为了客观评价Wind Energy Resource Assessment System /CMA(简称WERAS/CMA)系统(CTL方案)和将其中的客观分析法改成四维同化系统(简称FDDA方案)对既受狭管效应影响、又受湖陆风影响的阿拉山口和达坂城-小草湖风区起伏下垫面中的风能资源数值模拟的优劣,根据2009年7、10月和2010年1、4月 12UTC的NCEP再分析资料以及同期CMACAST下发的WMO各种常规观测资料开展了风场预报效果对比实验。结果表明:(1)对复杂区域而言,两种方案比过去单纯只用中尺度模式进行风场模拟的平均相对误差至少减小10%;(2)总体而言,两种方案对70 m高度处的风速模拟误差要大于30、50、100 m处的误差,在受多种环流尺度影响区域,模式在刻画平均风速/风向频率廓线方面的缺陷均极其相似;(3)在70 m高度上,两种方案5 m·s^-1以内的风速平均相对误差可达60%～130%,>5 m·s^-1的误差可控制在15%以内;对受湖陆风影响区域的模拟误差明显偏大,误差大小与湖陆风效应的季节变化有关; (4)两种方案均能抓住70 m左右高度上不同等级风速段的气候背景,对达坂城风区5～15 m·s^-1风速段的Ts预报评分可达0.6～0.7,对阿拉山口和小草湖风区≤5 m·s^-1风速段的Ts预报评分分别可达0.6～0.7和0.9左右。然而,对达坂城风区≤5 m·s^-1风速段的Ts预报评分仅0.3～0.4;(5)两种方案对所有风区需采取停机保护措施的、15 m·s^-1以上强风预报的Ts评分仅在0.4～0.6;(6)同一测风塔不同高度上,FDDA方案对风的预报效果不一定总优于CTL方案,但在70 m高度上,FDDA总体略优于CTL;即使同一风区,各个测风塔之间两种方案的预报效果也是因局地多尺度环流影响的不同或因预报的高度不同或预报季节的不同而异,这种预报误差差异的机理还有待探究。

To objectively evaluate the abilities of simulating/forecasting wind energy resource by the method of objective analysis in WERAS/CMA system (hereafter referred to as CTL scheme) and objective analysis replaced by four-dimensional data assimilation (FDDA) at complex land surface patter and rough terrain, such as Alashankou and Dabancheng-Xiaocaohu wind area which is influenced by both the effect of narrow pipe and lake-land breeze, the two schemes are analyzed and contrasted each other based on 12UTC NCEP reanalysis data and WMO conventional observation data issued by CMACAST during 2009 (July and October) and 2010 (January and April). The results indicate that: (1) For complex region, both of them are all reliable in detailed assessing and forecasting wind energy resources showing that the average relative error at least reduced by 10% compared to the precious method which utilized meso-scale model only. (2) But on the whole, they don't only have a very similar defect in describing the profile characteristic of average wind speed/wind direction frequency at the areas under the control of multiple micro-scale circulation, but also show greater error at 70 m than that at 30, 50 m and 100 m. (3) At about 70 m level, the predictive mean relative error for the ≤5 m·s^-1 is up to 60%-130%, for the >5 m·s^-1 is less than 15%. And the error is greater apparently at the areas of lake-land breeze effect and it is concerned with the changing of season. (4) At about 70 m, all of them can simulate climate background of different wind speed threshold value well, exhibiting that the threat scores for the 5-15 m·s^-1 intervals is up to 0.6-0.7 at Dabancheng wind area, and for the ≤5 m·s^-1is 0.6-0.7 at Alashankou and 0.9 or so at Xiaocaohu wind area. However, the threat scores for the ≤5 m·s^-1 interval is only 0.3-0.4 at Dabancheng wind area. (5) For all wind regions, the threat scores of the two schemes at about 70m are only 0.4-0.6 in predicting strong wind of above 15 m·s^-1 when the wind turbine generator need to stop or take some safeguard procedures. (6) Under the control of various local multi-scale circulations, for different heights of the same mask, the FDDA scheme does not seem to always provide better forecasting effect than the CTL scheme except at about 70 m height at which FDDA scheme is slightly better the CTL, even at the same wind area or different seasons, the forecasting effects are different from each other among various masks. So the physical mechanism why exists this kind of strange feature should be further illustrate in the future.