SeaWiFS Radiometric and Temperature Calibrations

The SeaWiFS calibration has been very stable and predictable for most of the mission to date, as evidenced by the remarkable temporal consistency of the nLw. Recently, analyses of the interannual repeatability of the nLw have shown measureable deviations from the mission trends. Starting in late 2005, the global nLw averages have exhibited downward trends of a few percent, primarily in the green bands. In addition, the fit of the operational calibration to the lunar measurements in the NIR bands has deteriorated over this same period. These effects motivated a re-analysis of the lunar calibration data over the mission to develop improved radiometric and temperature corrections.

The changes to the calibration and the rationale for each are as follows:

  1. The counts-to-radiance conversion coefficients for all bands have been reverted to the values measured by the instrument builder during the initial prelaunch calibration of SeaWiFS in 1993. The current conversion coefficients, which were measured by NIST during the recalibration of SeaWiFS prior to launch in 1997, give rise to trends with wavelength in both the lunar calibration residuals from the USGS photometric model of the Moon and in the vicarious gains computed from MOBY data. For the ocean color data, the changes to the conversion coefficients for Bands 1 through 7 are offset by corresponding changes to the vicarious gains. The lunar residuals and the vicarious gains are spectrally flatter with the 1993 conversion coefficients. This change in coefficients does not impact the on-orbit calibration of SeaWiFS over time. View additional details.

  2. The reference temperature used for the temperature corrections to all bands has been changed from 20 C to 16 C. The reason is that 20 C is outside of the range of the average focal plane temperature measurements observed during data collection, and changes to the temperature correction were introducing discontinuities in the calibration. A reference temperature of 16 C is well within the observed temperature range for both the lunar and ocean observations. The counts-to-radiance conversion coefficients have been renormalized to the 16 C reference temperature. View additional details.

  3. The lunar calibration residuals for Bands 1--5 do not show sufficient correlations with focal plane temperature to warrant using corrections derived from on-orbit data, so the focal plane temperature corrections for these bands have reverted to their prelaunch measured values. The lunar residuals for Bands 6 and 7 show corrleations with focal plane temperature that warrant using a single temperature correction coefficient derived from the lunar data for each band. The lunar residuals for Band 8 show a change in the focal plane temperature dependence over the mission, warranting the use of two temperature corrections for this band, with the temperature correction epoch boundary in mid 2005. View additional details.

  4. The radiometric calibration function that is used to fit the lunar calibration time series has been revised. As the SeaWiFS mission has gone on, the optimum time constant for the long-period exponential has become increasingly longer, with the fitting function approaching a linear function of time. For bands 1-2, the long-period behavior shows departures from linearity with time, so the current fits use simultaneous decaying exponential functions of time with time constants of 200 and 3200 days. For bands 3-8, the long-period behavior is still consistent with a linear function of time, so the current fits use a decaying exponential function of time with a time constant of 400 days in conjunction with a linear function of time. Ongoing analyses of bands 3-8 will look for the onset of departures from linearity. View additional details.

  5. The Gain 3 drift correction has been revised. The lunar calibration data are collected at non-standard gains (Gain 4 for Band 1 and Gain 3 for the other bands) to keep the radiance below the knees of the bilinear gains on the lunar disk. For the operational calibration, a temporal correction was computed to the Gain 3/1 ratio for Bands 7 and 8 and applied to the lunar measurements. For the new calibration, this correction has been recomputed and will be applied only for Band 7, as the previous Band 8 correction was determined to be a digitization artifact in the gain calibration data. In addition, this correction is now included in the calibration table as a separate function distinct from the overall radiometric calibration. View additional details.

Updated SeaWiFS lunar calibration time series


Image generated August 4, 2010 using 141 lunar cals collected through July 26, 2010.

The SeaWiFS lunar calibration time series, with the updated radiometric and temperature calibrations, is shown in the above figure. The fits to the time series, shown as the solid line for each band, were computed using the 140 lunar calibrations obtained through June 2010, prior to the satellite orbit raising maneuver. As of July 2010 one post-orbit raising calibration has been collected, which is consistent with the trend lines of the fits. This plot does not include the Gain 3 drift correction, since that correction only concerns the application of the radiometric calibration to the ocean data.

Some issues remain with the new SeaWiFS calibration

  1. The orbit rasing maneuver for the satellite, which occurred in early July of 2010, is designed to reverse the drift of the orbit node away from local noon. The node should start to drift back toward a noon node crossing time. The firt post-orbit raising lunar calibration does not show any change in the radiometric performance of the instrument, as is expected. The effects of the changing orbit node will continue to be monitored as additional lunar calibrations are obtained.

  2. The Band 6 global averages show a small upward trend for most of the mission. This trend may result from the residual uncertainties in the Band 7 Gain 3 correction. It should be noted that the light levels measured in Band 6 are an order of magnitude smaller than those measured in Bands 4 and 5 and that Band 6 is vicariously calibrated against a very small in situ signal. Consequently, small errors in the instrument calibration will tend to be exaggerated in the Band 6 anomlay plots.

It may be possible to further improve the instrument calibration as additional lunar measurements are taken; however the new calibration is believed to represent the best current knowledge of the calibration based on the lunar and gain calibration data.