Upwelling in the Eastern Subtropical North Atlantic Ocean

UPWELLING IN THE EASTERN SUBTROPICAL NORTH ATLANTIC OCEAN Tesis Doctoral presentada por Maria V. Pastor Mollà Dirigida por el Dr. Josep Lluís Pelegrí Llopart y la Dra. Jaime Palter Programa de Doctorado: Oceanografía Departamento de Física. Facultad de Ciencias del Mar UNIVERSIDAD DE LAS PALMAS DE GRAN CANARIA El Director, La Directora, La Doctoranda, Barcelona, a 11 de Julio de 2011 Abstract Coastal upwelling in the eastern margin and offshore curl-driven upwelling in the southeastern margin, make the subtropical Northeast Atlantic a region of major primary productivity. When examining a broad zonal area, from the coast to 40◦W, we find that the upward transport of nutrients due to offshore curl-driven upwelling becomes the main control on productivity. Nevertheless, despite its relatively small zonal extension of about 100 km, coastal upwelling extends its impact towards the open ocean through offshore Ekman transport and convergence of the meridional flow at Cape Blanc (21◦N). Analysis of hydrographic data from spring 1973 and fall 1975 shows an export from the coast to the open ocean of 2.9 Sv of water mass and 53 kmol s-1 of nitrate during spring and 0.6 Sv and 3 kmol nitrate s-1 during fall in the area south of Cape Blanc. It is fundamental to improve our understanding of the distribution of the different water masses of the upper thermocline, as they carry the nutrients that will reach the euphotic layer and sustain primary production. In the study region, central waters of northern and southern origin meet at the Cape Verde frontal zone. Northern waters are nutrient-poor and oxygen-rich while local southern waters are nutrient-rich and oxygen-poor. Here, intense double diffusive mixing enhances horizontal heat transfer, thus the front appears as a smooth feature in terms of temperature, but resembles a barrier in terms of other properties such as salt or nutrients. The application of an Optimum Multiparameter analysis to hydrographic data collected during November 2007 and November 2008 shows a sharp front separating the central waters of northern and southern origin. In contrast, at intermediate layers, the transition between Mediterranean Water and Antarctic Intermediate Water is smoother. i Resumen El afloramiento costero en el margen este y el aforamiento por rotacional del viento, o bombeo de Ekman, en el margen sureste hacen de esta zona del Atlantico Norte subtropical una región de principal importancia por su alta productividad. Examinando una extensiva área, desde la costa hasta 40◦O, encontramos que el transporte de nutrientes por bombeo de Ekman ejerce un control principal en la productividad. Por otro lado, aunque la escala zonal del afloramiento costero es de tan solo unos 100 km, su influencia se extiende hacia océano abierto gracias al transporte de Ekman y a la convergencia del flujo meridional en Cabo Blanco (21◦N). El análisis de datos hidrográficos correspondientes a la primavera de 1973 y el otoño de 1975 muestran una exportación de costa a océano abierto de 2.9 Sv de agua y 53 kmol s-1 de nitratos durante la primavera, y de 0.6 Sv y 3 kmol de nitratos s-1 durante el otoño en el área al sur de Cabo Blanco. Entender la distribución de las diferentes masas de agua en la termoclina superior es fundamental, ya que transportan los nutrientes que van a aflorar a la capa fótica y mantener una elevada producción primaria. En la región de estudio, aguas centrales de origen norte (pobres en nutrientes) y sur (ricas en nutrientes) se encuentran en la zonal frontal de Cabo Verde. Aquí, mezcla por doble difusión intensifica la transferencia horizontal de calor. Como consecuencia, el frente aparece como una fuerte barrera en cuanto sal y nutrientes, pero suavizado en cuanto a la temperatura. La aplicación de un Análisis Multiaramétrico Óptimo a un conjunto de datos hidrográficos recogidos durante noviembre de 2007 y noviembre de 2008 muestra el frente como una transición rápida entre las dos masas de agua centrales. Sin embargo, en las capas intermedias la transición entre aguas Mediterránea y Antártica es más suave. ii Acronyms AAIW Antarctic Intermediate Water AC Azores Current CC Canary Current Chl:C chlorophyll to Carbon ratio CORE Common Ocean-Ice Reference Experiment CSIC Consejo Superior de Investigaciones Científicas CUC Canary Upwelling Current CVFZ Cape Verde frontal zone EBUS Eastern Boundary Upwelling Systems ENSO El Niño-Southern Oscillation GD Guinea Dome GFDL Geophysical Fluid Dynamics Laboratory IIP Instituto de Investigaciones Pesqueras ITCZ Intertropical Convergence Zone MC Mauritania Current MOM4 Modular Ocean Model version 4 MW Mediterranean Water NACW North Atlantic Central Water NADW North Atlantic Deep Water iii NAO North Atlantic Oscillation NEC North Equatorial Current NECC North Equatorial Countercurrent OMP Optimum Multiparameter PU Poleward Undercurrent SACW South Atlantic Central Water SLP Sea Level Pressure SSH Sea Surface Height SST Sea Surface Temperature TOPAZ Tracers for Ocean Phytoplankton with Allometric Zooplankton iv Contents Abstract i Resumen ii Acronyms iii 1 Introduction 1 1.1 Background ................................ 1 1.2 Objectives and thesis outline . . . . . . . . . . . . . . . . . . . . . . . 15 2 Water and nutrient fluxes 19 2.1 Introduction................................ 20 2.2 Dataset................................... 25 2.3 Watermasses ............................... 27 2.3.1 Property-property diagrams . . . . . . . . . . . . . . . . . . 27 2.3.2 Vertical distributions . . . . . . . . . . . . . . . . . . . . . . . 31 2.3.3 Horizontal distributions . . . . . . . . . . . . . . . . . . . . . 35 2.4 Mixing in the Cape Verde frontal system . . . . . . . . . . . . . . . . 37 2.4.1 Large scale instability . . . . . . . . . . . . . . . . . . . . . . 37 2.4.2 Doublediffusion ......................... 39 v 2.5 Water and nutrient fluxes . . . . . . . . . . . . . . . . . . . . . . . . 45 2.5.1 Referencelevel .......................... 46 2.5.2 Dynamicheight.......................... 47 2.5.3 Geostrophic velocities and nitrate fluxes . . . . . . . . . . . . 49 2.6 Water and nutrient balances . . . . . . . . . . . . . . . . . . . . . . . 51 2.6.1 Alongshore transports . . . . . . . . . . . . . . . . . . . . . . 51 2.6.2 Cross-shore transports . . . . . . . . . . . . . . . . . . . . . . 54 2.6.3 Closing the balances . . . . . . . . . . . . . . . . . . . . . . . 55 2.7 Conclusions ................................ 59 3 Meridional changes in water properties 63 3.1 Introduction................................ 64 3.2 Datasets .................................. 67 3.3 Spatial distribution of water properties . . . . . . . . . . . . . . . . 69 3.4 Water types and optimum multiparameter analysis . . . . . . . . . 76 3.5 Distribution of water masses . . . . . . . . . . . . . . . . . . . . . . . 81 3.6 Discussion and conclusions . . . . . . . . . . . . . . . . . . . . . . . 85 4 Physical drivers of interannual chlorophyll variability 89 4.1 Introduction................................ 90 4.2 Satellite data, model output and methods . . . . . . . . . . . . . . . 94 4.3 Satellite −modelcomparison...................... 98 4.4 Sea surface height and chlorophyll . . . . . . . . . . . . . . . . . . . 102 4.5 Mechanisms of chlorophyll variability . . . . . . . . . . . . . . . . . 106 4.5.1 Nutrientsupply..........................108 4.5.2 Coastal upwelling versus offshore upwelling . . . . . . . . . 112 4.6 Discussion and conclusions . . . . . . . . . . . . . . . . . . . . . . . 115 vi 5 Conclusions and outlook 119 5.1 Conclusions ................................119 5.2 Futureresearch ..............................123 6 Resumen en español 127 6.1 Introducción y objetivos de la tesis . . . . . . . . . . . . . . . . . . . 127 6.1.1 Introducción............................127 6.1.2 Objetivos y resumen de la tesis . . . . . . . . . . . . . . . . . 143 6.2 Flujos de masa y nutrientes . . . . . . . . . . . . . . . . . . . . . . . 146 6.3 Cambios meridionales en las propiedades de las masas de agua . . 151 6.4 Variabilidad interanual de la clorofila . . . . . . . . . . . . . . . . . 159 6.5 Conclusiones y trabajos futuros . . . . . . . . . . . . . . . . . . . . . 170 6.5.1 Conclusiones ...........................170 6.5.2 Trabajosfuturos..........................174 Appendices A Adiabatic approximation 179 B Double diffusion and the Turner angle 181 C MOM4 and TOPAZ 183 C.1 ModularOceanModel..........................183 C.2 TOPAZ...................................184 References 187 vii Chapter 1. Introduction ¬: ¬: ¬: Cape Verde Cape Blanc Cape Timiris Cape Bojador Cape Juby Cape Ghir AC CUC WINTER NECC MC CC NEC Canary Islands Cape Verde I. Azores I. Madeira ¬: ¬: ¬: ¬1 ¬1 ¬1 ¬1 SUMMER Cape Verde Cape Blanc Cape Timiris Cape Bojador Cape Juby Cape Ghir Canary Islands Cape Verde I. GD NEC NEC MC NECC AC CC/CUC NECC Azores I. Madeira Figure 1.2: Schematic map of the main surface currents in the eastern North Atlantic Ocean during summer and winter. Blue bands along the coast mark the occurrence of intense upwelling during each season; during summer the Guinea Dome (GD) is also shown in blue. The dashed gray line marks the Cape Verde frontal zone. Current labels: Azores Current (AC), Canary Current (CC), Canary Upwelling Current (CUC), Mauritania Current (MC), North Equatorial Current (NEC), North Equatorial Countercurrent (NECC). The Poleward Undercurrent flows north as a subsurface current along the upper slope, its core typically at some 200 m depth, in many instances reaching the ocean surface. 6 Chapter 1. Introduction Figure 1.3: Temperature-salinity digram showing water mass types in the central, intermediate and deep layers of the Atlantic Ocean, from Sverdrup et al. (1942). coastal jet and CUC reach lowest latitudes as a southward flow along the African coast between capes Blanc and Verde (Lázaro et al., 2005), providing a connection between the northern and southern waters (Mittelstaedt, 1991). The strengthening of the trades at these lower latitudes cause the NECC to weaken and meander, and the MC reaches only south of Cape Verde (Fig. 1.2). The cyclonic flow system around the Guinea Dome is masked by the westward Ekman drift at the surface, but Siedler et al. (1992) reported that the dome still exist in the subsurface layers. Water masses The surface layer in the eastern subtropical North Atlantic (approximately the upper 100 m) is characterized by high salinities, high dissolved oxygen and low nutrient concentrations (Fraga and Manríquez, 1974). A particular feature of the North Atlantic subtropical gyre is the subsurface salinity maximum (e.g. Figs. 7 Chapter 1. Introduction 2.4 and 2.5). Dense high-salinity surface waters, produced by the excess evaporation over precipitation, subduct due to Ekman transport convergence and winter convection, and are carried into the large-scale circulation of the subtropical gyre (Defant, 1936; Bauer and Siedler, 1988). The water masses of the permanent thermocline are the central water masses (Sverdrup et al., 1942). These waters originate in the subtropical oceans of both hemispheres, where surface waters converge and Ekman puming is negative, and spread towards the equatorial regions. They are recognized in a TS-diagram by nearly linear TS-relationships (Fig. 1.3). The central waters extend from approximately 100 to 700 m. NACW have their source in the North Atlantic surface convergence (subducting) zone and reach subsurface waters at lower latitudes through the thermocline circulation (Sarmiento et al., 1982; Kawase and Sarmiento, 1985). Only water masses formed in the subducting zone during late winter and early spring may escape the surface layers to be injected in the permanent thermocline; during the rest of the year, waters subducted remain in the mixed-layer as the escape velocities caused by Ekman pumping are less than the seasonal advance of the mixed-layer thermocline. The vertical extension of the NACW in lower latitudes will be determined by the densest winter-outcrop isopycnal within the subducting zone, near σθ= 27.3 (Kawase and Sarmiento, 1985; Reid, 1994). Emery and Meincke (1986) subdivided the Central Water of the North Atlantic (NACW) into Eastern (ENACW) and a Western (WNACW) types. The division reflected different formation regions, south of the Subarctic Front for WNACW and south of the Iceland-Faroe Front for ENACW. Conversely, Tomczak and Godfrey (1994) argued that the temperature-salinity changes observed 8 Chapter 1. Introduction across the ocean basin result from environmental variability within the formation region. Meanwhile, Emery and Meincke (1986) identified only one type of South Atlantic Central Water (SACW), formed near the Brazil-Malvinas Confluence region of the Subtropical Convergence. Gordon et al. (1992) and Sprintall and Tomczak (1993) showed that central water formed in the Subtropical Convergence of the Indian Ocean was an important contributor to the thermocline of the Atlantic Ocean, entering the Atlantic basin via the Agulhas Current. SACW travels through the South Atlantic thermocline into the equatorial current system and the tropical region of the North Atlantic. The Cape Verde frontal zone (CVFZ) constitutes the boundary between NACW and SACW and corresponds to the southern limit of the North Atlantic thermocline recirculation (Stramma and Siedler, 1988; Zenk et al., 1991; Arhan et al., 1994). The CVFZ stretches southwest from 20◦N off the coast of Africa to the Cape Verde Islands, and then acquires a more zonal orientation as it progressively diffuses out towards the western side of the basin. Both central water masses occupy the same density range, with NACW being saltier and warmer than the SACW. As a result the front is density-compensated and is prone to multitude of intrusions, filaments and lenses (Zenk et al., 1991; Pérez-Rodríguez et al., 2001; Pastor et al., 2008). The intermediate layer in the study region, approximately between 700 and 1500 m, is occupied by Antarctic Intermediate Water (AAIW) and Mediterranean Water (MW). AAIW is formed in the Subantarctic Front and carried with the subtropical gyre of the South Atlantic towards the tropics (Suga and Talley, 1995). The AAIW is easily identified by low salinities with a salinity minimum centered at about 800 m depth. Two pathways transport AAIW to the eastern North Atlantic, one is through the western boundary current system and the Azores 9 Chapter 1. Introduction Current (Kawase and Sarmiento, 1985; Tsuchiya et al., 1992). The second path is through the eastern margin along the African coastline (Machín and Pelegrí, 2009), with a maximum northward penetration during fall (Machín et al., 2010). MW is formed in the Mediterranean Sea and enters the Atlantic through the Strait of Gibraltar; from here, it spreads north and southwards, influencing the whole North Atlantic (Worthington, 1976). Its characteristic high salinity and temperature signature is observed between about 600 and 1500 m depth. Both intermediate waters meet at about 32◦N (e.g Fig 3.5), with AAIW occupying a depth range slightly shallower than MW. Source water properties of intermediate waters are modified by mixing with water above and below, and their TS properties in the study region appear as local extreme TS values that deviate substantially from the values found in their formation region (Fig. 1.3). North Atlantic Deep Water (NADW) is found in the layer below, between 2000 and 4000 m approximately. NADW in the eastern North Atlantic is formed mainly by the Iceland Scotlan Overflow Water, modified by Labrador Sea Water and Lower Deep Water (McCartney, 1992; Dickson and Brown, 1994; van Aken, 2000). Interannual variability and climatic modes Interannual variability in coastal upwelling has mostly been related to the North Atlantic Oscillation (NAO) and to El Niño - Southern Oscillation (ENSO) events. The NAO index is defined as the anomalous difference in sea-level air pressure between the Iceland Low pressure system and the Azores High pressure system during the winter season (December through March). An increase in the index 10 Chapter 1. Introduction implies a stronger Azores High and increased easterly flow over NW Africa, leading to an increase in the wind driven coastal upwelling. Analyzing satellite SST from 1982 to 2001, Santos et al. (2005) reported a decadal scale shift of upwelling regime intensity from weak upwelling in 1980’s to intense in the 1990’s, linked to a shift in the NAO index. Meiners (2007) and Meiners et al. (2010) also found a positive correlation between the NAO index and the trade winds when studying impacts of climate variability on black hake dynamics in the Northwest African coast. Specifically, the NAO index could explain 53% of the variability in the meridional component of the wind stress off the coast of