1Any rise in sea level will have adverse impacts (such as coastal erosion and flooding) depending on the time scale and the magnitude of the rise and the human response to it (Paskoff, 1993). Recent climatic models summarized by the Intergovernmental Panel on Climate Change have predicted a significant warming and global sea-level rise for the 21st century (IPCC, 2001), that is expected to increase the flooding risk on low coasts, like most of the Gulf of Lions. Flooding risk is particularly high during atmospheric storms (Bouligand and Pirazzoli, 1999). The overwash and erosion phenomena are strengthened when the sea surge occurs during a strong astronomical high tide (Pirazzoli et al., 2006). In a low-lying area, in association with shallow waters offshore, like most of the Gulf of Lions, strong surges can cause flooding and damage. It is therefore important to better understand the possible regional and local trends of sea surges and of the associated meteorological forcing factors.
2The purpose of our paper is to analyse the meteorological conditions and the flooding risk in the Gulf of Lions associated with sea surges, defined as the difference between the observed sea level and the astronomical tide level. The Gulf of Lions is the North-Western Gulf of the western Mediterranean Sea (Fig. 1). It extends over 200 km, from Port-Vendres to Marseilles and shows two major coastal orientations: an east-west component on the eastern part and a north-south component on the western part with some local discrepancies to this general model (Fig. 1). The coastal area is mostly composed of sand and low beaches (excepted at both ends of the Gulf, near Marseilles and Port-Vendres), opened to shallow water, and is densely populated (Sabatieret al., 2004).
3When each sea surge is considered, the primary forcing is associated with the passage of extra-tropical storms (Pirazzoli, 2000; Pasaric and Orlic, 2001; Svensson and Jones, 2002; Trigo and Davies, 2002; Moron and Ullmann, 2005). Travelling mid-latitude low pressure systems act to raise the sea level below them, but this effect is quite weak in semi-enclosed basins like the Mediterranean Sea. The most important meteorological factors are the associated winds. The Mediterranean Sea is not on the main storm-track of the European and North Atlantic areas (Rogers, 1997) but travelling or stationary low-pressure systems can occur there (Alpertet al., 1990). The Gulf of Lions is also widely open to onshore southerly winds.
4This study analyses firstly the spatial characteristics of extreme sea surges around the Gulf of Lions. Sea-surge height variations at hourly and daily time scales are analysed at three stations (from west to east; Port-Vendres, Sete and Grau-de-la-Dent; Fig. 1) between 1986-1995. Then, the study focuses on the meteorological conditions associated with sea surges around the Gulf of Lions. Barometric conditions and local-scale winds associated with extreme sea surges are then studied. Linear trends and slow variations in the frequency and length of sea surge related-winds are analysed from 1949 to 2003 to investigate the long-term variations of these forcing factors. Finally, the return time of extreme sea surges is analysed to study the flooding risk associated with sea surges. The aim is to improve our understanding of the possible influence of climate change on the evolution of regional and local meteorological conditions associated with sea surges and flooding risks around the Gulf of Lions.
Fig. 1 - Migration of the bars
Localisation des marégraphes et des stations météorologiques.
(in grey, the location of the seawall).
5This work analyses hourly sea-level height measured at three tide-gauge stations along the Gulf of Lions (Fig. 1): Grau-de-la-Dent – GD – (43°3N - 5°E), Sete – SE – (43°1N - 3°4E) and Port-Vendres – PV – (43°01N - 3°02E) from 1986 to 1995 (Fig. 1; Table 1). These time series have been produced through the digitisation of the original paper records (marigrams) of sea levels using an integrated and automated tool kit for the digitization, transformation and validation of marigrams called NUNIEAU (NUmerisation des NIveaux d’EAU) (Ullmannet al., 2005). As sea surges occur mostly in winter from October to March (Bruzzi, 1996), this paper focuses on this wintertime period. Sea-level height data are expressed in hours UT+0 and in the same altimetric reference, i.e. the Nivellement general de la France (NGF).
6The hourly astronomical tides and sea surges at GD, PV and SE are computed using the POLIFEMO software (Tomasin, 2005). This software allows to compute the astronomical tide everywhere at hourly time-scales. POLIFEMO performs a least-squares fit (Foreman, 1977) of the available data with seven constituents, i.e. M2, S2, N2, K2, K1, O1, P1 (respectively “mean lunar period”, “mean solar period”, “major elliptic lunar period”, “luni-solar declinational semi-diurnal period”, “luni-solar declinational diurnal period”, “major lunar period”, “major solar period”). We verified for each year, that the sum of all hourly positive and negative sea surges is null. The astronomical tide is weak in the Mediterranean Sea and oscillates between +15 cm and -15 cm in this part of the basin (Ullmann et al., 2006). In the following, the term “extreme surge” corresponds to a sea surge ≥ 60 cm.
