Evaluation of models for prediction of diffuse solar radiation and comparison with satellite values

The diffuse radiation data are one of the important parameters required in solar energy projects. However, this parameter has not been measured in everywhere of Turkey. In the present study, models in the open literature for the determination of horizontal diffuse radiation with clearness index are investigated and, new models are developed for thirteen provinces over the Eastern Anatolia Region (EAR) of Turkey where the diffuse radiation values are not existing. In the monthly average, these models have a relation between the diffuse coefficient (H-d/H-o) and the diffuse fraction (H-d/H) with the daily clearness index. The most appropriate models based on NASA Surface meteorology and Solar Energy (NASA-SSE) values are determined. For this, the diffuse radiation data taken from the NASA-SSE and calculated from the models are compared for the locations mentioned in the study. The validation of the correlations obtained to predict the diffuse radiation over the EAR for which experimental data are not available is firstly tested by using the NASA-SSE satellite data. The performances of the models are indicated with the statistical test methods as mean absolute bias error (MABE), root mean square error (RMSE), the mean bias error (MBE), and mean relative error (MRE). The results indicate that some models are reasonably well to forecast the diffuse radiation. In addition, the diffuse radiation predictions of the new correlations obtained in the study are compared with satellite data. When the diffuse radiation values taken from the NASA-SSE and calculated from the models are compared, it is seen that the Model 4 for Agri, Bingol, Bitlis, Elazi.g, Erzincan, Hakkari, Igdir, Kars, and Mus, Model 14 for Erzurum, Model 20 for Malatya and Tunceli, and Model 11 for Van indicate the best performance. According to the RMSE values of twenty models in the literature and the NASASSE, the best result among all provinces is obtained from the Model 11 with 0.2385 MJ m(-2) day(-1) for Van.

Journal

JOURNAL OF CLEANER PRODUCTION

Year

2022

Volume

374

ISSN

0959-6526

eISSN

1879-1786

DOI

10.1016/j.jclepro.2022.133892

WoS UID

WOS:000867574700001