Characterization of present and future precipitation through bivariate copulas and its use in risk assessment of infrastructure
conference paper
Extreme rainfall is often investigated through Depth-Duration-Frequency (DDF) or Intensity-Duration-Frequency curves (IDF). These describe either rainfall depth or intensity as a function of duration for given return periods or probabilities of exceedance. This standard approach relies on a linear regression based on a limited number of observations. In this paper a different approach for the characterization of precipitation is proposed. We use rain gauges data from more than 30 measuring stations, containing data from 1951 onwards in the Netherlands. The data corresponds to measurements of rain amount (mm) corresponding to rain duration (in fractions of 6 minutes). With this data periods of continuous rain (which are defined as showers) are studied. Three parametric models are analyzed in order to characterize the joint distribution of rain amount and rain duration per shower. These parametric models are the Gaussian, Gumbel and Clayton copula. These are selected in order to capture different patterns of dependence including tail dependence. The results show that the Gumbel copula (upper tail dependence) provides the best description of the data of interest. Spatial and temporal variations are discussed in terms of the data and in terms of Structured Expert Judgment. For this last step a panel of 8 experts on the subject was gathered and a structured elicitation through Cooke’s model was performed. Finally the use of the statistical model in risk analysis is briefly exemplified with unavailability of transport due to flooding of tunnel. © 2015 Taylor & Francis Group, London.
Topics
TNO Identifier
954665
ISBN
9781138028
Publisher
CRC Press/Balkema
Source title
Proceedings of the 25th European Safety and Reliability Conference, ESREL 2015, 25th European Safety and Reliability Conference, ESREL 2015, 7 September 2015 through 10 September 2015
Place of publication
London
Pages
4295-4302
Files
To receive the publication files, please send an e-mail request to TNO Repository.