中国不同气候区逐日直散分离模型适用性分析及通用计算模型优化

利用中国5个气候区59个气象台站1981——2010年的日值气象数据,对比分析8个散总比和3个散射系数直散分离模型在中国不同气候区的适用性。采用判定系数（R2）、均方根误差（RMSE）、平均绝对误差（MABE）、平均误差（MBE）和全局性能系数（GPI）5个误差评价指标,确定各气候区最适宜的模型形式。以该模型为基础,建立适用于中国不同气候区的散总比和散射系数逐日直散分离通用模型。结果表明,除线性形式散射系数模型精度较差外,其他模型计算精度均较高,散总比和散射系数模型平均R2分别为0.85和0.62;基于晴空指数-日照百分率的二次多项式散总比模型和基于日照百分率三次多项式散射系数模型在不同气候区精度均最高;以该模型为基础建立中国不同气候区散总比和散射系数逐日直散分离通用模型,其平均R2分别为0.89和0.70。

SUITABILITY ANALYSIS OF DAILY DIRECT DIFFUSE SEPARATE MODELS OF SOLAR RADIATION ESTIMATION IN DIFFERENT CLIMATIC ZONES OF CHINA AND GENERAL COMPUTATION MODEL OPTIMIZATION

Using the daily meteorological data during 1981-2010 from 59 meteorological station in five different climatic zones of China, 8 diffuse fraction and 3 diffuse coefficient models for solar diffuse radiation estimation were compared and analyzed to determine the suitability of these models in different climatic zones. The five evaluation indexes, including determination of coefficient （R²）, root mean square error （RMSE）, mean absolute bias error （MABE）, mean bias error （MBE） and global performance index （GPI）, were adopted to obtain the optimal suitable model in different climatic zones. Based on these models, the diffuse fraction and diffuse coefficient generalized models suitable for different climatic zones of China were developed. The results show that the models, except for linear diffuse coefficient models, performs well in estimating diffuse solar radiation. The average R² for diffuse fraction and diffuse coefficient are were 0.85 and 0.62, respectively. The clearness index- and sunshine-based second polynomial diffuse fraction model and sunshine-based third polynomial diffuse coefficient models yield highest accuracy in each zone. Based on these two models, the diffuse fraction and diffuse coefficient generalized models are developed with average R² of 0.89 and 0.70.