Oceanographic conditions in North and Northwest Iberia and their influence on the Prestige oil spill

Transcription

1 Marine Pollution Bulletin 53 (2006) Oceanographic conditions in North and Northwest Iberia and their influence on the Prestige oil spill M. Ruiz-Villarreal a, *, C. González-Pola c, G. Diaz del Rio a, A. Lavin d, P. Otero a,e, S. Piedracoba e, J.M. Cabanas b a Instituto Español de Oceanografía (IEO), Centro Oceanográfico de A Coruña, Muelle de Ánimas, s/n, A Coruña, Spain b Instituto Español de Oceanografía, Centro Oceanográfico de Vigo, Cabo Estay, Canido, s/n, Vigo, Spain c Instituto Español de Oceanografía, Centro Oceanográfico de Gijón, Avda. Príncipe de Asturias 70bis, Gijón, Spain d Instituto Español de Oceanografía, Centro Oceanográfico de Santander, Promontorio de San Martín, Santander, Spain e Grupo de Oceanografía Física, Facultade de Ciencias, Campus Lagoas, Universidade de Vigo, Vigo, Spain Abstract Oceanographic conditions at the time of the Prestige oil spill (November 2002) and following months are analyzed based on a set of hydrographic cruises. The ship sank off one of the flanks of the Galician Bank, an offshore seamount, and a major oil spill drifted to the N and NW Iberian coast mainly driven by dominant winds. Coastal circulation was characterized by freshwater plumes and the poleward slope current, and could have affected the fate of the oil spill and influenced stranding places. Seasonal evolution of oceanographic conditions in this particular year is compared with the long-term average and reveals specific features that need to be taken into account in studies of the impact of the oil spill on populations. Spring conditions commenced earlier than other years in the Southern Bay of Biscay, contrastingly in western Iberia. The lack of subsurface intrusion of subtropical waters suggests a low intense penetration of the poleward current in Spanish Biscay slopes. In western Iberia, the slope poleward current observed in late autumn weakens and is exported off slope during upwelling pulses in the spring, with no strong intrusion of the poleward current on the slope at the time of the spring bloom. A description of current velocities near the wreck on the Galician Bank is obtained after the analysis of a mooring line. Ó 2006 Elsevier Ltd. All rights reserved. Keywords: Prestige oil spill; Poleward slope current; River plumes; Galicia; Cantabrian Sea; Galician Bank; Spring transition 1. Introduction On 13 November 2002 the Prestige, a single-hulled tank steamer with more than 77,000 metrics tons of heavy fuel oil, was reported in trouble, when a severe storm associated with strong winds and extreme sea conditions (Balseiro et al., 2003) hit Galicia (NW Iberian Peninsula, see geographical information in Fig. 1). The tanker started leaking oil and was towed offshore in a route first northwards and then southwestwards. After 6 days of route under swell, the vessel sank on 19 November on the south-western flank of the Galician Bank, an offshore seamount. The oil spill * Corresponding author. Fax: address: manuel.ruiz@co.ieo.es (M. Ruiz-Villarreal). affected more intensively the Galician coastal area, which is a well known productive ecosystem and it constituted a risk to the Rias Baixas, estuaries with intense mussel culture. Fuel was stranded during the first few days of the crisis near Cape Finisterre, especially after the ship approached the coast on 14 November, although the greatest amount of fuel was released when the ship sank on 19 November (in the following this spill will be referred to as the main spill). This fuel was driven to the coast mainly by winds and stranded in several places along the Galician coast from 29 November on. Oil continued leaking from the ship until major holes in the hull were sealed in January Multiple oil slicks drifted with wind into the southern Bay of Biscay, in the following also referred to as the Cantabrian Sea (see Fig. 1), and its impact was felt all along the X/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved. doi: /j.marpolbul

2 M. Ruiz-Villarreal et al. / Marine Pollution Bulletin 53 (2006) o N 44 o N 43 o N 42 o N G.Bank Wreck Galician Area G.Rias C. Finisterre Sisargas I. C. Ortegal Coruña Vigo Miño R. Cantabrian Sea St 6 Gijón Santander C. Peñas Duero R. 41 o N 12 o W 10 o W 8 o W 6 o W 4 o W 2 o W 45 o N 25-Sep to 09-Nov, Dic, o N D 43 o N C B 42 o N Demersales0902 A Prestige o N 45 o N 44 o N 43 o N 42 o N 41 o N Jan, 2003 Prestige Mar to 11-Apr, 2003 Pelacus o N 44 o N 43 o N 42 o N 41 o N 22-Mar to 08-Apr, 2003 HidroPrestige o W 10 o W 8 o W 6 o W 4 o W 2 o W 26-Sep to 27-Oct, 2003 Demersales o W 10 o W 8 o W 6 o W 4 o W 2 o W Fig. 1. Cruises with hydrographical sampling during the analyzed period (see also Table 1). The upper panel shows the position of the wreck, the sections performed during RadProf Cruises (white dots) and the regular coastal sampling sections (IEO Radiales Project) in the area. Marked isobath lines are 200, 1000 and 4000 m. Cantabrian coast and French Atlantic coastal zones during the following months. Several attempts to forecast the drift, and the spreading and stranding of the Prestige oil spill were performed during the crisis and afterwards (Montero et al., 2003; Daniel et al., 2004; Hackett, 2004). These exercises put forward the effect of currents on the drift of oil slicks, since wind forcing does not suffice for adequate prediction of stranding points. Furthermore, dispersion in the offshore area can also be affected by large scale circulation and by mesoscale features as (Hackett, 2004) results indicate. As we will illustrate, the main features of coastal circulation at the time of the accident were river plumes and the slope poleward current. Model products used in those simulations were not focused on coastal circulation and sensibly differ in the predicted current fields. Although they simulate a poleward slope current, none of them introduces river plumes. In the present work, we will examine available observational data to help in the analysis of the possible impact of oceanographic conditions on the fate of the Prestige oil spill. Fuel drifted at the surface, but additional fuel was found in bottom slicks during surveys on the Galician shelf, with maximum fuel concentrations near Cape Finisterre (see Sanchez et al., 2006; Serrano et al., 2006). Subsequent surveys tracked this bottom fuel and found that apparently it was dispersed northwards until April In the following months, its concentration was reduced. Since the effect that oceanographic conditions

