The present work illustrates the methodology followed to generate a high spatial (9 km) and high temporal resolution (10 min) global solar irradiance assessment, based on a numerical weather prediction model, for the south of Sonora region in Mexico and its validation with observational data. At the same time a comparison with an ERA-Interim output data was performed in order to determine if downscaling was necessary. The methodology used starts with obtaining the mean radiation year in order to strongly reduce computational cost. Each day of the mean radiation year defines the initial and boundary conditions of the simulations. The simulation outputs were used to create the monthly and the annual irradiation maps. The grid cells are compared with the corresponding observation and the precision of the model is evaluated. The correlation of the model with the observation data is higher than 0.88. The rRMSE of the model during the fall and winter is observed to be lowered than 6.4%, but the rRMSE increases during the spring and summer. The results show that the downscaling using the configuration selected was correct. (C) 2015 Elsevier Ltd. All rights reserved.