On the Determination of Triggering Factors of Coastal Chalk Cliff Collapses in Upper Normandy

(1) (1) (2) (3) (1) Session GM8.3 LETORTU Pauline , CADOR Jean-Michel , LE GENDRE Romain , QUENOL Hervé , COSTA Stéphane EGU 2011 – 1 - 206 Contact: (1) GEOPHEN-UMR LETG 6554 CNRS, University of Caen Lower Normandy, Caen, France Pauline LETORTU (2) IFREMER, Laboratory Environment Resources of Normandy, Port-en-Bessin, France GEOPHEN-UMR LETG 6554 CNRS, , University of Caen Lower Normandy, France (3) Laboratory Costel, Department of Geography, University of Rennes 2-Upper Brittany, Rennes, France Email: [email protected]

Brighton Hastings Dunkerque Bay of Authie Bay of Somme Calais Le Hourdel Isle of Wight INTRODUCTION Bay of Authie RESULTS The English Channel Bay of Somme Cayeux-sur-Mer Bay of Somme Abbeville Somme Le Hourdel The English Channel Le Tréport The English Channel Cayeux-sur-Mer 0.19 m/yr Dieppe Saint-Valéry-en-Caux Ault-Onival Ault-Onival Fécamp Le Havre Le Tréport Rouen 0.20 to 0.51 m/yr Le Tréport Bay of Seine Seine Criel-sur-Mer Ailly cape Penly Caen Berneval-sur-Mer Criel-sur-Mer Bracquemont Bresle The coastline of Upper Normandy (France) is made up of a 130 km long chalk ArquesDieppe study area Hautot-sur-Mer To determine triggering factors, we applied a principal component analysis (PCA) over a 10-day period. Though the role of rainfall, tidal range, and wind seem to be prepon- Quiberville-Sainte-Marguerite-sur-Mer SottevilleSaint-Aubin-sur-Mer Penly Scie Yères Ailly cape PaluelSaint-Valéry-en-Caux Berneval-sur-Mer Eaulne 0.13 to 0.19 m/yr Bracquemont Eletot Saâne Béthune Dieppe Senneville-sur-Fécamp cliff in the central and the oriental parts of the English Channel shoreline N Fécamp Durdent Hautot-sur-Mer Quiberville-Sainte-Marguerite-sur-Mer derant, the factor analysis points to the high complexity of the process of collapse triggering (figure 10). Antifer cape SottevilleSaint-Aubin-sur-Mer (figure 1). Spatial variability of cliff recession occurs due to layer variations in Etretat Varenne PaluelSaint-Valéry-en-Caux Fig. 4: A collapse in Dieppe La Heve Eletot observed (ESTRAN) cape Le Havre Senneville-sur-Fécamp Based on the PCA, excess of water and frost/thaw cycles prove to be the main continental factors triggering coas- Seine N Fécamp Statistical Analysis of Collapses local lithology (figures 2 and 3) or the influence of natural and man-made obs- Rouen Bay of Seine Antifer cape 0 10 km Etretat tal cliff collapses. A climate index [3] summing up these two main continental factors (figures 11 and 12) has tacles to the longshore drift. Such cliff retreat is locally different and its rate can Hydrology Urban area Altitude Seine Main river Major town 0-100 m Identification of Collapsed Volumes Distribution Identification of Triggering Factors The English Channel Secondary city 100-200 m been created. reach up to one meter per year. > 200 m La Heve Le Havre cape Discretizing our Statistical Population into 4 Classes: Principal Component Analysis (PCA) Fig. 1: Location of the Upper Normandy Rouen 0-200 m3 coastline 3 Due to collapses (figures 4 and 5) threatening settlements close to the coast- Bay of Seine 200-1,400 m Biplot (axes F1 and F2: 46,32 %) THE INDEX CREATION 12 0 10 km 1,400-10,000 m3 AVG rainfall 3 8 SD rainfall line, better understanding of this rocky coast erosion is greatly needed. Le Havre Etretat Yport Fécamp St. Valéry en Caux Varengeville Dieppe Penly Le Tréport Lithology Rate of coastal retreat Urban area more than 10,000 m SD water surplus AVG significant 100 m or deficit wave height SD significant 50 m Major town 4 AVG minimum wave height 0 m Santonian Chalk Area of fast retreat temperature AVG water 1) Identification of dry, wet, and transition periods by calculating the water balance over a 10-day period AVG afternoon Paleogene Coniacan Chalk Secondary city surplus or deficit Area of slow retreat 0 tidal range Campanian AVG daily Fig. 5: A collapse in Saint-Martin- AVG thermal Santonian Major Fitting to Statistical Laws maximum wind speed

