Chargement en cours


Where and When

31/10/2019 11:00 - D-104


Jennifer Montano de l'Université d'Auckland 


Predicting shoreline evolution is one of the key challenges in coastal studies with far-reaching economic and societal implications. Yet, the capability of models to predict shoreline erosion/accretion is limited at both the short (storms) and long (multi-annual) temporal scales, which has inspired development of a growing variety of models based on different approaches and techniques. Changes in the shoreline occur over temporal scales ranging from seconds (e.g., individual waves) to hours (e.g., storms), months (e.g., seasonal wave energy modulation) and decades (e.g., wave climate). Generally, during storms, the shoreline moves landward (erosion), while during calm periods the shoreline moves seaward (accretion). Shoreline changes occurring over much larger timescales (decadal to centennial) can be the result of other factors like longshore sediment transport gradients, changes in sediment supply and sea level rise (SLR), among others.

We present a new methodology to predict shoreline position based on Complete Ensemble Empirical Mode Decomposition (CEEMD). CEEMD decomposes the time series into a finite set of “intrinsic mode functions” (IMFs) which represent different temporal scales. This method overcomes limitations of Fourier-based methods for time series analysis (e.g. FFT and wavelet techniques) since the method was designed to analyse non-linear and non-stationary phenomena, identifying also processes and patterns hidden in the data (Huang et al., 1998; Torres, 2011).

The different temporal scales (IMFs) found in the drivers (sea level pressure fields SLPF and/or waves) are linked to the IMFs of the shoreline position (e.g. annual variability in the shoreline position can be predicted by using the annual variability in the drivers). Shoreline position is computed as the sum of all individuals scales (e.g., monthly, annual, bi-annual and trends).  We applied the new methodology to some of the longest datasets of daily shoreline evolution, currently available, Tairua beach, New Zealand (18 years) and Narrabeen, Australia (11 years). Time-scales longer than annual were better predicted through SLPF while faster oscillations were better predicted using wave characteristics (wave height, period and direction). Shoreline predictions were compared with the ShoreFor model (Davidson et al., 2013) displaying improvements, showing that shoreline response can be decomposed and predicted using an approach that isolates the main temporal scales.

En poursuivant votre navigation sur ce site, vous acceptez nos conditions d'utilisation notamment l’utilisation de cookies afin d'améliorer la qualité de vos visites et réaliser des statistiques.
Mentions légales / Politique de confidentialitéX