3 222 M. Ruiz-Villarreal et al. / Marine Pollution Bulletin 53 (2006) have on this dispersion process is not clear, a review of oceanographic conditions is relevant. Studies of the impact of an oil spill on ecosystems usually proceed by establishing a reference state in populations from which to compare the observed distribution. In the affected area, there are time series of biological parameters that can be used to try and assess the impact of the oil spill in the ecosystem (Sanchez et al., 2006; Serrano et al., 2006; Varela et al., 2006). These time series show that variability in marine populations is strongly correlated to variability in circulation at different scales, including interannual variability. Consequently, the description of circulation during the time of the oil spill must be complemented with a comparison of the actual conditions to mean conditions, so that possible effects of natural variability can be isolated from the perturbation introduced by the oil spill. The main objective of the paper is therefore the characterization of the hydrographical conditions during the accident and the subsequent spreading period, that can be relevant for the analysis of the evolution of the oil spill and for the assessment of its impact on the ecosystem. Special attention will be devoted to the following spring season, due to its great relevance for the ecosystem. The paper will be structured as follows: first a short review of meteorological and oceanographic characteristics of the area is presented, second, the dataset of hydrographical cruises will be presented after a note on the climatic conditions from autumn 2002 to autumn 2003 and used for gaining information about the evolution of hydrographical conditions and currents that year. Finally, a short review of conditions around the Galician Bank focusing on variability of currents and thermohaline properties near the Prestige wreck will be performed. 2. Meteorological and oceanographic characteristics of the area The affected area is located in the North Atlantic Eastern Boundary Current (EBC) system, highly influenced by seasonality of winds. Around Portugal and Galicia, this EBC is called the Portugal Current System, and it is a region of weak circulation bounded south by the Azores current and north by the North Atlantic current, part of the subpolar gyre. Large scale atmospheric forcing is determined by the location of the Azores High. The induced prevailing geostrophic winds define the first order picture of variability of circulation in the EBC area (Wooster et al., 1976) with two mean seasons: upwelling season from April to September and downwelling season from October to March. There is interannual variability in the intensity of the seasons and in the timing of the transition between them. The system is also subjected to event scale variability (scales from 3 to 14 days), i.e. upwelling pulses in the downwelling season, downwelling pulses and relaxation of upwelling in the upwelling season (e.g. Blanton et al., 1984). This eastern boundary system is not as well sampled or understood as other EBCs such as the California Current system, but the main features of circulation have been described. A schematic diagram summarizing circulation in the mean downwelling and upwelling seasons is presented in Fig. 2. The broad and weak equatorward Portugal Current characterizes surface offshore circulation (see e.g. Wooster et al., 1976 and recently Martins et al., 2002). There is evidence of a subsurface countercurrent transporting warm and salty subtropical Eastern North Atlantic Central Waters (ENACWst) polewards. Our evidence of the poleward current relies on geostrophic velocities estimated from hydrographical surveys (Frouin et al., 1990; Haynes and Barton, 1990; Fiuza et al., 1998; Oliveira et al., 2004), with further evidence from Lagrangian drifters (Haynes and Barton, 1990), currentmeters (Pingree and Le Cann, 1990; Haynes and Barton, 1990; Vitorino et al., 2002a) and modeling (Coelho et al., 2002; Peliz et al., 2003). Statistics of a historical currentmeter dataset in the Iberian west margin performed during the OMEX project (Huthnance et al., 2002; Coelho et al., 2002) reveal a prevailing northward flow over the northern Portuguese and Galician slopes and adjacent ocean, with poleward current intensity decreasing northwards, and a southward flow in upwelling months in surface layers over the shelf and slope. During autumn and winter, the poleward current flows over the slope, when apparently it is surface intensified (Oliveira et al., 2004; Peliz et al., 2003) and its magnitude is of the order of 20 cm/s. A subsurface maximum of salinity and temperature is linked to the presence of the poleward current, and this intrusion has characteristics of ENACWst. Additionally, the SST warm anomaly observable in satellite imagery in the area during autumn and winter months is related to the existence of the poleward current (Frouin et al., 1990). The poleward current shows an interannual variability on intensity and penetration in the Cantabrian Sea (Lavin et al., 1998; Sanchez and Gil, 2000; Cabanas et al., 2003). The penetration of the poleward current in the Cantabrian Sea is usually first observed around Christmas, and this phenomenon has been called Navidad (Spanish for Christmas ) (Pingree and Le Cann, 1992b). The Navidad has been linked to SST anomalies in satellite pictures and years with a warm surface anomaly have been referred to as Navidad years (Garcia-Soto et al., 2002). Intense Navidad years have been associated with swoddy (slope water oceanic eddies) formation in the Cantabrian Sea (Pingree and Le Cann, 1992a; Garcia-Soto et al., 2002). Coastal features observed specially in the downwelling season are freshwater fronts, associated with run-off from local rivers and observable in satellite pictures. Main freshwater inputs in the area (from 40 N) are the Douro River, the Minho River, local rivers in the Galician Rias Baixas and some local rivers with reduced catchment area along the Cantabrian coast. The change in the wind regime in spring triggers the upwelling of central water and provides the system with potential energy through elevation of isopycnals. This could be associated with the appearance of mesoscale structures in the west coast (Fiuza et al., 1998) and in the

