發展弱海氣偶合資料同化系統改進颱風預報

Abstract

本研究利用一套區域海-氣-波浪全面偶合模式建立一個ensemble optimal interpolation (EnOI)的弱偶合資料同化系統來預報颱風。目前此系統同化Argo和NTU浮標的溫鹽資料以及衛星海表溫來更新海洋模式裡的溫鹽結構，另用更新後更接近真實情況的溫鹽狀態透過區域模式內的偶合機制改變大氣和波浪模式的模擬。 本研究針對2016年強烈颱風尼伯特(Typhoon Nepartak)經過西北太平洋的個案設計一系列的6-hour-cycled偶合模式模擬探討如何利用弱偶合資料同化系統改進颱風預報。模擬結果顯示同化Argo的溫鹽資料能改善整個此表層以下的海洋溫鹽結構。同時同化NTU1和NTU2浮標的溫鹽資料能有效改善上層海洋的溫鹽狀態、修正附近海表溫的值，因此當颱風接近浮標時便能改善其強度預報。同化衛星資料可以很好的修正海表溫的分布以及尼伯特颱風的強度，缺點是數據的覆蓋情形會變動。而尼伯特的結構和大小則是在同化所有觀測資料後模擬的最好，因為同時有更好的海表溫分布以及更正確的上層海洋溫鹽數值。除了改進六小時預報(同6-hour-cycled simulation)，此弱偶合資料同化系統在經過7 DA cycle後亦可改善24小時颱風強度預報。總結以上，我們所建立的弱偶合資料同化系統在消耗較少量計算資源的情況下就能改進颱風的強度預報。

A weakly coupled data assimilation (DA) system based on the ensemble optimal interpolation (EnOI) is developed within a regional air-sea-wave-coupled model for the typhoon forecast. This system assimilates ocean temperature and salinity profiles from Argo, NTU buoys, and satellite Sea Surface Temperature (SST) in the ocean model component to update the ocean thermal structure. The updated ocean states can then feedback to atmosphere and wave models through the coupler. The 6-hour-cycled DA experiments demonstrate how the weakly coupled DA system improved the forecast of super typhoon Nepartak (2016) in west North Pacific. The results show that assimilating only Argo profiles can correct the subsurface ocean thermal structure. Assimilating both NTU1 and NTU2 buoy’s data can better correct the upper 500m thermal structure locally so as the typhoon’s intensity when the typhoon approached the buoys. Assimilating only satellite SST can well correct the horizontal SST distribution so as the typhoon’s intensity. The typhoon’s predicted structure and size are also improved after all the available observations are assimilated owing to the better SST pattern and upper ocean thermal structure at the same time. Besides 6-hour forecast, the 24-hour forecast of the typhoon Nepartak’s intensity is also improved after 7 DA cycle. To sum up, the weakly coupled DA system can improve the forecast of typhoon Nepartak’s intensity and even its size cost-effectively.