Papers »
Clustered separated peaks-over-threshold simulation—estimation and uncertainty quantification of hurricane sandy’s return period
The proposed Clustered Separated Peaks-over-threshold Simulation (CSPS) method for estimating storm surge return periods performs a Monte Carlo simulation of storm surge activity based on statistics derived from tidal gauge data. The CSPS separates the data represent hourly observed water level into three independent components to reflect the fact that different physical processes govern different components of water level—sea level rise and tidal cycle are treated deterministically and storm surge is treated stochastically. Consecutive surge exceedances above a chosen threshold are identified as clusters and the set of cluster maxima are fitted to the generalized Pareto distribution. As in traditional POT, threshold exceedance occurrences form a Poisson process in time. The time evolution of storm surge over finite surge duration is simulated deterministically at each stochastic occurrence of the Poisson process. This enables construction of sample storm surge time histories longer than the original dataset. A bootstrapping algorithm quantifies uncertainty by resampling the original exceedance cluster data and constructing alternate sets of sample time histories.Application of the CSPS to tidal gauge data collected at the Battery, New York in lower Manhattan from 1948 to 2013 indicates that the return period of Hurricane Sandy’s peak water level is 80 years with a 95% confidence interval of 35-505 years, significantly lower than other published return periods. The CSPS also shows that the 100-year water level is 5.46 meters above the station datum and that with 1 meter of sea level rise— all other climatological conditions held constant—this water level would become the 50-year event. These results suggest that storm surge hazard in lower Manhattan has, until now, been systematically underestimated.
Author(s):
Madeleine Lopeman
Columbia University
United States
George Deodatis
Columbia University
United States
Guillermo Franco
Guy Carpenter
United Kingdom