4 M. Ruiz-Villarreal et al. / Marine Pollution Bulletin 53 (2006) Fig. 2. Schematic circulation in the area during typical upwelling (spring and summer) and typical downwelling (autumn winter) season. Offshore current is dominated by the Portugal current, which is represented by a gray dotted arrow. Note that a typical season is a simplification and the system is subject to event variability that can dominate the response of the system. See Section 4.2 for details on spiciness (Flament, 2002). Cantabrian Sea, where a complex dynamics of fronts and eddies has been described in spring and summer (Gil et al., 2002; Lavin et al., 1998). Mesoscale structures have implications in the distribution of nutrients and in the spring phytoplankton bloom (Bode et al., 1996, 2002; Gil et al., 2002; Lavin et al., 2006). In spring, the poleward current weakens on the slope (Fiuza et al., 1998; Lavin et al., 2006). Surface atmospheric warming originates a shallow thermal stratification in coastal waters (Lavin et al., 1998), that can be broken in upwelling pulses, specially in the west coast. A characteristic phenomenon in summer, especially in the west coast, is the presence of filaments of colder water that extend from the coast into the ocean (Barton et al., 2001; Roed and Shi, 1999). Autumn is the season of the reversal of prevailing winds and conditions favor the development of the poleward flow on the slope (Sanchez and Gil, 2000; Cabanas et al., 2003; Lavin et al., 2006). 3. Dataset The dataset comprises data from routine hydrographical lines and cruises and from multi-disciplinary surveys organized to provide information about the impact of the spill. Details of the cruises are provided in Table 1 and the location of CTD stations in the different cruises is shown in Fig. 1. The temporal and spatial coverage allowed us to obtain a picture of the evolution of oceanographic conditions from the time of the oil spill to the following spring in the affected area. Data from IEO (Instituto Español de Oceanografía) programs with systematic hydrographical CTD sampling from the late 1980s can be used to establish a reference state for the area. IEO monitoring activities in the area include: monthly Spanish Standard Sections that obtain CTD, nutrient and phyto and zooplankton data in Vigo, Coruña, Cudillero, Gijon and Santander (Valdes et al., 2002); annual fishery stocks monitoring cruises in spring (Pelacus, e.g. Bode et al., 2002) and autumn (Demersales, e.g. Sanchez and Serrano, 2003), that provide us with CTD casts in the shelf and the slope; and Spanish Deep Standard Sections, that have sampled twice a year since 2003 three hydrographic lines: 43 N, 8 W and W from the shelf to open ocean. Several hydrographical cross-sections from the shelf offshore to a depth of more than 1000 m were performed in a late autumn winter situation during the Prestige Plataforma cruises (Serrano

5 224 M. Ruiz-Villarreal et al. / Marine Pollution Bulletin 53 (2006) Table 1 Summary and information about cruises used Cruise Dates Vessel Area No. of CTDs DEMERSALES September 9 November 2002 RV Cornide de Saavedra Shelf-slope of Galicia and Southern Bay of Biscay Prestige Plataforma December 2002 RV Cornide de Saavedra Shelf-slope of Galicia, from Vigo to Sisargas (Costa da Morte) Prestige Plataforma January 2003 RV Vizconde de Eza Shelf-slope of Galician coast, from Sisargas to Coruña (Costa da Morte) Hidroprestige March 8 April 2003 RV Cornide de Saavedra N from the shelf to 13 W Main purpose 128 Historical autumn time series of bottom trawls to assess demersal fisheries stocks 27 To determine the extension of oil spill on the shelf bottom and preliminary evaluation on the spill impact on benthic communities 8 To add some stations to the west of the area sampled in Prestige Plataforma To study hydrography, circulation and biochemistry in the affected area, including the near wreck zone Radprof April 2003 RV Cornide de Saavedra 8 W and W sections 65 Spanish Deep Standard Sections Pelacus03 19 March 11 April 2003 RV Thalassa Shelf-slope of Galicia and Southern Bay of Biscay 167 Historical spring time series for estimation of biomass of pelagic fisheries in the NW and N Iberian shelf Hidroprestige June 2003 RV Cornide de Saavedra N (Galician slope to Galician Bank) Radprof September 2003 RV Cornide de Saavedra 43 N, 8 W and W sections Demersales September 27 RV Cornide de Saavedra Shelf-slope of Galicia and October 2003 Southern Bay of Biscay 7 Maintenance operation of the mooring line on the Prestige wreck 59 Spanish Deep Standard Sections 117 Historical autumn time series of bottom trawls to assess demersal fisheries stocks et al., 2006). A hydrographical cruise with an extended grid (Hidroprestige0303) sampled spring conditions from the Galician shelf to the Galician Bank. Some complementary information was obtained from the reports of the Rayo Buoy network of the Spanish agency Puertos del Estado ( that measure currents, temperature and salinity in several points of the Galician and Cantabrian slope area. Some of the activities during the period following the sinking of the Prestige were aimed at improving our knowledge about currents near the wreck in the western flank of the Galician Bank to support the design of operations for an eventual extraction of fuel remaining in the shipwreck. A currentmeter line was moored at N W (near the Prestige wreck) in the southwestern slope of the Galician Bank and maintained at that place from 29 March to 17 September From this mooring and from different CTD casts around the Galician Bank, we have gained information about water mass variability around the bank and especially about current variability near the wreck. 4. Results: evolution of hydrographic conditions from autumn 2002 to autumn Meteorological conditions from summer 2002 to autumn 2003 Upwelling index (UI) has been traditionally used as a first order characterization of circulation in EBC systems. In our area, UI computed from geostrophic winds in a 2 2 cell centered at 43 N11 W(Blanton et al., 1984) has been used as representative of the effects of large-scale winds on circulation. Daily UI in that cell is represented in Fig. 3, where UI during CTD casts in Galicia of different cruises are highlighted. Although event variability is outstanding, the two main seasons described in the introduction are evident in the figure. During the summer season, upwelling dominates and a succession of upwelling pulses is interrupted by periods of relaxation of winds or downwelling. Summer 2002 was a rather typical summer with moderate upwelling in the western Galician coast. In the Cantabrian area, air temperatures were extremely cold in

