Cloud-Base Height Estimates Using a Combination of Meteorological Satellite Imagery and Surface Reports

Abstract

This paper describes how the combination of a satellite-derived cloud classification with surface observations can improve analysis of cloud-base height. A cloud-base retrieval that combines a cloud classification derived from visible and infrared satellite data with surface reports of cloud base is investigated. A method using the satellite classification to interpret the surface data is compared with a more traditional distance-weighted approach of interpolating the surface data. Cloud-height observations from the U.S. surface synoptic network were merged with a cloud classification of GOES-8 imager data for 235 test images from June 1996. Surface cloud-base height reports were withheld on a revolving basis and used as truth for the cloud-base height predictions from the satellite-based method. The comparison was limited to cloud-base heights of less than 10?000 feet because of biases in cloud-base height reporting at higher altitudes. Results indicate that fusion of the satellite cloud classification with surface cloud-base height reports yields a superior estimate of cloud-base height versus an estimate using only interpolated surface data. This is true even though the surface-only method was given the advantage of always being spatially closer to the control site. Performance improvement is more significant for broken and overcast conditions. In addition, the use of a simple textural measure, derived from the satellite cloud classification, causes the satellite-assisted method to outperform the surface-only method by an even wider margin.