The accuracy of satellite-estimated surface solar irradiance and the relationship between irradiance at pairs of locations and distance between the pairs were examined. Daily measured and satellite-estimated irradiances were available for a 2–3 year period from 20 Texas locations. For the distance analysis, irradiances were normalized as a percentage of daily potential surface irradiance.
In comparing the measured and satellite-estimated irradiances, the bias (the mean of satellite-estimated minus measured irradiance differences for the period of record) increased from ±1 MJ m−2 d−1 in the northern two-thirds of Texas to 2 MJ m−2 d−1 in the southern portion. The root-mean-square errors (RMSEs) between irradiances averaged 2.5 MJ m−2 d−1, which is 12% of the mean annual statewide potential surface-irradiance. Errors were at a maximum in the winter and a minimum in the summer.
Between measured irradiances at pairs of locations, RMSEs increased rapidly for the first 200 km distance between the pairs and increased at a lesser rate thereafter. Errors were reduced for five-day averages. Correlations of normalized measured or satellite-estimated irradiances between pairs of locations decreased exponentially as a function of the distance between the pairs. This decrease was less for five-day averages than for daily values. The correlation decrease as a function of distance was greater in this study than those previously reported and was likely the result of normalization.
These results are relevant to individuals interested in the accuracy of satellite-estimated irradiances, as well as those interested in interpolating measured irradiances from an existing network, or establishing an irradiance measurement network.