MM5、WRF-RUC及T639模式对山东沿海风力预报分级检验

摘要

In order to objectively evaluate the wind forecast ability of different models in Shandong coastal areas,a wind force forecast verification method,based on the precipitation grading forecast verification,is defined to test the daily strongest wind speed forecast by MM5,WRF-RUC and T639 models in totally 9 coastal areas of Shandong province.Results show that all the models overestimate the weak wind while underestimate the strong one; the T639 model's wind forecast is the lightest so as to be better for wind force scale below 4,but can hardly forecast for wind force scale above 8.MM5 and WRF-RUC model's wind forecasts are obviously better than the T639 model's for the wind force scale above 4,and WRF-RUC's forecast ability is the best.However,all the models display a similar weakness:underestimating stronger wind.The consistency verification of all the models' analysis field and the 24 h wind forecast with respect to the observation shows that the MM5 and WRF-RUC's wind forecasts are nearly the same as the observation for the wind force scale below 5,while for the wind over force 6,forecasts can be either stronger or weaker than the observation.All in all,the WRF-RUC model' wind forecast is the best in the three models for coastal wind forecast,while the MM5 model is a little unstable in the daily maximum wind speed forecast.%为客观评价不同的数值模式对山东沿海风的预报性能,结合中国气象局降水分级预报评分办法,定义了一种风力预报分级检验办法.对MM5、WRF-RUC和T639模式在山东沿海9个精细化海区代表站的日最大风速预报进行了检验,结果发现:各模式普遍存在对于小风天气预报偏大、大风天气预报偏小的特点.T639模式风力预报偏弱,因此,对于4级以下的风预报评分较高,而对于8级以上大风几乎没有预报能力.MM5和WRF-RUC模式对于4级以上的较强风力的预报结果明显好于T639模式,其中WRF-RUC模式预报准确率稍高于MM5模式,但风力越大,各模式均漏报越多.各模式分析场以及24 h风力预报与实况的一致性检验表明:5级以下的风力,MM5和WRF模式预报风力与实况基本为一致,但对于6级以上的大风,MM5模式预报较分散,WRF模式预报更接近实况风力.综合各模式对于风力预报的平均绝对误差,WRF-RUC模式预报误差最小,具有较高的参考价值.MM5模式预报准确率稍低于WRF-RUC模式,且存在一定的不稳定性.