投落送與QuikSCAT資料同化對WRF模擬之影響

An Impact Study of Dropsonde and QuikSCAT Data on WRF Simulations

簡芳菁(Fang-Ching Chien) ； 謝章眉(Chang-Mei Hsieh)

《大氣科學》 36卷3期 (2008/09) Pp. 217-247

摘要

本文使用WRF模式三維變分同化方法，針對2006年碧利斯颱風、凱米颱風及6月中旬梅雨鋒面三個個案，將投落送資料、QuikSCAT風場資料及傳統觀測資料同化於WRF中尺度模式中，藉以了解同化投落送與QuikSCAT資料對模式初始場及後續數值模擬的影響，並透過敏感度實驗來進行影響探討。
結果顯示，碧利斯颱風在加入投落送資料與QuikSCAT資料後，對模擬6至72小時的平均路徑誤差除分別有52 km及16 km的改進外，對雷達回波強度之模擬也有所改善，而凱米颱風在加入上述兩種資料後，對模擬路徑誤差之整體改進則分別達到71 km及42 km；6月中旬梅雨鋒面個案同化該兩種資料後之雷達回波模擬強度也有所改善，而與觀測較爲接近。
從影響實驗研究發現，同化投落送及QuikSCAT資料後，碧利斯颱風個案於模擬後期具有較強的南風分量，使其模擬路徑不至於像未同化該兩種資料者過於偏南；凱米颱風個案同化投落送資料後於模擬後期具有較弱的南風分量，而同化QuikSCAT資料後則具有較強的南風分量，使其模擬路徑不至於像未同化投落送及QuikSCAT資料者過於偏北及偏南，而與觀測路徑較爲接近。雷達回波強度方面，碧利斯颱風及6月中旬梅雨鋒面個案之模擬前期同化該兩種資料後，有較大的水氣混合比，因爲有較大的溼度及較多雲水含量的緣故，使其模擬有較強的最大回波值，與觀測較爲接近。

關鍵字

投落送 ； QuikSCAT ； WRF模式三維變分同化

並列摘要

This study uses the Weather Research and Forecasting (WRF) model and its variational data assimilation (WRF-Var) system to examine the impact of dropsonde and QuikSCAT data on numerical simulations. Three cases, including Typhoon Bilis (2006), Typhoon Kaemi (2006), and a Mei-yu rainfall event, are tested. These cases are chosen because there were special data collections of dropsonde observations.
The results show that the simulated track of Bilis has about 39% and 16% improvement averaged during the 72 h forecasts after the assimilation of dropsonde and QuikSCAT data, respectively. The simulation of radar reflectivity is also improved. Similarly, the simulated track of Kaemi is improved by 26% and 21% by assimilating dropsonde and QuikSCAT data, respectively. However, the data assimilation appears to have no improvement in terms of radar reflectivity simulation for the Kaemi case. As for the mid-June 2006 Mei-yu case, the assimilation of both kinds of data can help the simulation of radar reflectivity.
Further investigations show that the assimilation of dropsonde and QuikSCAT data produces stronger southerly environmental winds in the Bilis case, resulting in a better simulated track than the runs without these data. In these runs, the simulated tracks are too far to the south compared with the observed. In the Kaemi case, the tracks of simulations without dropsonde/QuikSCAT data assimilation were shifted to the north/south of the observed track, respectively. With the assimilation of dropsonde/QuikSCAT data, the simulated environmental flow becomes more northerly/southerly, resulting in better tracks. The data assimilation also helps to produce better water vapor mixing ratio in the simulations of the Bilis and Mei-yu cases. As a result, the radar reflectivity is better captured in the simulations.

並列關鍵字

Dropsonde ； QuikSCAT ； WRF-Var