Mauritania and Senegal between 1960 and 2004. Roy and Reason (2001) investigated links between the Multivariate ENSO Index and SST anomalies and wind stress anomalies off Northwest Africa (between 10◦and 20◦N). Their work showed that warm events in the Pacific during fall and early winter (El Niño conditions) lead to a relaxed state of the windinduced upwelling on the eastern side of the Atlantic basin and to warm conditions being observed along the coast of West Africa during late winter and spring. ENSO events also influence the Atlantic equatorial current system and the Guinea Dome. Lázaro et al. (2005) observed an intensification of the NECC during spring 1997 and 1998, and of the Guinea Dome during summer 1997, associated with the premature northward displacement of the ITCZ reported by Enfield and Mayer (1997), and coinciding with a La Niña event in the Pacific. Trends in coastal upwelling have also been identified by various authors, although they are sometimes contradictory. Bakun (1990) identified a significant increase in upwelling-favorable wind stress in a location off Northwest Africa at 28◦N from 1946 to 1981. The intensified winds in this site coincided with consis11 Chapter 1. Introduction tent increases in other sites of the world’s major coastal upwelling systems. They hypothesized that, in a global warming scenario, as atmospheric greenhouse gas concentrations increase, there would be an intensification of coastal upwelling due to an increase in the land-sea pressure gradient. Using proxy temperature data derived from sedimentary records extending back 2500 years, McGregor et al. (2007) inferred an anomalous and unprecedented increase in coastal upwelling off Cape Ghir during the 20th century. Conversely, analyzing QuickScat wind data from 2000 to 2007, Demarcq (2009) found a decreasing trend in the meridional component of the wind stress off Northwest Africa. However, this may only reflect interannual to decadal variability superimposed on a longer time scale trend. Primary production The coastal upwelling region off Northwest Africa is one of the four major Eastern Boundary Upwelling Systems (EBUS) of the world’s oceans. EBUS display high productivity and important fishery yields (Pauly and Christensen, 1995). The subtropical Northeast Atlantic has the largest active zone of all main EBUS, defined as the area where chlorophyll concentrations are above 1 mg m-3 (Carr, 2002). It is the second most productive EBUS, with an annual primary production of 0.33 Gt of Carbon per year, after the Benguela EBUS in the South Atlantic (Carr, 2002). Chlorophyll concentrations provide an indirect measurement of phytoplankton abundance and thus numerous efforts have been made to develop an algorithm to derive primary production from remote sensed chlorophyll (Behrenfeld 12 Chapter 1. Introduction 0.03 0.1 0.3 1 3 CHL (mg m ) -3 40 30 20 10 40 30 20 10 -30 -20 -10 -30 -20 -10 SUMMER WINTER SPRING FALL Cape Blanc Cape Blanc Cape Blanc Cape Blanc Figure 1.4: Mean seasonal chlorophyll concentrations (mg m-3) for winter (JFM), spring (AMJ), summer (JAS) and fall (OND). Black contours mark the 0.2 and 1 mg m-3 chlorophyll isolines. Data corresponds to the SeaWiFS satellite sensor for the years 1998 to 2007. 13 Chapter 1. Introduction and Falkowski, 1997; Carr, 2002; Marra et al., 2003). Figure 1.4 shows seasonal chlorophyll concentrations derived from 10 years of SeaWiFS satellite chlorophyll data. The region between 24◦N and the Strait of Gibraltar displays weak seasonal variability. Chlorophyll concentrations above 1 mg m-3 are confined to the shelf, despite the year-long upwelling occurrence. The weak offshore extension of chlorophyll in this region may be caused by nutrient limitation, as the negative wind stress curl that defines the subtropical gyre depresses the nutricline (Lathuilière et al., 2008). The region between 18 and 24◦N also shows a weak seasonality, but in this case high chlorophyll concentrations extend far offshore all year round. This latitudinal band includes, as the most predominant feature, the Cape Blanc giant filament (Gabric et al., 1993). During summer and fall, intense offshore transport off Cape Blanc occurs as the result of convergence of the southward CUC and the poleward MC. In winter and early spring, as upwelling reaches further south towards Cape Verde, convergence and offshore transport are reduced but still present (Pelegrí et al., 2006). Filament-like structures that advect chlorophyll offshore can also be seen off the major capes, such as Cape Ghir, where primary production reaches values of 5 g C m-2 year-1 (García-Muñoz et al., 2005; Pelegrí et al., 2005). Between 18◦N and the Strait of Gibraltar mesoscale features, like filaments and eddies, and wave processes contribute to the dynamical and biochemical variability in the region, but they are not discussed here. An extensive analysis of mesoscale processes is given by Barton et al. (1998) and citetBarton1998a and an excellent review on propagating waves is provided by Hagen (2001). The region south of 18◦N presents a large offshore extension of chlorophyll 14 Chapter 1. Introduction during winter and spring, when coastal upwelling bring nutrient rich waters to the surface. During summer, the strengthening of the NECC and the offshore displacement of the positive Ekman pumping area uplifts the upper layers of the thermocline, and by fall high primary production can be observed in the GD area (10◦N 22◦W approximately). Pelegrí et al. (2006) propose a mechanism by which the subsurface layers in the GD maintain a high nutrient concentration level. The GD develops during summer and fall and nutrients that reach the euphotic layer are utilized. During winter, upwelling-favorable trade winds extend south of Cape Blanc, and the GD relaxes. The vertical cell associated to coastal upwelling needs a supply of subsurface waters from the interior ocean, replenishing the nutrient levels of the GD region. 1.2 Objectives and thesis outline This Ph.D. thesis contains three central chapters in scientific article format, preceded by a general introduction and ended by general conclusions. A spanish summary of the thesis follows, which includes the objectives, methodology, main findings and conclusions. A wide range of data sources have been examined. Historical data have been reanalyzed. We also examine newly acquired hydrographic data and available remote sensed properties such as chlorophyll concentrations and sea surface height. Finally, an ocean general circulation model coupled to a state of the art biogeochemistry model provides insight into physical mechanisms driving variability in biochemical properties. 15 Chapter 2. Water and nutrient fluxes The steady-state connection between the offshore and coastal-upwelling fronts is controlled by three main factors: the location of the Cape Verde frontal zone, the intensity and latitudinal extension of coastal upwelling, and the size and location of the Guinea Dome open-ocean upwelling area. One of the main features in this region already described in the Introduction is the Cape Verde frontal system. It stretches southwest from Cape Blanc to the Cape Verde Islands, and effectively separates relatively new (salty, warm, nutrient-poor, and oxygen-rich) NACW from the older (fresh, cold, nutrient-rich, and oxygen-poor) SACW. Running meridionally along the coast we find an additional front between the cold upwelling belt and the warmer offshore waters. South of the Cape Verde frontal zone we find a cyclonic circulation around the Guinea Dome. The dome and associated circulation move offshore, towards the central Atlantic, during summer; in winter Ekman pumping intensifies and moves east (Nykjaer and Van Camp, 1994), merging with the coastal upwelling zone. When considering the main controls on the steady-state dynamics, it is important to keep in mind that the mean field also experiences remotely-forced interannual and interdecadal variations. Arfi (1985) used local data at 20◦N to suggest that upwelling intensified from the 60’s to the 70’s. Roy (1991) also presented results supporting that the 70’s was a period of relative intense upwelling which decreased in the 80’s. More recently satellite-derived sea surface temperature (SST) measurements have shown that the 80’s was a period of little interannual variability of the upwelling index (difference in SST between the coast and the open ocean) off Northwest Africa, with oscillations of a few tenths of a degree (Nykjaer and Van Camp, 1994; Hernández-Guerra and Nykjaer, 1997), but 22 Chapter 2. Water and nutrient fluxes that there existed a major shift by the end of the 80Õs into the mid 90Õs, with a change in the upwelling index greater than one degree (Santos et al., 2005). These changes are quite important as compared with the mean upwelling indexes for the 18 to 26◦N band, which range between 1 and 2◦C (Nykjaer and Van Camp, 1994; Hernández-Guerra and Nykjaer, 1997; Santos et al., 2005). The interannual oscillations in the upwelling index appear to be associated to individual North Atlantic Oscillation (NAO) events while the interdecadal variations may be related to sustained NAO events during several consecutive years (Santos et al., 2005). Our period of interest (March-April 1973 and October-November 1975) had moderate three-month averaged NAO indexes (Climate Prediction Center, http://www.cpc.ncep.noaa.gov), which are characteristic for the mild intensification observed during the 70’s. The above brief description of the main circulation patterns immediately points at several key issues that control the fluxes and large-scale patterns of central waters along Northwest Africa. The Cape Verde frontal system is characterized by sharp mesoscalar intrusions of both NACW and SACW (named interleaving after Barton and Hughes, 1982), how effective is it as a barrier between northern and southern waters? The coastal upwelling front stretches meridionally until the Cape Verde front, beyond in winter, is there a connection between these two frontal systems? The coastal transition zone is characterized by vertical upwelling cells and ubiquitous filaments that effectively transfer nutrient-rich waters into the nutrient-depleted surface waters of the subtropical gyre, where and at what rates these mass and nutrient exchanges take place? In this work we have recovered the data of four historical cruises to produce two large-scale synoptic data sets. These data sets are extensively analyzed to 23 Chapter 2. Water and nutrient fluxes 24W 22W 20W 18W 16W 14W 16N 18N 20N 22N 24N 26N ATLOR II-III March-April 1973 Atlor II March Atlor III April C. VERDE BOJADOR C. BLANC 50 m N S L Spring 50 m 0 24W 22W 20W 18W 16W 14W 16N 18N 20N 22N 24N 26N C.VERDE BOJADOR C.BLANC Atlor VI October Atlor VII November N S L 500 m 50 m Fall Figure 2.1: Hydrographic stations and selected sections for the spring (left) and fall (right)data sets. The legend for the sections is as follows: N, north; S, south; L, along-slope. investigate the above questions, so that we can improve our understanding and quantification of the water-mass and nutrient connections in the central waters all along Northwest Africa. In section 2.2 we briefly describe the original data, and how it has been assembled to produce spring and fall data sets, and in section 2.3 we plot the data to show the hydrographic conditions in the area during these two seasons. In section 2.4 we explore what processes are responsible for enhancing mixing at the frontal system, in section 2.5 we use the data to infer the surface and subsurface dynamic fields in the region, and in section 2.6 we compute the along and cross-shore mass and nutrient transports. 24 Chapter 2. Water and nutrient fluxes 2.2 Data set The data used in this work were collected by IIP as part of four different hydrographic cruises: Atlor II (March 1973), Atlor III (April 1973), Atlor VI (October 1975), and Atlor VII (November 1975). These cruises stretched along the African coastline, between 16.5 and 26.1◦N (Fig. 2.1) , and the data were compiled and published in Cruzado and Manríquez (1974); Fraga and Manríquez (1974); Manríquez and Rucabado (1976); Manríquez and Fraga (1978). Cruises Atlor III and Atlor VI covered the northern continental shelf and upper slope, limited by Point Durnford (23.6◦N), Cape Bojador (26.1◦N), the coast and the 500 m isobath. In these northern cruises the station spacing was about 30 km, and samples were taken down to 500 m (or to the bottom if shallower) at standard depths. Cruises Atlor II and Atlor VII ran from 17◦N/16.5◦N (Atlor II/VII) to Cape Blanc (21◦N) in the region from the continental slope to the 22◦W meridian, and had several additional stations as far north as 23◦N. In the southern cruises the mean distance between stations was approximately 80 km, and samples were taken down to 1000 m (or to the bottom), also at standard depths. During all cruises samples were taken with either Niskin or Hydro-Bios bottles. Temperature was obtained using Thermoschneider Wertheim/Main protected and unprotected thermometers, corrected using Hidaka’s equation (Keyte, 1965). Conductivity was obtained using a Hytech 6220 induction salinometer, and salinity was calculated using UNESCO’s (1966) equations. Dissolved oxygen was determined using the Winkler method (Strickland and Parsons, 1968), and nutrients were measured using a Technicon autoanalyzer. Other details on instrumentation and some preliminary analysis for these data sets may be found in Ballester et al. (1972) (Atlor II and At25 Chapter 2. Water and nutrient fluxes lor III) and Manríquez and Rucabado (1976) (Atlor VI and Atlor VII). Due to their location and dates, the four cruises may be combined into two data sets (each with about 80 hydrographic stations) with similar regional coverage, roughly from north of Cape Verde (17◦N) to Cape Bojador (26◦N). The first data set comprises Atlor II and Atlor III, and will hereafter be referred as spring (March-April 1973). The second data set includes Atlor VI and Atlor VII, and will hereforth be referred as fall (October-November 1975). These data sets allow the intercomparison of two synoptic situations for opposite seasons. One limitation for such an integrated study is the distinct zonal coverage of the data, as the original northern and southern cruises were designed to study the upwelling transition zone north of Cape Blanc and the Cape Verde frontal system, respectively. Nevertheless, all cross-shore sections extend well beyond the continental slope, so that we may expect they have good coverage of the coastal upwelling jet and adjacent interior ocean. Figure 2.1 shows the location of all hydrographic stations, as well as three selected sections that will be used to illustrate the vertical distribution of hydrographic and chemical properties. The northern section (N, about 25.5◦N) extends from the shelf to the middle slope (100 km long), the southern section (S, 21◦N) stretches some 500 km offshore from the continental slope, and the along-shore section (L, from 16 to 26◦N) runs some 1200 km roughly along the 500 m isobath. 26 Chapter 2. Water and nutrient fluxes 2.3 Water masses 2.3.1 Property-property diagrams Figure 2.2 illustrates the potential temperature-salinity (θ-S) relationships during both seasons, as obtained using all available stations (hereafter, when talking about temperature we will actually refer to potential temperature, calculated using the surface as the reference level). In this figure, as well as in Figure 2.3 , we have included straight lines that define NACW and SACW (Tomczak, 1981). The hydrographic stations may be divided in three groups, following a Northto-South θ-S transition from NACW to SACW: northern, frontal, and southern. Northern stations, with NACW characteristics, are found north of Cape Blanc in both data sets (Fig. 2.2 , left). There is also one station in each set at the latitude of Cape Blanc (20◦N, 500 km offshore in spring; 21◦N, 300 km offshore in fall) that displays NACW characteristics. Frontal stations, with transitional characteristics between NACW and SACW, are located off Cape Blanc all the way from the coast to deep waters, and yet further south in deep waters (Fig. 2.2 , center). Southern stations, with SACW origin, are found only south of Cape Blanc. These are mainly located over the continental slope, probably reflecting the influence of the poleward undercurrent, although in spring they reach further offshore and two slope stations display mixed characteristics at several depths (Fig. 2.2 , right). The fall data (Fig. 2.2 , bottom) shows high dispersion near the sea surface because of the presence of the seasonal mixed layer. During this season there are also some low salinity near-surface values that correspond to the three southernmost stations (located between latitudes 16.5 and 17◦N, and 75 to 200 km offshore), which 27 Chapter 2. Water and nutrient fluxes Figure 2.2: Spring (top) and fall (bottom) potential temperature-salinity diagrams displaying NACW (left), SACW (right), and intermediate (center) characteristics. The lines correspond to the two central water masses, as explained in the text. The insets indicate the location of the stations for each diagram. 