Analyzing satellite temperature retrievals over dust-laden fields in the Northeast Atlantic

SST retrieval methods attempt to provide corrections for attenuation due to molecular scattering of water vapor and background aerosol fields. However, negative biases in the retrieved SST arrive when anomalous aerosol content are injected into the atmosphere. Tropospheric aerosols, such as those produced by transport of terrestrial dust seem to have a persistent impact, but at the regional level, that may result in SSTs depressions of 1K [Nalli and Stowe, 2002]. This presents a problem because for most research and operational applications an accuracy of +/- 0.3 K or better over a length of 100km and a timescale of days to week must be attained. [Houghton2001, Nalli and Stowe, 2002]. The problem is not solely restricted to retrievals of fields with aerosols, but also in the incorporation of these fields into Numerical Weather Prediction (NWP) models.

We reevaluate the problem of physically retrieving accurate and precise SSTs, using hyperspectral IR radiative transfer, but now with emphasis on reducing errors associated with aerosol absorption and scattering. We propose applying a physical multi-spectral solution [after Nalli and Smith, 2003], while revisiting the problem of surface emissivity and atmospheric transmission to account for elevated dust fields. Unlike previous models that focused exclusively on emissivity, this model accounts for both quasi-specular emissivity and reflection, while the transmissivity is corrected by applying aerosols fields in forward model calculations with the Community Radiative Transfer Model (CRTM). The results of these corrections are validated against observed measurements from the eastern Atlantic Ocean, which is dominated by Saharan dust throughout most of the year and that is also a genesis region for Atlantic tropical cyclones. These observations are obtained from the NOAA Aerosols and Ocean Science Expeditions (AEROSE) and PIRATA Northeast Extension (PNE) buoys network. We believe that the improved physical SST methodology, has the potential to allow for improved representation of the geophysical state under dust laden conditions.