Cloud-Property Retrieval Using Merged HIRS and AVHRR Data

Abstract

A technique is developed that uses a multispectral, multiresolution (MSMR) method to improve the overall retrieval of mid-to high-level cloud properties by combining HIRS sounding channel data with higher spatial resolution AVHRR radiometric data collocated with the HIRS footprint. Cirrus cloud radiative and physical properties are determined using satellite data, surface-based measurements provided by rawinsondes and lidar, and aircraft-based lidar data collected during the First ISCCP (International Satellite Cloud Climatology Program) Regional Experiment (FIRE) in Wisconsin during the months of October and November 1986. HIRS cloud-height retrievals are compared to ground-based lidar and aircraft lidar when possible. Retrieved cloud heights are found to have close agreement with lidar for thin cloud, but are higher than lidar for optically thick cloud. The fact that the retrieved cloud height is higher than lidar for optically thick cloud is probably due to the attenuation of the lidar signal before the signal reaches through the cloud, while the satellite is viewing the cloud from above. AVHRR visible (0.63-?m) and infrared (11-?m) radiances are analyzed to determine the cloud emittances and reflectances collocated with each HIRS pixel. The bidirectional reflectances from the AVHRR visible-channel data are corrected for solar direct and diffuse surface reflection to isolate the cloud reflectance. The individual AVHRR pixel emittances are calculated using the cloud-top temperature derived from the HIRS cloud-retrieval analysis. The results of the reflectance-emittance relationships derived in this fashion are compared to theoretical scattering model results for both water-droplet spheres and randomly oriented hexagonal ice crystals. It is found that the assumption of 10-?m water droplets is inadequate to describe the reflectance-emittance relationship for the ice clouds seen here. Use of this assumption would lead to lower cloud heights using the ISCCP approach. The theoretical results show that use of hexagonal ice-crystal phase functions could lead to much improved results for cloud retrieval algorithms using a bispectral approach.