Aerosol Impact on Sea Surface Temperature Retrievals: Sensitivity and Modeling

Sea Surface Temperatures (SSTs) are an important measure of our current weather and climate, as well as an essential variable in both short and long term weather forecasting. Reliant on passive sensors, SST retrieval techniques are influenced by changes in atmospheric composition, including aerosols. Many empirically derived retrieval algorithms are based on matching Top of Atmosphere (TOA) Brightness Temperatures (BTs) from the Advanced Very High Resolution Radiometer (AVHRR) to buoy measurements during clear-sky conditions. As such, data is cloud-cleared to remove cloud-contaminated data. However, small, but influential Aerosol Optical Depths (AODs) data cannot be flagged as contaminated and the algorithms incorrectly calculate a cold SST due to the radiometer sensing the cooler, elevated aerosol layer temperature.

The Santa Barbra DISORT Radiative Transfer model is used to quantify the effects of aerosol contamination on retrieved TOA BTs. The calculated radiances are spectrally averaged BTs over each channel and used to calculate a SST using the Naval Oceanographic Office AVHRR algorithms. Using a radiative transfer model allows for the examination of SST changes due to varying AOD, height of an aerosol layer, and the satellite zenith angle (or viewing angle).

Six idealized height distribution types are used to emulate the wide range of observed aerosol distributions. Satellite zenith angles are varied from 0 to 60 degrees.

Moreover, the optical properties of aerosols, including single scatter albedo, asymmetry parameter, and the mass extinction coefficient are varied over the range of observed values.

Daytime SST retrievals are known within 0.5°C (Kilpatrick et al. 2001). Our analysis shows that errors greater than the retrieval uncertainty are observed for AODs greater than 0.25. Based on the AERONET data from Capo Verde (16.73°N , 22.94°W), 65% of the days during the boreal summer are found to have AOD greater than 0.25. Unfortunately, this seasonal peak in dust activity coincides with the active tropical cylogensis season for the region, making accurate SSTs even more vital for prediction purposes.

Annually at Capo Verde, an AOD greater than 0.5 is found to occur, on average, 61 days. Our analysis shows that in the Saharan Air Layer off the east coast of Africa a dust AOD of 0.5 causes errors in SST retrievals ranging from -1.0°C to -4.9°C. Bearing in mind daytime SST retrievals are considered to be accurate within 0.5°C the inaccuracy due to aerosol loading can be up to 10 times the accuracy threshold.

This sensitivity analysis will allow us to evaluate the potential for radiance correction with aerosol forecast models as a next generation approach to SST retrieval. Beyond just SSTs, it is necessary to explore aerosol effects on other atmospheric properties retrieved from infrared instruments, including radiances, which are assimilated directly into numerical weather prediction models with no correction.