Nadal-Caraballo, N.C.; Melby, J.A., and Gonzalez, V.M., 2016. Statistical analysis of historical extreme water levels for the U.S. North Atlantic coast using Monte Carlo life-cycle simulation.

A statistical analysis of extreme water levels was performed for 23 locations throughout the U.S. North Atlantic coast. Extreme value analysis, which focused on historical observations, was followed by the application of a Monte Carlo life-cycle simulation methodology. This study was part of a broader effort to quantify coastal flooding hazards in this region. Twenty-three stations were selected based on location and record length, meeting the requirement of a minimum of 30 years of hourly water-level measurements. Monthly maxima data were also used to complement the hourly water-level observations. The use of available water-level data was maximized through the development of partial duration series that combined both monthly maxima and hourly data. A generalized Pareto distribution was used to fit combined partial duration series corresponding to each of the 23 locations. The Monte Carlo life-cycle methodology was used to simulate 10,000 cycles of 100 years each, effectively extending the record lengths of extreme events through statistical simulation. A bootstrapping technique was used as part of Monte Carlo life-cycle simulation in order to develop probability distributions of extreme water levels, including mean, as well as 10% and 90% nonexceedance confidence limits (equivalent to an 80% confidence interval). Water-level probabilities determined in this study were compared to results from a previous effort where the generalized extreme value distribution was used to fit monthly maxima data.