28 Chapter 2. Water and nutrient fluxes Figure 2.3: Spring (top) and fall (bottom) property-property plots. Dots, triangles and crosses correspond, respectively, to stations displaying NACW, SACW, and intermediate characteristics in the temperature-salinity diagrams. The solid lines correspond to the two central water masses, as explained in the text. In the right panel the dotted line has the slope of the oxygen-nitrogen stoichiometric Redfield ratio. 29 Chapter 2. Water and nutrient fluxes likely respond to the influence of fresh water from the Senegal River discharge at 16◦N. In Figure 2.3 we have used exactly the same groups (northern, frontal and southern waters, indicated with dots, crosses and triangles, respectively) as in Figure 2.2 to illustrate the distribution of potential temperature, dissolved oxygen concentration (O2) and nitrate concentration (NO3) as a function of salinity (S), and of O2as a function of NO3. In this figure we have used all available stations but exclude the data points for the top 50 m, as the upper-mixed layer displays surface warm-water characteristics (Fraga and Manríquez, 1974) and is both nutrient-exhausted as a result of biological activity and oxygen-rich through contact with the atmosphere (Lamb, 1984). On these plots we have again drawn straight lines that define the NACW and SACW. As source water values for O2 are not included in Tomczak (1981), we have obtained these from the oxygen concentration values of those stations that closely match each water-mass line in the θ-S diagram. The procedure consists in making a linear regression to the salinityoxygen values, from which we then obtain the oxygen concentrations that correspond to the extreme salinity values defined by Tomczak (1981). The O2values obtained in this manner are comparable with results from Klein and Tomczak (1994), and the procedure, when repeated for NO3, shows good agreement with the water type definitions in Tomczak (1981). The straight lines in Figure 2.3 correspond to the water type values obtained with the above procedure for O2, and Tomczak’s (1981) values for the other parameters. Table 2.1 summarizes the property values that characterize the northern and southern central source water types. The left panel of Figure 2.3 (θ-S diagram) shows that water characteristics 30 Chapter 2. Water and nutrient fluxes change rather smoothly between northern (NACW) and southern (SACW) ends. The central panels (O2-S and NO3-S diagrams) of Figure 2.3 also illustrate that most data points lie between the NACW and SACW lines but the frontal domain appears to have more sparse coverage, with transition stations (as determined from the θ-S diagram) now preferentially shifting towards either water mass. We will come back to this issue in the next section, where we propose that this feature is the result of intense double diffusion in the Cape Verde frontal system (Zenk et al., 1991). The right (O2-NO3diagram) panel in Figure 2.3 shows the lines for the two water masses almost as if placed one after the other. The reason is that oxygen and nitrate are not fully independent variables, as nitrate increases/diminishes through oxidation/photosynthesis. The two lines, however, do not have the same slope as the two water masses have very different ages (Poole and Tomczak, 1999) , the slope for the NACW being similar to that expected from the oxygen-nitrate stoichiometric Redfield ratio (Takahashi et al., 1985). Finally, it is worth mentioning that the diagrams involving either dissolved oxygen or nutrients display substantial scatter around those lines that characterize these water masses. This is probably the result of enhanced primary production in the upwelling region. 2.3.2 Vertical distributions In Figures 2.4 and 2.5 we present the spring and fall distributions of T, S, σθ, O2, and NO3, down to 400 m, along sections N, S, and L (Fig. 2.3 ). Section N corresponds to pure NACW. During both seasons the T and NO3isopleths raise towards the coast, down to 100 m during fall and to 200 m in spring. The S and O2 31 Chapter 2. Water and nutrient fluxes C. Blanc C. Blanc C. Blanc C. Blanc C. Blanc C. Blanc C. Blanc cyclonic anticyclonic 25ºW 20ºW 15ºW 15ºN 20ºN 25ºN 25ºW 25ºW 25ºW 25ºW 20ºW 15ºW 15ºN 20ºN 25ºN 20ºW 20ºW 20ºW 15ºW 15ºW 15ºW 25ºW 20ºW 15ºW 25ºW 20ºW 15ºW Figure 2.8: Sequence of five-day mean SST images spanning from January 11 to 31, 2003 (top) and from January 21 to February 5, 2004 (bottom). 38 Chapter 2. Water and nutrient fluxes Blanc, something characteristic of winter months (Pelegrí et al., 2006). These images indicate the presence of mesoscalar structures (cyclonic/anticyclonic with cold/warm cores) along the coastal upwelling front, north and south of Cape Blanc, and suggest they are advected along this upwelling front until reaching the Cape Verde frontal system. Convergence of southern and northern waters at the frontal zone produces offshore export of the relatively cold upwelling waters. The offshore advection of these upwelled waters results in the development of a giant filament of relatively cool waters, clearly visible in the SST images (Gabric et al., 1993). As the Cape Verde front is density-compensating the interleaving of water masses at this frontal region may either arise from the kinetic energy of the flow converging off Cape Blanc, or simply reflect the offshore advection of the mesoscalar structures developed along the upwelling front, both north and south of Cape Blanc. These mesoscale meanders and vortices would penetrate to depths characteristic of the baroclinic coastal jet (about 200 m), which is typically about half the vertical scale of most features in the Cape Verde front (some 400 m, Figures 2.4 and 2.5 ), so that they certainly can only explain part of the frontal variability. In either case, the ultimate origin of interleaving would be wind forcing over the boundary region in the eastern subtropical and tropical gyres. 2.4.2 Double diffusion The unstable vertical distribution of salt in the Atlantic makes its upper ocean favorable to double diffusion in the form of salt fingers (Klein and Tomczak, 1994; St Laurent and Schmitt, 1999; You, 2002). This situation is enhanced at the frontal 39 Chapter 2. Water and nutrient fluxes Ocean Data Vie w 20 40 60 80 300 200 100 0 75 90 45 75 6 0 Ocean Data Vie w 20 40 60 80 300 200 100 0 45 45 60 Ocean Data View 100 200 300 400 45 45 45 45 75 60 60 Ocean Data View 100 200 300 400 75 75 45 45 45 45 60 60 Ocean Data View 0 200 400 600 800 1000 45 75 45 45 75 60 60 60 60 75 75 75 Ocean Data View 0 200 400 600 800 1000 1200 45 45 60 60 60 60 Distance [km] Depth [m] Africa Africa North South NSL Figure 2.9: Spring (top) and fall (bottom) distributions of the Turner angle (between 45◦and 90◦) on sections N (left), S (center), and L (right). Notice the change in horizontal scale between adjacent columns (scale halves changing column to the right). system, as salty NACW overlies relatively fresh SACW, so we wonder if double diffusion may be responsible for the relatively smooth distribution of temperature as compared with other hydrographic properties such as salinity and nutrients. In order to explore this idea we have computed the vertical and horizontal distributions of the Turner angle (see Appendix B for more details on the Turner angle). For the computation of the Turner angle we have used 50 m as the vertical difference interval (between 0 and 300 m) to obtain the mean state of double diffusion, similar to other studies for the Atlantic Ocean (Schmitt, 1990; You, 2002). Salt fingering may take place when this angle is between 45◦and 90◦, with moderate fingering occurring for angles between 60◦and 75◦, and strong fingering (s-f) prevailing for angles between 75◦and 90◦. Figure 2.9 presents the vertical distribution of the Turner angle for the same vertical sections as in Figures 2.4 and 2.5 . The distribution clearly suggests that sf is associated with the frontal system. Section S displays s-f angles in the offshore 40 Chapter 2. Water and nutrient fluxes Figure 2.10: Spring (left) and fall (right) distribution of the Turner angle at 225 m. area during spring, and at the two ends of the section during fall. Section L shows s-f angles in the northern half, below 100 m during spring and below 200 m during fall. The coastal upwelling band in the northern area, characterized by Section N, also has high Turner angles between 100 and 200 m depth. The reason for this is the presence of the subsurface salinity maximum, which makes this depth range prone to double diffusion. This may actually be an effective mechanism for diapycnal mixing, helping to maintain the vertical circulation cell in this upwelling zone. Figure 2.10 shows the horizontal distribution of the Turner angle at 225 m, which is close to the depth level (200 m) chosen for the horizontal distributions of properties (Figs. 2.6 and 2.7 ) and has good coverage of the frontal system. We could actually choose any horizon between 100 and 300 m and the results would not change too drastically. During spring Turner angles above 60◦, and even above 75◦, are found far off Cape Blanc and in the southernmost areas. During fall, on the other hand, most of the frontal system off Cape Blanc displays high 41 Chapter 2. Water and nutrient fluxes Turner angles. Again here we may appreciate that the northern coastal upwelling band is characterized by Turner angles greater than 60◦. Let us propose a simple mechanism that would favor the epipycnal (alongisopycnal or, because of the density compensating character of the front, nearhorizontal) smoothing of the temperature field in frontal regions, with warm saline waters overlying cool fresh waters (Figure 2.11). In this figure we only illustrate the thermal field, although the initial isotherms and isohalines are coincident. Suppose the central portion of the front (the most unstable one) undergoes intense fingering, so that downward (downgradient) salt and heat fluxes take place (Ruddick and Gargett, 2003). The heat flux decreases more rapidly with depth than the salt flux, as heat diffuses epipycnally (gray arrows in Fig. 2.11), away from the frontal region. In the warm-water side of the frontal region this flux would oppose the epipycnal down-gradient heat flux (horizontal black arrows in Figs. 2.11 a and 2.11 c), while in the cold-water side the two fluxes would add up. As a result the temperature gradients in the warm-water side would remain essentially unchanged but those in the cold-water side would decrease substantially. The enhancement of horizontal heat diffusion in the frontal system would conceal the original water masses in θ-S diagrams, in opposition to their unmasked presence in other property-property plots (such as NO3-S where the characteristics of a water parcel remain closer to those of the original water masses). The property-property plots in Figure 2.3 confirm the presence of NACW in the surface layers of several slope stations south of Cape Blanc, something not visible in the more diffusive θ-S distributions (Figs. 2.2 and 2.3 ). In order to illustrate the above ideas we have prepared Figure 2.12 , which 42 Chapter 2. Water and nutrient fluxes z cross-frontal coordinate, x z = const, or ρ const a ≅ b T S T, S z T 1 T 4 > T 3 T 5 > T 4 T 2 > T 1 T 3 > T 2 d-g s-f s-f x c d-g T Figure 2.11: (a) Schematics of frontal region with horizontal isopycnals (thin dotted lines) and sloping isotherms (thick lines). The initial (dotted) and final (solid) temperature distributions are shown. (b) Salt (S) and heat (T) vertical fluxes in the frontal region, illustrating the presence of vertical heat convergence. (c) Initial (dotted) and final (solid) temperature distribution on isopycnal illustrating strong-fingering (s-f) and ordinary down-gradient (d-g) heat fluxes. 43 Chapter 2. Water and nutrient fluxes Figure 2.12: Spring (top) and fall (bottom) nitrate-salinity (left) and potential temperature-salinity (right) diagrams. Only data points with intermediate water characteristics are included. Squares, triangles and asterisks correspond to data points with σθapproximately given by 26.5, 27.0 and 27.5, respectively (as explained in the text), and all other points are shown as dots. 44 Chapter 2. Water and nutrient fluxes shows the NO3-S and θ-S diagrams during spring and fall, but now including only the frontal stations (according to their distribution in the θ-S diagram). Here we have not included measurements at depths less than 50 m, in order to remove the effects of the surface mixed layer. In this figure we differentiate the data points with densities close to σθ= 26.5, 27.0 and 27.5 (σθ= 26.5 actually corresponds to all data points 26.45 <σθ<26.55, and similarly for the other two values). In order to illustrate how properties change along a density surface, we fit a regression line to each density value. At the mesoscale the differences between both diagrams can only arise from different diffusive heat behavior. The figures illustrate that in the NO3-S diagram the data points remain close to the two original water masses while in the θ-S diagrams they are distributed rather evenly along the straight lines. The lack of intermediate nitrate and salinity values indicates that these properties undergo much less epipycnal diffusion than temperature. 2.5 Water and nutrient fluxes The vertical upwelling cell drives nutrient-rich subsurface waters to the baroclinic zone, where they become accommodated by along-shore water and nutrient fluxes all the way to the Cape Verde front. In this manner the upwelling jet, if deep enough, may play a decisive role in the along-shore advection of the northern water-masses. The Cape Verde frontal system is usually thought to be the southern limit reached by the northern waters. This system moves seasonally several degrees (Stramma and Siedler, 1988), but atmospheric forcing has a much greater latitudinal oscillation, resulting in large changes in the SACW geostrophic 45 Chapter 2. Water and nutrient fluxes Figure 2.13: Geostrophic velocities across section S in fall, relative to (a) 200, (b) 300, (c) 400, and (d) 500 m. fluxes south of the Cape Verde front. In this section we explore the central water and nutrient paths in the region. 2.5.1 Reference level In order to calculate the geostrophic velocity field we need to select a proper reference level. For our data sets, where many stations are located in relatively shallow waters, we have to compromise between deepness, so that the reference velocities are indeed small, and shallowness, so that our sections reach down to the selected reference level. Once a reference level is chosen, we may extend it 46 Chapter 2. Water and nutrient fluxes to shallower stations over the slope and shelf, through the method described in Csanady (1979). The main limitation of this method is that it requires the bottom densities to assure a zero bottom alongshore velocity, which need to be interpolated from the available values. In Figure 2.13 we illustrate, as an example, the geostrophic velocities in section S during fall, with four different reference levels (200, 300, 400, and 500 m). There are significant differences between the velocity field calculated using the 200 m and 300 m reference levels. The differences, however, decrease greatly between 300 m and the deeper reference levels. This situation repeats itself for most transects (not shown), for which reason we have selected the 300 m as an appropriate reference level. However, use of such a shallow reference level over the slope may lead to underestimates of the poleward undercurrent. 2.5.2 Dynamic height Figure 2.14 shows distributions of dynamic height (referenced to 300 m) at the sea surface and 100 m, for both the spring and fall data sets. From the dynamic height differences we may infer the gross transport patterns (per unit depth). These have been sketched over the dynamic height distribution to provide an overview of the nearsurface regional circulation. Each arrow corresponds to a flux of 3.7 ×103m2s−1or, equivalently, to a water transport of 0.37 Sv over a 100 m deep water-column. The flow patterns are rather complex, similar to those suggested by the sequences of SST images in Figure 2.8 . A recurrent pattern for both seasons is the offshore-inshore cyclonic recirculation that occupies the whole region between 47 Chapter 2. Water and nutrient fluxes transports take place off Cape Blanc, 1.0 Sv for water-mass and 14 kmol s-1 for nitrate. In the southern area the transports revert sign, their maximum value being 1.8 Sv and 38 kmol s-1. During fall we find alternation of northward and southward transports in both the northern and southern areas. Despite the southward water-mass transport predominates over most northern sections we find that the mean nitrate transport is very close to zero, this being possible as northward flow usually takes place associated to the poleward undercurrent, where NO3is greater than near-surface. Only in the southernmost three sections we find predominant northward mass and nutrient transports, with maximum values of 0.5 Sv and 9 kmol s-1 off 19◦N. The presence of northward transport in this southern region during both seasons (substantially greater in spring) reflects the intensification of the poleward undercurrent. 