Best possible:

• the computed bottom reflectance(s) is the best possible estimate of the actual reflectance of shallow bottoms at the time of imaging that may be derived from the imagery used.

Scaling: bottom reflectances are scaled from 0 for black bottom to 200 in image B, or to 250 in spectral image LBS, for the brightest type of shallow water bottom that exists in the area covered by the imagery.

• This scaling scheme is relative to the actual range of bottom brightness which occurs in the shallow area under study.

• A more rigorous scaling scheme would require the computation of actual calibrated reflectances in physical units using the sensor system's gain settings. This is done in 4SM for Landsat 8 and WV2 images.

Error on computed bottom reflectance is an exponential function of error on computed bottom depth.

• Reliable : computed bottom reflectances are fairly reliable from null depth down to approximately halfway to the maximum modelisable depth.

• Far less reliable: but they are far less reliable over the remaining depth range particularly if the water turbidity is locally higher like along some beaches with breaking waves or in certain polluted harbors.

The N-bandscase

• N spectral values of the bottom reflectance are estimated for each shallow water pixel, thus describing the spatial variations, over the shallow area, of the spectral signature of this bottom in the wavebands used.

• This is only possible for bands which exhibit bottom detection: == an average bottom reflectance is computed for these bands.

• All other bands are assigned the average bottom reflectance.

The 2-bands case

• One only value of an average bottom reflectance is estimated for each shallow pixel, thus describing the spatial variations of the bottom's "brightness". This assumes the bottom to be spectrally neutral, i.e. a shade of gray.

The 1-band case : a constant value is written in image_B : Arcachon    Tarawa