6 M. Ruiz-Villarreal et al. / Marine Pollution Bulletin 53 (2006) Prestige trawl and spill 1st Spring upwelling pulse Demersales02 Prestige0103 HP0303 Demersales03 Upwelling Prestige1202 Pelacus03 Downwelling Fig. 3. Upwelling index at 43 N11 W from summer 2002 to autumn 2003 with indication of cruises referred to in the text. the spring summer At the Santander meteorological observatory, summer 2002 is recorded as the fifth coldest summer since 1960 and the coldest summer since Consequently, summer SST in the Cantabrian area was 2 C below the average summer cycle (Hughes and Lavin, 2003). In 2002, the autumn transition, defined as the time of reversal from upwelling dominance to downwelling dominance, started in September and was followed by a typical autumn characterized by a succession of downwelling events. From September 2002 to January 2003, the rainfall was higher than average in the Galician Cantabrian area. The Prestige trawl and spill took place during downwelling conditions. During the spreading period the only upwelling pulse was end November and early December, coincident with the arrival of the main spill to the Galician coast. In January and February, some days of upwelling dominance were followed by downwelling conditions that lasted almost a month from mid February. The first upwelling pulse in spring 2003 took place at the end of March. This pulse was followed by several days of southerly winds during April and a succession of weak upwelling pulses was characteristic of May to July, interrupted by downwelling pulses. The only intense upwelling pulses in spring took place in May and at the end of June. April, March and May 2003 showed increased temperatures and precipitation lower than average, except on the Galician West coast at the end of March and April, when rainfall was higher than average. As we will see in Section 5, spring winds in the Cantabrian area were anomalous for the season. Summer 2003 upwelling in the Galician west coast was weak compared to 2002 and to other years. Noteworthy, during this summer is a heat wave in August 2003 that affected the whole Galician Cantabrian region. The autumn transition in 2003, contrastingly with autumn 2002, was not clearly defined and downwelling dominance was not observed until November Oceanographic conditions in Galicia The Prestige accident: downwelling season, autumn and winter 2002 The first cruise in our dataset is Demersales02 in late September 2002, which surveyed the Cantabrian Sea in October. Horizontal salinity distributions at 5 and 100 dbar for this and other cruises are plotted in Fig. 5. Demersales02 cruise was performed during prevailing downwelling conditions after the autumn transition. A thermal stratification (not shown), typical of summer early autumn conditions, is observed in surface layers. Freshwater plumes are already noticeable near the Galician Rias Baixas in salinity at 5 dbar and a subsurface slope salinity maximum more intense in southernmost stations is clear at 100 dbar (Fig. 5). The existence of freshwater fronts (cold fringe near river mouths) and the surface SST warm anomaly associated with the poleward current in the Galician West coast can be seen in satellite pictures during autumn 2002 months (see SST on 6 December in Fig. 4). We have few subsurface observations at the time of the crisis, but some weeks after the accident, the Prestige-Plataforma cruises (see Table 1 and Fig. 1) sampled the Galician shelf and slope during downwelling conditions similar to those prevailing during Prestige trawl and spill (see Fig. 3). The first Prestige Plataforma cruise occupied the shelf and slope in the area most affected by the oil spill (Fig. 1) in December 2002 under downwelling conditions (Fig. 3). The CTD survey started at the shelf outside the ria de Vigo on 15 December and six cross-shore CTD transects were performed one per day. These transects covered

7 226 M. Ruiz-Villarreal et al. / Marine Pollution Bulletin 53 (2006) Fig. 4. SST on 9 December Freshwater plumes can be traced as zones of lower temperature near river mouths (see Fig. 1 for location of main rivers). Wind was upwelling favorable during this day. Image provided by the satellite receiving station at IEO Santander ( from the inner shelf around 100 m to a depth of more than 1000 m (with CTD data down to around 500 m). Haline stratification dominates surface layers: in the horizontal sections at 5 dbar (Fig. 5), freshwater from local rivers can be seen in the shelf with a frontal structure with saltier offshore waters around the 200 m isobath. Offshore salinity in the northernmost sections (for strong downwellingfavorable winds) is higher. Local river run-off is more intense in the southern sections, where run-off from the Minho River and the Rias Baixas Rivers provides more freshwater input to the shelf, whilst from Finisterre to Coruña, freshwater inflow is more reduced. Waters seawards are saltier and a salinity maximum can be observed on the slope in the 100 dbar horizontal slice, with higher salinities in the southernmost sections. Intruded waters are warmer and saltier than in Demersales02 and have characteristics of ENACWst. The across-shore extension of this intrusion is around 20 km. There is variability among the different transects in the salinity values in the intrusion zone, but the maximum salinities are observed in the station with depth closer to 200 m (upper slope) in all sections. We have computed dynamic height and geostrophic velocities referred to 450 m. Vertical sections of geostrophic velocities (not shown) in the sections with stations deeper than 400 m (Pontevedra, Corrubedo and Sisargas; A, B and D, respectively in Fig. 1(b)) show a surface intensified poleward flow in the outer shelf, with velocities of the order of 20 cm/s. Several days after the PrestigePlataforma1202 cruise, between 15 and 22 January, a reduced area in the Costa da Morte (coast from Cape Finisterre to A Coruña) northeast from that sampled in the previous cruise was studied in cruise PrestigePlataforma0103. Eight CTD casts were performed during the cruise, four of them at a depth less than 100 m between Finisterre and Coruña and four in the midshelf between Sisargas and Coruña. Conditions were light southerly winds associated with the relaxation of an upwelling event on 13 January. Horizontal salinity maps in Fig. 5 indicate the existence of a subsurface salinity maximum on the slope like that found in Prestige1202. The temperature in upper layers is lower than in Prestige1202 and a slight thermal inversion (of around 1 C) is present in surface layers. Surface currents measured by Silleiro buoy (off Vigo) of Rayo network show that flow was poleward during December and part of January. However, the intensity of this poleward flow was less than on previous year: mean velocity in December was 19.1 cm/s in 2001 and 8.5 cm/s in During part of January and February 2003 the flow was southwards in average, contrastingly to the usual winter conditions of poleward flow. This feature is likely associated with the upwelling pulses during those days (Fig. 3), since surface currents respond strongly to meteorological events.

