Evaluation of Solar Radiation from MERRA, MERRA-2, ERA-Interim and CFSR Reanalysis Datasets Against Surface Observations for Multan, Pakistan
Abstract
The solar radiation data measured by ESMAP for Multan in South Punjab, Pakistan is compared with four satellite reanalysis datasets (MERRA, MERRA-2, ERA-Interim, and CFSR). The validation of satellite estimates of solar radiation against surface measured data is performed on biases, RMSE and correlation coefficient. All reanalysis datasets overestimte the GHI but only ERA-Interim showed both underestimation and overestimation. The MBE for four datasets ranges from 18.82 W/m2 to 34.05 W/m2 for solar radiation. The monthy time series showed that all reanalysis datasets showed good trend in months having no clouds and rainings but in the months having clouds and rainings each reanalysis dataset showed a different trend. Clouds variables are the main reasons of overestimation and underestimation of GHI by reanalysis datsets. The bias in clearness index ranges from 0.05 to 0.11 for four reanalysis datasets. All reanalysis datasets overestimate the daily clearness index except ERA-Interim. ERA-Interim similar to the solar radiation results both overestimates and underestimates the daily clearness index. The estimates from MERRA are less accurate among all datasets whereas ERA-Interim is accurate in terms of both statistical parameters for time series trend and clearness index. NCEP-CSFR showed the best trend with a constant bias.