2.6.2 Cross-shore transports Along-shore convergence/divergence, as calculated from the difference in alongshore transport through consecutive normal-to-shore sections, traduces in crossshore export/import. In Fig. 2.18 (right) we present these values for the northern and southern areas of both data sets. During both seasons the northern area experiences little net water-mass and nutrient exchange between the boundary region and the deep ocean. The southern area exports substantial amounts of watermass and nitrate during spring, with maximum values of 1.7 Sv for water-mass and 27 kmol s-1 just south of Cape Blanc (20.5◦N). During fall the southern area shows a rapid latitudinal transition from substantial export at 20.5◦N (0.7 Sv for water-mass and 7 kmol s-1) to import at 18.5◦N (0.4 Sv for water-mass and 5 kmol 54 Chapter 2. Water and nutrient fluxes s-1), and the net exchange is, once again, quite small. Despite we lack good station coverage for the central area, we may obtain the watermass and nutrient convergence/divergence for each area by simply considering the along-shore transports in four cross-shore sections, those located at the latitudinal ends of the three areas (18◦N, 21◦N, 23.5◦N and 26◦N). For sections flanked to the east by the African coastline, as those in the northern and (to a lesser degree) central areas, along-shore convergence/divergence traduces in cross-shore export/import (across section L). In the southern area, however, our cross-shore sections do not quite reach the slope so there may also be exchange with the continental slope and shelf. During spring the northern area has small convergence/divergence values while the central area is characterized by moderate water and nutrient divergence (Table 2.2), or import from the deep ocean to the coastal transition zone (0.8 Sv and 14 kmol s-1). The southern area, on the other hand is characterized by very intense water and nitrate convergence, 2.8 Sv and 53 kmol s-1, or export to the deep ocean. During fall there is again little along-shore divergence in the northern area. The central area, however, has moderate water-mass and nutrient convergence. As in spring, the southern area is characterized by export to the deep ocean, but these values are one order of magnitude smaller (Table 2.2). 2.6.3 Closing the balances Cross-shore exchange may be accommodated through both the geostrophic and Ekman contributions. The geostrophic cross-shore water-mass and nitrate transports are calculated in the northern and southern areas by integrating the geostrophic 55 Chapter 2. Water and nutrient fluxes Table 2.2: Water (Sv) and nitrate (kmol s−1) transports, integrated for each of the three areas, for both the spring and fall cruises Water Mass Nitrate Spring/Fall Spring/Fall North GT 0.52 / 0.44 2.7 / 2.9 C/D-0.09 / 0.17 0.5 / 1.9 Central C/D0.77 / -0.17 13.5 / -2.6 South GTL1.15 / -0.22 19.2 / -7.9 GTO-1.63 / -1.10 -31.1 / -12.5 GTO−GTL-2.78 / -0.88 -50.3 / -4.5 C/D-2.84 / -0.56 -52.8 / -3.0 GT stands for cross-shore geostrophic transport (negative offshore) through the alongshore sections. C/D stands for convergence/divergence (negative/positive), that result in cross-shore transport (negative offshore), calculated as the difference in geostrophic along-shore transport between adjacent cross-shore sections. South of Cape Blanc we present the geostrophic transport across both sections L (GTL) and O (GTO). velocity and nitrate fluxes through the meridional sections delimiting these areas: section Lnin the northern area and sections Lsand O in the southern area (the latter is an open-ocean along-shore section, roughly parallel to section Ls, that stretches some 350 km from 18 to 21◦N, Fig. 2.7 ). Table 2.2 compares these vertically and spatially integrated geostrophic cross-shore exchanges, with the above along-shore convergence/divergence values. The agreement is reasonably good, with differences in volume and nitrate transports being typically about 0.5 Sv and 2 kmol s-1. This gives support to our choice of reference level, and grants confidence to our estimates of import-export in the central area, calculated solely from the along-shore convergence/divergence. The observed differences between along-shore convergence/divergence and crossshore geostrophic transport in the northern and southern areas may par56 Chapter 2. Water and nutrient fluxes Table 2.3: Mean (and standard deviation) nitrate concentration (mmol m−3) in the upper 50 m of the water column, averaged per area, during spring and fall Spring Fall North 0.47 (±0.43) 0.81 (±1.01) Central 7.25 (±2.29) 7.20 (±4.23) South 3.82 (±2.99) 2.55 (±2.74) tially be justified because of the Ekman contribution. We do not have actual wind values for the region during the cruises but, as a reference, we may use the mean monthly Ekman transports per unit along-shore distance reported by Nykjaer and Van Camp (1994). As most Ekman transport takes place in the surface mixedlayer we calculate the associated nitrate transport multiplying water transport by mean nitrate concentration of the surface mixed-layer. For this calculation we use, as a gross estimate, the mean concentration values in the uppermost 50 m for each area (Table 2.3). The wind-induced transport is indeed of the order of the observed differences in water transport (Table 2.4 ). This transport is greater in spring, the spring-fall difference being greatest in the central and southern areas. During spring the water-mass Ekman transport in the southern area is much smaller than in the northern area, but the nutrient Ekman transport is almost the same because of the presence of nutrient-rich surface waters. Given the approximations involved (shallow reference level, sections of limited length, and mean rather than synoptic winds) the water-mass balance closes reasonably. Note, for example, that an error in reference velocity of 0.01 m s-1 for a section 100 km long and 300 m deep corresponds to 0.3 Sv. Further, the wind-induced transport has a quadratic dependence on the surface winds so that 57 Chapter 2. Water and nutrient fluxes Table 2.4: Estimates of Ekman water (Sv) and nitrate (kmol s−1) transports integrated for each area, for both the spring and fall cruises (sign convention as in Table 2.2) Water Mass Nitrate Spring/Fall Spring/Fall North CD −GT -0.61 /-0.27 -2.2 / -1.0 mean wind -0.32 / -0.15 -0.1 / -0.1 Central mean wind -0.51 / -0.24 -3.7 / -1.7 South C/D−GTO+GTL-0.06 / 0.32 -2.4 / 1.5 mean wind -0.43 / -0.09 -1.6 / -0.2 These values are obtained both as the difference between along-shore convergence/divergence and cross-shore geostrophic transports, C/D - GT, or from the spring and fall mean Ekman transports per unit length in the region, mean wind (Nykjaer and Van Camp, 1994). In the latter case nitrate transport is obtained multiplying the water transport between adjacent stations by the mean nitrate concentration in the upper 50 m of the water column and integrating along-shore. doubling the wind causes the transport to quadruple, i.e. the use of a standard formula (Wu, 1980) in our region shows that an increase in the wind speed from 4 to 8 m s-1 causes an increase in transport from 0.6 to 2.4 Sv. On the other hand, nitrate shows some greater unbalances possibly as a result of nutrient utilization within the same upwelling region. This is also reflected in the standard deviations in top 50 m in Table 2.3. The fair agreement is due to several conditions that are specific for our study region. First, the coastal upwelling jet is the predominant velocity signal in the region, which does not extend vertically beyond some 200 m. Thus, the crossshore vertical upwelling cell that connects the interior ocean and the coastal-jet zone does not reach greater depths. Second, this coastal jet does not extend offshore beyond the internal radius of deformation (of about 100 km, equal or less 58 Chapter 2. Water and nutrient fluxes than our normal-to-shore stations), so that it is justified to speak about a Canary Upwelling Current (Pelegrí et al., 2005, 2006). And third, the Ekman contribution is relatively small as compared with along-shore convergence/divergence, although it may be important in the nutrient balance of the southern area as a result of the high surface-nutrient concentrations therein. 2.7 Conclusions We have combined data from four hydrographic cruises during the 1970s in order to produce two seasonal data sets (spring 1973 and fall 1975) with meridional coverage off Northwest Africa from 17 to 26◦N. The region is divided in three areas: southern (18-21◦N), central (21-23.5◦N), and northern (23.5-26◦N). The data sets have been used with three main purposes: to describe the hydrography of the zone, to examine mixing processes at the frontal system, and to estimate water mass and nutrient export/import between the deep ocean and the coastal transition zone. The whole region is characterized by shallow (down to no more than 200 m) coastal upwelling and by the presence of a rather abrupt transition from North Atlantic to South Atlantic Central Waters (NACW to SACW) in the Cape Verde frontal system. The frontal position, defined as the intersection of the S = 36.0 and σθ= 26.5 surfaces, is located roughly off Cape Blanc but moves north/south during fall/spring. The analysis of two sequences of SST images suggests that frontal interleaving may partly originate as mesoscale features in the coastal upwelling front, which converge off Cape Blanc and become transported westwards. 59 Chapter 2. Water and nutrient fluxes Property-property diagrams with stations nearby the Cape Verde frontal system illustrate that it effectively behaves as a barrier to all properties but heat. We have examined these diagrams for four properties (potential temperature, salinity, nitrate, and dissolved oxygen) and have found that in all diagrams (except those using temperature) just a few data points have properties intermediate between NACW and SACW, instead most points remain grouped around the original water masses. The temperature is a striking exception, as its distribution in vertical sections and in the θ-S diagrams appears as if the diffusion coefficient for heat was much greater than for other properties. The spatial distribution of the Turner angle shows that the frontal system is prone to the existence of double-diffusion in the form of salt fingering. We propose that doublediffusion is responsible for enhanced horizontal heat diffusion that results in a relatively smooth distribution of temperature across the front, as compared with other properties. The density field is used to estimate the near-surface geostrophic flow patterns in the region. The reference depth used for these calculations is 300 m, which a simple sensitivity analysis suggests can capture the major elements of the near-surface flow. During both seasons the flow between Cape Blanc and Cape Bojador is cyclonic. West and south of Cape Blanc the flow pattern changes greatly between seasons. West of Cape Blanc the flow is weak and directed southsouthwest during spring and alternates in direction during fall, while between Cape Blanc and Cape Verde the spring flow is along-shore to the north and weakens in fall. Nine cross-shore sections in the coastal transition zone are used to estimate the along-shore water and nutrient transports. From these values we then calculate 60 Chapter 2. Water and nutrient fluxes the along-shore convergence/divergence or, equivalently, the exchange between the upwelling region and the deep ocean in three areas (northern; 23.5−26◦N; central: 21−23.5◦N; southern: 18−21◦N). These values are finally explained in terms of cross-shore geostrophic and wind-induced contributions, the former calculated using along-shore sections and the latter estimated with monthly meanwinds. The overall picture is that presented by the geostrophic fields in Figure 2.14, modified by surface Ekman transport associated to the seasonal varying pattern of northeasterly winds. The northern area, and the central area during fall, experience geostrophic import and wind-induced export that add up to small net exchange with the deep ocean, each no more than 0.5 Sv and 3 kmol s-1. The central area during spring imports water and nutrients, as geostrophic inflow dominates over wind-induced outflow, of about 0.8 Sv and 14 kmol s-1. The southern area exports 2.9 Sv and 53 kmol s-1 to the deep ocean during spring, and 0.6 Sv and 3 kmol s-1 to the deep ocean during fall. 61 Chapter 3 Meridional changes in water mass distributions∗ Abstract An Optimum Multiparameter Analysis is applied to a data set in the eastern boundary of the North Atlantic subtropical gyre, gathered during the month of November of two consecutive years and spanning from 16 to 36◦N. This data set covers over 20◦of latitude with good meridional and zonal resolution over the whole coastal transition zone. The contribution from six water types in the depth range between 100 and 2000 m is solved. In the 100 to 700 m depth range the central waters of southern and northern origin meet abruptly at the Cape Verde Frontal Zone. This front traditionally has been reported to stretch from Cape Blanc, at about 21.5◦N, until the Cape Verde Islands, but in our case it penetrates as far as 24◦N over the continental slope. South of this latitude we actually find a less saline and more oxygenated variety of southern central water, which ∗submitted to Ciencias Marinas as M.V. Pastor, J. Peña-Izquierdo, J.L. Pelegrí, and A. MarreroDíaz. Meridional changes in water properties off NW Africa during November 2007/2008. 63 Chapter 3. Meridional changes in water properties Latitude [ºN] Distance [km] Depth [m] Depth [m] Figure 3.2: Distribution of salinity in color scale and dissolved oxygen as black isolines along sections (a) Meridional, (b) Front, (c) South, and (d) North. White isolines represent those isopycnals, σθ= 26.46 and 27.14, delimiting the central water mass layer. 70 Chapter 3. Meridional changes in water properties Depth [m] Depth [m] Latitude [ºN] Distance [km] Figure 3.3: Distribution of phosphates in color scale and silicates as black isolines along sections (a) Meridional, (b) Front, (c) South, and (d) North. White isolines represent those isopycnals, σθ= 26.46 and 27.14, delimiting the central water mass layer. 71 Chapter 3. Meridional changes in water properties with AAIW finding its northernmost expansion in late fall and MW stretching to the south during winter (Machín and Pelegrí, 2009; Machín et al., 2010). Further deep we find waters of northern origin, here generically grouped under the North Atlantic Deep Waters (NADW) denomination. The top layer, from the surface down to the potential density level σθ= 26.46 or approximately the upper 100 m of the water column, displays highly variable θ-S characteristics as the result of coastal upwelling and also due to the presence of mesoscalar variability in the coastal transition zone. The mesoscale features are themselves the result of instabilities in the coastal upwelling jet (e.g. Pelegrí et al., 2005; Pastor et al., 2008) as well as the disturbance of the CC flow by the islands, prominently the Canary archipelago (Sangrà et al., 2005, 2007, 2009; Machín et al., 2006). In general this surface layer has relatively high salinity and dissolved oxygen values and is depleted in nutrients (Figs. 3.2 and 3.3 ). A particular feature of the eastern North Atlantic subtropical gyre is the subsurface salinity maximum between 50 and 100 m (Figure 3.4), which is formed due to an excess of evaporation over precipitation. It sinks down to the corresponding density levels due to Ekman transport and winter vertical convection. Throughout spring and summer there is surface capping of this salinity anomaly, which then spreads horizontally over long distances (Bauer and Siedler, 1988). Immediately below and down to some 600 m (26.46 <σθ<27.14) we find the two dominant central waters of our domain, NACW and SACW. The confluence of NACW and SACW occurs at the CVFZ, which is located in the North Atlantic because of the year-long northern position of the Intertropical Convergence Zone (ITCZ, or the Earth’s thermal equator). The Cape Verde frontal system is characterized by an abrupt transition in temperature, salinity, dissolved inorganic nu72 Chapter 3. Meridional changes in water properties 35 35.5 36 36.5 37 2000 1500 1000 500 0 Salinity [psu] Depth [m] 5 10 15 20 25 θ [ºC] 0 1 2 3 4 5 Oxygen [ml/l] 0 0.5 1 1.5 2 2000 1500 1000 500 0 PO4 [μmol/l] Depth [m] 0 5 10 15 20 25 SiO4 [μmol/l] x E5 NACW, MW E41 Interleaving E54 SACW*, AAIW E70 SACW, AAIW Figure 3.4: Vertical distribution of potential temperature θ, salinity S, phosphate PO4, Silicate SiO4, and dissolved oxygen O2, at selected stations displaying different water characteristics. The dots and crosses not only serve to identify the different stations, they also indicate the sampling depths for inorganic nutrients. The location of stations E5, E41, E54 and E70 is shown in Figure 3.1. 