This study proposes a methodology to compute solar radiation values on the Earth surface on a monthly and annual basis and taking into account the relief, which is not an usual approach although it has enormous importance for environmental studies. It consists of two steps: a computational, physically-based model (potential solar radiation) and a refining through meteorological data (interpolated solar radiation). The potential solar radiation model allows computing the incident energy assuming uniform atmospheric standard optical density. The only information required is a Digital Elevation Model (DEM, 180 m resolution in our case) and a number of meteorological stations providing global solar radiation data (74 in our case). This model, implemented into the MiraMon Geographic Information System (GIS), takes into account astronomical, atmospheric and geographical factors. Atmospheric effects are considered through correctors obtained by comparing the potential values with the observed at the meteorological stations. The resulting monthly correctors have been spatially interpolated to build up anomaly maps, which allow to obtain a refined solar radiation cartography. The potential model has proven to be very precise as it gives results close to the interpolated radiation values, allowing a conversion to interpolated solar radiation maps with errors of 7.3% (March), 6.1% (June), 6.4% (September) and 13.1% (December) computed from a leave-one-out cross-validation. These results are always better than those obtained by classical interpolators, and still better if the process is run on the rugged terrain areas. A sensitivity analysis estimates the influence of the DEM quality on the model, concluding that an RMS error of 8 m in the DEM heights represents an RMS error of 133 kJ·m −2 ·day −1 (mean monthly) on the results. Finally, we have applied the model to obtain the 12-monthly (based on daily average) global solar radiation maps of Catalonia (NE of Iberian Peninsula) as well as the annual one.