Academic Commons

Theses Doctoral

Stochastic Analysis Of Storm-Surge Induced Infrastructure Losses In New York City

Hwang, Yunji

Hurricanes are among the most catastrophic types of natural hazards, with the potential to cause serious losses in lives and property. While hurricanes rarely have a huge impact on the New York City area, they do have the potential to cause major damage to the city's transportation infrastructure. This research will deal with two main considerations--fragility curves and exceedance curves of vulnerable points in that infrastructure. The primary objective of this study is to provide a model for predicting future hurricane related storm surge patterns and for estimating possible levels of damage from future events in order to develop planning strategies to mitigate against possible damage. The first step is to describe the frequency of past storm surge events in New York City from 1920 to 2012 and determine a probability distribution for hurricane hazard about the maximum daily and yearly storm surges. The second step is to estimate potential probabilistic models by looking at the empirical data on storm surges in New York City. The last step is to concentrate on the reliability assessment for several infrastructures subjected to hurricane loading and storm surges. No significant studies have been conducted using the available empirical data on storm surge heights in New York City, despite the fact that since an observation station was installed in the Battery, New York in 1920, daily and yearly maximum water levels at that location have been documented by the National Oceanic and Atmospheric Administration (NOAA). Considering the available daily maximum sea water levels from 1920 to 2012 yields a total of 31,148 data points (2,394 days of maximum height data are unfortunately missing); 92 data points of maximum sea water levels are also available. This is the first study to utilize the nearly century's worth of empirical data obtained by the observation station at the Battery. Extensive goodness of fit testing (including the use of various probability papers) is performed on the empirical daily maximum sea water level data. It is concluded that the daily maximum sea water levels at the Battery from 1920 to 2012 follow closely a logistic distribution, with a mean value of 8.10 feet and a coefficient of variation (COV) of 9.63%. The methodology of analyzing the yearly maximum sea water levels is quite similar to that used for the daily sea water levels (and the analysis is performed independently). It is found that the yearly maximum sea water levels at the Battery from 1920 to 2012 follow closely a generalized extreme value (GEV) distribution with a mean value of 10.72 feet and a COV of 10.07%. Then, applying exact and asymptotic Extreme Value Theory, the parent GEV distribution is used to determine the probability distributions for maximum sea water levels over a range of different multi-year periods including 1, 10, 50, 100, 200, and 500 years. Finally, the total volume of flood-vulnerable infrastructure is generated and flood damage probabilities when related to the established probability distributions for sea water levels are considered. The flood vulnerabilities of different parts of the built infrastructure in New York City are studied; specifically, the subway system and the tunnel system. The concept of fragility curves is used to express these vulnerabilities. Conclusions and recommendations are provided for estimating losses probabilistically over different periods, retrofitting and strengthening the infrastructure to reduce future potential losses, and determining repair priorities. This is very useful for cost-benefit analysis.

Geographic Areas


  • thumnail for Hwang_columbia_0054D_11356.pdf Hwang_columbia_0054D_11356.pdf application/pdf 3.38 MB Download File

More About This Work

Academic Units
Civil Engineering and Engineering Mechanics
Thesis Advisors
Deodatis, George
Ph.D., Columbia University
Published Here
May 23, 2013