73 Chapter 3. Meridional changes in water properties trients and dissolved oxygen. These properties are substantially different as a result of their different origin, the former being relatively young waters (therefore well oxygenated and nutrient poor) formed at the northern edge of the relatively saline and warm North Atlantic subtropical basin while the latter being much older waters of a more remote origin (the subtropical convergence of the South Atlantic). The CVFZ has been said to stretch southwest from near Cape Blanc towards the Cape Verde Islands (see inset in Fig. 3.1, Zenk et al., 1991; Pastor et al., 2008). In our observations, however, the front appears to begin significantly north of Cape Blanc, at about 24◦N (Figs. 3.2a, 3.3a). Along the CVFZ there is interleaving, or intrusions of NACW and SACW at different longitudes and depths (Zenk et al., 1991), favored by the density-compensating character of the temperature and salinity fields. This interleaving is clearly observed in Figures 3.2b and 3.3b, where high salinity-oxygen and low nutrient concentrations, characteristic of NACW, alternate with low salinity-oxygen and high nutrient SACW concentrations. A remarkable feature of the SACW are the relatively low dissolved oxygen values, which respond to the high primary production in the surface layers (after upwelling in coastal waters and within the GD) combined with the long recirculation times in this eastern basin, precisely around the GD. This is sharp clear in the vertical profiles of those stations within (E41) and south (E54 and E70) the CVFZ, where we find that SACW are characterized by a minimum in oxygen between about 100 and 500 m. At these depths nutrients are high, although their maximum corresponds to intermediate waters (see below). An interesting feature is apparent in station E54, located over the slope along transect South. Down to 74 Chapter 3. Meridional changes in water properties about 400 m this station shows relatively high dissolved oxygen concentrations and it is substantially less saline than the surrounding SACW, suggesting a less diluted southern variety. We will return to this issue in the following sections. Below the central stratum we find the intermediate layers, here again formed by waters of southern (AAIW) and northern (MW) origin. In general, those stations near the Strait of Gibraltar show a pronounced salinity maximum between about 1000 and 1500 m, associated to the presence of MW (Fig. 3.2a,d). The high salinity, low nutrient values observed 200 km offshore in the northern section (Figs. 3.2d and 3.3d) are indicative of a preferential offshore path for MW. The transition in this stratum appears to be more progressive than in the overlaying central waters, although there are instances of isolated highly-saline Mediterranean lenses in the northern end of the domain. Such an instance was found in one of the offshore stations during the CANOA07 cruise, just southwest of section North (not shown). The 1200 m salinity maximum (S = 36.5) in station E5 is the clear signature of MW in the northern stations (Fig. 3.4), yet not approaching the S = 38.4 Mediterranean outflow salinity values. As we progress south, the intermediate salinity maximum gets eroded and instead the AAIW characteristics dominate. As in the central stratum, those waters of southern origin are relatively fresh, cool, and nutrient/oxygen rich/poor as compared with those of northern origin. This again reflects the relatively long residence time of the southern waters combined with the influence of the warm and salty Mediterranean outflow. AAIW appears to propagate north close to the slope, reaching at least until the Canary Islands, and further nrth in some instances (Machín and Pelegrí, 2009; Machín et al., 2010). The low oxygen and high nutrient values in the southernmost stations are likely a result of the long 75 Chapter 3. Meridional changes in water properties path of these waters before reaching the eastern North Atlantic subtropical gyre. However, it is remarkable that the absolute oxygen minimum occurs within central waters (200 to 400 m) while the absolute nutrient maximum is found much deeper (800 to 1200 m). The different vertical extension of the low-oxygen and high-nutrient layers is an imprint of the high oxygen concentration of this waters at origin, i.e. despite their long circulation time and intense remineralization they retain relatively high oxygen values. Within the deeper layers we find NADW at all latitudes, with meridional changes substantially smaller than those within the intermediate and upper layers. Nevertheless, Figs. 3.2a and 3.3a suggest that the Canary Islands behave as an obstacle to the propagation of these deep waters, as there are significant changes in salinity, nutrients and oxygen at depth across the archipielago. 3.4 Water types and optimum multiparameter analysis Optimum multiparameter (OMP) analysis is a tool to analyse the water mass mixture in a water sample. The method calculates the contributions from the original water masses, called source water masses or water types, to the water sample. The water type contributions to each data point are obtained by finding the best linear mixing combination in a multi-parameter space (e.g. temperature, salinity, oxygen and inorganic nutrients) that leads to the observed values. This is done by minimizing the residuals between predictions and observations, in a non-negative least-squares sense (Mackas et al., 1987; Tomczak and Large, 1989). 76 Chapter 3. Meridional changes in water properties The solution of the OMP analysis includes two physical constraints: the contributions from all sources add up to one (mass conservation), and all contributions must be non-negative. In this work we use temperature (T), salinity (S), phosphate (PO4), silicate (SiO4) and oxygen (O2) to resolve the following linear system of mixing equations: ∑ i xiθi=θobs +Rθ(3.1) ∑ i xiSi=Sobs +RS(3.2) ∑ i xi(PO4)i= (PO4)obs +RPO4(3.3) ∑ i xi(SiO4)i= (SiO4)obs +RSiO4(3.4) ∑ i xi(O2)i= (O2)obs +RO2(3.5) ∑ i xi=1 (3.6) where θi,Si,(PO4)i,(SiO4)i,(O2)iare the values for each source water mass and θobs,Sobs,(PO4)obs,(SiO4)obs,(O2)obs are the observed values, the last equation being for mass conservation. Before resolving the system, the water type matrix is normalized to commensurate the different variables, and different weights are applied to each variable. The weights are calculated following Tomczak and Large (1989), Wj=σ2 j δjmax (3.7) 77 Chapter 3. Meridional changes in water properties Table 3.1: Source water mass values used in the OMP analysis for potential temperature θ(◦C), salinity S, phosphate PO4(µmol/l), Silicate SiO4(µmol/l), and dissolved oxygen O2(ml/l) θS PO4SiO4O2 NACWU18.65 36.76 0.25 0.36 4.79 NACWL11.00 35.47 1.05 5.65 4.26 SACWU15.25 35.70 1.41 6.92 1.51 SACWL9.70 35.18 1.94 14.08 1.64 SACW∗12.08 35.27 1.62 9.31 1.21 AAIW 6.50 34.90 2.02 22.55 2.73 MW 11.74 36.50 0.67 7.20 4.42 NADW 2.50 34.94 1.40 34.80 5.71 Weight 1 0.86 0.21 0.14 0.15 where σjis the standard deviation of the water type matrix for variable j, a measure for the ability of variable jto resolve differences in water mass content; and δjmax is a measure of the environmental variability of the variable jthat characterizes the water type, here estimated as the largest variance in variable jfor the definition of any water type. Weights are then normalized to temperature, i.e. a weight of one is given to temperature and values less than one to the other variables, and a weight of 10 is assigned to the last equation in order to emphasize mass conservation (Table 3.1). An essential component of OMP is the definition of the water type matrix. We follow the approach of defining values in the vicinity of the study region rather than at the remote areas of water mass formation. In this way we minimize remineralization effects on inorganic nutrients and dissolved oxygen, so that we may assume that phosphates, silicates and oxygen are approximately independent and conservative. The values used are given in Table 3.1. The OMP 78 Chapter 3. Meridional changes in water properties analysis is applied to data in the density range 26.46 <σθ<27.82. Data with density less than the upper limit have been excluded as their properties may be altered by atmospheric and biogeochemical processes. Six source water masses may be discerned in the study area. The upper part of the water column, down to σθ= 27.14, is dominated by the central water masses of northern (NACW) and southern origin (SACW). Here we have used those θS characteristics defined by Tomczak (1981) using hydrographic data from the region 20◦to 26◦N. Due to their formation process, the θ-S relationship for central waters is defined by a straight line (Mamayev, 1975). Therefore, in order to characterize a central water mass in the density range 26.46 −27.14 we require two source water types, i.e. for each water mass we need upper and lower end members (NACWU, NACWL, SACWU, SACWL). An additional water type of southern origin is detected in the southernmost part of the study region, on stations along the continental slope above 200 m depth, in waters having a relative salinity and temperature minimum and oxygen maximum. This variety, here named SACW∗, has been previously identified as a regional SACW variety from the tropical region (Fraga and Manríquez, 1974; Voituriez and Chuchla, 1978; Manriquez and Fraga, 1982). The θ-S characteristics for SACW∗have been defined using those southern stations that display a salinity minimum in the central water mass layer. Two water masses are found at intermediate layers: MW and AAIW. The θS values for MW have been derived from the WOCE hydrographic climatology (Gouretski and Koltermann, 2004), by searching the temperature and salinity values which correspond to the salinity maximum in an intermediate layer within a region bounded by latitudes 12 and 49◦N and from the coast to 1500 km offshore. 79 Chapter 3. Meridional changes in water properties least as far north as 20◦N in summer and weakening and only reaching 17◦N during winter (Peterson and Stramma, 1991; Siedler et al., 1992; Lázaro et al., 2005). Here we have used the diluted regional variety, SACW, to define those waters south of the front (Tomczak, 1981, 1984; Klein and Tomczak, 1994). In this manner the front shows up sharply in the changes of water mass composition, from northern to southern characteristics. If we had chosen this second variety, SACW∗, to define the dominant water masses in this region (e.g. Manriquez and Fraga, 1982) then the front would have appeared as a less abrupt contrast between northern and southern water percentages. Our work has also shown the complexity of the coastal transition zone, particularly over the continental slope. During our measurements there was significant coastal upwelling as far south as 16◦N but the water mass distribution did not show a southward penetration of NACW through a coastal upwelling jet. Instead, the CVFZ was mainly located north of an imaginary line between Cape Blanc and the Cape Verde Islands, particularly along the continental slope where it stretched as far as 24◦N. This is likely the result of the penetration of southern waters as a result of the PUC, which appears to be an important phenomenon in the area. Actually, it is possible that the PUC accommodates the SACW∗variety and transfers it along the slope beyond the CVFZ. Below 600 m and as deep as 1500 m we find two intermediate water masses, again of northern (MW) and southern (AAIW) origin. The transition between MW and AAIW in the intermediate levels is much less abrupt than in the central water layer and occurs north of the Canary Islands. In our observations MW are centered near 1200 m, slightly deeper than AAIW, which are centered at depths of about 1000 m. The observed large meridional penetration of AAIW 86 Chapter 3. Meridional changes in water properties takes place because our cruises were done in November: Machín and Pelegrí (2009) and Machín et al. (2010) have reported a strong seasonal signal in the northward penetration of AAIW, which surpasses the Canary Archipelago and finds its northernmost extension during late fall. North of the Canary Islands there are Mediterranean Eddies (Meddies, we actually found one in the CANOA07 cruise), which closely correspond to the undiluted MW type. Below somewhere between 1300 m, in the southernmost stations, and 1700 m, in the northernmost stations, we find the 50% NADW contour, which defines the predominance of this deep water mass. This contour deepens smoothly with latitude except at the location of the Canary Islands where it suddenly plunges some 100 m, suggesting that the deep waters reaching both sides of the archipelago have followed significantly different paths. 87 Chapter 4 Assessing physical drivers of interannual chlorophyll variability∗ Abstract Interannual chlorophyll variability and its driving mechanisms are evaluated in the eastern subtropical North Atlantic, from 10 to 24◦N, and from the coastline to 40◦W. Nine years of SeaWiFS data are compared with the output from an ocean general circulation model coupled to a state of the art biogeochemistry model. The model’s skill at simulating seasonal and interannual chlorophyll variability in the study region during the SeaWiFS era is established. We then assess the drivers of chlorophyll variability in the model during the last half of the twentieth century by looking at changes in nutrient supply. A weak positive correlation between chlorophyll concentrations and stratification in the sunlit layer (top 80 m) is found (r=0.13). To evaluate the driving mechanisms, a nutrient budget is calculated in the study region within the euphotic layer. We find that ∗in preparation as M.V. Pastor, J.B. Palter, J.L. Pelegrí and J.P. Dunne, 2011. Beyond stratification: assessing physical drivers of interannual chlorophyll variability in the eastern subtropical North Atlantic. 89 Chapter 4. Physical drivers of interannual chlorophyll variability diffusive mixing of nutrients to the euphotic zone only explains 9% of the variability in chlorophyll concentrations, while advective nutrient fluxes exert a more important control on chlorophyll variability. In particular, vertical advection has the strongest correlation and has also the largest fluxes and anomalies. Two processes are primarily responsible for the upward flux of nutrients to the euphotic layer that sustain a chlorophyll response. One is coastal upwelling, determined by the alongshore component of the wind stress. It occurs in a narrow meridional band (about 100 km wide) adjacent to coast, but further influences the offshore domain through horizontal Ekman transport and convergence of the meridional flow. Offshore upwelling driven by positive wind-stress curl during fall provides the second nutrient source. The curl-driven vertical velocities are smaller than the coastal upwelling but extend over a broader area. We find that variability in the position of the boundary separating the upwelling and downwelling domains, mainly during winter months, is critically important in setting the offshore extension of the high chlorophyll region in the eastern subtropical North Atlantic. 