Abstract
In this paper, a framework for identifying tsunami inundated areas using an innovative Generalized Improved Fuzzy Kohonen Clustering Network (GIFKCN) is proposed. GIFKCN hybridizes the Kohonen clustering network with Generalized Improved Fuzzy Partitions FCM (GIFP-FCM) algorithm to build a more efficient and effective neuro fuzzy classifier. GIFKCN classifier combines the advantages of both a neural network and fuzzy systems. A number of spectral indices are computed and the mean values of these indices are used to train the GIFKCN classifier. The novel classifier was applied to identify March 2011 Tohoku tsunami inundated areas in Ishinomaki city. The performance of the classifier is satisfactory with high overall accuracy and Kappa coefficient.