On the Importance of Wave Simulation Techniques for Forecasting Shoreline Change

Dylan Lawrence Anderson1, Jose A.A. Antolinez2, Fernando J. Mendez2 and Peter Ruggiero3, (1)Oregon State University, College of Engineering, Corvallis, OR, United States, (2)University of Cantabria, Ciencias y Tecnicas del Agua y del Medio Ambiente, Santander, Spain, (3)Oregon State University, Corvallis, OR, United States
Abstract:
Global climate change is projected to alter large-scale atmospheric circulation, storm tracks, and consequently the regional wave climates produced by these patterns. Since shorelines naturally evolve towards dynamic equilibrium with the local wave climate, any redistribution of wave energy has the potential to result in morphological changes equal to or greater than those induced by sea-level rise over the next several decades. Because nearly all state of the art coastal modeling frameworks require a representation of the wave climate as input, the development of methodologies that create realistic wave climate scenarios is necessary to forecast possible shoreline change.

Here we use a simple, one-line shoreline change model to assess the importance of wave simulation techniques on shoreline modeling. Our study site, the U.S. Pacific Northwest, exhibits significant seasonal to multi-decadal shoreline variability along relatively straight embayed beaches. One-line models, which calculate spatial gradients of alongshore sediment transport as a function of wave energy flux and angle, can represent this temporal variability if the wave input time series accurately represents the chronology and joint-probabilities of heights, periods, and directions.

Because dynamically downscaling waves from general circulation models is computationally expensive, we explore several statistical input-reduction techniques for constructing time series that capture realistic seasonal to multi-decadal variability and the chronology of storm events. Methods include continuous-time Markov Chains, data mining techniques, fitting of non-stationary distribution functions, auto-regressive logistic models, and trivariate copula dependence structures formed from correlating observed wave records with coincident sea level pressures. The wave climates produced by each method places an emphasis on either the chronological progression or the joint probabilities of the wave parameters or both. We assess variability in forecasted shorelines resulting from each statistical downscaling approach to determine the sensitivity of one-line model results to wave simulation technique, and whether our techniques can discern the signal of multi-decadal variability from their aleatoric uncertainty.