There is a growing demand for the estimation of solar energy potential at high latitude locations. This study compares four datasets; Cloud, Albedo, Radiation dataset Edition 2 (CLARA), Surface Solar Radiation dataset - Heliosat Edition 2 (SARAH), ECMWF Reanalysis 5 (ERAS) and Arctic System Reanalysis v2 (ASR) on high latitude locations. Global horizontal irradiance (GHI) from these datasets is compared with in-situ ground-measurements over multiple locations in Norway. The first two datasets are mainly based on satellite estimation of solar radiation, while the latter two are based on a combination of a weather-prediction model, satellite data, and other observations. The datasets are evaluated against quality-controlled in-situ measurements of solar radiation from pyranometers. Overall, CLARA, SARAH, and ERAS show moderate errors, while those of ASR are considerably larger. Monthly averages of global horizontal irradiance have mean absolute deviation (MAD) of 5.6 Wm(-2), 5.0 Win(-2), 6.8 Wm(-2), and 16.1 Wm(-2) for CLARA, SARAH, ERAS, and ASR, respectively. Seasonal error analysis of these datasets reveals that CLARA and SARAH have low errors in all seasons. The datasets are classified into clear-sky, intermediate-cloudiness, and overcast categories, by using two thresholds of cloudiness based on the ratio of radiation at ground to its corresponding clear-sky value (clear-sky index). The categories obtained from satellite and reanalysis data are then compared against estimates based on corresponding in-situ observations; this analysis shows that both CLARA and SARAH perform better than ERAS and ASR for these categories. SARAH and CLARA perform similarly in all types of conditions, but a gradual increase in errors for an increase of cloudiness is observed for ERAS and ASR. Yearly energy analysis shows that CLARA performs better than other datasets for locations above latitude 65 degrees N, and SARAH performs better in locations below 65 degrees N. A further analysis is performed to assess the cloud sensing abilities of ERAS. On a shorter time scale, there are errors due to inaccurate representation of clouds, however on longer time scales Le. months and years, these errors are considerably reduced. ERAS is observed to overestimate TCWC (the total cloud water content defined as the mass of water and ice in a cloud) in clear-sky and intermediate-cloudy categories, while in overcast category it is underestimated. Generally, an overestimation of solar radiation is observed in reanalysis and an underestimation is observed in satellite methods.