Return Period and Risk of Extreme Sea Levels under Non-Stationarity

Friday, 19 December 2014: 10:53 AM
Jayantha Obeysekera, South Florida Water Management, West Palm Beach, FL, United States and Jose D. Salas, Colorado State University, Fort Collins, CO, United States
Heavily populated urban centers and natural areas located in low lying coastal regions are highly vulnerable to sea level extremes. Current practice of using probabilistic methods for predicting future sea level extremes is based on concepts of return period and risk that have been developed under stationarity conditions. Recently, similar concepts under non-stationary conditions have been developed and a paradigm shift is needed to incorporate the newer concepts into practice. Current projections of future sea levels include varying degrees of acceleration with significant uncertainty and often alternative scenarios are considered on a risk-based framework for planning coastal infrastructure. Extension of the traditional concepts of return period and risk into a non-stationary framework is demonstrated by assuming a probability distribution of extreme values that is parameterized as a function of time. Specifically, the Generalized Extreme Value distribution with time-varying parameters has been applied for fitting the distribution of extreme sea level data. The scenario based projection of mean sea level is linked to the location parameter of the extreme value distribution. This allows the estimation of return period and risk associated with future sea levels in a changing world where mean sea level is expected to be accelerating. Confidence intervals for the projections of extreme sea levels using the new paradigm are also presented. The methodology presented here as demonstrated using the data from few tide gages can be used for planning and design of coastal infrastructure in a changing climate. The presentation ends with a discussion of a number of relevant issues involved in considering non-stationarity in planning and design of coastal projects.