according to the typhoon track moving to the Korean Peninsula, was the positive effects to manage water resources. Finally, there were a limitation for the improvement of typhoon track and intensity forecast according to simply multiplied bogussing data. Therefore, it is needed to consider both typhoon itself (bogussing) and synoptic field near typhoon (data assimilation) for the accurate typhoon forecast.

Keywords: Data Assimilation, 3D-Var (3 Dimensional Variational Method), K-PPM (K-water Precipitation Prediction Model), Typhoon forecast, Bogussing