7Three-hourly wind direction and speed measurements from nearby stations were provided by Météo-France at Port-Vendres (1949-2003), Sete (1949-1996) and Cap Couronne – CC – (1961-2004) (43°2N – 5°05E) (Fig. 1). Wind directions are recorded in 20° increments calculated clockwise from the geographical North and correspond to the average direction from which the wind blows during the ten minutes before the recording. Wind speed corresponds to the average speed (in m/s) and is measured over the same time interval as the direction. Local wind directions and speeds during extreme sea surges have been firstly analyzed on the common period between wind and surge height data (1986-1995). Analyses have been extended on the whole available period at each station with trends in wind speed and frequency computed firstly as linear regressions and with low-pass filtered variations to investigate the long-term evolution.
8European Center for Medium Weather Forecast (ECMWF) Reanalyses (ERA-40) mean sea level pressure – SLP – at 00h, 06h, 12h and 18h have been extracted from the ECMWF web site (http://www.ecmwf.int) from October 1, 1957 to March 31, 2002. The composite of the mean sea level pressure for each 6 hourly observations, when sea surges ≥ 60 cm were computed for the following time lags: 48, 24 and 0 hours. When a surge event ≥ 60 cm encompasses several hours, the hour when the highest sea surge was considered as the hour 0.
Table 1 - Location and data characteristics of the tide gauge.
Coordonnées et caractéristiques des enregistrements marégraphiques.
9The correlations of hourly and daily maximum sea surges have been calculated between the three tide gauge stations. The positive correlations are always significant at the one-sided 1% level and the amount of common variance between two stations is almost always above 50% (Table 2).
10The temporal evolution of sea surge at an hourly time scale is displayed for the mean of the 5, 10 and 20 highest sea surges observed at the three stations (Fig. 2). Sea surge > 60 cm almost always occured at the same time amongst the three stations (Fig. 2a). Furthermore, sea surge above 40 cm lasts always at least 12 consecutive hours and usually more than 1 day at the three stations (Fig. 2a,b,c). The phases as well as the amplitude of the sea surges are highly consistent amongst the three stations. (Fig. 2). Sea surge episodes seem to be highly consistent in the Gulf of Lions. Nevertheless, some differences between the three stations associated with local-scale effects are observed when individual events such as the highest sea surges are analysed (Fig. 3).
Fig. 2 - Mean hourly sea surge height variations at the three stations for two days before to two days after the (a) the 5 highest, (b) the 10 highest and (c) the 20 highest sea surges recorded at GD for the wintertime period from 1986-1995.
Variations horaires moyennes des surcotes aux trois stations durant les deux jours avant et les deux jours après les (a) 5 plus fortes surcotes, (b) les 10 plus fortes surcotes et (c) les 20 plus fortes surcotes enregistrées à GD pour les hivers de la période 1986-1995.
Table 2 - Correlation between hourly sea surges (upper part) and daily maximum sea surges (lower part) for the 1986-1995 period between Grau-de-la-Dent, Sete and Port-Vendres.
Corrélations entre les surcotes horaires (haut du tableau) et entre les maximums journaliers (bas du tableau) au GD, SE et PV sur la période 1986-1995.
11During the 48 hours before the maximum sea surges at GD, SE and PV, a low pressure system tends to remain stationary near the Bay of Biscay and intensifies, particularly after 24 hours (Fig. 3). This low pressure system usually reaches its minimum SLP over the Bay of Biscay one day before the highest surge. Moreover, it is always associated with high pressure over Central Europe (Fig. 3). The E-W gradient across France is rather strong from one day before the maximum surges. These SLP patterns during sea surges are consistent at all three stations and favour strong onshore winds, from SW to SE sectors, and the piling up of water in the northern edge of the Gulf of Lions.
Fig. 3 - Composite of SLP calculated at time lags of 48, 24 and no lag for the 32 sea surges ≥ 60 cm at GD, the 54 sea surges ≥ 60 cm at SE and the 29 sea surges ≥ 60 cm at PV. Isobar contours are drawn in hPa.
Composites des SLP moyennes calculées 48, 24 et 0 heures avant les 32 surcotes ≥ 60 cm a GD, les 54 surcotes ≥ 60 cm à SE et les 29 surcotes ≥ 60 cm à PV. Les isobares sont tracées en hPa.
12For each hourly surge greater or equal to 60 cm, the direction of local wind has been extracted at the nearest meteorological station at the time and during the preceding 5 hours of the surge (to have at least two three-hourly wind measurements for each hourly surge value). At GD and SE stations, more than 80 % of sea surges ≥ 60 cm are associated with southerly onshore winds blowing from 100° to 120° and mostly from 120° (Figs. 4a-b). According to the direction of the coast (west – east for GD and south-west – north-east for SE) southward opened on the Mediterranean Sea (Fig. 1), winds blowing from 100° to 120° at SE and GD stations are onshore winds tending to push the water toward these coasts and leading to a local surge peak.
