Direct atmospheric transmittance and sky radiance were measured spectrally at the coastal site of Scripps Institution of Oceanography pier in La Jolla, California, during the winters of 1993 and 1994. Direct atmospheric transmittance was also measured aboard R/V Wecoma and on Catalina Island during the 1994 California Cooperative Oceanic Fisheries Investigation winter cruise. The data were analyzed to (1) verify whether the aerosol models selected by Gordon and Wang <1994> for the sea-viewing wide field of view sensor are adequate for ocean color remote sensing from space and (2) identify what type of in situ atmospheric optics measurements should be performed to verify atmospheric correction algorithms. Aerosol optical thickness at 870 nm was generally low at La Jolla, with most values below 0.1 after correction for stratospheric aerosols. Values were even lower offshore (R/V Wecoma, Catalina Island), and no systematic correlation was found between aerosol characteristics and meteorological conditions. Therefore a mean background model, specified regionally, may be sufficient for ocean color remote sensing from space. For optical thicknesses above 0.1, two modes of variability characterized by ¿ngstr¿m coefficients of 1.2 and 0.5 and corresponding to tropospheric and maritime models, respectively, were identified in the measurements. The aerosol models selected by Gordon and Wang <1994> allow one to fit, within measurement inaccuracies, the derived values of ¿ngstr¿m coefficient and pseudo phase function (the product of single-scattering albedo and phase function), key atmospheric correction parameters. Importantly, the pseudo phase function can be derived from measurements of the ¿ngstr¿m coefficient. This means that shipborne Sun photometer measurements made at the time of satellite overpass are usually sufficient to verify the atmospheric correction of ocean color. ¿ 1998 American Geophysical Union