Ocean Climate of the South East Atlantic Observed from Satellite Data and Wind Models N.J

Home , Angola Current, Benguela current, Cape Agulhas, Guinea Current, South Atlantic Current

Progress in Oceanography 59 (2003) 181–221 www.elsevier.com/locate/pocean

Ocean climate of the South East Atlantic observed from satellite data and wind models N.J. Hardman-Mountford a,∗, A.J. Richardson b, 1, J.J. Agenbag c, E. Hagen d, L. Nykjaer e, F.A. Shillington b, C. Villacastin e a Plymouth Marine Laboratory, Prospect Place, West Hoe, Plymouth, Devon PL1 2PB, UK b Oceanography Department, University of Cape Town, Rondebosch 7701, Cape Town, South Africa c Marine and Coastal Management, Private Bag X2, Rogge Bay, 8012 Cape Town, South Africa d Insitute for Baltic Sea Research Warnemuende, Seestrasse 15, 19119 Warnemuende, Germany e Institute for Environment and Sustainability, Joint Research Centre, I-21020 Ispra, Va, Italy

Revised 8 September 2003; accepted 14 October 2003

Abstract

The near-coastal South East Atlantic Ocean off Africa is a unique and highly dynamic environment, comprising the cool Benguela Current, warm Angola Current and warm Agulhas Current. Strong coastal upwelling and the Congo River strongly influence primary production. Much of the present knowledge of the South East Atlantic has been derived from ship-borne measurements and in situ sensors, which cannot generally provide extensive spatial and temporal coverage. Similarly, previous satellite studies of the region have often focused on small spatial areas and limited time periods. This paper provides an improved understanding of seasonal and interannual variability in ocean dynamics along the South East Atlantic coast of Africa using time series of satellite and model derived data products. Eighteen years of satellite sea surface temperature data are complimented by 7 years of sea level data. Three years of chlorophyll a data illustrate the seasonal biological response, but the time series is not of sufficient length for investigating interannual variability in chlorophyll biomass. Modelled wind fields are used to describe atmospheric forcing of the surface ocean. This is the first synoptic-scale description of the South East Atlantic from a suite of large spatial coverage, long time series products. Previous studies of seasonal and interannual variability in the region are reviewed and used to interpret key oceanographic features and processes identified in the satellite data. Key findings of this study are:

1. Descriptions of seasonal and interannual variability from these data show climate forcing of the South East Atlantic coast of Africa from both the northern and southern boundaries. Bimodal seasonal signals of equatorial origin propagate poleward along the Angolan coast, while the trade winds and events in the Agulhas region dominate the Agulhas Bank and Southern Benguela. The Northern Benguela is a mixed regime, under the influence of forcing from both directions. 2. The Benguela Ninõ years of 1984 and 1995 are clearly observed in sea surface temperature and sea level anomalies and correspond to anomalously weak southerly winds at the equator. These conditions were also observed in 1999, suggesting this too may have been a Benguela Ninõ year. 3. Consideration of putative Ninõ-type events in the equatorial Atlantic from this and other studies suggests that the

∗ Corresponding author. Tel.: +44-1752-633100; fax: +44-1752-633101. E-mail address: [email protected] (N.J. Hardman-Mountford). 1 Present address: Sir Alister Hardy Foundation for Ocean Science, The Laboratory, Citadel Hill, Plymouth PL1 2PB, UK.

0079-6611/$ - see front matter  2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.pocean.2003.10.001 182 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181–221 frequency of these events is much higher than previously estimated and may be similar to the frequency of El Ninõ- Southern Oscillation (ENSO) events in the Pacific Ocean. Furthermore, years of anomalously strong southerly winds at the equator occur during Pacific ENSO years.

