Reprocessing #1: MODIS/Aqua f/Q Normalization

The f/Q correction accounts for the spectrally-dependent, non-isotropic structure of the subsurface light field, which is not addressed in other bidirectional reflectance corrections (see also the Gothic-R term described in Morel et al. 2002). The f/Q correction depends on sensor and solar zenith angles, their relative azimuth angle, and in-water constituents. With regards to the latter, Case-1 conditions are implied and the chlorophyll concentration is used to approximate the marine inherent optical properties (e.g., the volume scattering function). Please refer to Morel et al. 2002 (and references therein) and the OBPG BRDF Evaluation Web page for additional details and analyses.

Questions for Consideration

Are satellite-retrieved chl concentrations of adequate quality to correct the radiances? One simple analysis suggests that variation of +10 and +25% in retrieved chl changes the f/Q correction factor by -0.5 and -1.5%, respectively.
Is the chl range with which the f/Q model was derived appropriate for global application? If applied globally, how are Case 2, high chl Case 1, and turbid areas affected? How well does the model interact with the current NIR correction?
Is the existing suite of wavelengths with which the f/Q model was derived (MERIS) sufficient for application to SeaWiFS and MODIS? Is interpolating between wavelengths adequate?
Even if the correction improves satellite comparisons at Level-3, does the community want the correction applied full time (e.g., distributed Level-2 products)?

Evaluation Results

The observations made in the table below are highlights from selected analyses. The community is strongly encouraged to peruse the following complete analysis results:

The OBPG has compiled a very brief list of notable PROS and CONS for consideration.

PROS CONS
a theoretical need for the additional normalization limited Case 1 water model design
convergence of SeaWiFS and MODIS-Aqua level-3 time series undesirable operation, modifies Lw using modeled chl
few visual inconsistencies in level-2 files paucity of level-2 verification analyses
paucity of peer-reviewed alternatives inability of community to verify model in the field
improved residual scan dependence

Highlighted observations

In theory, for the most extreme sensor and solar geometries, the f/Q correction will modify radiances by -40 to 10%. In reality, MODIS/Aqua radiances would be modified by -10 to 5%.
Level-2 validation results for radiances and chlorophyll are statistically similar with and without the correction, although field data for this activity is sparse.
Residual seasonality in the open ocean MODIS/Aqua-SeaWiFS time series comparisons removed. Significant bias to water-leaving radiances in coastal subset introduced.
Overall improvement in the open ocean MODIS/Aqua-SeaWiFS latitudinal comparison with the correction applied. Significant latitudinal bias evident without the f/Q correction.

Additional comments

Field data used in calibration and validation activites also require application of this correction. Similarly, existing empirical chl algorithms do not have corrected radiances, although one simple analysis suggests the resulting uncertainty is minimal.

References

Morel, A., and B. Gentilli, 1996: Diffuse reflectance of oceanic waters. III. Implication of bidirectionality for the remote-sensing problem. Appl. Opt., 35, 4850-4862

Morel, A., D. Antoine, and B. Gentilli, 2002: Bidirectional reflectance of oceanic waters: accounting for Raman emission and varying particle scattering phase function. Appl. Opt., 41, 6289-6306

18 January 2005, P. Jeremy Werdell, NASA Ocean Biology Processing Group, SSAI

The f/Q correction accounts for the spectrally-dependent, non-isotropic structure of the subsurface light field, which is not addressed in other bidirectional reflectance corrections (see also the Gothic-R term described in Morel et al. 2002). The f/Q correction depends on sensor and solar zenith angles, their relative azimuth angle, and in-water constituents. With regards to the latter, Case-1 conditions are implied and the chlorophyll concentration is used to approximate the marine inherent optical properties (e.g., the volume scattering function). Please refer to Morel et al. 2002 (and references therein) and the OBPG BRDF Evaluation Web page for additional details and analyses.

Questions for consideration

Are satellite-retrieved chl concentrations of adequate quality to correct the radiances? One simple analysis suggests that variation of +10 and +25% in retrieved chl changes the f/Q correction factor by -0.5 and -1.5%, respectively.
Is the chl range with which the f/Q model was derived appropriate for global application? If applied globally, how are Case 2, high chl Case 1, and turbid areas affected? How well does the model interact with the current NIR correction?
Is the existing suite of wavelengths with which the f/Q model was derived (MERIS) sufficient for application to SeaWiFS and MODIS? Is interpolating between wavelengths adequate?
Even if the correction improves satellite comparisons at Level-3, does the community want the correction applied full time (e.g., distributed Level-2 products)?

Evaluation results

The observations made in the table below are highlights from selected analyses.
The community is strongly encouraged to peruse the following complete analysis results:
The OBPG has compiled a very brief list of notable PROS and CONS for consideration.

PROS CONS
a theoretical need for the additional normalization limited Case 1 water model design
convergence of SeaWiFS and MODIS-Aqua level-3 time series undesirable operation, modifies Lw using modeled chl
few visual inconsistencies in level-2 files paucity of level-2 verification analyses
paucity of peer-reviewed alternatives inability of community to verify model in the field
improved residual scan dependence

Highlighted observations

In theory, for the most extreme sensor and solar geometries, the f/Q correction will modify radiances by -40 to 10%. In reality, MODIS/Aqua radiances would be modified by -10 to 5%.
Level-2 validation results for radiances and chlorophyll are statistically similar with and without the correction, although field data for this activity is sparse.
Residual seasonality in the open ocean MODIS/Aqua-SeaWiFS time series comparisons removed. Significant bias to water-leaving radiances in coastal subset introduced.
Overall improvement in the open ocean MODIS/Aqua-SeaWiFS latitudinal comparison with the correction applied. Significant latitudinal bias evident without the f/Q correction.

Additional comments

Field data used in calibration and validation activites also require application of this correction. Similarly, existing empirical chl algorithms do not have corrected radiances, although one simple analysis suggests the resulting uncertainty is minimal.

References

Morel, A., and B. Gentilli, 1996: Diffuse reflectance of oceanic waters. III. Implication of bidirectionality for the remote-sensing problem. Appl. Opt., 35, 4850-4862

Morel, A., D. Antoine, and B. Gentilli, 2002: Bidirectional reflectance of oceanic waters: accounting for Raman emission and varying particle scattering phase function. Appl. Opt., 41, 6289-6306


18 January 2005, P. Jeremy Werdell, NASA Ocean Biology Processing Group, SSAI