颱風路徑數值預報模式初始場處理：兩種方法對模式預測誤差之比較

The Initialization Process of the Numerical Typhoon Track Forecasting Model: Comparisons of the Forecast Errors from Two Bogus Methods

葉天降(T. C. Yeh) ； 陳得松(D. S. Chen) ； 黃康寧(K. N. Huang) ； 彭順台(M. S. Peng) ； 謝信良(S. L. Shieh)

《大氣科學》 27卷2期 (1999/06) Pp. 171-189

摘要

由於觀測資料不夠完整，現今颱風路徑數值預報作業模式仍需引用虛擬處理方式，使模式有比較佳之初始場。本文以1996年10個颱風中之52個案，透過中央氣象局颱風路徑預報模式之模擬，比較兩種颱風初始處理方法對模式路徑預測誤差之影響。而此兩種初始處理方法中，一種是1996年中央氣象局當時作業採用之方法，另一種則是所要測試之新方法，此方法包含引用類似於Kurihara et al.（1993）之渦旋濾波技術、加入由相當正壓所模擬由地球自轉效應造成之非對稱渦旋、以及容許觀測資料透過分析過程再修改所加之渦旋。測試之結果顯示，不同之初始處理方法造成模式路徑預測結果有顯著的差異，引用新的初始處理方法使51個24小時路徑預測之平均誤差減小12公里，而使47個48小時路徑預測之平均誤差減小148公里。進一步之分析顯示，引用新的初始處理方法後，對部份移動速度較持續的颱風，其24小時預測誤差有增大的現象。

關鍵字

颱風路徑預報 ； 模式初始場處理

並列摘要

For lack of complete observations, a special vortex bogus procedure is usually employed for most of the operational forecast model to begin the typhoon track forecast. This study compares the impact of two vortex bogus procedures to the model track forecast errors. Two methods studied are the operational system used at the Central Weather Bureau in 1996 and a new vortex bogus procedure. The new vortex bogus procedure includes a vortex filtering scheme similar to Kurihara et al. (1993), implanting a wave number one asymmetric beta gyre generated by an equivalent barotropical model, and allowing the observations to modify the bogused vortex through the objective analysis. From the analysis of 52 simulations by using the Typhoon Track Forecasting Model/Central Weather Bureau, we found the model track forecast is significantly affected by the vortex bogus procedure. Applying the new vortex bogus procedures reduce the mean 24 hours track forecast error by 12 km, and reduce the mean 48 hours track forecast error by 148 km. We also found that for some cases, which the CLIPER (CLImatology-PERsistence) forecast errors are small, the 24 hours forecast error increased when the new vortex bogus procedure is applied.

並列關鍵字

Typhoon track forecasting ； Model initialization process