Online Program Home
My Program

Abstract Details

Activity Number: 488 - Nonstationary and Anisotropic Spatial Processes
Type: Contributed
Date/Time: Wednesday, August 1, 2018 : 8:30 AM to 10:20 AM
Sponsor: Section on Statistics and the Environment
Abstract #330672 Presentation
Title: Nonstationary Flood Frequency Analysis: a Mixed and Pooled Approach
Author(s): Philip Yates* and John Grego
Companies: DePaul University and University of South Carolina
Keywords: Floods; Hydrology; Finite Mixture Models; Changepoints; Nonstationarity

The primary method to model flood flows is flood frequency analysis. Typical methods model annual maximum flows as a stationary process, and the 1% chance flood is computed at the 0.99 quantile (99th percentile) of an assumed underlying distribution. Nonstationarity in a flood series may occur for a variety of reasons. External disturbances, either due to natural and non-natural changes in climate, greatly impact the assumption of stationarity of a flood series. For this analysis, a method for flood frequency analysis is proposed for nonstationary data. The first part of this analysis would be to attempt to detect a changepoint in the flood series. The flood series can then be discretized into known components, so a pooling weight can be calculated as the proportion of the series belonging to each component. Within each component, a test will be performed to assess whether a finite mixture model is needed to estimate that component's flood series' distribution. The point estimate of the 0.99 quantile and its estimated standard error will then be computed.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2018 program