Bivariate Hydrologic Risk Assessment of Flood Episodes using the Notation of Failure Probability

Shahid Latif,Firuza Mustafa

doi:10.28991/cej-2020-03091599

Abstract

Floods are becoming the most severe and challenging hydrologic issue at the Kelantan River basin in Malaysia. Flood episodes are usually thoroughly characterized by flood peak discharge flow, volume and duration series. This study incorporated the copula-based methodology in deriving the joint distribution analysis of the annual flood characteristics and the failure probability for assessing the bivariate hydrologic risk. Both the Archimedean and Gaussian copula family were introduced and tested as possible candidate functions. The copula dependence parameters are estimated using the method-of-moment estimation procedure. The Gaussian copula was recognized as the best-fitted distribution for capturing the dependence structure of the flood peak-volume and peak-duration pairs based on goodness-of-fit test statistics and was further employed to derive the joint return periods. The bivariate hydrologic risks of flood peak flow and volume pair, and flood peak flow and duration pair in different return periods (i.e., 5, 10, 20, 50 and 100 years) were estimated and revealed that the risk statistics incrementally increase in the service lifetime and, at the same instant, incrementally decrease in return periods. In addition, we found that ignoring the mutual dependency can underestimate the failure probabilities where the univariate events produced a lower failure probability than the bivariate events. Similarly, the variations in bivariate hydrologic risk with the changes of flood peak in the different synthetic flood volume and duration series (i.e., 5, 10, 20, 50 and 100 years return periods) under different service lifetimes are demonstrated. Investigation revealed that the value of bivariate hydrologic risk statistics incrementally increases over the project lifetime (i.e., 30, 50, and 100 years) service time, and at the same time, it incrementally decreases in the return period of flood volume and duration. Overall, this study could provide a basis for making an appropriate flood defence plan and long-lasting infrastructure designs. Doi: 10.28991/cej-2020-03091599 Full Text: PDF

Highlights

Nowadays, flood events are characterized as one of the most severe and disastrous naturally occurring hydrologic consequences across the world, and the risk of their occurrence will increase in the future due to the global and regional climate-changing scenario [1,2,3]
Few commonly used onedimensional parametric family functions, such as Gamma-3P, Generalized Extreme Value (GEV), Generalized Gamma-3P, Inverse Gaussian-2P, Johnson SB-4P, Lognormal-2P and Weibull-2P, are selected and tested as candidate models, and all the distribution fitting procedures are carried out using the Easyfit software (MathWave Technologies 2004–2017)
The selected copula was employed in deriving the joint return periods (RPs) for both the ‘OR’ and ‘AND’ case for a different possible combination of flood characteristics

Summary

Introduction

Flood events are characterized as one of the most severe and disastrous naturally occurring hydrologic consequences across the world, and the risk of their occurrence will increase in the future due to the global and regional climate-changing scenario [1,2,3]. The potential damage could likely be a function of several intercorrelated flood characteristics, such that ignoring the spatial dependency among them might contribute to underestimation of the uncertainty distributed over the estimated flood design quantiles, and often demands more flood characteristics, i.e. based on the joint probability density function (or JPDF) and joint cumulative distribution functions (or JCDF) [13] This is especially so from the prospect of hydraulic or flood defence infrastructure design procedures, where the accountability of multivariate design parameters could be a feasible desire [14, 15]. Numerous studies have performed the copula-based multivariate joint analysis of the flood characteristics and have estimated the design variable quantiles under different notations of return periods, i.e., based on joint distribution, conditional joint distribution or Kendall’s distribution [14, 15, 19 and references therein]

Methods

Results

Conclusion