Monitoring land-surface snow conditions from SSM/I data using an artificial neural network classifier

Previously developed Special Sensor Microwave/Imager (SSM/I) snow classification algorithms have limitations and do not work properly for terrain where forests overlie snow cover. In this study, the authors applied unsupervised cluster analysis to separate SSM/I brightness temperature (T/sub B/) observations into groups. Six desired snow conditions were identified from the clusters; both sparse- and medium-vegetated region scenes were assessed. Typical SSM/I T/sub B/ signatures for each snow condition were determined by calculating the mean T/sub B/ value of observations for each channel in the corresponding cluster. A single-hidden-layer artificial neural network (ANN) classifier was designed to learn the SSM/I T/sub B/ signatures. An error backpropagation training algorithm was applied to train the ANN. After training, a winner-takes-all method was used to determine the snow condition. Results showed that the ANN classifier was able to outline not only the snow extent but also the geographical distribution of snow conditions. This study confirms the potential of using cluster means in ANN supervised learning, and suggests a nonlinear retrieval method for inferring land-surface snow conditions from SSM/I data over varied terrain.