The next requirement is to determine what kind of spectral variation of scattering might be expected from the aerosol in question. To achieve this, Gordon and Wang calculated - for a range of aerosol types (nine in total), including marine, tropospheric and coastal - scattering as a function of 1, and for each aerosol type e(17, 18) was then obtained using eqn 7.8. The estimated value of e(17, 18) from eqn 7.11 will generally be found to fall between two of the model aerosol values. It is then assumed that the general relationship, e(1j, 18) also falls between those for the two selected aerosol models proportionately in the same manner as was found for e(17, 18). The pas(1i) values for the set of SeaWiFS bands in the visible region can then be calculated from the measured value of pas(18).
Since we are assuming that in the visible region multiple scattering is significant, what we now need is [pa(1) + par(1)] rather than pas(1). In fact a near-linear relationship exists between [pa(1) + par(1)] and pas(1), but differing from one aerosol type to another. Once again, the two aerosol models that bracket the aerosol under study are used, proportionately, to provide the relationship by which, using lookup tables, [pa(1) + par(1)] can be obtained from pas(1). Thus, for each individual pixel in the scene, [pa(1) + par(1)] can be calculated, and with the separately calculated values of pr(1) and t(1), using a rearranged form of eqn 7.5 (omitting sunglint)
Was this article helpful?