Evolution of the MERIS / OLCI Bright Pixel Atmospheric Correction: Accounting for Glint, and Spectral Backscatter.
Gerald, Moore1; Jean-Paul, Huot2; Constant, Mazeran3

The bright pixel atmospheric correction (BPAC, Moore et al 1999) has been part of the operational processor for MERIS on ENVISAT since its launch, and is implemented for the proposed OLCI processor. The BPAC consists of a coupled atmosphere-hydrological model that is parameterised by the Angstroem exponent, the properties of pure water and those of particulate matter in the near infra-red (NIR). The BPAC has been used in the operational processor for the MERIS, in order to correct for the Antoine and Morel (1999) atmospheric correction for excess NIR reflectance in case 2 or other waters, where backscatter from particulates results in a significant NIR signal. As part of the evolution for the potential 4th reprocessing of the MERIS archive, and for the OLCI on Sentinel 3 the BPAC has been substantially upgraded.
A new version of the BPAC has been developed for the 4th MERIS reprocessing; the revised algorithms uses an analytical solution, and in addition to modelling the spectral signature of particulates and atmosphere, it accounts for the specular scatter from sun glint.
Hitherto the glint component in MERIS and other ocean colour sensors has been estimated from the wind-field and a Cox and Munk or similar model of wave slope. This approach has disadvantages in terms of operational forecasting where the true wind field in not know in near real time. The modelled wave slope assumes that the wave slope has reached a steady state and that there are no effects due to land. For the coastal zone, where the BPAC can provide NIR backscatter estimates, the wave-slope and wind-field can be influenced by the nearby by both the land topography and the local bathymetry. Example images are shown of retrieved glint in coastal inlets, and around islands where noticeable wakes are visible. The retrieved glint is also compared with that derived from a Cox and Munk mode;
The BPAC produces estimates of the spectral backscatter in the NIR, although these are not available as routine MERIS products. In contrast to the visible spectral regions currently used for geophysical algorithms, the NIR / Red spectral region shows relatively little CDOM absorption, thus the spectral backscatter provides a robust estimate of TSM. The TSM product is using in-situ data from the North-Western Baltic Sea, a mesotrophic region in the south-west of Portugal and the NOMAD dataset. Using the NOMAD data, the potential to use the BPAC to provide a long term European time series of backscatter using MERIS is demonstrated.
The spectral backscatter derived from the BPAC enables estimates of spectral absorption to be made in the red spectral region. The estimates of absorption are used to provide estimates of chlorophyll absorption, and the derived chlorophyll, since the pigment / packaging effect is considerably reduced at red wavelengths. The chlorophyll derived is validated using in-situ data from the North-Western Baltic Sea and the NOMAD dataset.