透過您的圖書館登入
IP:18.119.104.238
  • 期刊

臺灣地區WRF颱風系集降雨機率預報之評估、校正與經濟價值分析-第一部分:預報評估

Evaluation, Calibration and Economic Value Analysis of the Probabilistic Precipitation Forecasts from WRF Ensemble Prediction System in Taiwan Area- Part 1: Forecast Evaluation

摘要


本研究主要是針對臺灣地區的颱風,利用中央氣象局發展的WRF系集預報系統(WEPS),進行定量降雨機率預報;最終目的在於提供更具參考價值的系集颱風降雨預報產品,並讓使用者藉由系集機率預報做出最佳化的決策,以得到最大的經濟價值。研究成果擬分述於三篇論文中,包括系集降雨機率預報之評估、校正與經濟價值分析。本論文(第一部分-預報評估)的研究重點在於評估WEPS的品質良窳,並利用WEPS產生颱風降雨機率預報,詳細校驗其預報表現。研究結果顯示:WEPS的系集散度可良好地反應預報的不確定性。颱風降雨機率預報具有相當好的區辨能力,但有明顯的預報偏差,可信度和區辨能力均隨著降雨門檻增加而降低。不同區域的校驗結果顯示:颱風降雨預報在陸地區域的可信度與區辨能力均優於海洋區域,這與做為真實場的雷達降雨估計特性有關。此外,平地的可信度優於山區,但山區的區辨能力優於平地。整體而言,山區的預報技術優於平地,這是因為地形鎖定效應造成颱風降雨在迎風斜坡上的可預報度高於平地,因而WEPS在山區的預報技術和區辨能力相對較佳。Mann-Whitney檢定顯示山區的預報能力優於平地具有統計上的顯著性。

並列摘要


This study aims to develop the probabilistic quantitative precipitation forecasts (PQPFs) associated with typhoons by utilizing the WRF ensemble prediction system (WEPS) developed by Central Weather Bureau (CWB). The ultimate goal is to provide more valuable precipitation forecast products associated with typhoons based on the WEPS and help users optimize their decision making and obtain the maximum economic value by using the PQPFs. The research results will be displayed in three articles, including evaluation, calibration, and economic value analysis. In this paper (part I - Evaluation of PQPFs), we focused on the evaluation of the quality of WEPS, generated PQPFs from WEPS, and verified the performance of PQPFs in details. The ensemble spread can well represent the forecast uncertainties. The PQPFs have good discriminating ability but also obvious biases. The reliability and discrimination decrease with increasing precipitation thresholds. Verification results from different areas show that the reliability and discrimination over land areas surpass those over adjacent sea areas. This is related to the characteristic of radar quantitative precipitation estimation (QPE) used as ground truth. In addition, the reliability over plain areas is better than that over mountain areas, but the discrimination over mountain areas excels that over plain areas. Taken as a whole, the forecast skill over mountain areas surpasses that over plain areas. This is because the terrain-locking effect leads to higher predictability of typhoon rainfall over windward slopes than over plain areas. Therefore, the discrimination and forecast skill of WEPS over mountain areas are better. The non-parametric Mann-Whitney test indicated that the greater forecast ability over mountain areas than over plain areas is statistically significant.

延伸閱讀