Multiclass support vector classification to estimate typhoon rainfall distribution

Research output: Contribution to journalArticlepeer-review

10 Citations (Scopus)


The prediction of typhoon rainfalls provides important information for mitigating disasters. Different from models based on regression, this study proposes a classification methodology for typhoon rainfall estimation. A multiclass support vector classification with a one-against-one scheme is applied to develop the model. Using important typhoon characteristics, including the minimum central pressure, maximum central wind velocity, cyclonic radius and moving track as input variables, the multiclass classification model is able to predict the classes of rainfall depth, duration and type. The methodology combines the predicted classes of three rainfall parameters to estimate typhoon rainfall distribution. Calibration and validation results pertaining to data from 98 typhoon rainfall events in Alishan, Taiwan, show the ability of the proposed classification methodology to estimate typhoon rainfall distribution. Moreover, the proposed typhoon rainfall estimation model can potentially be updated into a real-time forecasting model by including relevant variables.

Original languageEnglish
Pages (from-to)110-121
Number of pages12
JournalDisaster Advances
Issue number10
Publication statusPublished - 2013 Oct

All Science Journal Classification (ASJC) codes

  • Geography, Planning and Development
  • Safety, Risk, Reliability and Quality
  • Environmental Science (miscellaneous)
  • Earth and Planetary Sciences (miscellaneous)


Dive into the research topics of 'Multiclass support vector classification to estimate typhoon rainfall distribution'. Together they form a unique fingerprint.

Cite this