WRF模式对新疆中部地面总辐射预报性能的检验  

Forecast Examination of Surface Global Horizontal Irradiance over Middle of Xinjiang Using WRF Model

作　　者：辛渝[1] 王澄海[2] 沈元芳[3] 陈洪武[1] 

机构地区：[1]新疆气候中心,新疆乌鲁木齐830002 [2]兰州大学大气科学学院/半干旱气候变化教育部重点实验室,甘肃兰州730000 [3]中国气象科学研究院灾害天气国家重点实验室,北京100081

出　　处：《高原气象》2013年第5期1368-1381,共14页Plateau Meteorology

基　　金：科技部2010年度公益性行业(气象)科研专项(GYHY201006036);中国气象局2010年度工作任务"太阳能光伏发电预报系统研究"(气预函[76]号)

摘　　要：以2009年1，4，7，10和12月全球GFS预报场资料为初值场，对新疆达坂城一小草湖风区进行风能预报建模的同时，也对周边辐射气候各异的吐鲁番、乌鲁木齐和焉耆站的辐射预报效果进行了检验。结果表明：（1）大多数情况下，模式中的逐时向下短波曝辐量能预报出各地总辐射的日变化趋势，两者的平均绝对误差百分率存在上午大、午后小。以及日出时刻的平均绝对误差百分率一般是日落时刻的3倍～4倍以上的特点。（2）当吐鲁番、乌鲁木齐站的日总辐射与日天文总辐射比〉0．4时，预报的日平均绝对误差百分率在25％以内；而〈0．4时，模式预报便无参考价值。焉耆站辐射预报的偏差大小对这一比值却不敏感。（3）对于吐鲁番站而言，10月预报效果最佳，4月次之；对于乌鲁木齐、焉耆站而言，则是4月预报效果最佳，10月次之；各地冬季月份的预报偏差均最大，主要与冬季的散射辐射与总辐射之比为一年最大和模式对冬季冰雪效应的预报不足有关，全年不宜采用同一套参数化方案或同一套统计订正方程进行辐射预报。（4）辐射预报场偏差的季节变化特征和日变化特征与风场预报的偏差特征一致，两者紧密相关．Based on the historical weather forecasts adopting WRF model driven by GFS initial field in Jan- uary, April, July, October and December of 2009 at Dabancheng Xiaocaohu wind farm where wind power prediction model was established according to the combination of dynamic and statistical methods, the fore- cast accuracy of downward short wave radiant exposure （SWREDOWN） replacing global horizontal irradi- ance （Eg） was also examined at Turpan, Urumqi and Yanqi meteorological stations. The results are as fol- lows： （1） Mostly, the hourly SWREDOWN in WRF model can display the variation tendency of E on surface. The mean absolute percentage error （MAPE） of hourly Eg prediction is smaller in afternoon, second in morning, and the bias at sunrise is more than 3~4 times larger as compared to sunset. （2） At Turpan and Urumqi station, the MAPE of daily Eg prediction is less than 25~ on the condition that the ratio of the Eg on surface to the horizontal extraterrestrial irradiance is more than 0. 4, otherwise the prediction is worthless, while it is not sensitive to the ratio at Yanqi station. （3） At Turpan station, the best forecast effect of Eg is in October, second in April, while at Urumqi and Yanqi stations, the best is in April, sec- ond in October, and the worst are all in winter months, because during the period there exist the biggest ratio of scatter irradiance to Eg on surface through a year and shortage of predicting ice and snow effect in model, so it is improper to adopt a same set of parameterization scheme or MOS equation on predicting Eg. （4） There is some closed co-relationship between the irradiance and wind field because their seasonal and diurnal bias variation characteristics are similar.

关 键 词：WRF模式 总辐射预报 新疆中部 

分 类 号：P422.1[天文地球—大气科学及气象学]