8 M. Ruiz-Villarreal et al. / Marine Pollution Bulletin 53 (2006) Fig. 5. Salinity at 5 dbar and 100 dbar for the cruises of Fig. 1. The fields are computed by using optimal interpolation method with isotropic correlation decay of 0.4. Data is represented where random mean squared error for the analyzed field is lower than 0.4. Marked isobath lines are 200 and 1000 m.

9 228 M. Ruiz-Villarreal et al. / Marine Pollution Bulletin 53 (2006) The horizontal gradients seen in salinity plots can be associated with event scale variations (upwelling, downwelling or relaxation of upwelling) that induce variation of conditions between different cross sections. In this sense, there is evidence of the fast (hours) response of plumes, and consequently of the front between the poleward current and river plumes, to wind variations (e.g. Garcia-Berdeal et al., 2002). Under downwelling conditions like those during the Prestige trawl and spill, the plume is confined to the coast and the flow is northwards in the surface freshwater layer. Velocities in the plume can get very high (almost 1 m/s) under conditions of high volume of freshwater in the shelf combined with strong downwelling winds during storms, as is evident in numerical simulations in April 2001 (Ruiz-Villarreal et al., 2005), a year with several river floods during winter and spring. Also noticeable is that plumes can intrude in the rias during downwelling events, and some reports of the intrusion of the Minho River have been reported in the ria de Vigo (Mouriño and Fraga, 1982). Upwelling pulses or relaxation of downwelling conditions originate a displacement of the plume offshore into the poleward domain and a reduction of velocities in the freshwater layer. In Fig. 4, the plume can be seen to extend in the offshore direction in a day (6 December 2002) of upwelling favorable wind (note the upwelling pulse at the beginning of December in Fig. 3) Spring transition 2003 Conditions during the following spring and summer affect the response of the ecosystem, because it is the time of the spring phytoplankton bloom and also many species have their spawning season from early spring on. Measurements were taken during two cruises in the spring: Pelacus03 and Hidroprestige0303. In Pelacus03, that sampled the Galician shelf from 19 to 29 March, maximum CTD depths were 200 m and the outermost station in radial transects was located at the outer slope at a depth of more than 1000 m. The presence of freshwater in the Galician area is clear in the surface salinity map (Fig. 5). Surface salinity values are lower and the offshore extension is higher than during Prestige1202. The first upwelling pulse in spring that year took place during the cruise at the end of March, and upwelling induced the surface offshore migration of river plumes. Additionally, horizontal salinity plots from Pelacus03 indicate that saltier waters from the poleward current have retreated from the shelf and slope and higher salinity waters are only found in the southernmost stations. In the broader area cruise Hidroprestige0303, similar features are observed: freshwater plumes extending to the offshore area and higher salinity to the south of the domain. Offshore stations were occupied in a series of meridional lines starting near the slope from 26 March to 1 April. Separation between stations was 20 nm (37 km) in the deep area and 5 nm (9.3 km) in the shelf and slope. The Galician shelf and slope were surveyed from 1 April starting from the northernmost stations as zonal radial transects until N, when rough sea state associated with northerly winds forced a change in sampling order, and shelf stations to the south were performed before slope stations (3 4 April and 4 5 April, respectively, see also Alvarez-Salgado et al., 2006). The transition from winter to spring conditions in the shelf and slope is illustrated in Fig. 6, where late autumn conditions during Prestige Plataforma1202 are compared with spring conditions during Hidroprestige0303. The map of potential density at 25 dbar depicts the across-shore structure: in late autumn lighter waters associated with freshwater fronts can be observed near the coast and a zone of higher density is found on the slope. In spring, a density gradient between lighter water closer to the coast and denser offshore water is seen, although density is lower than in December. This distribution is also influenced by the different meteorological conditions during both cruises: downwelling in autumn and upwelling in spring. Under downwelling conditions, the plume is confined near the coast and the freshwater layer reaches deeper than under upwelling conditions, when the plume displaces offshore and gets shallower. The map of dynamic height at 5 dbar shows a poleward flow during the December cruise, with geostrophic velocities of the order of cm/s. In spring, the poleward flow is not observed at the shelf and slope and a zone of less intense poleward flow is seen offshore. Although care need to be taken with non-synopticity of CTD stations (see also Alvarez-Salgado et al., 2006), apparently a cyclonic eddy is present to the SW of the slope with a lighter core but warm and salty water similar to some found in May 1995 by Fiuza et al. (1998). From the water characteristics of this eddy we can put forward that it was originated on the slope. Vertical sections of density show sloping of isopycnals towards the coast and vertical sections of velocity indicate southward geostrophic flow, which is masked at the surface by the freshwater plume. Some of the stations in HP0303 offshore from the slope have properties closer to ENACWst, and especially those in the zone of the cyclonic eddy. Intrusions of ENA- CWst are clearly identifiable in spiciness plots. Spiciness is a state variable sensitive to isopycnal thermohaline variations (Flament, 2002) and consequently useful for the analysis the slope intrusion of ENACWst we have described. Warmer and saltier waters have larger spiciness. In Fig. 6c, spiciness computed at the r 0 = 26.9 isopycnal (where the intrusion is found to be maximal) characterizes the transition between autumn and spring conditions. The intrusion of ENACWst on the slope during Prestige1202 can be traced by the spiciness maximum near the coast. This is in contrast with the conditions during spring: lower spiciness is found towards the coast and the higher spiciness waters lie offshore. The use of spiciness allow us to put forward that observed features can be associated with the upwelling of the colder and less salty deeper waters that induce the migration of higher spiciness ENACWst offshore. The rest of the spring season along the Galician coast can be inferred from the evolution of upwelling index we

