The Temporal Distortion Index (TDI): a new procedure to analyze solar radiation forecasts

In this work a new point of view to evaluate prediction models is presented. It is captured by mean of a novelty dissimilarity measure among time series, the Temporal Distortion Index (TDI), which compiles a new methodology to evaluate and control the solar radiation prediction models. This methodology complements the traditional verification measures found in the literature by adding the evaluation of the impact that time misalignments produces in the forecast accuracy. This new measure of error will allow a deeper knowledge of the prediction model behaviour besides a bi-criteria perspective to the problem of comparing different forecasts. The information about temporal features of the forecasts could play a key role in tasks as combination of different prediction models, Concentrating Solar Power (CSP) plants operation or energy grid integration.