Methods for Estimating Regional Skewness of Annual Peak Flows in Parts of the Great Lakes and Ohio River Basins, Based on Data Through Water Year 2013

Scientific Investigations Report 2019-5105
By:  and 

Links

  • Document: Report (3.32 MB pdf)
  • Table: Table 1 (99.5 KB xlsx) - Streamgages in parts of the Great Lakes and Ohio River Basins considered for use in regional skew analysis
  • Figures:
    • Figure 1A (5.25 MB pdf) - Map of study area in the Great Lakes and Ohio River Basins showing 4-digit hydrologic units
    • Figure 1B (2.41 MB pdf) - Map of study area in the Great Lakes and Ohio River Basins showing locations of streamgages used in skew analysis
    • Figure 2 (1.94 MB pdf) - Map showing the pseudo record lengths of streamgages in the Great Lakes and Ohio River Basins used in the regional skew analysis
    • Figure 3 (1.95 MB pdf) - Map showing unbiased station skew of streamgages in the Great Lakes and Ohio River Basins used in the regional skew analysis
    • Figure 5 (1.95 MB pdf) - Map showing residuals from constant model of skew for 368 streamgages in the Great Lakes and Ohio River Basins used in the regional skew analysis
  • Data Release: USGS data release - Annual peak-flow data, PeakFQ specification files and PeakFQ output files for 368 selected streamflow gaging stations operated by the U.S. Geological Survey in the Great Lakes and Ohio River basins that were used to estimate regional skewness of annual peak flows
  • Open Access Version: Publisher Index Page
  • Download citation as: RIS | Dublin Core

Abstract

Bulletin 17C (B17C) recommends fitting the log-Pearson Type III (LP−III) distribution to a series of annual peak flows at a streamgage by using the method of moments. The third moment, the skewness coefficient (or skew), is important because the magnitudes of annual exceedance probability (AEP) flows estimated by using the LP−III distribution are affected by the skew; interest is focused on the right-hand tail of the distribution, which represents the larger annual peak flows that correspond to small AEPs. For streamgages having modest record lengths, the skew is sensitive to extreme events like large floods, which cause a sample to be highly asymmetrical or “skewed.” For this reason, B17C recommends using a weighted-average skew computed from the station skew for a given streamgage and a regional skew. This report generates an estimate of regional skew for a study area encompassing most of the Great Lakes Basin (hydrologic unit 04) and part of the Ohio River Basin (hydrologic unit 05). A total of 551 candidate streamgages that were unaffected by extensive regulation, diversion, urbanization, or channelization were considered for use in the skew analysis; after screening for redundancy and pseudo record length greater than 36 years, 368 streamgages were selected for use in the study. Flood frequencies for candidate streamgages were analyzed by employing the Expected Moments Algorithm, which extends the method of moments so that it can accommodate interval, censored, and historic/paleo flow data, as well as the Multiple Grubbs-Beck test to identify potentially influential low floods in the data series. Bayesian weighted least squares/Bayesian generalized least squares regression was used to develop a regional skew model for the study area that would incorporate possible variables (basin characteristics) to explain the variation in skew in the study area. Twelve basin characteristics were considered as possible explanatory variables; however, none produced a pseudo coefficient of determination greater than 5 percent; as a result, these characteristics did not help to explain the variation in skew in the study area. Therefore, a constant model having a regional skew coefficient of 0.086 and an average variance of prediction (AVPnew) (which corresponds to the mean square error [MSE]) of 0.13 at a new streamgage was selected. The AVPnew corresponds to an effective record length of 54 years, a marked improvement over the Bulletin 17B national skew map, whose reported MSE of 0.302 indicated a corresponding effective record length of only 17 years.

Suggested Citation

Veilleux, A.G., and Wagner, D.M., 2019, Methods for estimating regional skewness of annual peak flows in parts of the Great Lakes and Ohio River Basins, based on data through water year 2013: U.S. Geological Survey Scientific Investigations Report 2019–5105, 26 p., https://doi.org/10.3133/sir20195105.

ISSN: 2328-0328 (online)

Study Area

Table of Contents

  • Abstract
  • Introduction
  • Methods
  • Results and Discussion
  • Summary
  • Acknowledgments
  • References Cited
  • Appendix 1. Assessment of a regional skew model for parts of the Great Lakes and Ohio River Basins by using Monte Carlo simulations

Additional publication details

Publication type Report
Publication Subtype USGS Numbered Series
Title Methods for estimating regional skewness of annual peak flows in parts of the Great Lakes and Ohio River Basins, based on data through water year 2013
Series title Scientific Investigations Report
Series number 2019-5105
DOI 10.3133/sir20195105
Year Published 2019
Language English
Publisher U.S. Geological Survey
Publisher location Reston, VA
Contributing office(s) WMA - Integrated Modeling and Prediction Division
Description Report: vi, 25 p.; Data Release; 6 Figures
Country United States, Canada
Other Geospatial Great Lakes and Ohio River Basins
Online Only (Y/N) Y
Additional Online Files (Y/N) Y