Temporal scaling analysis of irradiance estimated from daily satellite data and numerical modelling

- The temporal variability of the global irradiance estimations coming from satellite and numerical model has been compared.
- A scaling analysis considering different periods has been carried out.
- A different behaviour is obtained for each climate according to the scale considered.
- Variations of irradiance measured in several stations have been compared with the estimations.

The temporal variability of global irradiance estimated from daily satellite data and numerical models has been compared for different spans of time. According to the time scale considered, a different behaviour can be expected for each climate. Indeed, for all climates and at small scale, the persistence decreases as this scale increases, but the mediterranean climate, and its continental variety, shows higher persistence than oceanic climate. The probabilities of maintaining the values of irradiance after a certain period of time have been used as a first approximation to analyse the quality of each source, according to the climate. In addition, probability distributions corresponding to variations of clearness indices measured at several stations located in different climate zones have been compared with those obtained from satellite and modelling estimations.For this work, daily radiation data from the reanalysis carried out by the European Centre for Medium-Range Weather Forecasts and from the Satellite Application Facilities on climate monitoring have been used for mainland Spain. According to the results, the temporal series estimation of irradiance is more accurate when using satellite data, independent of the climate considered. In fact, the coefficients of determination corresponding to the locations studied are always above 0.92 in the case of satellite data, while this coefficient decreases to 0.69 for some cases of the numerical model. This conclusion is more evident in oceanic climates, where the most important errors can be observed. Indeed, in this case, the RRMSE derived from the CM-SAF estimations is 20.93%, while in the numerical model, it is 48.33%.Analysis of the probabilities corresponding to variations in the clearness indices also shows a better behaviour of the satellite-derived estimates for oceanic climate. For the standard mediterranean climate, the satellite also provides better results, though the numerical model improves significantly. In fact, for continental climate, both sources offer similar outcomes.