The corresponding retrieval algorithm
log S = 3.318 + 1.953 log y (4)
appears to yield accurate results in the interval .1<chloroph (mg/m3) «f.
Eqs (2) and (4) exhibit similar accuracies (around + 30 $)
over the whole validity range. =
An iterative procedure was used to determine the atmospheric
correction to the nemoxely measured radiances, through the equation
LW{(A) = [L(A) — LR(A)] — e(A, 670)
(L(670) — LR(670) — LW(670)] (5)
where L(A) = measured (apparent) radiance, LW (A) and LR (A) = contributions
to L (^) due £o the water upwelling radiance and to molecular scattering in
the atmosphere, € (3,670) = ratio of the contributions to the apparent
radiances at wavelength ^ and 670 nm due to scattering by the atmospheric
aerosols. The quantity € (3,670) was inferred from CICS data pertaining to a
selected clean-water area in the scene, where the LW (X) values can be computed
with adequate accuracy.
The iterative computation neglected in the first instance
LU (670) in Eq. (5). The corresponding first onder estimate of the sediment
content (Eqs (2) ox (4)) was used to caleulate the LW (670) term of Eq.(5) gor
the second onder calculation. The procedure was found to be effective, yielding
nesidual errors in the retrieved sediment concentration below 10%.
It is remarked that the proposed method does not depend upon
Limiting assumptions about the homogeneity of the atmospheric conditions over
the CZCS image, due to the proven Low sensitivity to such conditions of the
terme (A,670).
In agreement with the predictions of the quoted sensitivity
analysis, the standard chlorophyll retrieval algorithm
log chl. = A + Blog[R(520)/R(550)) (6)
was shown to be very sensitive to the sediment concentration (at Least for the
chlorophyll content range .1 - 3mg/m3 of the waters considered).
An expression similar to Eq (6), but operating on the variable
Z = [R(520)/R(550)] + a[R(550) — R(670)] (7)
with «~6uas tested and found to be Little affected by sediment (over the range
1. 10 g/m3), while yielding a similar accuracy.
712
