Derek M. Burrage

Whitecap (WC) formation due to waves breaking on a wind-roughened sea surface facilitates the exchange of mass, momentum, and energy across the air–sea interface. While approximate analytical electromagnetic (EM) models paired with surface gravity wave spectra have been used to predict surface roughness emissivity enhancement, these methods do not reveal details of the response to foam, breaking wave, and WC geometry. They also ignore possible coupling between the roughness and WC emissivity effects. We report the application of a full-wave finite-difference time-domain (FDTD) EM model to investigate the separate and combined emissivity effects of specific surface roughness profiles, associated WC fields, and overlying spray. The model solves Maxwell's equations directly for an arbitrary free space and dielectric configuration. It is applied to multiple dielectric layers representing foam and spray overlying flat and rough sea surfaces. The foam layer profiles are adapted from Anguelova's L-band radiative transfer model and the rough surface is a statistical realization of the Kudryavtsev gravity wave spectrum. The model is also used to investigate the secondary effect of sea state on emissivity for a given mean square slope, which is the primary factor governing rough surface emissivity enhancement. The accuracy and precision of the FDTD model emissivity estimates and the detectability of WCs using L-band radiometry are assessed under various wind conditions, including those of tropical storm and category 1 hurricane strength. The prospects for performing Monte Carlo simulations for stronger winds and deterministic simulations of breakers with WCs of various void fractions, shapes, and scales are also considered.

The East China Sea (ECS) is often obscured from space in the visible and near-visible bands by cloud cover, which prevents remote sensing retrieval of optical properties. However, clouds are transparent to microwaves, and satellites with L-band radiometers have recently been put into orbit to monitor sea surface salinity (SSS). Previous studies have used the mixing of fluvial colored dissolved organic matter (CDOM) near coasts, where the mixing is approximately conservative over short time scales, to estimate SSS. In this study, the usual relationship between CDOM and salinity in the ECS has been used in reverse to estimate CDOM from remotely sensed SSS in the ECS and compare that CDOM with MODIS data. The SSS data used are 7 day composites from NASA’s Aquarius/SAC-D satellite which has an L-band radiometer. The challenges in using this approach are that 1) Aquarius SSS has coarse spatial resolution (150 km), and 2) the ECS has numerous anthropogenic sources of radiofrequency interference which adds noise to the L-band signal for the SSS retrievals. Despite the limits in the method, CDOM distribution in the ECS can be estimated under cloudy conditions. In addition to all-weather retrievals, an additional advantage of the approach is that the algorithm provides an estimate of CDOM absorption that is unaffected by the spectrally similar detritus absorption that can confound optical remote sensing estimates of CDOM.

The recent and imminent launch of the SMOS and Aquarius satellites carrying microwave L-band radiometers provides an opportunity to map Sea Surface Salinity (SSS) globally with an expected error <; 0.2 psu. However, the accuracy of retrieved SSS depends critically on brightness temperature (Tb) corrections for sea surface roughness (SSR) effects. This paper assesses the performance of representative roughness correction models when compared with published data, and applied to recently-acquired airborne L-band radiometer data. One type of model currently being used to process SMOS data combines a wind-driven gravity wave spectrum that describes SSR, with an electromagnetic (EM) model that determines microwave emissivity, to predict the Tb roughness increment relative to the flat sea response. We find that selection of both the spectral and emissivity models strongly influences the resulting (~1 K) Tb errors. We conclude that more accurate modeling of short wavelength spectral components and their EM influence is needed, to reduce these errors to acceptable levels.