西北地区地表太阳总辐射计算模型适用性评价

Evaluation on applicability of global solar radiation calculation model in Northwest China

为有效提高中国西北地区Rs预报精度,选取西北地区11个代表性气象站点1959—2015年逐日气象数据,评价了4种基于日照时数的Rs计算模型(Angstrom-Prescott,Ogelman,Bahel和Louche模型)和2种基于温度的Rs计算模型(Hargreaves和Bristow-Campbell模型)在西北地区4个分区(温带大陆性高温干旱区、温带大陆性干旱区、高原大陆性半干旱区和温带季风半干旱区)的适用性.结果表明:6种模型在西北地区的Rs模拟值与实测值均在 P< 0.001水平具有统计学意义;基于日照时数的 Rs计算模型(R^2介于0.901~0.903)精度高于基于温度的模型(R^2介于0.695~0.719);其中,基于日照时数的模型中Bahel模型的精度最高,其R^2, MAE,MRE,RMSE和NSE 分别为0.903,1.624 MJ/(m^2 ·d),15.7%,2.298 MJ/(m^2 ·d)和0.902;基于温度的模型中Bristow- Campbell模型精度最高,其值分别为0.719,2.851 MJ/(m^2 ·d),30.7%,3.959 MJ/(m^2 ·d)和0.713.因此,为有效提高西北地区Rs日值和月值预报精度,在仅有温度资料时推荐使用Bristow-Campbell模型,在仅有日照时数资料时推荐使用Bahel模型.

To effectively improve the prediction accuracy of Rs in Northwest China, the daily climate data collected from 11 representive meteorological stations during 1959-2015 were used to estimate Rs . Four kinds of sunshine-based models (Angstrom-Prescott, Ogelman, Bahel and Louche models) and two kinds of temperature-based models (Hargreaves and Bristow-Campbell models) were evaluated in four sub-zones, which are the temperate continental high temperature-arid zone, the temperate continental arid zone, the plateau continental semiarid zone and the temperate monsoon semiarid zone. The results show that the estimate R s results of each model has a significant correlation with the measured value at the 0.001 level. Generally, the applicability of sunshine-based model (with R^2 ranging from 0.901 to 0.903) is better than that of temperature-based model (with R^2 ranging from 0.695 to 0.719). Among the 4 sunshine-based models, Bahel model shows the best performance, with R^2 of 0.903, MAE of 1.624 MJ/(m^2 ·d), MRE of 15.7%, RMSE of 2.298 MJ/(m^2 ·d) and NSE of 0.902. The most accurate temperature-based model is Bristow-Campbell model, with R^2 of 0.719, MAE of 2.851 MJ/( m^2 ·d ), MRE of 30.7%, RMSE of 3.959 MJ/(m^2 ·d) and NSE of 0.713. Ove-rall, the Bahel model is recommended to estimate daily and monthly Rs when only the sunshine duration data are avai-lable in Northwest China and the Bristow-Campbell model is recommended to estimate R s when only temperature data are available.