Ruddick et al. Applied Optics, 39, 897-912 (2000), have extended the standard SeaWiFS atmospheric correction algorithm for use over turbid coastal and inland waters; however, their assumption of a spatially homogeneous constant ratio for the water-leaving reflectances normalized by the sun-sea atmospheric transmittance at 765 and 865 nm, under the simple bb/α model, can result in a significant inaccuracy for highly turbid water. Using the more accurate bb/(α+bb) model, which, more realistically, does not assume spatial homogeneity ratio, the calculated corresponding perturbation for this assumption is several times larger than what Ruddick et al. evaluated under bb/α model. Such perturbation-related error could exceed 10% in highly turbid water with R(8) > 0.05. We suggest using the modified assumption, 1/R(8)=α0/R(7)
+(1-α0)/f; instead of Ruddick's assumption, in SeaWiFS atmospheric correction algorithms for turbid waters. The Maclaurin seriers expansion of such modified assumption is expressed as, R(8)=α0)-1[R(7)+(α0-1)α0-1f1R(7)2+(α0-1)2α0-2f2R(7)3+.....]; -1<(α0-1)α0-1f1R(7)<1
Calibration of α0, an IOPs related regional parameters, is determined on an image-by-image basis from regressive empirical relationship of scatterplot of the-Rayleigh-corrected reflectances for these two bands. The calculated corresponding perturbation for Ruddick's second assumption is several times larger than what Ruddick et al. evaluated under simple bb/α model. Such perturbation-related error could reach 20% for the highly turbid water over the Pearl River estuary.