10 M. Ruiz-Villarreal et al. / Marine Pollution Bulletin 53 (2006) Fig. 6. Comparison of sampled late autumn conditions to early spring ones. A transition from a strong shelf-slope current and density gradient from December 2002 (Prestige Plataforma1202) to March 2003 (Hidroprestige0303) is evident. (a) Potential density at 25 db. (b) Geopotential height and geostrophic velocity field at 5 dbar relative to 450 dbar. Extrapolation of dynamical heights from nearby offshore stations has been performed. This method might bias currents over the shelf, but the difference in the poleward flow between the two cruises is unambiguously present. Objective analysis is performed as in Fig. 5. (c) Spiciness at r 0 = Marked isobath lines are 200 and 1000 m. reviewed in Section 4.1, which indicates that upwelling during that spring was not very intense Summer and autumn 2003 Not many data are available during summer 2003 to further analyze the seasonal evolution of the system that year. Summer upwelling along the west coast was not very intense. This can be induced from the evolution of upwelling index and from data from the Vigo standard section in the ria de Vigo and adjacent shelf and is also corroborated by inspection of SST satellite images. The most characteristic feature was the high temperatures associated with a heat wave in August and to the reduced upwelling this year. Temperatures in surface layers during summer were high with peaks inside the ria de Vigo of 20 C (3 C more than previous year and from the average of the last 25 years). Nutrient levels were lower than other years during spring and summer due to the less intense upwelling, with less entrance of subsurface waters rich in nutrients (Fundación Provigo, 2003, 2004). In September 2003 cruises (Radprof0903 and Demersales03) a thermal surface stratification is developed again except in near-shore stations in Radprof0903, which were under the influence of the intensified upwelling near Cape Finisterre (Blanton et al., 1984). There is no clear hydrographical signature of the poleward current on the slope stations, contrastingly in Demersales02. However, in Demersales03, salinity is higher in Galicia than in the Cantabrian Sea and the across-shore structure is similar. In the Vigo standard section, observed temperatures in autumn are around 3 C lower than the same temperatures in 2002, which supports the view of a reduced poleward intrusion during early autumn Inspection of UI in Fig. 2 shows that the autumn transition took place in 2003 later than in 2002, and October 2003 was still a month of upwelling prevailing conditions. Another feature observed in salinity plots is that there is less freshwater in the Cantabrian Sea than in 2002, associated with the more reduced rainfall during September and October Oceanographic conditions in the Cantabrian Sea As we illustrated in previous section, a clear intrusion of ENACWst in Galicia with a subsurface maximum on the

11 230 M. Ruiz-Villarreal et al. / Marine Pollution Bulletin 53 (2006) slope and associated with a poleward geostrophic flow was observed in autumn winter Hydrographical data in the Cantabrian Sea during that period indicate that the entrance of the slope current in the southern Bay of Biscay was weaker and less persistent than in other years. This entrance is linked to a peak signal on salinity and temperature in slope waters that can be distinguished from other interannual trends. From the analysis of the historical data available from Santander Station 6 (the best sampled slope station in the area, see González-Pola et al., 2005), we find no signal of slope (below the mixing layer) salty and warm water arrivals during the winter spring following the Prestige spill (Fig. 7). Station 6 is located on the slope at a depth of 850 m, but only 6 km from the 200 dbar isobath. This is the station where the signal of poleward intrusion is higher and offshore stations of the hydrographical line rarely show signals of this intrusion. Years with clear peaks in salinity and temperature were 1995, 1996, 1998 and 2002 (the last one exhibits the most intense signal), and less clear in During most of these years, the signal persisted for 2 or 3 months and could also be detected on the outer shelf. This subsurface intrusion was neither observed in the Gijon Standard section (shelf stations repeated since 2001), although it was recorded during winter However, signals on the shelf are more difficult to consider as indicative of poleward current arrival because advected waters entraining the shelf encounter resident waters affected by the mixing layer development. The mixing layer reaches the shelf bottom at some time between December and January and causes high variability on the upper waters heat and salt properties from one year to another. Therefore heat and salt budgets during the previous summer stratification and the development of the mixing layer may blur the signals of poleward intrusion in the shelf. The results of the Pelacus03 during spring confirm the lack of poleward intrusion during spring 2003, and especially if comparing them with the hydrographical distribution in spring 2002 (Nogueira et al., 2004b), where the warm and salty waters can be seen on the slope. In Pelacus03, freshwater fronts are seen in the easternmost part of the Cantabrian Sea and upwelled waters to the east of Cape Peñas could be related to the early commencement of spring conditions this year in the Cantabrian Sea (see discussion in Section 5). Sea surface temperatures recorded in Santander hydrographical line in summer 2003 were the warmest in the time series ( ) and values were 1 C above the mean from June to October, a fact probably induced by the heat wave in August we mentioned in Section 4.1. The thermocline was shallow; heat stored in the upper layer was high but below the mixed layer values remained around average. The mean potential temperature in the upper 300 m presented high values in the second part of the year, but not as high as those during 1997 and Climatically, in most of the North Atlantic during 2003, temperature and salinity have been reported higher than long-term average with record values in some regions (Hughes and Lavin, 2004). In autumn 2003, salinity values at 5 dbar and 100 dbar (Fig. 5) are higher than those in Demersales02, related to the higher precipitation in autumn Near the wreck: the Galician bank The Galician Bank is a submarine bank located about 200 km offshore from the Galician coast north of 42 N before the Iberian coast changes NS orientation to WE orientation (see Fig. 1). The shallow part has a typical depth of around 700 m and minimum depth is 500 m. The depth in the adjacent ocean to the west is more than 5000 m. Between the Bank and the Iberian Peninsula there is a channel with depths ranging to 3500 m followed by the slope and a narrow shelf. Many studies have reported the interaction of seamounts with oceanic currents and have described phenomena like amplification of tidal currents, internal tides, excitation of trapped waves, enhanced mixing, turbulence and internal waves, eddy generation and retention of material (see a recent review in White and Mohn, 2004). The impact of the Galician Bank in circulation in the NE Atlantic has been recognized in several studies (Maze et al., 1997; Coelho et al., 2002; Colas, 2003). Fig. 7. Time series of slope water properties in Santander Station 6. We have taken m because this layer is below the mixed layer in all years and therefore advective effects can be more clearly isolated. Correlated clear peak signals on salinity and temperature are seen in winters 1995, 1996, 1998 and 2002 (the last the most intense signal).

