The adaptation of modeled solar radiation data with coincident ground measurements has become a standard practice of the industry, typically requested by financial institutions in the detailed solar resource assessments of solar projects. This practice mitigates the risk of solar projects, enhancing the adequate solar plant design and reducing the uncertainty of its yield estimates. This work presents a procedure for improving the accuracy of modeled solar irradiance series through site-adaptation with coincident ground-based measurements relying on the use of a regression preprocessing followed by an empirical quantile mapping (eQM) correction. It was tested at nine sites in a wide range of latitudes and climates, resulting in significant improvements of statistical indicators of dispersion, distribution similarity and overall performance: relative bias is reduced on average from -1.8% and -2.3% to 0.1% and 0.3% for GHI and DNI, respectively. Relative root mean square deviation is reduced on average from 17.9% and 34.9% to 14.6% and 29.8% for GHI and DNI, respectively; the distribution similarity is also improved after the site-adaptation (KSI is 3.5 and 3.9 times lower for GHI and DNI at hourly scale, respectively). The methodology is freely available as supplementary material and downloadable as R-package from SiteAdapt.

中文翻译：

---

建模的太阳辐射数据的站点自适应：SiteAdapt程序

建模的太阳辐射数据与同时进行的地面测量值的匹配已成为行业的标准做法，这通常是金融机构在太阳能项目的详细太阳能资源评估中所要求的。这种做法降低了太阳能项目的风险，增强了适当的太阳能发电厂设计并减少了其收益估算的不确定性。这项工作提出了一个过程，该过程可通过站点适应与地面同时发生的测量（依赖于回归预处理，然后进行经验分位数映射（eQM）校正）相结合的方式来提高建模的太阳辐照度序列的准确性。在广泛的纬度和气候的九个站点进行了测试，从而显着改善了分散，分布相似性和整体性能的统计指标：GHI和DNI的相对偏差平均分别从-1.8％和-2.3％降低到0.1％和0.3％。GHI和DNI的相对均方根偏差平均分别从17.9％和34.9％降低到14.6％和29.8％；站点适应后，分布相似性也得到了改善（在小时尺度上，GHI和DNI的KSI分别低3.5和3.9倍）。该方法可作为补充材料免费获得，也可以从SiteAdapt下载R-package。