4.1 Introduction The eastern subtropical North Atlantic accommodates one of the most important eastern boundary current ecosystems in terms of total annual primary production (Carr, 2002; Carr and Kearns, 2003). Elevated primary production sustained by nutrient-rich subsurface waters upwelled in eastern boundary regions support high fish catches. Altogether, the four major coastal upwelling ecosystems account for 20% of global marine fish catch despite occupying about 1% of the global ocean (FAO, 2009). Understanding how productivity in these areas may vary in a changing and variable climate has important ecological and socioeco90 Chapter 4. Physical drivers of interannual chlorophyll variability a 40 30 20 10 -60 -40 -20 40 30 20 10 40 30 20 10 -60 -40 -20 -0.5 0 0.5 Chlorophyll difference (mg m ) -3 a c b d e Satellite Model winter - summer Feb 2000 - Feb 1998(1959_1968) - (1977_1986) Figure 4.1: Maps illustrating chlorophyll (mg/m3) variability at three timescales. Seasonal variability (a,b) is shown as winter minus summer chlorophyll concentrations. Interannual variability (c,d) is shown as February 2000 (February with maximum satellite averaged chlorophyll) minus February 1998 (February with minimum satellite averaged chlorophyll). Interdecadal variability (e) is shown as chlorophyll averaged during the decade 1959 to 1968 minus the average from 1977 to 1986. Left panels correspond to SeaWiFS data, right to model output. Dark grey contour shows the 0.2 mg/m3isoline of chlorophyll for winter, February 2000 and the period 1959 to 1968. Light grey contour shows the same isoline for summer, February 1998 and the period 1977 to 1986. 91 Chapter 4. Physical drivers of interannual chlorophyll variability nomic implications given the expected increase in fish demand (Delgado et al., 2003). The eastern subtropical North Atlantic is the most spatial and temporally variable of the four major eastern boundary current ecosystems (Carr, 2002). The basin’s eastern limb shows one of the strongest zonal gradients of chlorophyll in the world’s ocean (e.g. Fig. 2 in McClain, 2009). The large-scale wind-stress field drives downwelling in the subtropical gyre, which results in a deep pycnocline and nutricline, and consequent low surface chlorophyll (<0.07 mg/m3). Along the basin’s eastern margin, the wind forcing causes divergence of the horizontal ocean currents as a result of both Ekman transport along the coast and offshore curl-driven Ekman pumping (McClain and Firestone, 1993). The upward velocities induced by this coastal Ekman transport and Ekman pumping bring nutrient-rich waters to the euphotic layer, through a combination of alongisopycnal and diapycnal transfer, sustaining high chlorophyll concentrations at the basin’s eastern margin (Pelegrí et al., 2006). In a warming environment, conditions in the subtropical gyre and in the coastal upwelling region may evolve differently. Coastal upwelling rates have been projected to increase due to an increased land-sea pressure gradient (Bakun, 1990; Bakun et al., 2010). On the other hand, increased stratification in oligotrophic gyres (a possible consequence of warming) is thought to reduce upward mixing of nutrients and decrease photosynthesis (McClain et al., 2004; Gregg et al., 2005; Behrenfeld et al., 2006; Polovina et al., 2008). However, the link between declines in surface chlorophyll and increased stratification has most often been inferred from correlations between chlorophyll and sea surface temperatures, leaving the underlying physical mechanisms unexplored. Furthermore, a debate exists about the recently observed 92 Chapter 4. Physical drivers of interannual chlorophyll variability variability being a climate-change induced trend (Polovina et al., 2008; Irwin and Oliver, 2009) or part of a multidecadal oscillation (Martinez et al., 2009). A recent study of satellite ocean color data and output from three biogeochemical models suggested distinguishing between climate-change driven trends and natural variability will require about 40 years of continuous satellite chlorophyll measurements (Henson et al., 2010). Surface chlorophyll in the eastern subtropical North Atlantic has a marked seasonal cycle; yet, interannual variability can be as large or larger than the seasonal changes (Fig. 4.1). In this paper, we use 49 years of output from an ocean general circulation model, coupled to a state of the art biogeochemistry model, to explore the physical mechanisms that drive interannual chlorophyll variability in this region. The use of models allows in-depth study of several hypothesized physical drivers of chlorophyll variability such as changes in stratification and the large scale wind field. The model’s skill at simulating chlorophyll spatial and temporal variability in the study region is first tested by comparing its output to nine years of Sea-viewing Wide Field-of-View Sensor (SeaWiFS) data (section 4.3) and satellite measurements of sea surface height (section 4.4). We examine links between chlorophyll and sea surface height to explore the premise that sea surface height gives an insight to possible mechanisms explaining chlorophyll variability. Finally, we asses the drivers of chlorophyll variability in the model output during the last half of the twentieth century. Specifically, we test various hypothesis in section 4.5 through the assessment of each term in a nutrient budget. The conclusions are presented in section 5, including a discussion about the role of subtropical gyre boundary shifts. 93 Chapter 4. Physical drivers of interannual chlorophyll variability 4.2 Satellite data, model output and methods We use Level 3 SeaWiFS monthly chlorophyll downloaded from oceancolor. gsfc.nasa.gov at 9 km resolution for the period November 1997 to December 2007. Absolute dynamic topography (SSH), produced by Ssalto/Duacs and distributed by Aviso, with support from Cnes, is downloaded from http://www. aviso.oceanobs.com/duacs/. It corresponds to merged data from Topex/ Poseidon and ERS satellites. We averaged the original weekly data onto monthly, then smoothed the 1/3◦spatial resolution with a 3x3 (1◦x 1◦) boxcar filter to reduce mesoscale features. The general ocean circulation model used is Version 4 of the Geophysical Fluid Dynamics Laboratory’s Modular Ocean Model (MOM4) (Griffies et al., 2008), forced with the Common Ocean-Ice Reference Experiment (CORE) data set (Griffies et al., 2009; Large and Yeager, 2009). We use version 2 of the CORE reanalysis effort, which includes six-hourly interannual varying meteorological fields for the period 1958 - 2006 (10 m air temperature, humidity, air density, zonal wind, meridional wind, and sea level pressure). Daily varying shortwave and longwave radiative fluxes are available from 1983 and monthly varying precipitation since 1979. Prior to those years, data are filled with the climatological annual cycle. Continental runoff is available as a climatological annual mean. The ocean model has fifty levels in the vertical direction, a longitudinal resolution of 1◦and a latitudinal resolution varying between 1◦in the extratropics and 1/3◦on the equator. This simulation uses the Boussinesq approximation and therefore steric effects do not influence the sea level height. Steric effects include the expansion or contraction of the water column due to temperature or salinity changes. MOM4 94 Chapter 4. Physical drivers of interannual chlorophyll variability has the ability to respond to atmospheric loading. However, for various reasons, most coupled models do not actually apply the sea level pressure (SLP) on the ocean. One reason is that interpolating SLP to the ocean can be fraught with error, especially next to land-sea boundaries where land has high elevation. Therefore, the model does not consider variations in SLP, so that all changes in SSH respond to the ocean internal dynamics. The simulation was initialized from hydrographic mean properties taken from the World Ocean Atlas 2001 (Conkright et al., 2002). It was spun-up for 348 years with forcing from a climatological year calculated using the mean CORE data during the years 1958-1977 before the final loop with the 49 years of CORE interannual variability was integrated. The biogeochemical component is given by the model Tracers for Ocean Phytoplankton with Allometric Zooplankton (TOPAZ), which simulates prognostically all major nutrient elements (N, P, Si and Fe). The ecosystem is based in three classes of phytoplankton. The small class dominates when growth rates are low; this size class resists sinking. Large phytoplankton represent diatoms and other phytoplankton that bloom and sink quickly. Finally, diazotrophs fix nitrogen directly. Phytoplankton growth rates are modeled as a function of variable chlorophyll to carbon ratios and colimited by nutrients and light. A more detailed description can be found in Appendix C. We use the distance from the coast to the 0.2 mg/m3isoline of chlorophyll (DCHL) to characterize the size of the high chlorophyll region between the oligotrophic subtropical gyre and the African coast. Larger distances indicate a larger area with chlorophyll higher than 0.2 mg/m3. Studies on chlorophyll variability in coastal upwelling systems generally have defined the areas of high productivity as those having >1 mg/m3(e.g. Nixon and Thomas, 2001; Carr, 2002; 95 Chapter 4. Physical drivers of interannual chlorophyll variability 2006), and using aerosol optical depth as a proxy for dust (from SeaWiFS data), we found that 1998 was the dustiest year during the SeaWiFS 1997 - 2007 period (not shown). In the model, wet and dry dust deposition fluxes are prescribed from the monthly climatology of Ginoux et al. (2001) and thus no interannual variability exists. This is an advantage for the purpose of this study, as it allows us to focus on chlorophyll variability caused by ocean dynamics. Even excluding the very high positive anomaly mentioned above, the width of the high chlorophyll region varies interannually by about 500 km for satellite and 600 km in the model output, comparable to the seasonal cycle of 780 km. Qualitatively, the anomalies show similar patterns; the high chl region is largest from 1999 to 2004 and decreases from 2005 to 2008. The anomalies of the high chlorophyll region averaged from 10 to 24◦N in the model and SeaWiFS are significantly correlated beyond the 1% level (Fig. 4.3). The shrinking of the high chlorophyll region in the latter years has been interpreted as a possible global warming related trend and correlated with a trend in sea surface temperature (Polovina et al., 2008). In the next section we use the model to place the variability over the SeaWiFS era in the context of several decades and asses the mechanisms driving this temporal variability. 4.4 Sea surface height and chlorophyll Satellite measurements provide global SSH data available during the SeaWiFS decade. We explore the relation between SSH and chlorophyll, as processes that may influence surface chlorophyll are likely reflected in SSH. We assume as a first approach that SSH variability reflects surface nutrient availability in the region, 102 Chapter 4. Physical drivers of interannual chlorophyll variability 98 99 00 01 02 03 04 05 06 07 10 12 14 16 18 20 22 24 −1500 −1000 −500 0 500 1000 1500 600 400 200 0 -200 -400 -600 98 99 00 01 02 03 04 05 06 07 10 12 14 16 18 20 22 24 −1500 −1000 −500 0 500 1000 1500 98 99 00 01 02 03 04 05 06 07 10 12 14 16 18 20 22 24 −1500 −1000 −500 0 500 1000 1500 D (km) CHL D (km) SSH Figure 4.4: Hovmöller plots of monthly anomalies for satellite-derived distance from the coast to (left) the 0.2 mg/m3isoline of chlorophyll and (right) the 37 cm SSH isoline, over the SeaWiFS data period (November 1997 to December 2007). because of the inverse relationship between SSH and thermocline depth (Stammer, 1997; Mayer et al., 2001), and the common coincidence of the thermocline depth and nutricline depth (Wilson and Coles, 2005; Signorini et al., 1999). The temporal correlation of chlorophyll and SSH anomalies in both the model and data substantiate this assumption (Fig. 4.2, bottom): when SSH is depressed, chlorophyll is high. Likewise, distances to the chlorophyll and SSH isolines are positively correlated (Fig. 4.4 and 4.5). A larger distance to the SSH isoline reflects lower SSH in the east, which is expected to be linked to a shallower thermocline and a shallower nutricline. Thus, when the distance to the SSH isoline is large we expect an increase in chlorophyll, and therefore a larger distance to the 0.2 mg/m3chlorophyll isoline. Using SSH as a proxy for thermocline depth essentially assumes that, for large scales, the surface ocean responds to its internal baroclinicity. This is equivalent to saying that, for the considered region, water mass is in balance. If the balance is divergent (convergent) the whole region 103 Chapter 4. Physical drivers of interannual chlorophyll variability 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 90 92 94 96 98 00 02 04 06 10 12 14 16 18 20 22 24 −1200 −600 0 600 1200 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 90 92 94 96 98 00 02 04 06 10 12 14 16 18 20 22 24 −1500 −1000 −500 0 500 1000 1500 D (km) CHL D (km) SSH Figure 4.5: Hovmöller plots of monthly anomalies for model-derived distance from the coast to (top) the 0.2 mg/m3isoline of chlorophyll and (bottom) the -18 cm SSH isoline, over the simulated period (1958 to 2006). 104 Chapter 4. Physical drivers of interannual chlorophyll variability would have a mean decrease (increase) in SSH. In line with these expectations, interannual variability in the width of the high chlorophyll region and the low SSH region show a remarkable agreement in the satellite observations for the period November 1998 to December 2007 (Fig. 4.4). Both indices show negative anomalies (SSH and chlorophyll isolines close to coast) in the first year of the time series and positive anomalies (enlargement of the high chlorophyll and low SSH region) in the following years, especially for SSH. Negative anomalies are larger in northern latitudes and progress southwards in the second half of the time series. The shift in the boundary between high and low chlorophyll waters is mirrored by a similar shift in the boundary between low and high SSH. The subtle shift in this boundary over the SeaWiFS era is related to the expansion of the oligotrophic gyre (Polovina et al., 2008). Placing the variability in the context of fifty years of reanalysis-forced model simulation, the recent trend appears to be embedded in a multidecadal oscillation (Fig. 4.5). As with satellite observations, there is a strong correlation between modeled chlorophyll and SSH distances (r=0.73, p<<0.01). South of 20◦N, the high chlorophyll region is highest from 1959 to 1976, coincident with positive anomalies of the distance to the -18 cm isoline of SSH. The chlorophyll and SSH time series are then followed by 23 years of consistent negative anomalies. From 1999 to 2004, chlorophyll and SSH show less agreement (Fig. 4.5). In these years, the width of the low SSH region is 1000 km above the mean between 13 and 18◦N, while the high chlorophyll region remains restricted to the coast. Low SSH is linked to the shoaling of the nutricline, but that relationship is weakest in the northern part of our domain. In 1999-2004 the anomalously large region of depressed SSH (high SSH distance) lies northward of its position in the 1959105 Chapter 4. Physical drivers of interannual chlorophyll variability 1964 period (Fig. 4.5). Because the correlation between nutricline depth and SSH is weaker in the northern part of the domain, the anomalously high SSH distances in the latter period do not correspond with an equally strong response in nutrient supply to the euphotic zone. Thus, the offshore extension of the high chlorophyll region responds more weakly to the enlargement of the northern portion of the low SSH region than an enlargement of the southern region. Two physical processes govern the time evolution of sea surface height in MOM4. One is convergence or divergence of horizontal currents, caused primarily by changes in the wind field. The second one is mass fluxes through the surface such as precipitation, evaporation, river runoff and ice melt. Again, no steric effects are included as the model conserves volume rather than mass. Therefore, the modeled changes in SSH respond predominantly to variability in the convergence or divergence of the horizontal velocities, as the precipitation minus evaporation and river run-off term has a low impact in the study area. 4.5 Mechanisms of chlorophyll variability The size of the high chlorophyll region, as defined by the distance between the coast and the 0.2 mg/m3chlorophyll isoline, is related to the chlorophyll concentration in the study region enclosed by the box in Fig. 4.2. In both model and data, the higher the chlorophyll concentration in the region, the greater the distance to the isoline (r=0.97 for model output, r=0.73 for satellite data, both p<<0.001). In the model, chlorophyll concentrations depend both on phytoplankton concentrations and on the chlorophyll to Carbon ratio (Chl:C) in their cells. Chl:C is a function of irradiance, iron (Fe) and growth rate in the model. 