Contents

1. Introduction ...... 183 1.1. Oceanographic and meteorological overview ...... 183 1.1.1. Meteorology ...... 183 1.1.2. Major currents and boundaries ...... 185 1.1.2.1. Angola Current ...... 185 1.1.2.2. Benguela Current ...... 185 1.1.2.3. Angola–Benguela front ...... 186 1.1.2.4. Agulhas Current ...... 186 1.1.3. Upwelling ...... 187 1.1.3.1. Gabon/Angola coastal system ...... 187 1.1.3.2. Benguela ...... 187 1.1.3.3. Agulhas Bank ...... 189 1.1.4. Interannual variability ...... 189 1.1.4.1. Benguela Nin˜os and tropical-origin anomalous events ...... 189 1.1.4.2. Agulhas Current intrusions ...... 190 1.2. Satellite remote sensing of the South East Atlantic Ocean ...... 190

2. Data and methods ...... 192 2.1. Data ...... 192 2.1.1. Sea surface temperature from satellite ...... 192 2.1.2. Sea level from satellite ...... 192 2.1.3. Chlorophyll a from satellite ...... 192 2.1.4. Surface winds from models ...... 192 2.2. Methods ...... 193 2.2.1. Monthly maps ...... 193 2.2.2. Latitude–time plots ...... 193 2.2.3. Annual sea surface temperature anomaly maps ...... 194 2.2.4. Time series of anomalies ...... 194 2.2.5. Spectral analysis ...... 194

3. Results and discussion ...... 195 3.1. General overview ...... 195 3.2. Seasonality ...... 198 3.2.1. Angola Current and Angola coastal upwelling ...... 198 3.2.2. Congo River plumes ...... 201 3.2.3. Benguela upwelling ...... 201 3.2.4. Agulhas Current and Western Agulhas Bank ...... 202 3.3. Interannual variability ...... 203 3.3.1. Equatorial forcing and Atlantic Nin˜o-type events ...... 203 3.3.2. Agulhas warming and warm water intrusions to the Southern Benguela ...... 214

4. Conclusions ...... 215 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181–221 183

1. Introduction

The near-coastal South East Atlantic Ocean off Africa is a highly dynamic environment (Shannon, 1985a; Shillington, 1998), unique in many respects. The cool Benguela Current is the only eastern boundary current to be bounded at both ends by warm water of tropical origin: the Angola Current in the north and the Agulhas Current to the south (Shannon & Nelson, 1996). Additionally, the Benguela Current system may possess the strongest sustained, locally wind-driven, coastal upwelling of any region of the world ocean, the Lu¨deritz upwelling cell of southern Namibia (Parrish, Bakun, Husby, & Nelson, 1983; Bakun, 1993). Furthermore, into this area flows the Congo River, the largest freshwater input to any eastern ocean boundary, and this river has a marked effect on the productivity of the region (Thomas, Carr, & Strub, 2001). Much of the present knowledge of the South East Atlantic has been derived from ship-borne measurements and in situ sensors, which cannot generally provide both extensive spatial and temporal coverage. Previous satellite studies of the region have often focused on small spatial areas and limited time periods. We are now in the position to view both physical (sea surface temperature and sea surface height) and biological (chlorophyll a) variables in the South East Atlantic from satellite over broad temporal and spatial scales. The aim of this paper is to provide an improved understanding of the ocean dynamics along the South East Atlantic coast of Africa by describing seasonal and interannual variability using satellite-derived data products. These products include 18 years of sea surface temperature (SST) data, 7 years of sea surface height data, and 3 years of ocean colour data. Additional information for interpretation is provided by 17 years of modelled wind data. This is the first synoptic-scale description of the South East Atlantic from a suite of long time series satellite products. It is hoped that the insights provided by this study will aid current scientific research programmes in the region, such as the Benguela Current Large Marine Ecosystem (BCLME) programme.

1.1. Oceanographic and meteorological overview

The major oceanographic features of the South East Atlantic are shown in Fig. 1.

