Statistical modeling of interannual shoreline change driven by North Atlantic climate variability spanning 2000–2014 in the Bay of Biscay

Geo-Mar Lett (2016) 36:479–490 DOI 10.1007/s00367-016-0460-8 ORIGINAL Statistical modeling of interannual shoreline change driven by North Atlantic climate variability spanning 2000–2014 in the Bay of Biscay A. Robinet 1,2,3 & B. Castelle 2,3 1 1 4 & D. Idier & G. Le Cozannet & M. Déqué & E. Charles 5 Received: 6 May 2016 / Accepted: 2 August 2016 / Published online: 9 August 2016 # The Author(s) 2016. This article is published with open access at Springerlink.com Abstract Modeling studies addressing daily to interannual coastal evolution typically relate shoreline change with waves, currents and sediment transport through complex processes and feedbacks. For wave-dominated environments, the main driver (waves) is controlled by the regional atmospheric circulation. Here a simple weather regime-driven shoreline model is developed for a 15-year shoreline dataset (2000–2014) collected at Truc Vert beach, Bay of Biscay, SW France. In all, 16 weather regimes (four per season) are considered. The centroids and occurrences are computed using the ERA-40 and ERA-Interim reanalyses, applying k-means and EOF methods to the anomalies of the 500-hPa geopotential height over the North Atlantic Basin. The weather regime-driven shoreline model explains 70% of the observed interannual shoreline variability. The application of a proven wavedriven equilibrium shoreline model to the same period shows that both models have similar skills at the interannual scale. Relation between the weather regimes and the wave climate in the Bay of Biscay is investigated and the primary weather regimes impacting shoreline change are identified. For instance, the winter zonal regime characterized by a strengthening of the pressure gradient between the Iceland low and the Azores high is associated with high-energy wave conditions * A. Robinet 1 BRGM, DRP/R3C, 45100 Orléans, France 2 CNRS, UMR 5805 EPOC, Pessac, France 3 Univ. Bordeaux, UMR 5805 EPOC, Pessac, France 4 CNRM/GAME, Météo-France, Toulouse, France 5 CNRS, CNES, IRD, Université Paul Sabatier, UMR 5566 LEGOS, Toulouse, France and is found to drive an increase in the shoreline erosion rate. The study demonstrates the predictability of interannual shoreline change from a limited number of weather regimes, which opens new perspectives for shoreline change modeling and encourages long-term shoreline monitoring programs. Introduction Sandy coasts are complex environments that are under increasing threat posed by anthropogenic pressures and climate change. Shoreline change is governed by myriad nonlinear physical processes interacting through complex feedbacks covering a wide range of spatial and temporal scales (Stive et al. 2002), challenging model developments. Although several complex process-based morphodynamic models have been developed in recent decades, simulations at large temporal scales, i.e., years, are still hardly reliable. Shoreline evolution on timescales from hours (cf. storms) to years has recently been simulated with fair skill using wave-driven empirical equilibrium-based models (e.g., Davidson and Turner 2009; Yates et al. 2009; Davidson et al. 2013; Castelle et al. 2014; Splinter et al. 2014a). These models can also reproduce the interannual shoreline variability that sometimes exceeds the seasonal variability (e.g., Castelle et al. 2014). However, model skills strongly depend on the availability and quality of wave data. The characteristics of waves reaching the coast depend strongly on remote surface atmospheric circulation (e.g., Bacon and Carter 1993; Young 1999; Woolf et al. 2002; Le Cozannet et al. 2011; Charles et al. 2012a; Martínez-Asensio et al. 2016). Because waves are the primary driver of shoreline change along most coastlines, interannual shoreline variability is expected to be related to interannual large-scale atmospheric dynamics. Therefore, directly using atmospheric conditions as inputs in shoreline models appears 480 as an appealing approach. This reduced-complexity strategy may also implicitly account for other drivers such as mean water level fluctuations (Ruggiero et al. 2001; Serafin and Ruggiero 2014). Using a simple approach, Kuriyama et al. (2012) revealed that about 45% of the interannual shoreline variability measured at a NW Pacific Ocean beach can be attributed to largescale climate fluctuations described through a combination of teleconnection pattern indices. Barnard et al. (2015) recently gave new evidence that large-scale atmospheric circulation patterns control unusual, local storm-driven shoreline change around the Pacific Basin, with enhanced erosion along the NW American coast and the SE Australian coast caused by extreme El Niño and La Niña, respectively. Studies focusing on NE Atlantic sandy coasts and climate variability have already highlighted the existence of a relationship between the North Atlantic Oscillation teleconnection (NAO) and the beach sand bar states (e.g., Masselink et al. 2014) or alongshore sediment transport (e.g., Silva et al. 2012; Idier et al. 2013). However, none of these studies addresses the potential link between the large-scale atmospheric circulation and shoreline variability. In addition, these studies used teleconnection pattern indices to characterize the largescale atmospheric circulation, as they are freely available online and easy to use. However, it is also possible to describe large-scale atmospheric circulation and its variability by so-called weather regimes. Weather regimes are recurrent and persistent atmospheric circulation patterns. They are usually identified by cluster analysis (Michelangeli et al. 1995) applied to daily fields of mean sea-level pressure or geopotential height (at a given pressure level) taken over an area of interest. Using this approach, the North Atlantic synoptic circulation can be accurately characterized, as atmospheric data located over the oceanic basin only are used for the weather regime computation (Cassou et al. 2004; Barrier et al. 2013, 2014). In this paper, a simple weather regime-driven shoreline model is implemented to investigate shoreline interannual variability at Truc Vert beach, Bay of Biscay, SW France. A set of 16 seasonal weather regimes (four per season) is computed for the North Atlantic Basin and the shoreline model is tested against a shoreline dataset covering a 15-year period from 2000 to 2014. The relation between weather regimes, waves and shoreline evolution, as well as the model skills are discussed. Geo-Mar Lett (2016) 36:479–490 seasonally modulated waves generated over the North Atlantic Ocean with a mean significant wave height Hs of 1.7 m, a mean peak wave period of 10.3 s and a dominant WNW direction (Castelle et al. 2015). Summer is characterized by the dominance of NW short waves whereas longer and larger waves coming from the WNW prevail in winter. Hs can episodically exceed 8 m during severe winter storms with a peak wave period often larger than 15 s (Ca (...truncated)