North Atlantic wind and wave climate: Observed extremes, hindcast performance, and extratropical recurrence intervals

摘要

An investigation of the extreme offshore wind and wave climate in the mid-Atlantic region has been conducted for the U.S. Bureau of Safety and Environmental Enforcement (BSEE). The overall objective of the project is to assist with the development of Metocean standards for offshore wind farm design, and establish a 100-yr extratropical wind speed and wave height climatology for the specific regions of interest. Specific accomplishments include evaluating and selecting a climatological data base to use for the study, establish a technique for performing the extremal analysis, and generating maps of 100-yr return period wind speeds and wave heights. Measured data from National Data Buoy Center (NDBC) and Scripps Coastal Data Information Program (CDIP) offshore stations were used to characterize the storm climate and to assess the strengths and weaknesses of two North Atlantic Ocean hindcasts. Hindcasts under consideration included the 20-yr USACE Wave Information Studies (WIS) with kinematically adjusted storm winds, and a new 30-yr WAVEWATCH III® hindcast using National Centers for Environmental Prediction (NCEP) Climate Forecast System Reanalysis Reforecast (CFSRR) winds. Using the latest available techniques for wave spectral partitioning, a sea and swell climatology for the region is presented. Robust error metrics quantify hindcast performance in replicating both the observed wave systems and event peak conditions. Each hindcast product was found to have specific strengths and weaknesses. Although the WIS hindcast is shown to have superior winds, the WIS and NCEP wave hindcast results are mixed. As the NCEP product is presently only available for research purposes, the WIS hindcast was selected for use in computation of the final map products. A total of four extremal analysis techniques were evaluated for assessment of extratropical wind and wave storm data. The Empirical Simulation Technique (EST) provided in [14] employs a life-cycle approach to extre- e value analysis. In contrast, the Generalized Pareto Distribution (GPD) [2], Weibull [11], and Generalized Extreme Value (GEV) [22] methods are parametric extrapolations to the data. To evaluate these approaches in our area of interest, extratropical Storm populations were identified at five test stations by applying both Peak over Threshold (POT) and Annual Maximum Series (AMS) techniques (for use with the GEV) to the NCEP 30-yr hindcast which is the longest hindcast currently available. Tropical storms were removed by an automated process linked to the North Atlantic Hurricane Database (HURDAT). The four extremal techniques were applied to the storm populations and comparisons made at the 50- and 100-yr recurrence interval. The selection of a final methodology will be described and resulting maps of the 50- and 100-yr recurrence interval extratropical wind speeds and wave heights will be presented.