HR: 17:15h
AN: OS24B-05 [Abstracts]
TI: Simultaneous Retrieval of Aerosols and Ocean Color Using the Linearized CAO-DISORT Model: Improved
SEAWiFS and MODIS Retrievals and Comprehensive Error Budgets
AU: * Spurr, R J
EM: rtsolutions@verizon.net
AF: RT SOLUTIONS Inc., 9 Channing Street, Cambridge, MA 02138
United States
AU: Stamnes, K H
EM: kstamnes@stevens-tech.edu
AF: Stevens Institute of Technology, Castle Point on Hudson
, Hoboken, NJ 03070
United States
AU: Li, W
EM: wli@stevens.edu
AF: Stevens Institute of Technology, Castle Point on Hudson
, Hoboken, NJ 03070
United States
AU: Eide, H
EM: heide@stevens.edu
AF: Stevens Institute of Technology, Castle Point on Hudson
, Hoboken, NJ 03070
United States
AU: Zhang, K
EM: kzhang@stevens.edu
AF: Stevens Institute of Technology, Castle Point on Hudson
, Hoboken, NJ 03070
United States
AU: Stamnes, J J
EM: JakobJ.Stamnes@fi.uib.no
AF: University of Bergen, Allegaten 55, Bergen, N-5007
Norway
AB:
The coupled atmosphere-ocean discrete ordinate radiative transfer model CAO-DISORT has been given a linearization capability;
we show that the entire radiation field is analytically differentiable with respect to any atmospheric or marine constituent
parameter. One call to the linearized CAO-DISORT model will simultaneously generate simulated upwelling and/or downwelling
radiance fields and analytic Jacobians for any viewing geometry and for any optical depth in either medium. The model also
has a new pseudo-spherical capability for dealing with atmospheric solar beam attenuation in a curved spherical-shell
atmosphere.
With the linearized CAO-DISORT code as the forward model, we use iterative classical least squares inversion methods to
perform simultaneous retrieval of aerosol and marine parameters. The retrieval is based on minimization of chi-square merit
functions appropriate to a direct comparison measured and simulated top-of-atmosphere SeaWiFS and MODIS radiances. The state
vector comprises two boundary layer aerosol parameters (total number density and bimodal fractional weighting) and one or two
marine parameters (chlorophyll concentration and additional CDOM absorption in coastal waters). Information from all
instrument channels is used in the inversion. We present an analysis of radiance residuals for a wide variety of SEAWiFS and
MODIS scenes; retrievals are considerably more accurate that those obtained using traditional atmospheric correction and
water-leaving radiance modeling approaches.
Error sources are four-fold: TOA radiance instrument errors, uncertainty about model parameters assumed known but not
actually retrieved, modelization errors due to forward model assumptions, and smoothing errors arising from a priori
regularization. For a representative set of scenarios, we use the CAO-DISORT linearization capability in conjunction with the
inverse code to generate contribution functions, averaging kernels and model parameter Jacobians, as required for the
computation of these error sources. We present a detailed error and sensitivity analysis for SEAWiFS and MODIS ocean color
and atmospheric aerosol retrievals; this is the first time that comprehensive error budgets have been drawn up for this
remote sensing application.
DE: 3359 Radiative processes
DE: 3360 Remote sensing
DE: 4262 Ocean observing systems
DE: 4264 Ocean optics (0649)
DE: 4275 Remote sensing and electromagnetic processes (0689, 2487, 3285, 4455, 6934)
SC: Ocean Sciences [OS]
MN: Fall Meeting 2005