1.1.1. Meteorology Oceanographic conditions in the South East Atlantic are largely controlled by basin-scale, ocean–atmosphere interactions over the South Atlantic. In the East Atlantic Ocean, the ‘meteorological equator’ or Intertropical Convergence Zone (ITCZ) is to be found several degrees north of the geographical equator. The climatic position of the ITCZ clearly shows an inclination from southwest to northeast (Citeau, Berge´s, Demarcq, & Mahe´, 1988). Low atmospheric pressure over tropical regions of the African continent causes a divergence of the southeast trade winds, creating the southwesterly monsoon winds along the Angolan coast. In the eastern equatorial Atlantic, seasonal and interannual changes in the position of the ITCZ reflect changes in the southwest monsoon and the southeast trade winds directly. Associated fluctuations are correlated with changes in sea level air pressure over the central South Atlantic, in both the position and intensity of the atmospheric high pressure cell that forms the South Atlantic Anticyclone. This anticyclone interacts with the equatorial low-pressure belt of the ITCZ in the north and continental low-pressure cells over southern Africa in the east to control the southeasterly trade winds along the west coast of southern Africa. Accordingly, it provides the driving force for the Benguela Current regime, which feeds into the South Equatorial Current as part of the basin-scale, wind-driven circulation (Peterson & Stramma, 1991). Intense solar radiation warms near-surface waters causing an accumulation of warm surface water in western equatorial zones (Wyrtki, 1982). Here, the current convergence results in positive sea level anomalies and downwelling accompanied by a deep pycnocline (Carton, 1994). Across the whole equatorial ocean, pressure gradients are established and maintained within superficial layers along and parallel to the equator and 184 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181–221

Fig. 1. Map of the South East Atlantic showing surface and near-surface currents, frontal zones, upwelling cells, major areas of freshwater input and bathymetry. EUC, Equatorial Under Current; SEC, South Equatorial Current; SECC, South Equatorial Counter Current; AnC, Angola Current; BOC, Benguela Oceanic Current; BCC, Benguela Coastal Current; SAC, South Atlantic Current; AgC, Agulhas Current; ABF, Angola–Benguela front; STF, subtropical front; STG, subtropical gyre; ACC, Antarctic Circumpolar Current. Modiﬁed after Shannon and Nelson (1996), Shannon (1985a), Peterson and Stramma (1991), Moroshkin et al. (1970). these currents convey warm water towards the equatorial margins of the eastern boundary current systems (Hagen, 2001). On the basin scale, zonal and meridional components of the southeast trade winds affect the ocean currents differently. The westward equatorial component (easterlies) balances the baroclinic pressure gradi- N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181–221 185 ents in the sea and maintains eastward ﬂowing downgradient motions, producing the observed system of equatorial undercurrents and counter currents (Hagen, 2001). Further south, the equatorward component of the southeast trade winds is responsible for the offshore Ekman transport in near surface layers near the eastern ocean boundary and causes the belt of coastal upwelling along the South African and Namibian coasts. Here, the climatic pattern in the corresponding windstress curl exhibits a cyclonic sign and related Ekman-pumping processes support the maintenance of coastal upwelling within offshore distances of several hundred kilometres (Bakun & Nelson, 1991). South of the continent, the mid-latitude westerlies cause wind reversals in winter, which cause deep mixing on the western Agulhas shelf. In summer, easterly winds on the Agulhas shelf cause coastal upwelling and a cool cyclonic ridge projects seawards from Cape Agulhas (Boyd & Shillington, 1994). Modulation of the summer southeasterly winds appears to be the primary forcing of coastal trapped waves that propagate polewards along the west coast of Namibia and South Africa (Shillington, 1998).

1.1.2. Major currents and boundaries 1.1.2.1. Angola Current The poleward Angola Current is a fast (40–50 cm sϪ1), narrow and stable geostrophic flow of warm (Ͼ24 °C), saline (36.4) water that reaches 250–300 m depth and meanders laterally and vertically over both the continental shelf and slope of the Angolan coast (Moroshkin, Bunov, & Bulatov, 1970; Dias, 1983a, b; Lass, Schmidt, Mohrholz, & Nausch, 2000). This current forms the eastern section of a regionally fixed, cyclonic gyre at about 12° S, 4–5° E (Angola Dome). In the upper 100 m, this feature seems to be formed mainly by the southeast branch of the South Equatorial Counter-Current and the southward-turning waters from the north branch of the Benguela Current, with only a moderate influx of waters originating north of the equator. However, in deeper layers, northern waters become pre- dominant in maintaining mixed water properties conveyed by the Angola Current to higher latitudes (Moroshkin et al., 1970). Little is known of the seasonality of the Angola Current. Dias (1983b) noted temporal changes in its velocity, with flow being stronger in March than July. During the winter and spring, very warm Angola Current water (27–30 °C) frequently retreats to the northwest of the Angola Dome and is replaced by slightly cooler waters (20–26 °C). This cooling is the result of advection from the main Benguela upwelling area (Feistel, Hagen, & Grant, 2004). Seasonal intrusion of warm, saline tropical waters from the Angolan coast into the Benguela upwelling region to the south has been described extensively (Boyd, Salat, & Maso´, 1987; Hart & Currie, 1960; Stander, 1964; O’Toole, 1980; Badenhorst & Boyd, 1980; Le Clus & Kruger, 1982; Boyd, 1983; Kruger & Boyd, 1984; Hagen, Feistel, Agenbag, & Ohde, 2001). Maximum southward penetration is seen in late summer and early autumn (Boyd et al., 1987), coinciding with intensification of the Angola Current and a partial relaxation in equatorward windstress along the Namibian coast (Boyd, 1987).

