A New Retrieval of Aerosol Optical Depth from Reflectance Ratios on Partly Cloudy Days: Development and Evaluation

The aerosol optical depth (AOD) is an important parameter for describing the Earth's radiation budget and is critical in the context of cloud-aerosol-precipitation problem. Extracting AOD from satellite-measured cloudy-sky reflectances represents a great challenge in remote sensing, where cloud-induced artifacts can be responsible for large (up to 140%) uncertainties in the satellite-derived AOD. We developed a new reflectance ratio (RR) method, which presents an effective way to reduce substantially the impact of the cloud contamination on the retrieved AOD. The RR method relies on the well-known fact that clouds reflect about the same fraction of sunlight regardless of the wavelength, and thus the reflectance ratio is less sensitive to the radiative effects of clouds than reflectance. We examined the performance of the RR method under partly cloudy conditions using a model-based study and demonstrated its ability to yield an accurate estimation of domain-averaged AOD (~10%). We also conducted a case study for a field of single-layer shallow cumuli observed on June 12, 2007 during the Cloud and Land Surface Interaction Campaign (CLASIC) and the Cumulus Humilis Aerosol Processing Study (CHAPS). The CLASIC and CHAPS took place in June 2007 over the U.S. Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) Southern Great Plains site. The case study involves application of the RR method to reflectance ratios measured by the MODIS Airborne Simulator for two pairs of wavelengths (660 and 470 nm and 870 and 470 nm) and comparison of the RR-retrieved AODs with those provided by ground-based Multi-filter Rotating Shadowband Radiometers (MFRSRs). The RR- and MFRSR-derived AODs are shown to be comparable (within 5%) for the 660-nm wavelength. The RR method overestimates and underestimates the MFRSR-derived AODs by 15-30% for 470 nm and 870 nm, respectively. In our presentation we will outline the RR method and show results from the sensitivity and case studies.