1380 km westward of SE, at PV, more than 60 % of sea surges ≥ 60 cm are associated with winds blowing from 160° to 200° (Fig. 4d). But it is interesting to note that over 10% of extreme surges are associated with northerly winds blowing from 300° to 340° winds, that is the “Tramontane” (Fig. 4d). The tide gauge at PV is located on a local north-west – south-east oriented coast opened northeastward to the Mediterranean Sea (Fig. 1). PV is thus mostly reactive, as the other stations, to southerly winds that lead to a relatively uniform regional sea surge, but also to particular northerly winds due to the local direction of the coast. In the latter case, a local surge is associated with a regional-scale northerly wind.
14The speed of local wind has been extracted at the time and during the preceding 5 hours of sea surges ≥ 60 cm. More than 80 % of them are associated with winds >10 m/s that obviously mostly contribute to the local surge-peak development (Fig. 5).
15In summary, according to the general orientation of the Gulf of Lions southern opening on the Mediterranean Sea, the three stations usually record regional-scale strong sea surges when a wind from ~ 100° to 200° is blowing above 10 m/s. The impact of strong southerly winds that cause water to pile up against the coast is therefore regionally consistent around the Gulf of Lions. But according to the local site exposure at PV, some of the strong sea surges at PV are associated with northerly winds.
Fig. 4 - Mean frequency (in %) of 3-hourly winds during sea surges ≥ 60 cm in bar and long-term climatology (1986-1995) superimposed as a full line at (a) CC for surges at GD, (b) SE for surges at SE and (c) PV for surges at PV.
Fréquence moyenne (en %) des vents tri-horaires pendant les surcotes ≥ 60 cm (histogramme) et la climatologie (1986-1995) en ligne pleine superposée à (a) GD, (b) SE et (c) PV.
Fig. 5 - Frequency (in %) of wind speed of surge-related local winds during sea surges ≥ 60 cm in bar and long-term climatology superimposed as a full line. (a) 100-120° winds at CC, (b) 100-120° winds at SE, (c) 160-200° winds at PV.
Fréquence (en %) des vitesses des vents associées aux surcotes ≥ 60 cm (histogramme) et la climatologie (1986-1995) en ligne pleine superposée. (a) pour les vents de 100-120° à CC, (b) pour les vents de 100-120° à SE et (c) pour les vents de 160-200° à PV.
16The seasonal frequency of local winds > 10 m/s related to sea surges ≥ 60 cm has been extracted at each station to investigate the long-term evolution of this forcing factor. The seasonal frequency of winds associated with sea surges ≥ 60 cm increases at CC, SE and PV (but only for the southerly components at PV which is largely predominant). Linear trends are respectively +0.083 %/year and 0.087 %/year for 100-120° > 10 m/s at CC and SE and +0.075 %/year for 160-200° > 10 m/s at PV; Fig. 6). The increasing linear trends are always significant at the 95% level and is mixed with irregular heterogeneous decadal-scale variability (Fig. 6). Despite the different available period and differential local rates, there is a significant regional-scale increase in the frequency of southerly winds > 10 m/s.
17The seasonal maximum length of consecutive three hourly wind observations associated with sea surges ≥ 60 cm increases at each station with significant trends (Fig. 7). A regional increase in the duration of southerly wind episodes seems to be generalized and uniform around the Gulf of Lions (Fig. 7).
18In summary, despite the different available period and different local rates, there is a significant regional-scale increase in the frequency and length of southerly surge related wind in the Gulf of Lions since ~1950 (Figs. 6-7).
Fig. 6 - Seasonal frequency of winds ≥ 10 m/s (in %) associated with sea surges ≥ 60 cm (circles) at (a) GD, (b) SE, (c) PV (160-200° only), with the low-pass filtered variations superimposed as full line and linear trend superimposed as dashed line. The significance level (p) of the slopes is estimated with a Student’s T test.
Fréquence saisonnière des vents ≥ 10 m/s (en %) associés aux surcotes ≥ 60 cm (cercles) à (a) GD, (b) SE et (c) PV (160-200° seulement), avec les variations à long terme en lignes pleines superposées et les tendances linéaires en lignes discontinues. La significativité de la régression est estimée avec un test T de Student.
Fig. 7 - Seasonal maximum durationh of consecutive three-hourly observations of wind ≥ 10 m/s associated with sea surges ≥ 60 cm at the three stations with the low-pass filtered variations superimposed as full line and linear trend superimposed as dashed line. The significance level (p) of the slopes is estimate with a Student’s T test.
Durée maximum saisonnière des séquences de vents tri horaire ≥ 10 m/s associés aux surcotes ≥ 60 cm pour les trois stations avec les variations à long terme en lignes pleines superposées et les tendances linéaires en lignes discontinues. La significativité de la tendance linéaire est estimée avec un test T de Student.