106 Chapter 4. Physical drivers of interannual chlorophyll variability In the study region, the seasonal variability of Chl:C is mostly governed by Fe input and growth (maximum values in winter, minimum in summer). Chl:C has also significant interannual variability which is positively correlated to chlorophyll concentrations (r=0.90). Thus, an increase in chlorophyll may indicate an increase in phytoplankton biomass and/or a higher Chl:C. TOPAZ uses the Geider et al. (1997) photoacclimation model which adjusts Chl:C to ambient conditions. When growing conditions are optimal, both growth rates (more biomass) and Chl:C (more chlorophyll) increase. Consequently, biomass and Chl:C are tightly related. Hereafter we will study changes in carbon biomass which do not depend on Chl:C to exclude any variability caused by changes in cellular Chl:C. Growth rates and biomass are modeled as a function of irradiance, nutrient availability and temperature. In our study region, irradiance and temperature are not dominant controls on biomass. Prior to 1983, the model irradiance is prescribed as a climatological annual cycle. From 1983 onwards, the model uses variability of incoming radiation, but we find no correlation to biomass. An increase in temperature would increase growth rates. In our study region, the average temperature in the top 80 m is negatively correlated to biomass; this correspondence is likely due to the relation between temperature and upwelling. Therefore, investigating the different paths of nutrient supply into this region and their relative importance will help elucidating mechanisms that cause biomass changes and variability in the size of the high chlorophyll region. 107 Chapter 4. Physical drivers of interannual chlorophyll variability 4.5.1 Nutrient supply In order to address interannual variability in nutrient supply, we consider each term in the PO4budget (Eq. 4.3) for the study region outlined in Fig. 4.2, above the average depth of the 1 Watt light layer (80 m). Studying these terms provides a mechanistic view of the physical controls on phytoplankton biomass. Because a number of previous studies have linked stratification variability with chlorophyll and biomass variability (McClain et al., 2004; Gregg et al., 2005; Behrenfeld et al., 2006; Polovina et al., 2008), we also compare our biomass time series with a common measure of stratification, the density difference between 200 m and the surface. However, this measure of stratification gives no valuable information in our region, as density variations at 200 m are small, and the metric thus reflects only density variability at the surface, predominantly driven by temperature changes. Higher temperatures at the surface are indeed correlated with lower biomass, but this gives little indication of the physical mechanisms responsible for the decrease. The density difference between the surface and the base of the euphotic zone (80 m) may be a more suitable measure of the stratification impacting the sunlit layer. The density difference between the surface and 80 m is actually positively correlated with biomass (r=0.18, p<<0.001) and surface chlorophyll concentrations (r=0.13, p<0.01), opposite to stratification exerting a leading control on the nutrient supply. Instead, this slight positive correlation suggests a distinct mechanism controlling phytoplankton variability that is not suppressed under increasing stratification, which we next search for among all the nutrient supply terms in the region. The PO4budget equation (Eq. 4.3) includes lateral advection, vertical advec108 Chapter 4. Physical drivers of interannual chlorophyll variability -5 0 5 -1000 0 1000 -5 0 5 -5 0 5 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 90 92 94 96 98 00 02 04 06 -5 0 5 D (km) ∂ zwPO x 10 4 (mol m ) -2 s-1 ∂ yvPO x 10 4 (mol m ) -2 s-1 mixing x 10 (mol m ) -2 s-1 a b c d 10 10 10 CHL biomass Biomass (mg C m ) -3 Figure 4.6: Time series of monthly anomalies averaged in the study region (latitudes 10 to 24◦N, from 40◦W to the coast, top 80 m depth). (a) Distance to the 0.2 mg/m3chlorophyll isoline, and phytoplankton biomass as the black line, (b) vertical PO4convergence, (c) meridional PO4convergence and (d) PO4input due to mixing. Anomalies of zonal PO4convergence are not shown as their magnitude is not notable as compared to meridional and vertical convergence anomalies. 109 Chapter 4. Physical drivers of interannual chlorophyll variability Figure 4.7: Correlation between (left) vertical PO4advection, wPO4, and vertical PO4gradient, ∂zPO4, and (right) vertical PO4advection, wPO4, and vertical velocity, w. tion and mixing. Anomalies of diffusive mixing of PO4only play a relatively important role in the months of February and March (Fig. 4.6d), when convection and vertical diffusion inject PO4into the northern part of this region. On an interannual timescale, anomalies of PO4diffusive mixing explain less than 4% of the biomass variability (or 6% of the changes in the distance to the 0.2 mg/m3chlorophyll isoline). Conversely, advection of PO4seems to be a key factor (correlation of 0.86 with biomass). Amongst vertical, meridional and zonal PO4convergence, we find that interannual variability of biomass is highest correlated to vertical convergence (r=0.82), although correlation to meridional convergence is also high (r=0.73). Interannual variability of the vertical PO4fluxes are caused by changes in vertical velocities rather than changes in the vertical PO4gradient, calculated as the PO4difference between 75 m and 85 m (Fig. 4.7). Most of the merid110 Chapter 4. Physical drivers of interannual chlorophyll variability -1 -0.5 0 0.5 1 wPO 4uPO 4vPO 4 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 90 92 94 96 98 00 02 04 06 -6 -4 -2 0 2 4 6 Transport (Sv) V coast Vcurl (mol m ) -2 s-1 advection *109 ∂ z∂ x∂ y b a Figure 4.8: Time series of monthly means averaged in the study region (latitudes 10 to 24◦N, from 40◦W to the coast, top 80 m depth). (a) Advective terms in the PO4budget (vertical ∂zwPO4, zonal ∂xuPO4, meridional ∂yvPO4), and (b) water volume upwelled due to offshore curl-driven upwelling (Vcurl) and coastal Ekman transport (Vcoast). ional transport of PO4into the study region takes place through the southern border (95% on a yearly average). This is consistent with southern waters having higher nutrient concentrations. Nevertheless, the magnitude of both the monthly means and the anomalies of vertical convergence are much larger than those of the meridional component (Fig. 4.6b and c, Fig. 4.8a), suggesting that upward advective fluxes are the most important supplier of nutrients into this region and the dominant control on phytoplankton biomass variability and the size of the high chlorophyll region. 111 Chapter 5 Conclusions and outlook 5.1 Conclusions How effective is the Cape Verde front as a barrier between the two central water masses? Central waters masses from northern and southern origin have different physico-chemical properties. NACW is warmer, more saline, and has less nutrients and more dissolved oxygen, as compared to SACW. The Cape Verde front represents the meeting of these two water masses in the eastern subtropical North Atlantic. The front has been described as a barrier between the two water masses where epipycnal and less frequently diapycnal mixing occurs (Tomczak, 1981; Zenk et al., 1991). Here we propose that double diffusive mixing enhances horizontal heat transfer resulting in the smoothing of temperature gradients across the front, as compared to other properties such as salt, nutrients or dissolved oxygen. 119 Chapter 5. Conclusions and outlook How does the coastal upwelling front interact with the Cape Verde frontal system? In addition to the Cape Verde front that separates central waters of northern and southern origin, a second front exists in the coastal region between newly upwelled waters, and the offshore waters. We propose that the interleaving frequently found in the Cape Verde front may originate as mesoscale features in the coastal upwelling front, which then converge off Cape Blanc and are transported westwards. Do water mass and nutrient exchanges between the coastal upwelling region and the open ocean display seasonal variability? Offshore transport along the NW African coast has its maximum near Cape Blanc, at about 21◦N. Convergence of the southward Canary Upwelling Current and the northward Mauritania Current during summer and fall is expected to produce intense offshore export. During winter and spring the along-slope flow convergence is expected to be substantially reduced. The data collected during spring 1973 and fall 1975 display large offshore transports off Cape Blanc but, contrary to expected, the maximum transport is found during spring in the area south of Cape Blanc. Does the position of the Cape Verde front vary in time? Traditionally, the Cape Verde front position has been defined by the location of the 36 isohaline at 150 m depth. Using this definition, the front is located at 21◦N during spring 1973, at 22.5◦N during fall 1975, and at 23.3◦N during fall 2008. A clearer picture of the front is given by the water mass percentages. Using 120 Chapter 5. Conclusions and outlook the 50% isoline of water mass content as the divider between waters of northern and souther origin, the front is seen to shift northwards with depth in fall 2008: its location is 24◦N in the top 250 m, and 25◦N below. This tilt may be caused by the southward NACW transport associated with the Canary Upwelling Current. What is the origin of the subsurface salinity minimum often found in the South Atlantic Central Water domain? The relative salinity minimum and oxygen maximum found at about 300 m within the SACW domain, and clearly visible in θS diagrams, made us wonder if this was the imprint of less diluted SACW of tropical origin. The distinction of an additional central waters type of southern origin in the OMP analysis has allowed to infer a propagation path for the less saline, oxygen rich variety (SACW*). This water mass could be advected northwards by the Poleward Undercurrent, while the traditional SACW would have already experienced some mixing with NACW. How is the along-slope latitudinal distribution of central and intermediate water masses off NW Africa? The Cape Verde frontal system represents a rapid transition between central waters of southern and northern origin. The less diluted SACW variety does not extend north from this front but some of the locally diluted is found north of the Cape Verde front along the slope. At intermediate levels the transition between Antarctic and Mediterranean waters is much smoother than in the central levels. Does the state-of-the-art numerical coupled MOM4/TOPAZ model properly reproduce the observed dynamics in the eastern subtropical North Atlantic? 121 Chapter 5. Conclusions and outlook The biogeochemistry model TOPAZ, coupled to the global ocean circulation model MOM4, has proven to reproduce the main dynamical features and the observed seasonal and interannual chlorophyll variability in the study region. Model output from a historical run forced with 49 years of interannually varying data is thus analyzed to investigate chlorophyll variability and its driving mechanisms in the eastern subtropical North Atlantic. What is the dominant nutrient transport amongst vertical advection, horizontal advection and diffusive mixing? An assessment of the nutrient budget in the euphotic layer (top 80 m) of the study region shows that vertical nutrient advection is the main driver of interannual chlorophyll variability, although northward transport at the southern boundary is also considerable. Diffusive mixing only explains 9% of the chlorophyll interannual variability. Is the variability in the vertical nutrient transport originated by changes in the velocity field or in the nutrient gradient? Interannual variability in the vertical nutrient transport at 80 m is dominated by changes in velocity, rather than changes in the nutrient reservoir. Which process plays a more decisive role in the upward nutrient transport, coastal upwelling or offshore upwelling? Two main processes bring nutrients to the euphotic layer in the eastern subtropical North Atlantic. One is the coastal upwelling; the second, Ekman pumping driven by wind stress curl. We have shown that offshore vertical velocities re122 Chapter 5. Conclusions and outlook lated to wind stress curl dominate over coastal velocities induced by along-shore winds in setting the area of chlorophyll concentration above 0.2 mg m-3. Do shifts in the southeast subtropical gyre boundary determine the area of high chlorophyll? The intensity of Ekman pumping in the eastern subtropical North Atlantic is the leading control on setting the interannual extension of the high chlorophyll region under study. Wind-driven shifts in the boundary between the upwelling (tropical gyre) and downwelling (subtropical gyre) domains are also a key factor that determine the extension of the high chlorophyll region. 5.2 Future research ∗The historical data analyzed in chapter 2 was digitalized from printed collections. This data will be made available in a public database for the scientific community use. ∗We would like to use an idealized model to represent the horizontal temperature smoothing across the Cape Verde front. For this purpose, we could use an exponential equation (modified from Rodríguez-Santana, 1997) to model the unequal property distributions across the front, and study its effect on property-property diagrams. In particular, this procedure should allow us to estimate the relative importance of isopycnal and diapycnal mixing leading to the water properties in the frontal region. 123 Chapter 5. Conclusions and outlook ∗Optimum Multiparameter analysis (OMP) has been used to infer mixing between water masses at the medium-scale of motion. The analysis assumes identical exchange coefficients for all properties (e.g. temperature, salinity, phosphate, dissolved oxygen). However, our results suggest that temperature diffuses faster than all other parameters in the Cape Verde frontal region. Future research should address the development of an OMP formulation that includes the effect of differential exchange coefficients. ∗The OMP technique establishes the contribution of the different water types from the distances between the data points in property-property spaces, implicitly assuming there are no differences between along-isopycnal or diapycnal mixing processes. These distances could be decomposed in alongisopycnal and diapycnal contributions, with relative weights in accordance to the size of the corresponding diffusive coefficients. ∗The nutrient content in the surface layer varies depending on the source water masses upwelling at the coast. A high-resolution physical-biological model would be useful to investigate the impact of the variability in the Cape Verde front position on primary production. ∗In a more recent analysis, Peña Izquierdo et al. (2011) have linked the water mass analysis presented in chapter 3 to the flow field over the continental slope. This approach allows to identify a connection between the central water masses, with SACW being the result of NACW and SACW* mixing in the Cape Verde frontal region. Furthermore, the northward propagation of SACW* is linked to the northward branch of the cyclonic circulation around the Guinea Dome. Due to the seasonally variable nature of the dome, it is 124 Chapter 5. Conclusions and outlook expected that transport of SACW* oscillates also during the year. Future directions should explore the seasonal variability of SACW* transport and its connection with the Guinea Dome circulation. ∗In chapter 4, model output has been analyzed to understand chlorophyll variability in the eastern subtropical North Atlantic. The same coupled ocean circulation-biogeochemical model, or a higher resolution model if available, could be used to investigate the Guinea Dome seasonal nutrient replenishment as suggested by Pelegrí et al. (2006). ∗The modeled chlorophyll interannual variability in the eastern subtropical North Atlantic could be examined in terms of large-scale variability in climate modes. It would be important to further explore the link between chlorophyll variability and various climatic indices such as the North Atlantic Oscillation or the Atlantic Meridional Mode. ∗We have seen that the extension of the high-chlorophyll area in the Cape Blanc region is related to large-scale processes, affecting the whole eastern subtropical gyre. We have viewed changes in this extension as arising from variations of the dynamics in both sides of the Cape Verde frontal system, specifically in the along-slope convergence, which is related to the intensity of coastal upwelling, and the offshore surface divergence, or positive Ekman pumping. But these two regions are not isolated, they are connected with upstream water sources, i.e. the Azores Current for the northern region and the North Equatorial Counter Current for the southern region. It remains to be studied what are the large-scale mechanisms that may con125 Chapter 5. Conclusions and outlook trol the intensity of these upstream water sources and what is their effect off NW Africa. 