12 M. Ruiz-Villarreal et al. / Marine Pollution Bulletin 53 (2006) Results from historical hydrographical cruises and climatological studies (Daniault et al., 1994; Maze et al., 1997; Iorga and Lozier, 1999) suggest the existence of two veins of Mediterranean Water (MW), one on the slope between the Galician Bank and the Iberian coast and the other that recirculates to the west of the Galician Bank. Main water masses at the area of the NE Atlantic where the Prestige sank have been described in the literature (e.g. reviews in van Aken, 2000; van Aken, 2001 or Lavin et al., 2006). CTD profiles over the wreck were performed on three different cruises (Hidroprestige0303, Hidroprestige and Radprof0903) and the profile of temperature, salinity and oxygen from a CTD cast during Hidroprestige0303 is shown in Fig. 8. Thermocline and ENACW waters occupy the water column until the relative salinity minimum at about 500 m. Under that minimum, influence of MW is seen in the increase of temperature and salinity and the decrease of oxygen. A relative maximum in oxygen is found at 1800 and it is associated with a layer of Labrador Sea Water (LSW). Under this level, there is NADW (North Atlantic Deep Water). The variability (standard deviation) found in the three casts is higher at MW (0.41 C, 0.11 in salinity) and LSW (0.32 C, 0.06 in salinity); it is lower in ENACW levels (0.07 C and 0.01 in salinity at 300 m) and it is very small at deep levels (<0.01 in salinity and 0.02 C in temperature). Maximum temperature and salinity at MW were obtained during the June cruise and minimum temperature and salinity at LSW during April. A description of currents near the wreck can be obtained from the analysis of the mooring line near the Prestige wreck ( N; W) in the southwestern slope of the Galician Bank at a depth of 3900 m. The mooring line was maintained at that place from 29 March to 17 September 2003, with a maintenance operation on 25 June The mooring included a total of eight Aanderaa mechanical currentmeters during the period and a RDI ADCP 300 Hz currentmeter at 75 m for almost one and a ENACW 1000 MW Pressure (dbar) LSW NADW Temp. (ºC) Sal Oxy. (ml l -1 ) Fig. 8. Calibrated profile of temperature (black solid line), salinity (black dotted line) and oxygen (gray line) from a CTD cast during Hidroprestige0303 near the wreck ( N and W) on 29 March Maximum depth reached was 3872 m (3939 dbar), a few meters above the bottom. The temperature measured at the deepest part was 2.50 C (potential temperature H = 2.16 C) and salinity was Bottom oxygen measured by Winkler was 5.51 ml/l.

13 232 M. Ruiz-Villarreal et al. / Marine Pollution Bulletin 53 (2006) Table 2 Statistics for velocities at the mooring line near the Prestige wreck ( N and W) from 29 March to 17 September 2003 a Rcm Nominal Depth Actual depth Mean speed Maximum speed EW velocity NS velocity rcm11354 * ± ± ± 6.4 rcm ± ± ± 6.2 rcm ± ± ± 5.1 rcm ± ± ± 5.5 rcm ± ± ± 13.2 rcm ± ± ± 3.1 rcm11445 * ± ± ± 3.3 rcm ± ± ± 2.4 a Depths are in meters and velocities in cm/s Mean east west and north-south velocities are given in a geographical reference frame positive eastwards and northwards. Standard deviations are given after mean values. Speeds are calculated for the entire period, but for currentmeters marked with *. Rcm at 3000 m was deployed from 29 March to 8 April and from 26 June to 17 September and Rcm at the surface was deployed from 26 June to 17 September. Table 3 Statistics for velocities measured by the ADCP deployed at the mooring line near the Prestige wreck ( N and W) from 29 March to 12 May 2003 a Depth Mean speed Maximum speed EW velocity NS velocity ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 6.8 a Depths are in meters and velocities in cm/s Mean east west and north south velocities are given in a geographical reference frame eastwards and northwards positive. Standard deviations are given after mean values. half month (29 March 12 May). The mean values of currents from the mooring line are given in Table 2 (RCMs) and Table 3 (ADCP) and a stick plot of the evolution of subtidal currents is shown in Fig. 9. The maximum velocities are observed in surface layers, with peaks of more than 45 cm/s in the upper layers measured by the ADCP and in currentmeters at MW levels (800 and 1100 m) where recorded extreme are 28 and 42 cm/s, respectively. Mean speed in upper layers (0 400 m) is of the order of 10 cm/s and reduces with depth (13.9 ± 7.4 cm/s at 20 m and 6.0 ± 3.7 cm/s at 400 m). Prevailing direction is NW until June with increasing inclination with north direction with depth (20 at 20 m and m/s cm/s cm/s cm/s cm/s cm/s cm/s cm/s cm/s Apr03 May03 Jun03 Jul03 Aug03 Sep03 Apr03 May03 Jun03 Jul03 Aug03 Sep03 Apr03 May03 Jun03 Jul03 Aug03 Sep03 Apr03 May03 Jun03 Jul03 Aug03 Sep03 Apr03 May03 Jun03 Jul03 Aug03 Sep03 Apr03 May03 Jun03 Jul03 Aug03 Sep03 Apr03 May03 Jun03 Jul03 Aug03 Sep03 Apr03 May03 Jun03 Jul03 Aug03 Sep03 Apr03 May03 Jun03 Jul03 Aug03 Sep03 NCEP wind 75 m 200 m 400 m 800 m 1100 m 2200 m 3000 m 3700 m Fig. 9. Stick plot of time series of currentmeters in the Prestige mooring ( N W), with up and right representing true north and east, respectively. Currents have been low passed and subsampled every 6 h. Depth of the currentmeters corresponds to nominal depth (see Table 2 for actual depths), except for the shallowest current panel, where data from ACDP layer centered at 40 m and data from the rcm11354 (actual depth 35 m) are plotted for the periods they were available. In the top panel, daily NCEP winds in the grid point centered at 42.5 N, 12.5 W are plotted.

