| description abstract | A high-quality dataset collected at an oceanographic tower was used to compare water-leaving radiances derived from simultaneous above- and in-water optical measurements. The former involved two different above-water systems and four different surface glint correction methods, while the latter used three different in-water sampling systems and three different methods (one system made measurements a fixed distance from the tower, 7.5 m; another at variable distances up to 29 m away; and the third was a buoy sited 50 m away). Instruments with a common calibration history were used, and to separate differences in methods from changes in instrument performance, the stability (at the 1% level) and intercalibration of the instruments (at the 2%?3% level) was performed in the field with a second generation Sea-viewing Wide Field-of-view Sensor (SeaWiFS) Quality Monitor (SQM-II). The water-leaving radiances estimated from the methods were compared to establish their performance during the field campaign, which included clear and overcast skies, Case-1 and Case-2 conditions, calm and roughened sea surface, etc. Three different analytical approaches, based on unbiased percent differences (UPDs) between the methods, were used to compare the various methods. The first used spectral averages across the 412?555-nm SeaWiFS bands (the part of the spectrum used for ocean color algorithms), the second used the ratio of the 490- and 555-nm bands, and the third used the individual (discrete) wavelengths. There were eight primary conclusions of the comparisons, which were considered within the context of the SeaWiFS 5% radiometric objectives. 1) The 5% radiometric objective was achieved for some in-water methods in Case-1 waters for all analytical approaches. 2) The 5% radiometric objective was achieved for some above-water methods in Case-2 waters for all analytical approaches, and achieved in both water types for band ratios and some discrete wavelengths. 3) The largest uncertainties were in the blue domain (412 and 443 nm). 4) A best-to-worst ranking of the in-water methods based on minimal comparison differences did not depend on the analytical approach, but a similar ranking of the above-water methods did. 5) Above- and in-water methods not specifically designed for Case-2 conditions were capable of results in keeping with those formulated for the Case-2 environment or in keeping with results achieved in Case-1 waters. 6) There was a significant difference between two above-water instruments oriented perpendicular with respect to the sun, but pointed in the same direction (best agreement) versus the opposite direction (worst agreement). 7) The overall intercomparison of all methods across Case-1 and Case-2 conditions was at the 9.1% level for the spectral averages, and at the 3.1% level for the band ratios (uncertainties other than those associated with implementing the individual methods account for 2%?4% and 1%?3% of these values, respectively). 8) A comparison with traditional regression analyses confirms the UPD conclusions. | |
