A Technique for Global Monitoring of Net Solar Irradiance at the Ocean Surface. Part II: Validation

Abstract

The present study constitutes the generation and validation of the first satellite-based, long-term record of surface solar irradiance over the global oceans. The record is generated using Nimbus-7 earth radiation budget (ERB) wide-field-of-view (WFOV) planetary-albedo data as input to a numerical algorithm designed and implemented for this study based on radiative transfer theory. Net surface solar irradiance is obtained by subtracting the solar radiation reflected by the ocean-atmosphere system (measured by satellite) and the solar radiation absorbed by atmospheric constituents (modeled theoretically) from the solar irradiance at the top of the atmosphere (a known quantity). The resulting monthly mean values are computed on a 9° latitude-longitude spatial grid for November 1978°October 1985. Because direct measurements of surface solar irradiance are not available on the global spatial scales needed to validate the new approach, the ERB-based values cannot be verified directly against in situ pyranometer data. Although the ERB-based annual and monthly mean climatologies are compared with those obtained from ship observations and empirical formulas, a comparison with long-term mean climatologies does not provide an assessment of the month-to-month accuracies achieved using the new technique. Furthermore, the accuracy of the ship-based climatologies is questionable. Therefore, the new dataset is validated in comparisons with short-term, regional, high-resolution, satellite- based records (which were generated using methods that in turn have been validated using in situ measurements). The ERB-based values of net surface solar irradiance are compared with corresponding values based on radiance measurements taken by the VISSR (Visible-Infrared Spin Scan Radiometer) aboard GOES (Geostationary Operational Environmental Satellite) series satellites during the TOGA (Tropical Ocean Global Atmosphere), Tropic Heat, and MONEX (Monsoon Experiment) field experiments. The rms differences are 14.5 W m?2 (i.e., 6.2% of the average VISSR-based value on monthly time scales) for the TOGA data comparison, 6.4 W m?2 (i.e., 2.5% of the average VISSR-based value on monthly time scales) for the Tropic Heat data comparison, and 16.8 W m?2 (i.e., 7.5% of the average VISSR-based value on monthly time scales) for the MONEX data comparison. The ERB-based record is also compared with an additional satellite-based dataset, focused primarily over the Atlantic Ocean, that was generated using radiance measurements from the Meteosat radiometer. On the basis of these validation studies, errors in the new dataset are estimated to lie between 10 and 20 W m?2 on monthly time scales.