Fuzzy time series for real-time flood forecasting

Chang Shian Chen, You Da Jhong, Wan Zhen Wu, Shien Tsung Chen

Research output: Contribution to journalArticle

7 Citations (Scopus)


This study applied fuzzy time series (FTS) analysis to develop a real-time flood forecasting model to forecast typhoon flood discharges. Two crucial factors that influence the performance of FTS are the partition of intervals of the variable and the defuzzification method. This study examined the effects of various interval lengths and two defuzzification methods, the centroid and the midpoint methods, on the model performance. Criteria of model completeness and consistency principle were used to determine the effective interval length, and analytic results showed that the midpoint method outperforms the centroid method. Two structures of forecasting models were proposed to make multiple-hour-ahead flood forecasts. Validation results from typhoon flood events in the Wu River in Taiwan showed that the proposed FTS model, which is novel in hydrologic forecasting, can effectively forecast flood discharges.

Original languageEnglish
Pages (from-to)645-656
Number of pages12
JournalStochastic Environmental Research and Risk Assessment
Issue number3
Publication statusPublished - 2019 Mar 1

All Science Journal Classification (ASJC) codes

  • Environmental Engineering
  • Environmental Chemistry
  • Safety, Risk, Reliability and Quality
  • Water Science and Technology
  • Environmental Science(all)

Fingerprint Dive into the research topics of 'Fuzzy time series for real-time flood forecasting'. Together they form a unique fingerprint.

  • Cite this