Atmospheric correction of ocean color imagery: Use of Junge power-law size distribution

Date of Award




Degree Name

Doctor of Philosophy (Ph.D.)



First Committee Member

Howard R. Gordon, Committee Chair


In this study, a new atmospheric correction algorithm capable of simultaneously retrieving aerosol and ocean optical parameters under the presence of both highly- and weakly-absorbing aerosols has been developed. It is suggested that the radiative properties of realistic aerosols can be well simulated with those resulting from the Junge power-law aerosol models. The use of the latter makes it possible to vary the atmospheric radiative properties continuously through a variation of the aerosol parameters. The atmosphere is assumed to consist of two, plane parallel and horizontally homogeneous, layers with the Fresnel reflecting bottom boundary. The radiative properties of the ocean water are assumed to be those of Case 1 waters. A system of non-linear equations is constructed for the radiances detected by a multi-band remote sensor and, is subsequently solved using non-linear optimization procedures. The algorithm's performance has been studied with simulated test data. It is shown that the aerosol single scattering albedo (pi0) and the pigment concentration (C) can be excellently retrieved to within 6% and 10% respectively even under the presence of the instrument calibration errors. However, because of significant differences in the scattering phase functions for the test and power-law distributions, large error is possible in the estimate of the aerosol optical thickness. The positive results for the pigment concentration C suggest that the detailed shape of the aerosol scattering phase function is not needed for the atmospheric correction of ocean color sensors. The relevant parameters are the single scattering albedo and the relative spectral variation of the optical depth. The vertical distribution of aerosols and a spectral variation of the aerosol's refractive index have adverse effects on the accuracy of retrievals. Fortunately, these cases are easily identifiable in the course of non-linear optimization procedure as long as the data "fit" objective function SLSQ becomes relatively large, i.e., 5--6%, in contrast to less than 1% in other test cases. The algorithm was incorporated into the SeaWiFS image processing system SeaDAS. The results demonstrate that the algorithm's performance is superior to the NASA Standard Atmospheric Correction algorithm. A significant advantage of the new approach is that realistic multicomponent aerosol models are not required for the retrieval of C.


Physics, Atmospheric Science; Physics, Optics

Link to Full Text