126 Chapter 6 Resumen de la tesis en español Afloramiento en el margen oriental del Atlántico Norte subtropical 6.1 Introducción y objetivos de la tesis 6.1.1 Introducción El margen oriental del Atlántico Norte subtropical es una región de elevado interés por dos razones principales. Primera, una fracción significativa de este área oceánica muestra una alta producción primaria que sostiene unos ricos recursos pesqueros. Las pesquerías constituyen una fuente importante de alimentos para el ser humano y un sector importante de la economía. La segunda razón es su conexión con el clima global. En la región oriental del Atlántico Norte subtropical, aguas subsurperficiales entran en contacto con la atmósfera e intercambian propiedades como calor y dióxido de carbono. En efecto, las masas de agua re127 Chapter 6. Resumen en español Figure 6.3: Digrama temperatura-salinidad con los tipos de masas de agua centrales, de las capas intermedias y profundas en el Océano Atlántico, tomado de Sverdrup et al. (1942). Masas de agua La capa superficial del Atlántico Noreste subtropical (aproximadamente los primeros 100 m) está caracterizada por altas salinidades, una alta concentración de oxígeno disuelto y bajas concentraciones de nutrientes (Fraga and Manríquez, 1974). Una característica particular del giro subtropical del Atlántico Norte es el máximo de salinidad subsuperficial (p.ej. Figura 2.4 y 2.5). Aguas superficiales densas con salinidad alta, producidas por el exceso de evaporación frente a precipitación, subducen debido a la convergencia por transporte de Ekman y a la convección invernal, y son transportadas a través de la circulación a gran escala del giro subtropical (Defant, 1936; Bauer and Siedler, 1988). Las masas de agua de la termoclina permanente son las masas de agua centrales (Sverdrup et al., 1942). Estas aguas tienen su origen en las regiones sub134 Chapter 6. Resumen en español tropicales de ambos hemisferios, donde las aguas superficiales convergen y el bombeo de Ekman es negativo, y se extienden hacia las regiones ecuatoriales. En un diagrama Temperatura-Salinidad (TS) destacan por su relación aproximadamente lineal (Figura 6.3). Las aguas centrales ocupan el rango de profundidad de aproximadamente 100 a 700 m. El Agua Central Noratlántica (NACW) tiene su origen en la zona de convergencia superficial del Atlántico Norte y alcanzan latitudes menores a través de la circulación termoclina (Sarmiento et al., 1982; Kawase and Sarmiento, 1985). Únicamente las masas de agua formadas en la zona de subducción a finales de invierno y comienzos de primavera pueden escapar de las capas superficiales y ser inyectadas en la termoclina permanente; durante el resto del año, las aguas subducidas quedan en la capa de mezcla ya que las velocidades de escape generadas por el bombeo de Ekman negativo son menores que el avance estacional de la capa de mezcla. La extensión vertical del NACW en latitudes más bajas vendrá determinada por el afloramiento invernal de la isopicna más densa en la zona de subducción, aproximadament σθ=27.3 (Kawase and Sarmiento, 1985; Reid, 1994). Emery and Meincke (1986) subdividió el NACW en dos tipos, Oriental (ENACW) y Occidental (WNACW). La división reflejaba distintas regiones de formación, al sur del Frente Subárctico para WNACW y al sur del Frente de Islandia-Faroe para ENACW. En cambio, Tomczak and Godfrey (1994), argumentan que los cambios de temperatura y salinidad observados a lo largo de la cuenca ocánica son el resultado de la variabilidad medioambiental dentro de la región de formación. Por otro lado, Emery and Meincke (1986) identifican sólo un tipo de Agua Central Sudatlántica (SACW), con formación cerca de la zona de convergencia subtropical de Brasil-Malvinas . Gordon et al. (1992) y Sprintall and Tomczak 135 Chapter 6. Resumen en español (1993) han mostrado que el agua central formada en la convergencia subtropical del Océano Índico representa una contribución importante de la termoclina del Océano Atlántico, introducida en la cuenca Atlántica a través de la Corriente de Agulhas. El Agua Central Sudatlántica (SACW) viaja a través de la termoclina del Atlántico sur hacia el sistema ecuatorial de corrientes y la región tropical del Atlántico Norte. La zona frontal de Cabo Verde (CVFZ) constituye la frontera entre el NACW y SACW y corresponde al límite sur de la recirculación termoclina del Atlántico Norte (Stramma and Siedler, 1988; Zenk et al., 1991; Arhan et al., 1994). La CVFZ se extiende hacia el suroeste desde los 20◦N en la costa de África hacia las Islas Cabo Verde, y entonces adquiere una orientación más zonal a medida que se va difundiendo progresivamente hacia el lado occidental de la cuenca. Las dos masas de agua ocupan el mismo rango de densidad, siendo el NACW más salada y cálida que el SACW. Como resultado, el frente está compensado en densidad lo cual propicia la generación de multitud de intrusiones y lentes (Zenk et al., 1991; Pérez-Rodríguez et al., 2001; Pastor et al., 2008). En la región de estudio, la capa intermedia está ocupada por Agua Intermedia Antártica (AAIW) y Agua Mediterránea (MW), aproximadamente entre los 700 y 1500 m de profundidad. El AAIW tiene su origen en el Frente Subantárctico y es transportada a través del giro subtropical del suratlántico hacia los trópicos (Suga and Talley, 1995). El AAIW puede ser fácilmente identificada por su baja salinidad, con un mínimo de salinidad centrado sobre los 800 m profundidad. Dos caminos transportan AAIW hacia el margen oriental Atlántico Norte, uno es a través del sistema de corrintes de frontera oeste y la Corriente de Azores (Kawase and Sarmiento, 1985; Tsuchiya et al., 1992). El segundo camino es a 136 Chapter 6. Resumen en español través del margen oriental a lo largo de la costa africana (Machín and Pelegrí, 2009), con una penetración máxima durante el otoño (Machín et al., 2010). El MW se forma en el Mar Mediterráneo y es introducida en el Atlántico a través del Estrecho de Gibraltar; desde donde se extiende hacia el norte y el sur, influyendo sobre todo el Atlántico Norte (Worthington, 1976). Su alta salinidad y temperatura características se pueden observar entre aproximadamente los 600 y 1500 m profundidad. Ambas aguas intermedias se encuentran sobre los 32◦N (p.ej. Figura 3.5), con el AAIW ocupando un rango de profundidad ligeramente más superficial que el MW. Las propiedades de las masas de agua intermedias originales están modificadas por mezcla con aguas de las capas superiores e inferiores, y sus propiedades TS en la región de estudio aparecen como valores extremos locales que se desvían sustancialmente de los valores encontrados en su región de formación (Figura 6.3). El Agua Profunda Noratlántica (NADW) se encuentra en la capa inferior, entre los 2000 y 4000 m aproximadamente. El NADW del Atlántico Nororiental está formada principalmente por Agua de Desbordamiento de Islandia-Escocia, modificada por Agua del Mar de Labrador y Agua Profunda Inferior (McCartney, 1992; Dickson and Brown, 1994; van Aken, 2000). Variabilidad interanual y modos climáticos La variabilidad interanual en el afloramiento costero ha sido relacionada principalmente con la Oscillation del Atlántico Norte (NAO) y con El Niño - Oscilación Sureña (ENSO). El índice NAO viene definido por la anomalía en la diferencia de la presión atmosférica a nivel del mar entre el sistema de bajas presiones de 137 Chapter 6. Resumen en español Islandia y sistema de altas presiones de Azores durante el invierno (de diciembre a marzo). Un aumento en el índice implica un aumento en las altas presiones de Azores y un aumento del flujo hacia el este sobre el noroeste africano, dando lugar a un aumento en el afloramiento costero generado por los vientos alisios. Analizando imágenes de satélite de temperatura superficial del mar (SST) entre 1982 y 2001, Santos et al. (2005) encontraron un cambio decadal en la intensidad del afloramiento pasando de un régimen de afloramiento débil en la década de los 80’s a un afloramiento intenso en los 90’s, asociado a un cambio en el índice NAO. Meiners (2007) y Meiners et al. (2010) también encontraron una correlación positiva entre el índice NAO y los vientos alisios al estudiar impactos de la variabilidad climática en la dinámica de la merluza negra en el noroeste de la costa africana. Específicamente, el índice NAO podía explicar el 53% de la variabilidad en la componente meridional de la tensión del viento en la costa de Mauritania y Senegal entre 1960 y 2004. Roy and Reason (2001) investigaron conexiones entre el Índice Multivariativo ENSO y anomalías en la SST y en la tensión del viento en la costa del noroeste africano (entre 10 y 20◦N). Su trabajo mostró que periodos cálidos en el Pacífico durante el otoño y pricipios de invierno (situación El Niño) daban lugar a un estado relajado del aforamiento inducido por viento en el lado oriental de la cuenca Atlántica y a eventos cálidos observados a lo largo de la costa oeste de África a finales de invierno y en primavera. Los eventos ENSO también influyen en el sistema ecuatorial de corrientes del Atlántico y en la dinámica del Domo de Guinea. Lázaro et al. (2005) observaron una intensificación de la NECC durante la primavera de 1997 y 1998, y del Domo de Guinea durante el verano de 1997, asociada a un desplazamiento prematuro hacia el norte de la ITCZ indicado por 138 Chapter 6. Resumen en español Enfield and Mayer (1997), y coincidiendo con un acontecimiento La Niña en el Pacífico. Varios autores también han identificado tendencias en el afloramiento costero, aunque en ocasiones las tendencias son contradictorias. Bakun (1990) identificó un aumento significativo en la tensión de viento favorable al afloramiento en un punto del noroeste africano situado en la latitud 28◦N entre 1946 y 1981. La intensificación de los vientos en este sitio coincidió con un aumento en otros lugares dentro de los principales sistemas de afloramiento costero del mundo. Este autor hipotetizó que, en un escenario de calentamiento global, con el aumento en las concentraciones de gases de efecto hinvernadero habría una intensificación del afloramiento costero debido a un aumento en el gradiente de presión entre la región costera y el mar. Utilizando datos de temperatura derivados a partir de registros sedimentarios que se extienden 2500 años atrás, McGregor et al. (2007) infirieron un aumento anómalo y sin precedentes en el afloramiento costero en Cabo Ghir durante el siglo XX. En cambio, analizando datos de viento de QuickScat entre 2000 y 2007, Demarcq (2009) encontró una tendencia al decrecimiento en la componente meridional de la tensión del viento en el noroeste africano. Aun así, estos resultados pueden reflejar únicamente variabilidad interanual o decadal superpuesta a una tendencia a escala de tiempo mayor. Producción primaria La región de afloramiento costero del noroeste africano es uno de los cuatro Sistemas de Afloramiento de Frontera Este (EBUS) en el océano global. Los EBUS muestran una alta productividad e importantes capturas pesqueras (Pauly and 139 Chapter 6. Resumen en español 0.03 0.1 0.3 1 3 CHL (mg m ) -3 40 30 20 10 40 30 20 10 -30 -20 -10 -30 -20 -10 VERANO INVIERNO PRIMAVERA OTOÑO Cape Blanc Cape Blanc Cape Blanc Cape Blanc Figure 6.4: Concentraciones de clorofila promedio (mg m-3) en invierno (EFM), primavera (AMJ), verano (JAS) y otoño (OND). Los contornos en negro marcan las isolíneas de 0.2 y 1 mg m-3 de clorofila. Los datos corresponden al sensor SeaWiFS entre 1998 y 2007. 140 Chapter 6. Resumen en español Christensen, 1995). El Atlántico Nordeste subtropical tiene la zona activa más grande de los cuatro EBUS principales, área definida como la región donde las concentraciones de clorofila superan el 1 mg m-3 (Carr, 2002). Es el segundo EBUS más productivo, con una producción primaria anual de 0.33 Gt de Carbono por año, después del EBUS de Benguela en el Atlántico Sur (Carr, 2002). La clorofila proporciona una medida indirecta de abundancia de fitoplancton y de ahí los numerosos esfuerzos en desarrollar un algoritmo para derivar producción primaria a partir de clorofila medida por satélite (Behrenfeld and Falkowski, 1997; Carr, 2002; Marra et al., 2003). La Figura 6.4 muestra las concentraciones de clorofial estacionales derivadas a partir de 10 años de datos satelitales del sensor SeaWiFS. La región entre 24◦N y el Estrecho de Gibraltar muestra una variabilidad estacional débil. Concentraciones de clorofila por encima de 1 mg m-3 están confiadas a la plataforma continental, a pesar de la existencia de afloramiento durante todo el año. La estrecha extensión costa afuera de la clorofila en esta región puede ser causada por limitación de nutrientes, ya que el rotacional negativo de la tensión del viento, característico del giro subtropical deprime la nutriclina (Lathuilière et al., 2008). La región entre 18 y 24◦N también muestra una estacionalidad débil, pero en este caso las altas concentraciones de clorofila se extienden considerablemente costa afuera durante todo el de año. La característica más predominante de esta banda latitudinal es el filamento gigante de Cabo Blanco (Gabric et al., 1993). Durante el verano y otoño, la convergencia de la CUC, que fluye hacia el sur, y la MC, que fluye hacia el norte, resulta en un intenso transporte hacia océano abierto. En invierno y principios de primavera, cuando el afloramiento alcanza más al sur de Cabo Verde, la convergencia y el transporte hacia océano abierto 141 Chapter 6. Resumen en español se ven reducidos pero están aún presentes (Pelegrí et al., 2006). Estructuras de tipo filamento que advectan clorofila costa afuera también puede ser observadas en los principales cabos, como Cabo Ghir, donde la producción primaria alcanza valores de 5 g C m-2 año-1 (García-Muñoz et al., 2005; Pelegrí et al., 2005). Entre los 18◦N y el Estrecho de Gibraltar elementos mesoscalares, como filamentos y remolinos, y procesos ondulatorios contribuyen a la variabilidad dinámica y bioquímica de la región, pero no trataremos estos temas en este estudio. Barton et al. (1998) y Barton (1998) han proporcionado un análisis extenso de los procesos mesoscalares y Hagen (2001) ha realizado una revisión excelente de la propagación de ondas. La región al sur de 18◦N presenta una gran extensión de la clorofila costa afuera durante invierno y primavera, cuando aguas ricas en nutrientes llegan a la superficie gracias al afloramiento costero. Durante verano, el fortalecimiento de la NECC y el desplazamiento costa afuera del área de bombeo de Ekman positivo eleva las capas superiores de la termoclina, y ya en otoño una alta producción primaria puede ser observada en el área del GD (localizado en 10◦N, 22◦O aproximadamente). Pelegrí et al. (2006) han propuesto un mecanismo por cual las capas subsurperficiales del GD mantienen un nivel alto de concentración de nutrientes. El GD se desarrolla durante verano y otoño y los nutrientes que alcanzan la capa fótica son utilizados. Durante el invierno, los vientos alisios favorables al afloramiento se extienden al sur de Cabo Blanco, y el GD se relaja. La celda vertical asociada al afloramiento costero necesita un suministro de aguas subsurperficiales desde el océano interior, este suministro reestablece los altos niveles de nutrientes de la región del GD. 142 Chapter 6. Resumen en español 6.1.2 Objetivos y resumen de la tesis Esta tesis contiene tres capítulos centrales en formato de artículo científico, precedidos por una introducción general y seguidos de unas conclusiones generales. Para realizar la tesis, se ha examinado una gran variedad de fuentes de datos. Datos históricos han sido reanalizados. También han sido examinamos datos hidrográficos recientemente adquiridos y propiedades medidas remotamente por sensores satelitales, como concentración de clorofila o altura superficial del mar. Finalmente, un modelo de circulación ocánica general acoplado a un modelo biogeoquímico de última generación ha permitido una mejor comprensión de los mecanismos físicos que generan variabilidad en propiedades bioquímicas. Capítulo 2 Datos hidrográficos históricos de cuatro campañas han sido combinados para producir dos conjuntos de datos, concretamente el conjunto de primavera de 1973 y el de otoño de 1975, que cubren la zona de transición costera entre las latitudes 17 y 26◦N. Los datos estaban disponibles en colecciones impresas y por ello se tuvieron que digitalizar antes de comenzar el análisis. La hidrografía de la zona se describe en cuanto a su temperatura, salinidad, nitratos y oxígeno disuelto. La mezcla entre aguas centrales del norte y del sur es explorada, enfatizando el papel especial de la doble difusión en aumentar la difusión horizontal de calor. Finalmente, se calculan intercambios de agua y nutrientes entre la zona de transición costera y el océano profundo, diferenciando entre la contribución geostrófica y la de Ekman . Las preguntas que se intentaran contestar en este capítulo son: 143