Atmospheric correction of MERIS data

The main steps in the atmospheric correction of MERIS data are

1. Cloud detection
2. Rayleigh correction - for molecular (Rayleigh) scattering and absorption by gases - e.g. ozone
3. Application of land / sea masks
   (aerosol correction takes place only over water - sea and large lakes)
4. Sun-glint identification
5. White-cap detection
6. Correction for scattering by aerosols (Mie scattering):

   detection of Case II water (needs a different aerosol correction algorithm) determine aerosol type (continental / marine), determine aerosol optical depth, calculate aerosol transmittance calculate signal contribution from aerosol scattering.

The two types of L2 ocean products arising from this processing are:

1. Water-leaving radiances reflectances in 14 bands (band 11, the oxygen absorption band is omitted).
2. Level 2 confidence flags, particularly PCD_1_13 and PCD_1_14 -for more on flags, see MERIS Level 2 Flags.

You can find more information about these processing steps in the MERIS Handbook, section 2.7 Level 2 Products and Algorithms. From the handbook website you may also download the ATBDs (Algorithm Theoretical Basis Documents) that describe the different processing steps and gives the scientific reasoning behind them. The handbook also contains a useful flow diagram on Level 2 processing over water.

Meanwhile, a few brief points to remember about atmospheric correction over water:

• Correction for molecular scattering (Rayleigh scattering) is easily handled, as only the illumination and observation conditions have to be known, although a knowledge of atmospheric pressure and wind speed at the sea surface is also needed for accurate calculations.

• Correcting for aerosol scattering is a much harder problem. The optical properties and aerosol distribution (vertically and through the field of view), are not known when processing an ocean color scene, but have to be inferred from the data. Aerosol correction starts by collecting information about aerosol from the radiances measured in near-infrared bands (> 700 nm) where the water-leaving reflectances are assumed to be zero over Case I waters. This knowledge is then extrapolated toward visible wavelengths.

• Atmospheric correction proceeds differently for Case II water, mainly because the strong scattering by sediment particles often found there water may interfere with the attempt to calculate aerosol content from the NIR bands. Turbid water flagging is therefore necessary before choosing which aerosol model to use.