Validation of a physical retrieval scheme of solar surface irradiances from narrowband satellite radiances

An algorithm is presented to derive the downwelling solar surface irradiance from satellite measurements of the 0.63-mu m reflectance, which explicitly accounts for variations in cloud optical depth and integrated water vapor. For validation, a long-term dataset of 40 356 pyranometer measurements and 1450 NOAA-14 Advanced Very High Resolution Radiometer (AVHRR) satellite scenes of the Netherlands is used. A mean overestimate of the satellite-retrieved irradiance by 7% is found, which is consistent with numerous other studies reporting positive biases of atmospheric transmissivities that are calculated by radiative transfer schemes in comparison with measurements. The bias can be explained by the calibration and measurement uncertainties of both the AVHRR and pyranometer. A strong solar zenith angle dependence of the bias is found when water clouds are assumed in the retrieval. Such a dependence is not observed for ice clouds. Currently, there is not enough information for a conclusive explanation of this behavior. Comparing individual pyranometer measurements at 30 stations within a region of about 150 km(2) averaged over 40 min, a large raise of 86 W m(-2) is found. If the average of all of the stations for a satellite overpass is considered instead, a much better accuracy is achieved (rmse of 33 W m(-2)). For monthly averages of all of the stations, the rmse is further reduced to 12 W m(-2). Evidence is presented that suggests that a significant fraction of the rmse in the comparison originates from the variability of the irradiance field, which limits the representativeness of the reference ground-based pyranometer measurements for the satellite-retrieved values.