Exatraction of track and intensity patterns from historical tropical cyclones using the self-orgazining map

Abstract

 Extraction of tropical cyclone (TC) patterns in terms of their track and intensity would be useful to reveal the relationship between feature of TC and its socio-enviromental impacts. However, extraction of typhoon patterns is not straightforward because observed typhoon data is comprised of multiple variables such as longitude, latitude and central pressure on typhoon track. This study extracts typical patterns of typhoon track and intensity by applying a nonlinear classification method, the Self-Organizing Maps. In this study, 60 years of best track data from 1951 to 2010 is transformed into the inputs for the SOM, so that each input represents the track and intensity of a typhoon. By employing the SOM to those inputs, we extracted nine track and intensity patterns from 1573 TCs during the past 60 years. The result indicated a track pattern was closely related to large decreases in central pressure. It was also confirmed that life cycle of TC was long when typhoon showed large decreases in central pressure.