區域系集預報系統研究：系集成員產生方式之評估

The Study of Regional Ensemble Forecast: Evaluation for the Performance of Perturbed Methods

李志昕(Chih-Hsin Li) ； 洪景山(Jing-Shan Hong)

《大氣科學》 42卷2期 (2014/09) Pp. 153-179

摘要

如何產生離散合理之系集預報成員，以提供足夠的預報離散，此為建立系集預報系統之首要工作。本研究將針對模式物理參數法擾動和初始條件擾動進行實驗，並分析其預報結果，找出最佳之系集成員產生方式，以建構有效的區域系集預報系統。本研究使用4種不同之系集成員產生方法進行WRF模式預報實驗。第一組實驗參考不同之模式積雲參數法、微物理參數法、邊界層參數法以及土壤模式等，產生40組以物理參數化擾動為主的系集成員進行數值模式預報實驗；第二組實驗利用不同之資料同化策略，產生10組以初始場擾動為主之系集成員進行預報實驗；第三組實驗使用WRF三維變分法之隨機擾動，產生32組系集預報成員；第四組實驗，結合第一組和第三組實驗，使用WRF三維變分法產生之初始場擾動，再搭配物理參數法擾動，產生18組系集預報成員。透過校驗分析顯示，第四組實驗在初始場以及模式預報場皆有擾動產生，使用此一方法製造之系集預報成員，能產生最大之預報離散，且於預報可信度和預報能力皆最佳，此為本研究建議之最佳系集成員產生方式。

關鍵字

系集預報系統 ； 預報離散

並列摘要

In order to represent the forecast uncertainties, the spread-enough ensemble products from an ensemble prediction system (EPS) is of most important. The goal of this paper is to understand the role of the physics perturbation and initial perturbation to configure a robust EPS. Four different one-month regional forecast experiments based on WRF model by using different perturbations were conducted in the paper. The first experiment with 40 members used physical perturbations, including the cumulus, microphysics, planetary boundary layer parameterization schemes, and land-surface models. The second experiment with 10 members use different data assimilation strategy. The 32 members of the third experiment were generated by adding random perturbations in the initial conditions. The last one combines the physical perturbations and random initial perturbations to get the 18 members. The results show the forth experiment have the best performance of spread and forecast skill due to the better represented perturbations both in the initial condition and through the model prediction.

並列關鍵字

Ensemble forecast system ； Spread