A stochastic event-based approach for flood estimation in catchments with mixed rainfall and snowmelt flood regimes

Valeriya Filipova,Deborah Lawrence,Thomas Skaugen

doi:10.5194/nhess-19-1-2019

Abstract

Abstract. The estimation of extreme floods is associated with high uncertainty, in part due to the limited length of streamflow records. Traditionally, statistical flood frequency analysis and an event-based model (PQRUT) using a single design storm have been applied in Norway. We here propose a stochastic PQRUT model, as an extension of the standard application of the event-based PQRUT model, by considering different combinations of initial conditions, rainfall and snowmelt, from which a distribution of flood peaks can be constructed. The stochastic PQRUT was applied for 20 small- and medium-sized catchments in Norway and the results give good fits to observed peak-over-threshold (POT) series. A sensitivity analysis of the method indicates (a) that the soil saturation level is less important than the rainfall input and the parameters of the PQRUT model for flood peaks with return periods higher than 100 years and (b) that excluding the snow routine can change the seasonality of the flood peaks. Estimates for the 100- and 1000-year return level based on the stochastic PQRUT model are compared with results for (a) statistical frequency analysis and (b) a standard implementation of the event-based PQRUT method. The differences in flood estimates between the stochastic PQRUT and the statistical flood frequency analysis are within 50 % in most catchments. However, the differences between the stochastic PQRUT and the standard implementation of the PQRUT model are much higher, especially in catchments with a snowmelt flood regime.

Highlights

The estimation of low-probability floods is required for the design of high-risk structures such as dams, bridges and levees
Floods with a 100-year return period are sometimes required for the design of levees and the design and safety evaluation of high-risk dams requires the estimation of flood hydrographs for the 1000-year return period and, in some cases, floods with magnitudes of up to the probable maximum flood (PMF)
We have presented a stochastic method for flood frequency analysis based on a Monte Carlo simulation to generate rainfall hyetographs and temperature series to drive a snowmelt estimation, along with the corresponding initial conditions

Summary

Introduction

The estimation of low-probability floods is required for the design of high-risk structures such as dams, bridges and levees. Flood mapping usually requires input hydrographs for flood events with return periods of up to 1000 years Methods for estimating these floods can be generally classified into three groups: (1) statistical flood frequency analysis, (2) the single design event simulation approach and (3) derived flood frequency simulation methods. When return periods that are longer than the observed record length are needed, the process requires extrapolation of the fitted statistical distribution This introduces a high degree of uncertainty due to the number of limited observations relative to the estimated quantile Significant progress has been made in methods for reducing this uncertainty by incorporating historic or paleo-flood data (Parkes and Demeritt, 2016), where available Another way to “extend” the hydrological record in order to reduce the uncertainty is to combine data series from several different gauges by identifying pooling groups or hydrologically similar regions, where this is possible.

Objectives

Findings

Discussion

Conclusion