Abstract
An algorithm is developed to rapidly estimate cloud properties for a large area from daytime imager data. In this context, a large area refers to a grid cell composed of many imager pixels. The algorithm assumes a gamma distribution to model the subgrid variability in the optical depth and estimates both the mean and the width of the horizontal distribution of optical depth. Optical depth in this study refers to a vertically integrated value at 0.63 μm. Mean values of the cloud-top effective particle radius and cloud-top temperature are also estimated. Retrievals were performed separately for ice and water cloud layers within a grid cell. Applications of this approach to data from NOAA's Advanced Very High Resolution Radiometer (AVHRR) are presented. Simulations indicate that this method performs well for all retrieved parameters except for thin clouds with very broad distributions of optical depth. Comparison of this approach versus rigorous pixel-level retrieval results for an actual scene with multiple cloud layers indicate that comparable performance is achieved with a two to three orders of magnitude increase in computational efficiency. This approach is being implemented into the Clouds from AVHRR (CLAVR) suite of cloud algorithms at NOAA. The computational efficiency of this approach will allow for efficient reprocessing of the entire data record of the AVHRR.
Corresponding author address: Andrew K. Heidinger, NOAA/UW CIMSS, 1225 West Dayton St., Madison, WI 53706. Email: Andrew.Heidinger@noaa.gov