1.1.2.2. Benguela Current To the south of the Angola Current lies the equatorward-flowing, cool Benguela Current system. This is composed of a mix of Indian and South Atlantic subtropical thermocline water, with significant contributions of saline, low oxygen tropical Atlantic water and cooler, fresher subantarctic water (Garzoli, Gordon, Kamenkovich, Pillsbury, & Duncombe-Rae, 1996). The Benguela Current system comprises a coastal branch, modulated by local weather processes, and an offshore oceanic flow, which forms the eastern limb of the South Atlantic subtropical gyre (Shannon, 1985a; Peterson & Stramma, 1991). Although the coastal and offshore components of this current system are separated to some degree by an offshore divergence over the continental slope, both flows are integral to the functioning of the system so most authors treat them together (Lutjeharms, 1977; Nelson & Hutchings, 1983; Parrish et al., 1983; Shannon, 1985a). The prevailing southeast trade winds are responsible for strong offshore Ekman transport and the upwelling of cool, subthermocline water along the Benguela coast between 16° S and about 34° S(Shannon, 1985a). 186 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181–221

The Benguela Current begins as a northward flow off the Cape of Good Hope, where it skirts the African west coast equatorward up to around 24°–30° S. Here, the larger oceanic component branches away from the coast, towards the northwest, and flows into the South Equatorial Current. However, one or two branches of the current do continue along the coast to meet with the Angola Current at the Angola–Benguela front near 16° S(Moroshkin et al., 1970; Wedepohl, Lutjeharms, & Meeuwis, 2000). There is a well-defined mean flow in the Benguela Current that is mostly confined near the African continent. Highest speeds occur in the south during summer and in the north during winter, following the seasonal migration of the trade winds. On its western side, however, flow is more transient and is characterised by large eddies shed from the Agulhas Retroflection, which are probably responsible for the spatial variations in transport seen here (Garzoli et al., 1996). Wedepohl et al. (2000) found that in the south the current has a width of 200 km, but as it flows northward it widens rapidly to 750 km in the north.

1.1.2.3. Angola–Benguela front The Angola-Benguela frontal zone is the convergence between the south- flowing, warm Angola Current and north-flowing, cold Benguela Current (Shannon, Agenbag, & Buys, 1987). The front is a permanent feature at the sea surface, maintained throughout the year in a narrow latitudinal band between 14° and 16° S, with a general west to east orientation at the coast, although this varies offshore (Meeuwis and Lutjeharms, 1990). Salinity data suggest the front may extend to 200 m depth, but it is particularly marked in both temperature and salinity in the upper 50 m (Shannon et al., 1987). It is termed a frontal zone because often two or three parallel fronts can be observed (Shannon & Nelson, 1996). The front fluctuates seasonally in latitude, seaward extent, temperature and strength. It is furthest south in summer and early autumn and furthest north in winter and early spring (Boyd et al., 1987). Boyd et al. (1987) associated this migration with the seasonal movement of the ITCZ in the equatorial Atlantic. Tem- peratures associated with the front exhibit a similar pattern of seasonal change. At the coast, the front is least defined in winter, a period coinciding with the coolest water off Namibia and stronger northward flow of Benguela water. It is most clearly developed during summer and autumn, when the warm-flowing southward Angola Current is at its maximum and coastal upwelling is at its weakest. On average, the front reaches its southernmost position in near-coastal zones and penetrates seawards up to distances of about 250 km, although traces of it can be found up to 1000 km offshore. Maximum seaward extent is in spring and summer (Shannon et al., 1987; Meeuwis & Lutjeharms, 1990).

