Jo Høkedal

Validation of MERIS water products and bio-optical relationships in the Skagerrak

The MERIS Level 2 Reduced Resolution products for Case 2 water available in September 2003, and conversion functions used in the reference model, have been validated against in situ data from the Skagerrak, collected during the summers of 2002 and 2003. The MERIS water‐leaving reflectance deviated less than 20% from the measured reflectances in the blue‐green part of the spectrum, but it had a tendency of being overestimated by up to 40% in the blue part of the spectrum and underestimated by up to 20% in the red part. The average relative deviation between the MERIS product for total suspended material (dry weight) and the in situ values was 30%. The MERIS values for chlorophyll a were on an average a factor 2 higher than the in situ values, and a new conversion factor should be used for the Skagerrak area. The absorption coefficients for the sum of yellow substance and bleached particles at 442 nm were underestimated by a factor of up to 10 by the MERIS product. The mean values of the spectral slopes of particle scattering and bleached particle absorption were close to the values of the reference model, while the observed slope of yellow substance was slightly lower than the model slope.

Spectral backscattering coefficient in coastal waters

Based on 186 field observations in the Oslo Fjord of irradiance and radiance and 105 laboratory measurements of beam attenuation, this analysis demonstrates that the ratio between the backscattering coefficient b pb and the scattering coefficient b p of the particles is wavelength dependent and not a constant value. The mean values of b pb/b p at the different wavelengths are close to Petzolds 0.017–0.019 from the San Diego Harbor at 515 nm. For all wavelengths and stations (630 observations) the mean value of b pb/b p is 0.020, the standard deviation of the dataset is 0.015, half of the ratios are greater than 0.020, more than 10% are greater 0.030 and about 5% greater than 0.040. The numerical magnitude of these ratios indicates that small particles are the cause of the latter deviations from the ‘Petzold case’. An example from the northern border of the Skagerrak shows a spectral shape of b pb that can be approximated by λ−1.54, λ being the wavelength, while b p is roughly proportional to λ−0.41. The spectral shape of b pb/b p becomes ∼λ−1.12. During a dinoflagellate bloom in the inner part of the Oslo Fjord direct measurements of b pb have revealed a slope of b pb∼λ−1.59. The spectral shape of the mean values of b p(λ) in the Oslo Fjord, based on the 105 observations, can be approximated by the function ∼λ−0.67, while the spectral shapes of b pb(λ) and b pb(λ)/b p(λ) display a greater variety of forms.

Validation of MERIS water quality products in Murchison Bay, Lake Victoria – Preliminary results

As a step forward to validating the retrieval algorithm associated with the MERIS (Medium Resolution Imaging Spectrometer) sensor onboard the European Space Agencys ENVISAT satellite, we compare scaling factors for (442) (pigment absorption coefficient at 442 nm) versus chl-a concentration, and for (scattering coefficient for particles at 440 nm) versus TSM (total suspended matter) concentration, as well as in situ measurements of chl-a and TSM concentrations in Murchison Bay, Lake Victoria, with MERIS level 2 products. Our comparisons show that MERIS chl-a data products underestimate the in situ chl-a concentration by a factor of about 0.6, and that MERIS level 2 products overestimate the in situ TSM concentration and the yellow substance absorption coefficient by a factor of about 2. Application of a local inherent optical property (IOP) scaling factor improves the retrieval provided by MERIS level 2 products. The MERIS level 2 products also underestimate the water-leaving reflectance. These discrepancies between retrievals from MERIS data and in situ measurements are due to the current imperfect atmospheric correction algorithm for inland waters. Therefore, a need arises to develop site-specific bio-optical models and to improve existing MERIS retrieval algorithms.

Spectral optical and bio‐optical relationships in the Oslo Fjord compared with similar results from the Baltic Sea

The spectral shape of the attenuation coefficient c in Baltic waters along the Swedish coast is almost identical to the one in the Oslo Fjord, and there are also strong similarities between these two areas for the absorption coefficient a and the scattering coefficient b as functions of c. However, while the single‐scattering albedo b/c in the Oslo Fjord on an average is 62% and 78% at 425 and 525 nm, respectively, the corresponding numbers for the Gulf of Gdansk are 54% and 61%. Another characteristic difference is that particle absorption plays a much more significant role in the Gulf of Gdansk than in the Oslo Fjord. Thus some parts of the Baltic exhibit similarities with the Oslo Fjord while others display differences, and it cannot be assumed a priori that algorithms from one area will have a general validity in the other. Algorithms for chlorophyll‐a and yellow substance as functions of colour indices, determined for the southern part of the Baltic Sea between the Pommerian Bight and the Gulf of Gdansk, are clearly not valid for the Oslo Fjord, but the result improves by changing their constants by a best‐fit procedure. We conclude that statistical algorithms for remote sensing are not generally interchangeable between the Baltic and the Oslo Fjord, and that the constants of the algorithms have to be locally tuned.