Upper Turonian Chalk (without flint) F2 (20,53 %) Coniacian amplitude SW fault -4 SD thermal SD daily Turonian aux-Buneaux amplitude maximum wind NE Middle Turonian Chalk (with flint) Fitting the power law distribution Fitting the Gumbel law distribution Cenomanian 3 3 AVG water table speed 0 10 km to class of volume 1,400-10,000 m to class of volume 200-1,400 m Lower Cretaceous -8 elevation SD minimum Source: lithology from [1] and rates of retreat from [2] (Gendarmerie Nationale) Collapse Power law Kimmeridgian 100 10 Collapse Gumbel law SD afternoon SD water tabletemperature 2) Evaluation of the intensity of dry and wet periods by discretizing each period tidal range elevation -12 Fig. 2: Geology in Upper Normandy Fig. 3: Lithology and retreat rate in Upper Normandy coastline [2] -20 -16 -12 -8 -4 0 4 8 12 16 10 5 F1 (25,79 %) coastline [1] Frequency Frequency 1 0 3) Characterization and addition of duration factor using a multiplier 1000 10000 1000 Biplot (axes F1 and F3: 36,07 %) 10 AVG water table AVG afternoon -5 8 elevation tidal range 0 AVG water surplus 3 Collapsed volume (m³) AVG rainfall SD afternoon Collapsed volume (m³) or deficit OBJECTIVES 6 SD rainfall tidal range 3 SD minimum SD water surplus Fitting observed frequency of collapse volume (1,400-10,000 m ) to the power law 4 temperature 4) Possible addition of frost value, taking into account the duration and intensity of the phenomenon 3 or deficit SD water SD significant and fitting observed frequency of collapse volume (200-1,400 m ) to the Gumbel law table elevation groyne 2 AVG thermal wave height amplitude AVG daily 0 SD minimum maximum wind speed groyne The statistical distribution of collapsed volumes was analysed. No scaling temperature SD daily -2

F3 (10,28 %) AVG minimum maximum wind The primary aim of this study is to determine the triggering factors of coastal cliff retreat. power law relationship, or any other laws tested, fit our data taken as a whole. temperature speed -4 By discretizing our statistical population into 4 classes of collapsed volumes, AVG significant -6 wave height we obtained two good statistical fits. The 1,400-10,000 m3 class fits the power Fig. 11: The index creation We seek to contribute to scientific discussion about the predominance of marine processes and/or sub-aerial weathering processes in the trig- shingle accumulation -8 law. There is a scaling power law relationship between event volumes of -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 ger of cliff collapses. Human actions must be taken into account because they modify longshore drift (figure 6) and can increase swell action 1,400-10,000 m3 and chalk production. This relationship implies that erosion F1 (25,79 %) is not dominated by any characteristic event [4]. The class of collapsed volu- The principal component analysis mes 200-1,400 m3 fits the Gumbel law. at the cliff foot with the implementation of artificial structures. for the axes F1-F2 and F1-F3 18

Fig. 10: Statistical analyses and results 16

14 The index in a wet context (figure 12) is a good indicator Fig. 6: Human interventions and their 12 Coefficient 0 effects on the longshore drift [2] Coefficient 1 of coastal collapses. Higher index coefficients correspond Coefficient 2 10 METHODOLOGY Coefficient 3 Coefficient 4 with greater numbers of collapses. Some of these seem to 8 Coefficient 5 Coefficient 6 be linked to runoff (figure 13). 6 Coefficient 7 This study is carried out in collaboration with the non-profit corporation, ESTRAN. Thanks to this partnership, we have a remarkable and unique database describing 332 cliff Collapse falls. Some parts of coastline are highly sensitive to retreat (figure 7). Observed coastal cliff collapses mainly occurred during winter (figure 8). 4 2 Bay of Somme Number of collapses from Saint-Valéry-en-Caux to Le Tréport since 2002 Le Hourdel 0 150

This index allows us to identify 10-day periods that are 2 2 2 2 3 3 3 3 4 4 4 4 5 5 5 5 6 6 6 6 7 7 7 7 8 8 8 8 9 9 9 9 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

