Optically-Thin Cirrus Cloud Radiance Bias in Satellite Radiometric Sea Surface Temperature Retrieval

Abstract:

Satellite-based retrievals of sea surface temperature (SST) are highly sensitive to the optical properties of the atmosphere, including clouds. Cloudy pixels, in particular, are screened in order to avoid potential retrieval contamination in their presence. Due to the lack of continuous in-situ observations across the global oceans, though, SSTs calculated from satellite radiances are often the most practical way to obtain a sufficient global estimate. Cloud clearing techniques struggle to flag cloudy retrievals from passive radiometers with cloud optical depths less than 0.3. These optically-thin clouds are almost exclusively cirrus. Corresponding radiance biases associated with unscreened cirrus can be significant due to their inherently cold cloud top temperatures.

To investigate frequency of such cloud contamination, 1-km SST observations over tropical oceans (±30° latitude) from the Moderate Resolution Imaging Spectroradiometer aboard NASA’s Aqua satellite (AQUA-MODIS) are collocated with cloud profiles from the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) instrument aboard NASA’s CALIPSO satellite. Potential SST biases based on radiance retrievals for MODIS, AVHRR and VIIRS are solved using a radiative transfer model (RTM) with integrated cirrus cloud properties of varying cloud top height and optical depth. Frequencies of occurrence for each cloud top height and optical depth from the collocated CALIOP/AQUA-MODIS data are superimposed upon the conceptual cloud SST radiance bias models to estimate potential net bias. Using the CALIPSO-MODIS collocations, clouds of all types are found to be present in the best quality AQUA-MODIS Level-2 data at a frequency of 25%, with over 90% of those clouds being cirrus. The RTM simulations suggest that when cirrus are present, the mean SST bias due only to cloud is over 0.6°C over the tropical oceans.