3 Novembre 2014

Polder 1 new radiometric calibration results

Following a characterization of an instrumental non-linearity defect and its correction inside the data level-1 processing (see Fougnié et al., 2001), a reappraisal of the radiometric calibration was required. In addition, improvements in some of our calibration methods have been made since the end of calibration phase.
The two main modifications in the calibration algorithm are:

    the use of a climatology (SeaWiFs 1999) to identify the oceanic sites that are the best for calibration over Rayleigh scattering. These sites have to be stable in terms of spatial and seasonal variation. The second objective of this climatology was to provide indicative values for pigments concentration and/or water leaving radiance that are more realistic than the previously used ones.

    the second improvement is the use of the 490 nm spectral band as pivot (instead of 565) for the inter-band calibration over sunglint. The absolute calibration of the 490 band is considered more credible than the absolute calibration of the 565 nm band.

    Different methods were considered to establish the new version of the radiometric calibration. Among them, the calibration over rayleigh scattering (absolute), the calibration over sunglint (interband), the calibration over clouds (interband), the cross calibration with SeaWiFS over desert sites (N1 referring to Barnes et al., 2001, and N2 referring to Eplee et al., 2000), and vicarious calibration using in-situ measurements (Fougnié et al., 1999). The main results for all these methods are:

    Ak443P443490565670765865
    Rayleigh0.9250.9300.9801.0461.007  
    Sunglint  0.980*1.0201.0211.0311.043
    Clouds 0.9600.980* 0.995  
    SeaWiFs_N10.8960.8980.9581.0411.0151.0350.999
    SeaWiFs_N20.9040.9050.9961.0501.0591.0940.999
    In-situ+ Rayleigh 0.9000.9451.0261.004  
    Synthesis of calibration results (* band used as pivot). Ak = Ak_in-flight/Ak_pre-flight with Ak defined by DN=Ak L where DN are the digits corresponding to the observed radiance L.

    Conclusion:

    • a relatively good consistency of all the methods: all the results are within ± 3%
    • a decrease of the sensitivity from the prelaunch value is evidenced for the 443 nm band. The uncertainty of absolute calibration in this spectral band is still higher than in the other bands.
    • the Rayleigh scattering calibration results are directly used for the bands 443, 443P, and 490
    • the other bands derived from the interband calibration with the 490 band using the sunglint
    • the new Version 3.0 calibration coefficient used to generate the level-1 products are

    Band443P443490565670763765865910
    Ak0.9250.9300.9801.0201.0211.0171.0311.0431.023

    The difference found between V3.0 and V2.0 coefficients is due to both non-linearity correction and algorithm improvement.

    References:

    Barnes, R. A., R. E. Eplee, G. M. Schmidt, F. S. Patt, and C. R. McClain, "The calibration of SeaWiFS, Part 1: Direct Techniques", Appl; Opt., vol. 40, pp. 36,6632-36,6700, 2001

    Eplee, R. E., and C. R. McClain, "MOBY data analysis for vicarious calibration of SeaWiFS bands 1-6, SeaWiFS Postlaunch Calibration and Validation Analyses, Part. 1, NASA Tech. Memo. 2000-206892, Hooker and Firestone, Eds., NASA GSFC, vol. 9, 43-50, 2000.

    Fougnie, B., P.Y. Deschamps, and R. Frouin, "Vicarious Calibration of the POLDER Ocean Color Spectral Bands using in-situ measurements", IEEE Trans. Geosci. and Remote Sensing, vol. 37, No. 3, pp. 1567-1574, 1999.