Abstract
The rainfall characteristics of flash floods are highly sensitive to the joint impacts of different rainfall event features. In this study, we propose an objective framework for identifying rainfall characteristics for flash floods and assessing the risks while considering the combined impacts of multiple rainfall characteristics. The flash flood events are first classified into different types in terms of flood intensity and process using fuzzy C-means clustering with a subtractive clustering algorithm. Strong association rules between rainfall indices and flood types are subsequently identified based on an association rule mining method. The rainfall indices that strongly affect flash flood processes are obtained based on these association rules. Based on the results of the association analysis, the risks of different types of flash floods under different combinations of key rainfall indices are evaluated based on the Bayesian formula and copula function. The association rule analysis with single and multiple rainfall indices demonstrates that the maximum 12-h rainfall intensity and total antecedent cumulative rainfall are the major rainfall characteristics that affect flash flood processes in the study area. We also examine the risk of flash floods under different combinations of maximum 12-h rainfall intensity and total cumulative rainfall. This study provides an effective and quantitative approach to improve the risk analysis of flash floods and advances its application in the risk management of future flash floods.
Original language | English |
---|---|
Pages (from-to) | 2015-2037 |
Number of pages | 23 |
Journal | Water Resources Management |
Volume | 38 |
DOIs | |
Publication status | Published - Apr 2024 |
Scopus Subject Areas
- Civil and Structural Engineering
- Water Science and Technology
User-Defined Keywords
- Flash floods
- Apriori algorithm
- Copula function
- Bivariate risk analysis
- Risk probability