1.1.2.4. Agulhas Current In contrast to the relatively broad current cores of the Benguela Current, a typical eastern boundary system, the Agulhas Current has relatively narrow current cores and forms a branch of the western boundary current system of the Indian Ocean. It has a high core velocity up to 2 msϪ1 (Grundlingh & Lutjeharms, 1979) and follows the eastern and southern continental slope of South Africa, conveying warm near-surface water (24–28 °C, Lutjeharms, Cooper, & Roberts, 2000) from the tropical Indian Ocean along the southern continental slope of Africa (Gordon, 1985). When it reaches the southernmost terminus of the continental shelf it retroflects and flows eastwards again as the Agulhas Return Current (Gordon, 1985; Quartly & Srokosz, 1993). Warm water filaments sometimes separate from the retroflected flow and are preferentially advected along the western edge of the Agulhas Bank and up the west coast of South Africa, following closely the position of the upwelling front around the southernmost upwelling cells (Bang, 1971; Lutjeharms & Stockton, 1987; Lutjeharms, 1996). Although the Agulhas Current flows strongly all year round, satellite SST and sea level data reveal marked seasonal variability (Matano et al., 1998; Weeks, Shillington, & Brundrit, 1998). Using altimeter data, Matano et al. (1998) found this variability to be at a maximum during the summer and at a minimum during the winter, attribu- ting it to seasonal changes in transport. The Agulhas Retroflection can be found, on average, around 15° E. However, there may be an early bifurcation of the current near 25° E during the summer months (Lutjeharms & van Ballegooyen, 1988; Matano et al., 1998). N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181–221 187

An interesting aspect of the Agulhas Retroflection is that, at the point of its retroflection, large (~200– 300 km average diameter), anticyclonic eddies, termed Agulhas Rings, are frequently pinched-off into the South Atlantic Ocean (Lutjeharms & van Ballegooyen, 1988; Boebel et al., 2003) at a rate of about 6 (range 4–12) eddies per year (de Ruijter et al., 1999; McDonagh, Heywood, & Meredith, 1999). These rings enclose pools of Indian Ocean water, more than 5 °C warmer but less saline than South Atlantic surface water at similar latitude (Gordon, 1985). The rings keep their distinctive thermal characteristics as far west as 5° E and as far south as 46° S(Lutjeharms & van Ballegooyen, 1988). Thus, these eddies contribute a major heat flux to the South Atlantic Ocean and form an essential component of the global overturning circulation (McDonagh et al., 1999), with a potentially strong influence on global climate patterns (Gordon, 1985). Cold-core (cyclonic) eddies have also been observed in the area of the Agulhas Retroflection and Cape Peninsula (Gordon, 1985; Lutjeharms & van Ballegooyen, 1988). These coexist with the anticyclones but have a slightly smaller diameter (~120 km) and drift across the northwestward migration path of the anticyclones (Boebel et al., 2003).

1.1.3. Upwelling 1.1.3.1. Gabon/Angola coastal system Along the coasts of Gabon and Angola, seasonal upwelling has been observed during July to September (Picaut, 1983; Shannon, 1985a), with cells identified at Cabinda and Luanda in Angola and at Pointe Noire in Gabon (Picaut, 1983; Lutjeharms & Meeuwis, 1987; Binet, Gobert, & Maloueki, 2001). Berrit (1976) concluded that north of 15° S winds were not favourable for Ekman upwelling and the upwelling observed had other causes. Studies of coastal upwelling north of the equator in the Gulf of Guinea came to a similar conclusion (Bakun, 1978; Picaut, 1983). It is generally considered that this non-wind-driven coastal upwelling, along both the Gulf of Guinea and Angolan coasts, is related to equatorial upwelling in the eastern Atlantic and can be explained by remote forcing from the western equatorial Atlantic, as first proposed by Moore et al. (1978). Rapid seasonal intensification of easterly winds in the western equatorial Atlantic generates equatorially trapped Kelvin waves that travel eastward along the equatorial wave guide, down zonal baroclinic pressure gradients (Katz, Hisard, Verstraete, & Garzoli, 1986), until they reach the west coast of Africa. There they diverge poleward, becoming trapped by the coast where they raise the thermocline in their wake, bringing cooler water near the surface. An elaboration of this theory includes the reflection of Rossby Waves back across the Atlantic in a westward direction, travelling both north and south of the equator (O’Brien, Adamec, & Moore, 1978; Adamec & O’Brien, 1978). Subsequent support for this mechanism has been provided by both modelling studies (e.g. Philander & Pacanowski, 1980; Busalacchi & Picaut, 1983; Philander & Pacanowski, 1986) and observations (e.g. Merle, 1978; Servain, Picaut, & Merle, 1982; Houghton, 1983; Picaut, 1983; Hough- ton & Colin, 1986; Verstraete, 1987, 1992).