19Following a suggestion made in a previous report (Pirazzoli, 2006), hourly surges have been tabulated to produce a normalized frequency distribution in vertical bands with a tabulating interval of 5 cm and the frequency distribution of positive surges has been tentatively quantified by exponential equations relating surge heights to return periods. The values obtained are plotted graphically in figure 8, where extreme sea surges have been extrapolated with the best fitting exponential functions y = 0.0005e-0.1156x for positive surges at GD, y = 0.0007e-0.0939x at SE and y = 0.0005e-0.1216x at PV. The resulting levels for return periods of 10 years (with ±5 cm accuracy) are > 80 cm for positive extremes at each station (Fig. 8; Table 3).
20The frequency of sea surges ≥ 40 cm (respectivelly ≥ 60 cm), is more than 15 hours (respectively 1 hour) per season (October to March) at each station (table 3). Furthermore, observed sea levels > 40 cm are “geomorphologically” active in these low lying and sandy beaches (Bruzzi, 1996). The astronomical tide height in this part of the Mediterranean Sea oscillates between -15 and +15 cm (Ullmannet al., 2006). For example, a height of +40 cm can result from a tide of +15 cm and a surge of +25 cm, or a tide of +10 cm and a surge of +60 cm, or even a tide of -15 cm and a surge of +65 cm. It means that sea surges ≥ 60 cm, that occures at least one hour per season, are mostly active in shore recession (Table 3). Despite different local frequencies and return times, sensitive to the local tide gauge exposure, the coastal area of the Gulf of Lions is a “geomorphologically” active zone subject to storm surges linked with shore erosion and flooding hazards (Ullmannet al., 2006).
Fig. 8 - Retun period of surge heights at GD, SE and GD from October to March (period 1986-1995).
Temps de retour des surcotes à GD, SE et PV calculés sur la période d’octobre à mars de 1986 à 1995.
Table 3: Frequency and return times of hourly surges at PV, SE, and GD during the October to March season.
Fréquence et temps de retours des surcotes horaires à PV, SE et GD calculés sur la période d’octobre à mars de 1986 à 1995.
21Wintertime (October to March) sea surges have been extracted from hourly sea level height recorded at three stations (Port-Vendres, Sete and Grau-de-la-Dent), almost evenly distributed along the Gulf of Lions (Fig. 1). Sea surges are largely in phase at hourly and daily time scales (Table 1, Fig. 2). The spatial coherence observed for 1986-1995 between the three stations suggests that the long-term sea surge rise at Grau-de-la-Dent in the whole 20thcentury (Ullmann et al., 2006) could be extrapolated to other sites in the Gulf of Lions.
22Sea surges around the Gulf of Lions mostly occur during the same barometric system showing a low pressur system on the Atlantic near the Bay of Biscay associated with high pressure over Central Europe leading to a regional-scale southerly wind piling up water against the coast (Figs. 3-4). But, strong local-scale sea surge > 60 cm could also occur at PV in association with northerly winds (i.e. Tramontane) because of the local exposition of the coast (Figs. 1-4). From an impact perspective, local coastal exposures to onshore winds are important to explain how local and regional winds may lead to a strong local surge-peak.
23The frequency and length of winds associated with surges > 60 cm for the 1986-1995 period are then analysed for the 1948-2003 period. The frequency and length of the southerly winds associated with a regional-scale sea surge increases almost uniformly across the Gulf of Lions (Figs 6-7). These changes could explain, at least partly, the increase in the frequency and intensity of sea surges since the middle of the 20th century. Moreover, since the end of the forties, a long-term rise of sea level pressure over Central Europe (15°-30° E; 40°-55°N) is synchronous with an increase in the relative seasonal frequency of barometric patterns associated with low pressure over the Atlantic between the Bay of Biscay and the British Isles and high pressure over Central Europe (Ullmann and Moron, 2006). This large scale sea-level pressure rise over Central Europe could explain, at least partly, the increase in the frequency and length of southerly regional winds favourable to sea surges in the Gulf of Lions.
24Sea surges ≥ 60 cm that occures at least one hour per season at each station, are most active in erosion processes such as shore recession (Table 3). Despite different local frequency and return times, sensitive to the local tide gauge exposure, the coastal area of the Gulf of Lions seems to be a “geomorphologically” active zone subject to storm surges linked with shore erosion and flooding risks.
25Recent trends of sea surge-related wind frequency and length and an accelerated global sea-level rise expected during the next decades (IPCC, 2001), wich would enable surges to affect the coast at higher levels, suggest a possible increase in shore erosion and of flooding risk on low coasts such as most of the Gulf of Lions. Moreover, the rising human pressure along this littoral increases its vulnerability linked with this rise.