A Meta-Analysis of Solar Forecasting Based on Skill Score

We conduct the first comprehensive meta-analysis of deterministic solar forecasting based on skill score, screening 1,447 papers from Google Scholar and reviewing the full texts of 320 papers for data extraction. A database of 4,758 points was built and analyzed with multivariate adaptive regression spline modelling, partial dependence plots, and linear regression. Notably, the analysis accounts for the most important non-linear relationships and interaction terms in the data. We quantify the impacts on forecast accuracy of important variables such as forecast horizon, resolution, climate conditions, regions' annual solar irradiance level, power system size and capacity, forecast models, train and test sets, and the use of different techniques and inputs. By controlling for the key differences between forecasts, including location variables, the findings from the analysis can be applied globally. An overview of scientific progress in the field is also provided.