1.1.3.2. Benguela As in other eastern boundary current systems, the Benguela is characterised by strong coastal upwelling and high productivity (Hutchings, Pitcher, Probyn, & Bailey, 1995; Hill et al., 1998; Thomas et al., 2001; Carr, 2002). The upwelling is induced by equatorward winds that ﬂow nearly parallel to the coast causing offshore Ekman divergence of surface waters. The main upwelling area is along the coasts of Namibia and South Africa, between 16° and 34° S, with intense upwelling cells found at several locations along the coast (Shannon, 1985a; Lutjeharms & Meeuwis, 1987). The strongest and most persistent upwelling cell is near Lu¨deritz on the southern Namibian coast (~27° S). Windstress is greatest here and sea temperatures are coldest (Shannon, 1985a; Lutjeharms & Meeuwis, 1987). Equatorward coastal winds and upwelling are strong throughout the year, with a slight maximum in spring and a slight minimum in autumn. The perennial intense windstress in this upwelling centre means shelf waters are well mixed all year round (Parrish et al., 1983; Bakun, 1993). The Lu¨deritz upwelling cell divides the Benguela system into northern and southern parts, forming both a physical and biological boundary (Shannon, 1985a; Pitcher, 188 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181–221

Brown, & Mitchell-Innes, 1992), although Agenbag and Shannon (1988) put the boundary slightly to the north at Meob Bay (24.5° S). Other strong upwelling cells are located near the boundaries of the Benguela system. The northernmost boundary has historically been identified as the Cape Frio or Cunene upwelling cell around 18° S (Copenhagen, 1953; Shannon, Nelson, & Jury, 1981; Nelson & Hutchings, 1983), although a number of observations have been made of upwelling as far north as 15° S(Hart & Currie, 1960; Stander, 1964; Parrish et al., 1983). The southernmost boundary was identified as the Cape Peninsula cell at 34° Sby Hart and Currie (1960) and Andrews and Hutchings (1980), although upwelling can extend as far south as Cape Agulhas (35° S, 20° E) so this could be regarded as a more appropriate boundary of the Benguela system (Shannon, 1985a). Certainly, there is a marked change in the wind field south of 35° S(Shannon, 1985a). The region between the Cape Peninsula and Cape Agulhas, the western Agulhas Bank, has characteristics of both the Benguela and Agulhas Current regimes and may be regarded as a transitional zone. A number of generally weaker upwelling cells are also located along the Benguela coast. In the Northern Benguela region (i.e. north of Lu¨deritz), cells are located along the northern Namibian coast around 20° S and near Walvis Bay (23° S) in central Namibia. In the Southern Benguela, additional cells are located near Hondeklip Bay (30° S) in northern South Africa, often called the Namaqua cell, at Cape Columbine (32° S) north of Cape Town and at Cape Agulhas (Shannon, 1985a; Lutjeharms & Meeuwis, 1987). A phenomenon associated with upwelling cells along the Benguela coast is the presence of shallow (50 m deep) upwelling filaments that may extend several hundreds of kilometres from the coast into the South Atlantic Ocean (Lutjeharms, Shillington, & Duncombe-Rae, 1991; Shillington, Hutchings, Probyn, Wald- ron, & Peterson, 1992). These are especially prevalent around the Lu¨deritz and Namaqua upwelling cells between 26° and 29° S(Lutjeharms et al., 1991; Shillington et al., 1990; Hagen et al., 2001). Despite the quasi-permanent nature of these upwelling cells, such filaments have typical life spans of a few days to several weeks and are generally oriented perpendicular to the coast (Shannon & Nelson, 1996). Irregularities in the alongshore windstress, coastal geometry, and topography of the continental shelf trigger non-linear instabilities in associated currents to support the generation of sporadically occurring filaments along the entire coast (Ikeda & Emery, 1984). These filaments are sometimes extended further offshore through interaction with northward moving Agulhas rings (Lutjeharms et al., 1991; Shillington et al., 1992). Between such filaments, there are zones of relatively undisturbed coastal dynamics. For instance, near Palgrave Point (20.5° S), between the major upwelling cells of Lu¨deritz and Cape Frio, there is a local minimum in windstress and offshore Ekman transport (Parrish et al., 1983). The intensity of wind mixing is weakest here during the winter (July–September), leading to strong stabilisation of the water column (Bakun, 1993). Upwelling fronts are located at the edge of the upwelling zone, often close to the shelf edge, although details of the structure vary between areas (Shannon, 1985a). Like upwelling filaments, these fronts can also be modified through interaction with northward moving Agulhas rings (Lutjeharms et al., 1991; Shil- lington et al., 1992). The shelf-edge front off the Cape Peninsula is particularly strong in summer, being enhanced by the upwelling inshore and advection of warm Agulhas Current water offshore (Shelton & Hutchings, 1990). Associated with this strong front is a geostrophic, equatorward jet current that flows along the line of steepest bathymetric gradients around the Cape Peninsula to Cape Columbine (Shelton & Hutchings, 1982). Boyd and Nelson (1998), using measurements along a cross-shelf transect, found that the seasonal changes in flows in this area were a consequence of the presence of Agulhas water, wind- forcing and barotropic shelf waves. In late winter-spring (August–October), strong north-westward flow was observed close inshore and by early summer this had formed a “classic” equatorward jet current. During summer, these circulation patterns became more complex, with the north-westward flow of the jet current located further offshore and marked southward flow inshore. By autumn, the front had moved far offshore due to increasing offshore Ekman transport (Boyd & Nelson, 1998). N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181–221 189

