Cloud Data Set for Neural Network Classification Studies

摘要

A large set of GOES imagery has been analyzed to generate a cloud data set forneural network classification studies. The classification scheme uses 13 classes of clouds. The cloud data base is large enough to allow cross-validation tests of automated cloud classification techniques. The methods used to acquire the data on cloud type and a brief overview of the planned automated classification experiments are presented. The appendix contains an inventory of the cloud data set.