Cloud base height estimates from combining a satellite cloud classification with surface reports

Surface cloud base height observations over the U.S. from June 1996 are used to examine two approaches to estimating cloud base height at a location without surface observations. An older, nearest-neighbor type, surface observations only approach is compared to a method which uses a satellite-derived cloud classification plus surface observations. GOES-8 infrared and visible imagery is used as input to the bispectral histogram cloud classifier. Errors from each method are compared. Results show that by combining the areal extent of satellite classifications with the surface reports, a superior estimate of cloud base is obtained. Textural measures obtained from the classification further improve the performance of the new combined technique.

Topic for discussion: How well do multispectral satellite signatures correspond to variations in cloud base?