|
Updated:
14 February 2003
Disclaimer:With all these enhancements and algorithms, it must be noted that the satellite sensor is broad band and the unique discrimination of surface materials is often limited. At best the presented algorithms are an indication of occurrence of the labelled features and some will be more definitive than others. While we have presented a powerful on-line remote sensing analysis tool, there are limitations of algorithms in terms of accuracy. An understanding of the properties of marine and coastal materials and their interaction to produce a complex signal at the satellite sensor is required to properly analyse the processed imagery. Variability in scene contents will always result in subjective and relative accuracies of imagery.
Water Depth AlgorithmEstimating water depth from optical reflectance data is valid for viewing bathymetric features but there are inaccuracies due to absorbing and reflecting materials both on the sea-foor and in the water column. For a homogeneous substrate and water-column, water depth is proportional to the logarithm of the radiance just below the water surface (Jerlov,1976). Band 1 penetrates deepest to generally greater than 20 metres water depth but is subject to errors due to the band 1 absorption of water-column and benthic chlorophyll. We chose to use bands 1,2 and 3 in a combination algorithm adapted from Bierwirth et al (1993), computationally less complex although effectively the same, which reduces the errors. This is:Water depth (is proportional to) [ln(b1)/-k1] + [ln(b2)/-k2] + [ln(b3)/-k3] where ki are generalised water attenuation coefficients for band i. Apart from the errors due to submerged vegetation and dark substrates, this algorithm (and any other) is virtually useless for determining water depths in areas where suspended matter is concentrated. |
|