Development of Regional Skew Coefficients for Selected Flood Durations in the Columbia River Basin, Northwestern United States and British Columbia, Canada

Scientific Investigations Report 2020-5073
Prepared in cooperation with the U.S. Army Corps of Engineers
By: , and 



Flood-frequency (hereinafter frequency) estimates provide information used to design, operate, and maintain hydraulic structures such as bridges and dams. Failures of these structures could cause catastrophic loss of property, life, or both. In addition to frequency estimates that use annual peak streamflow, frequency estimates of flood durations are required to safely and effectively operate the numerous dams in the Columbia River Basin of the northwestern United States, and British Columbia, Canada. Frequency studies rely on U.S. Geological Survey Guidelines for Determining Flood Flow Frequency (Bulletin 17C, published in 2018). A major consideration in estimating frequencies is the use of skew coefficients, which measure the asymmetry of flood flow distributions. Large uncertainties are associated with estimating the at-site skew coefficients directly from streamflow records, which are limited in length. Skew also is sensitive to extreme events for limited record lengths. Bulletin 17C recommends using regional skew coefficients to weight with the at-site skew estimate for more reliable frequency estimates. In this study, streamflow records from 313 unregulated U.S. Geological Survey streamgage sites and 97 regulated sites with naturalized streamflow records provided by the U.S. Army Corps of Engineers were used to develop regional skew models for the Columbia River Basin. The naturalized streamflow records were synthesized by removing regulatory components such as withdrawals and reservoir storage. Skew models were developed for 1-, 3-, 7-, 10-, 15-, 30-, and 60-day flood durations and used to estimate regional skew coefficients for the Columbia River Basin.

This report used Bayesian statistical regression methods to develop and analyze regional skew models based on hydrologically important basin characteristics. After examining a suite of available basin characteristics, mean annual precipitation had the strongest correlation to skew across the flood durations. Regional skew regression models were fit using mean annual precipitation for selected subbasins in the Columbia River Basin.

Suggested Citation

Lind, G.D., Lamontagne, J.R., and Stonewall, A.J., 2020, Development of regional skew coefficients for selected flood durations in the Columbia River Basin, northwestern United States and British Columbia, Canada (ver. 1.1, October 2020): U.S. Geological Survey Scientific Investigations Report 2020–5073, 48 p.,

ISSN: 2328-0328 (online)

Study Area

Table of Contents

  • Abstract
  • Introduction
  • Data Methods
  • Cross-Correlation Model of Concurrent Flood Durations
  • Flood-Frequency Analysis
  • Regional Duration—Skew Analysis
  • Summary
  • References Cited
  • Appendixes 1–3
Publication type Report
Publication Subtype USGS Numbered Series
Title Development of regional skew coefficients for selected flood durations in the Columbia River Basin, northwestern United States and British Columbia, Canada
Series title Scientific Investigations Report
Series number 2020-5073
DOI 10.3133/sir20205073
Edition Version 1.0: August 2020; Version 1.1: October 2020
Year Published 2020
Language English
Publisher U.S. Geological Survey
Publisher location Reston, VA
Contributing office(s) Oregon Water Science Center
Description Report: viii, 48 p.; 8 Tables; Data Release
Country United States, Canada
Other Geospatial Columbia River Basin
Online Only (Y/N) Y
Additional Online Files (Y/N) Y
Google Analytic Metrics Metrics page
Additional publication details