A Decision Tree Algorithm for Freeze/Thaw Classification of Surface Soil Using SSM/I

摘要

A decision tree algorithm was developed to classify the surface soil frozen/thawed status based on the microwave emission and scattering characteristics of the frozen/thawed soil, and the cluster analysis of samples from the frozen soil, the thawed soil, the desert and the snow. The decision tree works with five SSM/I channels (19V, 19H, 22V, 37V, 85V) and three dominant indices, including the scattering index, the 37GHz vertical polarization brightness temperature and the 19GHz polarization difference. The algorithm also takes the scattering effect of the desert and the precipitation into consideration. The classification results validated by the 4cm-depth soil temperature and the Map of the Geocryological Regionalization and Classification in China showed satisfied accuracy.