西藏日太阳总辐射估算模型的优化与时间尺度效应

根据2012～2014年拉萨站点的日太阳辐射观测数据,对3种日太阳总辐射估算模型(左大康、Prescoff和王炳忠等提出的估算模型)的估算结果和实际观测结果进行误差检验分析,结果表明:3种模型都通过了显著性检验,其平均绝对误差(MAE)分别为6.17 MJ/(m~2·d)、6.42 MJ/(m~2·d)和12.31 MJ/(m~2·d),均方根误差(RMSE)分别为1.60 MJ/(m~2·d)、1.65 MJ/(m~2·d)和2.78 MJ/(m~2·d),表明利用地理位置信息和日照时数的左大康估算模型对拉萨日太阳总辐射的估算效果优于Prescoff和王炳忠估算模型。利用左大康估算模型分别估算得到2015年拉萨、那曲和葛尔地区的日太阳总辐射值,并进行模型验证。通过分析拉萨1971～2015年的太阳总辐射变化和周期性特征,研究日太阳总辐射估算模型的时间尺度效应。按照11年和四季两种不同的时间尺度,利用线性回归法对日太阳总辐射估算结果和实际观测结果进行误差和相关性分析,得到经验参数a和b的值与时间尺度有一定关系。11年和四季两种时间尺度除夏季以外,其他时间拟合模型的精度均得到了提高。通过线性回归法得出45年的西藏日太阳总辐射优化估算模型,该模型估算日太阳总辐射的误差小于左大康模型,估算精度较高。

Estimation Model Optimization for Daily Total Solar Radiation in Tibet and Time-scale Effects

According to Lhasa stations solar radiation observation data during the period from 2012 to 2014, actual observations of daily solar radiation and estimation values from 3 estimation models, Zuo Dakang, Prescoff and Wang Bingzhong et al. estimation models, were tested and analyzed. The results showed that all of the 3 models can pass the significance test. The mean absolute error (MAE) were 6.17 MJ/(m²·d), 6.42 MJ/(m²·d) and 12.31 MJ/(m²·d), and the root mean square error (RMSE) were 1.60 MJ/(m²·d), 1.65 MJ/(m²·d) and 2.78 MJ/(m²·d), respectively, for Zuo Dakang, Prescoff and Wang Bingzhong et al. estimation models, indicating that the Zuo Dakang estimation model, which uses geographic location information and sunshine hours is better than Prescoff and Wang Bingzhong estimation models in estimating the total solar radiation in Lhasa. The daily solar radiations in Lhasa, Nagqu and Geer areas in 2015 were estimated using the Zuo Dakang estimation model and statistical verifications were employed. The changes and periodic characteristics of solar radiation in Lhasa from 1971 to 2015 were analyzed and the time scale effects of daily solar radiation estimation model were studied. According to time scales of 11 years and four seasons, the linear regression was used to analyze errors and correlations between daily solar radiation estimations and actual observations, the obtained empirical parameters, a and b, showed a certain relationship with the time scale, except of summer, the fitting accuracies for both time scales of 11-year and four-season were improved. Using linear regression, the optimal estimation model of solar radiation in Tibet for 45 years were obtained,R=R_a(0.296+0.704 S/S_L), the estimation error of daily solar radiation by the above model was smaller than by the Zuo Dakang model, and the estimation accuracy was also higher.