14 M. Ruiz-Villarreal et al. / Marine Pollution Bulletin 53 (2006) at 400 m). From July on current diminishes and prevailing direction turns SW. In the layer influenced by MW ( m), variability and maximum values are higher, and are associated with the MW vein that has been described on the western slope of the bank. The currentmeter at 1100 m lies in this MW layer and it exhibits the higher speeds in the mooring: mean of 18 cm/s with peaks of more than 40 cm/s. Semi-amplitude of the main tidal velocity harmonics M2 is around 5 cm/s at most depths, but gets maximum (10 cm/s) at 1100 dbar. Power spectra of currents peak at M2 frequency and it is an order of magnitude higher at MW levels than at the other levels, which might indicate the presence of internal tides. Several mesoscale events are evident and result from the intense dynamics at this depth and from the interaction of the bank with mean circulation. From the beginning of the mooring to the end of June, the current was directed to the north with mean values of 10 cm/s. In the 800 m currentmeter, velocities were well correlated with the 1100 m velocities, although velocities were smaller. At the end of June, previously to the mooring service, the current turns southwards. This change coincides with southward flow in currentmeters at 3000 m and 2200 m. This last one measured currents to the north in the first months of the deployment and it is located at a depth that can be influenced by LSW (see Fig. 8). In the second fortnight in August, a strong change in velocities is observed at 800 m. In the beginning velocities are directed to the NW and then the direction changes gradually to S. This change is accompanied by a change of 0.2 in salinity and 0.5 C in temperature. At 1100 m the change in velocity is smaller, but an increase in temperature of 0.5 C is evident. Although it must be studied in further detail, this data suggests that a meddy approached the area and it seems that the influence of that structure at 800 m extends to shallower levels: higher velocities are seen in currentmeters at 200 m and 75 m. Other marked changes in temperature at 1100 m are the 0.5 reduction around 10 June, coincident with a velocity of around 15 cm/s to the SW. In waters under MW levels velocities are slower, with average speeds of the order 3 ± 2.5 cm/s, and maxima up to 15 cm/s. Noticeable variations at deep levels are the mentioned change at 2200 m and the increased velocities to the north during August. We should note that we do not have currentmeters at LSW levels due to the malfunctioning of one currentmeter, but significant velocities and variability can be expected at that level. 5. Discussion In the presentation of the results, we have concentrated in describing the main features of circulation and their variability. We have characterized seasonal variability and commented on the specific features of this year related to average conditions. The results confirm some of the information known from the literature, but contrastingly with other studies, we have characterized the sequence during the same year and related it to interannual variability. In this section, we will discuss how the oceanographic conditions we have presented could be relevant for studying different aspects of the oil spill Implications of oceanographic conditions for oil transport The drift of marine oil spills is mainly driven by surface winds, although coastal circulation can play a role. Winds in the affected area at the time of the Prestige crisis and following spring from NCEP are plotted in Fig. 10. The progressive vector evolution of daily winds since 1996 is represented in two points (a) the NCEP grid point closest to Galicia and in (b) a grid point in the center of the Cantabrian Sea. Wind during November and December 2003 blew towards Galicia and was responsible of the transport of oil released at the time of the sinking towards the coast. Circulation in the offshore area is characterized by a rich mesoscale structure that can enhance the dispersion of oil in the offshore area, as Hackett (2004) puts forward. When the main spill approached the coast on 29 November, it was observed to remain offshore at the shelf in front of the Galician Rias without stranding (Montero et al., 2003). Although a change in wind occurred during those days (see Fig. 3 and Montero et al., 2003), the modeling efforts we reviewed in the introduction indicate that wind forcing alone is not sufficient for simulating the oil spill evolution and the stranding points. From our results we can extrapolate that at the time of the Prestige oil spill, dynamics in the shelf and slope were significant and typical of a late autumn winter situation: river plumes and the presence of the poleward current, both of which respond to meteorological events. On 28 November, the intense SW wind of previous days relaxed and consequently it is expected that the plume displaced offshore and velocities in the plume got reduced. These factors together with the enhanced dispersion in the frontal area are likely to have affected the drift of the oil spill. On the slope, the poleward current was present, and induced surface velocities of the order of cm/s. However, during those days of upwelling, the offshore surface extension of the plumes induced a reduction of surface velocities and interacted with the currents in the poleward current domain. None of the circulation models used to force the referred models of the drift of the Prestige oil spill considered freshwater plumes, although they consider the effect of a poleward current. Consequently, forcing oil spill drift models with circulation data from a high resolution model that realistically represents the poleward current, the river plumes and the event variability caused by changes in meteorological forcing is necessary for analyzing the impact of coastal circulation on the drift of the Prestige oil spill. From early December 2002, slicks drifted into the Cantabrian Sea. Fig. 10a shows that wind forcing was favorable to the drift of oil slicks into the Cantabrian Sea. Garcia-Soto (2004) infers the existence of a poleward flow in the Galician and Cantabrian slopes during December