Probabilistic formulation for storm surge predictions

Alessandro Contento, Hao Xu, Paolo Gardoni

Research output: Contribution to journalArticle

Abstract

The impact of several devastating hurricanes revealed the vulnerability of large areas of the U.S.A. to this catastrophic natural hazard. Therefore, especially considering the uncertainties in future climate and demographic conditions, exposure and vulnerability, there is a strong need to include storm surge predictions in hurricane risk analyses. In general, the models used for storm surge are unsuited for probabilistic studies that require a large number of simulations. This article presents a probabilistic formulation for storm surge predictions that are developed using the combination of a logistic model and a non-stationary random field. The proposed fully probabilistic formulation (i) provides the probability that a location is being flooded, (ii) considers the spatial correlation among the storm surge at different locations and (iii) predicts storm surge at locations that are different from those of the observations used for the model calibration and are not restricted to be alongshore. Such formulation can include the effects of climate change by calibrating the models with high-fidelity simulations that account for the effect of climate change performed using a realistic set of hurricanes. The proposed formulation is used to predict the storm surge in a specific geographic region near the Pamlico River in North Carolina.

Original languageEnglish (US)
Pages (from-to)547-566
Number of pages20
JournalStructure and Infrastructure Engineering
Volume16
Issue number4
DOIs
StatePublished - Apr 2 2020

    Fingerprint

Keywords

  • floods
  • Hurricanes
  • predictions
  • probabilistic models
  • simulation models

ASJC Scopus subject areas

  • Civil and Structural Engineering
  • Building and Construction
  • Safety, Risk, Reliability and Quality
  • Geotechnical Engineering and Engineering Geology
  • Ocean Engineering
  • Mechanical Engineering

Cite this