Statistical parameters as a means to a priori assess the accuracy of solar forecasting models

• Use of statistical parameter never used for the global radiation forecasting.
• A priori index allowing to optimize easily and quickly a clear sky model.
• New methodology allowing to quantify the prediction error regardless the predictor used.
• The prediction error depends more on the location and the time series type than the machine Learning method used.

In this paper we propose to determinate and to test a set of 20 statistical parameters in order to estimate the short term predictability of the global horizontal irradiation time series and thereby to propose a new prospective tool indicating the expected error regardless the forecasting methods used. The mean absolute log return, which is a tool usually used in econometrics but never in global radiation prediction, proves to be a very good estimator. Some examples of the use of this tool are exposed, showing the interest of this statistical parameter in concrete cases of predictions or optimizations. This study gives a judgment for engineers and researchers on the installation or management of solar plants and could help in minimizing the energy crisis allowing to improve the renewable energy part of the energy mix.