Evaluation of the accuracy of downward radiative flux observations at the sea surface

Observations of downward radiative flux at the sea surface generally contain uncertainty due to limited numbers of observations and limitations of auxiliary equipment. The lack of shading from direct solar radiation and ventilation systems causes bias or random errors. To evaluate the error of radiation measurements at buoys, downward shortwave and longwave radiative fluxes are compared with International Satellite Cloud Climatology Project (ISCCP), Japanese 55-year Reanalysis (JRA55), and Moderate Resolution Imaging Spectroradiometer (MODIS) retrieved model calculations of 3-h and daytime averages. Cloud masking is evaluated by a combination of MTSAT-1R and in situ observations. Coincident observations from a land-surface station located near the buoy observatories are compared with satellite and reanalysis products. The bias at buoys, compared with retrievals, approximately over- and under-estimate for longwave and shortwave fluxes, respectively. The bias at buoys is larger and smaller than the land by 23–34 W m⁻² for longwave and 13–51 W m⁻² for shortwave radiation using 3-h averages under clear-sky conditions. The differences in bias decrease when using daytime averages for longwave, but the difference for shortwave increases with daytime averages. To evaluate the effect of environmental factors on buoy observations, we compared rainfall, wind speed, and solar zenith angle with the biases. We found that rainfall and wind speed affect buoy pyrgeometers such that they overestimate the longwave flux. The cosine of solar zenith angle does not cause overestimation for longwave flux, and the effect of dome heating is small. The strong wind causes underestimation of the shortwave radiative flux due to tilting. The effect of wind is reduced when daily averages are used.