Develop a Generalized Skew Update and Distribution Shape for Texas for Texas Flood

This project will investigate and document generalized (regional) skew coefficients (RegSkew) and other measures of distribution shape in and near Texas. RegSkews, which are derived by procedures integrating sample at-site skew values obtained at many streamgages, are important for peak-streamflow frequency (flood frequency) analyses because of the high sampling variability. The measures of distribution geometry are normally derived from study of USGS instantaneous annual peak streamflow data and ancillary watershed properties. However, identification of applicable time periods of the USGS observational record is complex and critical for execution of this study, and the USGS peak-values database provides only qualitative information to this effect. New RegSkews are therefore needed for hydrologic design because previous studies (nationally c.1982; Texas c.1996) are outdated relative to Federal guidelines (England and others, 2018 [Bulletin 17C]). Future flood frequency analyses will inherently be more reliable and with decreased uncertainties when new RegSkews are used and in particular used with the Expected Moments Algorithm of Bulletin 17C and other settings. Bulletin 17C currently recommends Bayesian generalized least squares (B-GLS) concepts to estimate RegSkews because B-GLS reflects the precision of available estimates, their cross correlations, and the precision of the regional model. This project will report on the results of B-GLS for Texas. The complexity of the Texas flood hydrology, due to a broad spectrum of wide ranging climatic, rural to urban development conditions, and potential flood-flow regulation effects, requires further research of spatial and temporal trends in annual peaks and empirical distributions. Further RegSkew and other measures of distribution shape concepts and methods that incorporate machine learning and generalized additive models will be explored in this project to fully discern probability distribution shape and prediction for the distal tail estimation of flood frequency. The project will also produce products and training materials suitable for self-training and inclusion in workforce development facilitated training.

Language

  • English

Project

Subject/Index Terms

Filing Info

  • Accession Number: 01743926
  • Record Type: Research project
  • Source Agency: Texas Department of Transportation
  • Contract Numbers: 0-6977
  • Files: RiP, STATEDOT
  • Created Date: Jun 24 2020 1:50PM