BOUSSOLE: Vicarious calibration

Preliminary results

The table belowlists the computational aspects used in the OSOA model (geometry, aerosol refractive index, and optical depth) to predict the TOA radiance corresponding to MERIS observations. Only five matchups were available at the time of this preliminary analysis, depending on the quality of ground measurements and MERIS scenes. It should be noted that on 7 September 2002, the ground measurements made on the morning showed questionable data. Since the aerosol optical depth at 675nm and the Angström exponent were stable all along this day-- varied from 0.097 to 0.119, and varied from 1.45 to 1.69--the aerosol model derived from the measurements collected at the end of the day (at 1616GMT) was used, because it includes the maximum value of the scattering angle (150).

Date	Hmeris	ta	a	qs	qv	Dj	qd	HPPL	mr-jmi
07/11/02	10h00	0.134	1.34	29.5	7.7	25.3	157.2	08h39	1.46-j0.009
09/07/02	09h35	0.097	1.59	45.1	39.0	38.6	153.8	16h16	1.41-j0.000
09/26/02	09h38	0.041	1.29	50.5	35.7	45.4	146.4	10h27	1.55-j0.004
10/01/02	10h21	0.074	1.61	48.7	26.8	123.3	113.8	11h25	1.52-j0.011
10/12/02	10h21	0.037	1.02	56.0	39.1	49.4	140.7	11h22	1.57-j0.008

Computational aspects used in the OSOA radiative transfer model for the calculation of the top of atmosphere radiance in the MERIS geometry : aerosol optical depth at 675 nm ta(measured), Angström exponent a (measured), solar zenith angle qs (in degrees), viewing zenith angle qv (in degrees), relative azimuth Dj=js-jv between sun and satellite (in degrees), scattering angle qd (in degrees), aerosol model (complex refractive index mr-jmi) derived by NN. The time corresponding to MERIS overpass (Hmeris) and to the chosen sequence of principal plane measurements (HPPL) is indicated in GMT.

The table below reports the results of the comparison between MERIS TOA radiances and OSOA calculations. The agreement is very good and remains below 5%. Gordon (1998) showed that for a given absolute calibration uncertainty in the near IR, the calibration error progressively decreases with decreasing wavelength. Recently, Wang and Gordon (2002) demonstrated that a calibration uncertainty of 15% at 865nm leads to an acceptable calibration uncertainty smaller than 3% at 412nm. The results presented here are, therefore, within the accuracy requirements for the NIR band. Those results are still preliminary, however, because the number of available match-up data is too low and is not significant enough to draw rigorous conclusions about the MERIS calibration.

Date	LTOA -MERIS	LTOA-OSOA	DL/L (%)
July 11 2002	0.0044248	0.0042194	+4.64
Sept 07 2002	0.0042832	0.0044165	-3.11
Sept 26 2002	0.0027727	0.0028420	-2.50
Oct 01 2002	0.0027479	0.0027362	+0.42
Oct 12 2002	0.0027550	0.0027337	+0.77

Comparison between MERIS top of atmosphere radiances (LTOA) at 865 nm and OSOA radiative transfer calculations. Radiances are expressed in W m-2 nm-1 sr-1. The relative difference DL/L corresponds to the ratio (LMERIS-LOSOA)/LMERIS.

As a future activity, the vicarious calibration method will be extended to the visible bands. In the visible spectrum, the marine signal is not equal to zero, so the additional difficulty will be to get simultaneous in-water radiometric measurements in order to reconstruct the total TOA radiances as the sum of the atmospheric signal (calculated using the same technique as the NIR band) and the marine signal multiplied by a diffuse transmittance. To achieve this task, the continuous record collected by the BOUSSOLE buoy will allow to get information on the signal exiting the water and will increase the number of matchups needed to perform a relevant vicarious calibration of MERIS.