Influence Of Radar Data Assimilation On The Simulated Typhoon Path With Non-Homogeneous Distribution Of Data

For the limitation of actual radar number and locations and the uncertainty of typhoon track, it is difficult for the Doppler radars to capture the complete wind fields of typhoon. Consequently, only partial wind information of typhoon can be obtained and used in the numerical simulation. Under the condition, incomplete observations on the wind structure of typhoon will make much difference to the forecasts of typhoon track. In order to understand the influence of non-homogeneous distribution of radar data on the typhoon path prediction and try to solve this problem, the Typhoon Matsa is considered as the analysis case in this paper. WRF (Weather Research Forecast) model and ARPS (Advanced Regional Prediction System) model are employed to provide simulated data for producing radar data and the background fields for ARPS 3DVAR system, respectively. Using the simulated 9 radar data, a set of assimilation and forecast experiments are conducted with ARPS 3DVAR system and ARPS model to understand the impact of non-homogeneous distribution of radar data on the typhoon path prediction. On this basis, an approach is proposed in which the limited observation information is extended to fill the non-observed area. The corresponding numerical experiments are performed to test the feasibility and effectively of the approach. The main conclusions are as follows:(1) Before the typhoon landfall, the simulated path with WRF model and ARPS model is close to the best track. After landfall, the difference between the best track and simulated paths with two models increases with the forecast length. Compared with the real, the differences of typhoon path and the maximum wind speed near the typhoon center with WRF mode simulation are less than the results with ARPS model.(2) Doppler radar data assimilation has a positive effect on the typhoon path forecast. Radar radial wind data assimilation adjusts the wind field structure obviously. And compared the background field, it is closer to the "real" field; Radar reflectivity data can adjust the water vapor condition through the clouds analysis procedure. And the forecasting results on the typhoon path prediction with continuous multi-time radar data assimilation are better than that with single radar data assimilations.(3) Radar data assimilation under non-homogeneous distribution of radar data has an obvious impact on the typhoon path forecasts. Under the homogeneous distribution (that is to say, data can fully cover the typhoon flow area), the assimilation and forecast results is best. However, when incomplete radar observation is used in data assimilation, the forecast results of typhoon path are not the same, or even quite different.(4) Based on the 3DVAR analysis, an approach is proposed to use typhoon wind field model to fill the no-observed regional data. Preliminary numerical experiment results show that the method can improve the forecast skill of typhoon path to some extent.