Modelling solar radiation over complex terrains using monthly climatological data

Meteorological stations placed in remote areas (such as mountainous sites) often record only precipitation and air temperature data, summarised into monthly tables. There is therefore a need for appropriate methods to estimate solar radiation data from monthly inputs. It is important to know the quality and characteristics of the estimates made in order to understand what impacts the modelled data may have on the use to which they are meant. This paper evaluates the behaviour of two empirical models based on the difference between maximum and minimum air temperatures and compares results with a newly developed model, for 24 stations (11 for model calibration, the rest for model validation) in Italian sites. The new model (SRMCT, solar radiation model for complex terrains) replaces the minimum air temperature with a monthly function and uses rainy days number to correct the air temperature range. Comparisons were made using a fuzzy-logic-based multiple-statistics assessment approach (Irad). An appraisal of the temporal distribution of mean errors over a year was also performed. The new model produced the best overall estimates. The two air temperature-based models can be an alternative approach when only air temperature data are available, but at many sites they introduced large inaccuracies and seasonal patterns in the error distribution. The study demonstrates the appropriateness of both air temperature range and rain correction factors for monthly based solar radiation estimates. At the same time, it shows the value and importance of using a range of assessment statistics, even aggregated into an overall indicator, to evaluate model estimates.