Quality control and Methods for modeling daily global solar radiation (Case study: Gorgan, Iran)

ABSTRACT: Information of global solar radiation is required in many applications due to its essential role as a mainly diver for many biological processes on the earth surface. In spite of its significant, solar radiation is not widely measured compared to other meteorological variables due to the cost and maintenance and calibration requirements of the measuring equipment. This research, with Gorgan station, Iran, as a case study, aimed to quality control, calibrate existing models and develop new models for estimating missing global solar radiation data using commonly measured meteorological data and to propose a strategy for selecting the optimal models under different situations of available meteorological data. Using daily global radiation, sunshine hours and temperature covering the years from 2007-2008, we calibrate or develop and evaluated eight existing models and one new models. Validation criteria included intercept, slope, and coefficient of determination, mean bias and root mean square error. Result showed that quality control method is capable for identifying errors in measured solar radiation. The best result (R2=0.89) was derived from derived model in this research, which uses sunshine hours and temperature as predictors. The Bristow and Chambell model, which only uses temperature, was almost as good, explaining 87% of the solar radiation variance. Sunshine based models (Angsrom-Preskat; Akin-Oghlu and Esevit; Elagib and Mansell) provided less occurred results, of which the best one (MBE=0.055) is the Elagib and Mansell model. Generally, adding temperature terms into models which use sunshine hours, increase accuracy of model. We, thus, propose a strategy for selecting and optimal method for calculating missing daily values of global solar radiation: (1) when sunshine hour and temperature data are available, use Chen model and derived model in this research; (2) when only sunshine hour data are available, use Abdullah model and (3) when only temperature data are available use Bristow and Chambell model