The English Channel 150 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 0.19 m/yr / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / 1 4 7 0 1 4 7 0 1 4 7 0 1 4 7 0 1 4 7 0 1 4 7 0 1 4 7 0 1 4 7 0 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 1

Ault-Onival / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

100 conducive to coastal collapse in order to improve risk pre- 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.20 to 0.51 m/yr Le Tréport The English Channel Ailly cape (! Collapse Criel-sur-Mer Cayeux-sur-Mer (!(!(!(!(!(!(!(!(! (!(!(!(! 50 Ailly cape Penly (!(!(! 0.13 to 0.19 m/yr Berneval-sur-Mer (!(!(!(!(! Bracquemont (! Fig. 13: A collapse probably due to runoff Dieppe (!(! Number of collapses Fig. 12: Every 10 days from 2002 to 2009, the climate index was calculated in a wet context, Hautot-sur-Mer (!(!(!(! Saint-Aubin-sur-MerQuiberville-Sainte-Marguerite-sur-Mer (!(!(!(! (!(!(!(!(!! (! Sotteville (!(!(! ((!(!(!(!(!(!(!(!(!(!(!(!(! (! 0 Saint-Valéry-en-Caux (!(!(!(! ((!(!(!(!(!(!(!(! (! and the number of collapses was observed. Paluel (!(!(!(!(!(!(! (!(!(!(!(!(!(!(!! Summer Autumn Winter Spring !(!(!(!(!(!(! ((!(!(!(!(!(!(!(!(!(! (!(!(!(!(!(!(!(!(!(!(!(!(!(!(!(!(!(!(!(!(!(!(!(!(!(!!!! Eletot !((!( ((((!(!(!(!(!(!(!(!(!(!(!(!! Dieppe Senneville-sur-Fécamp N (!(!(! ((!(!(!(!(!(!(!!!(!!!!(!!((!(!!(!!!(!(!(!(! N Fécamp (!(! ((((!((!(((( Source: ESTRAN data (!(!(!(!(!(! Antifer cape (!(!(!(!(!(! Etretat (!(!(!(! Fig. 8: Number of collapses CONCLUSIONS (!(!(!(!(!(! (!(!(!(!(!(!(!(! from Saint-Valéry-en-Caux to Le Tréport Sainte-Marguerite Varengeville-sur-Mer Hautot-sur-Mer since 2002 La Heve Le Havre cape -sur-Mer It is difficult to classify the processes that trigger coastal chalk cliff collapses. Indeed, numerous combinations of factors, and the phenomena of process shifts, or even hysteresis Rouen Bay of Seine 0 3 km Collecting Data (ESTRAN) effects, draw a more complex link between triggering agents and coastal cliff falls. However, we have identified periods of high probability of triggering due to main conti- 0 10 km Rate of coastal retreat Urban area nental factors. This index will prove to be a useful tool in warning and protecting the human population. Area of fast retreat Major town !( Observed collapse Enriching these Data (Volume, Spatial Organisation of Collapse, Processes such as Rainfall) Area of slower retreat Secondary city Municipal boundary Source: Rates of retreat from [2] The English Channel References: Source: ESTRAN data Creation of the Database ESTRAN [1] Costa S., 1997. Dynamique littorale et risques naturels: L’impact des aménagements, des variations du niveau marin et des modifications climatiques entre la Baie de Seine et la Baie de Somme. PhD thesis of l’University of Paris I, 376 p. Fig. 7: Cliff collapses from Sainte-Marguerite-sur-Mer to Dieppe observed since 2002 [2] Costa S., 2000. Réactualisation des connaissances et mise en place d’une méthode de suivi de la dynamique du littoral haut-normand et picard. Final report, Prefecture of Picardy, Contrat de Plan Interrégional du Bassin de Paris (CPIBP). 103 p. Statistical Analysis of Collapses [3] Savouret E., Amat J-P., Cantat O., Filippucci P., 2010. « Chronique du temps vécu à Verdun en 14-18 », Actes de l’Association Internationale de Climatologie, international symposium of Rennes, « Risques et changement climatique », vol. 3, pp. 571-576. The analysis deals with the statistical definition of the relationships between processes and the collapse (figure 9). [4] Dewez T., Chamblas G., Lasseur E., and Vandromme R., 2009. Five seasons of chalk cliff face erosion monitored by terrestrial laser scanner: from quantitative description to rock fall probabilistic hazard assessment. Geophysical Research Abstracts, Vol. 11, EGU2009-8218, 2009, EGU Fig. 9: Methodological framework General Assembly 2009