由於風能為風速之三次方( 3 2 2 1 p = rV pr ),只要有3%的風速誤差就會造成10%的風能誤差,而一般氣象資料若有10%之觀測誤差,則10% 的風速誤差將會造成30% 的風能評估誤差,因此風速之觀測分類預報的準確度對風能評估與應用相當的重要。另一方面,由於風速與風向會依當地之地形、地貌與盛行風之方向而有極大之差異,對四周環海之台灣而言,由於水面與陸地之熱動力差異,近岸地形、地表、海陸風以及中央山脈地形等對風場影響很大。氣象模式之解析度要多高才能解析這些因素之影響。台灣位於歐亞大陸與西太平洋之交界,四面環海,地形複雜,對數值天氣預報而言無疑是一大挑戰,廣大洋面上觀測資料的不足,造成模式初始資料的誤差,再加上物理過程的不確定性及複雜地形可能引發的中小尺度天氣現象等,都會造成數值模式在進行短期天氣預報,特別是對風場預報時產生較大的偏差,因此我們可以嘗試使用系集預報的方法,利用多組不同系集成員的模擬預報實驗,期望能夠降低模式因為初始資料及物理過程的不確定性而造成誤差,以提升短期天氣預報,特別是風能預報的能力,本研究預計以三年為期,協助委辦單位逐步完成一中尺度數值天氣系集預報系統之建立,風力 /風能預報之應用評估以及驗證。本團隊過去多年參與中央大學、師範大學、台灣大學以及中央氣象局共同合作進行之動力模式降雨系集與降尺度預報實驗,對系集預報系統已累積不少之經驗,以這些經驗為基礎,本研究將集中在系集預報系統在風力/風能預報效果之評估以及驗證上,期望能夠經由長期測試評估,發展適用於台灣之風力系集數值預報系統,以及風能評估與預報之應用。 Taiwan is located in the borderline between the Asia continent and the Pacific Ocean. Its unique complex terrain and island characteristics present a great challenge to numerical weather prediction. The lack of observation over the surrounding ocean, the uncertainties on the model physics, and the mesoscale and microscale weather phenomena that result from the modification of synoptic scale flow by complex terrain could significantly downgrade the accuracy of a numerical simulation. Therefore, we will use ensemble forecast, by running many ensemble members with different settings, in an attempt to reduce the aforementioned uncertainties of the model. The purpose of this 3-year project is thus to establish a mesoscale numerical weather prediction network for Taiwan area. It will consist of several ensemble members that are individually maintained by NCU. NCU have the responsibility to produce daily weather forecasts and proceed vilification with surface measurement. These products will provide more guidance and set up an ensemble system for the short-range weather forecast. 研究期間:9901 ~ 9912