1.1.3.3. Agulhas Bank Although the southern extremity of the Benguela coastal upwelling is usually identified as either Cape Point or Cape Agulhas, upwelling also occurs east of Cape Agulhas (Schumann, Perrins, & Hunter, 1982; Shannon & Nelson, 1996; Lutjeharms et al., 2000). Over the Agulhas Bank, however, the wind-driven surface Ekman transport is mainly directed towards the coast due to prevailing westerlies. Nonetheless, substantial upwelling can occur at the continental shelf break, where the Agulhas Current flow impinges upon the shelf edge (Bakun, 1993; Lutjeharms et al., 2000). This has been particularly noted where the shelf widens along the path of the current (Bakun, 1993). Lutjeharms et al. (2000) described this upwelling as a distinct cell towards the eastern limit of the Agulhas Bank, although details of the mechanisms are not well understood. Classical Ekman coastal upwelling generally only occurs during periods of strong, persistent easterly winds, especially at prominent headlands (Schumann et al., 1982). Despite both upwelling and strong wind mixing over the Agulhas Bank, the vertical stratification of the water column is strong all year round (Parrish et al., 1983), with a continuous, strong thermocline that deepens towards the west (Largier & Swart, 1987). Temperature differences of 8–10 °C over depth intervals of as little as 10 m are not uncommon (Swart & Largier, 1987). The near-surface advection of Indian Ocean surface water appears to counteract the substantial local wind stirring to provide stable conditions (Largier & Swart, 1987; Bakun, 1993). These physical features result in the Agulhas Bank having higher levels of primary production than other major western boundary currents (Probyn, Mitchell-Innes, Brown, Hutchings, & Carter, 1994).

1.1.4. Interannual variability Interannual variability in the South East Atlantic has been extensively documented in sea temperature and salinity (e.g. Shannon & Agenbag, 1990), sea level (e.g. Brundrit, de Cuevas, & Shipley, 1987), winds (e.g. Shannon et al., 1992) and atmospheric pressure (e.g. Agenbag, 1996), although relatively few long- term studies have been undertaken. In general, power spectra analyses of long-term (multi-decadal) records from coastal stations worldwide exhibit three common time windows:

1. Fluctuations on the ocean–basin scale associated with the quasi-biannual oscillation (QBO), generally with periods from 24 to 26 months (Naujokat, 1986). 2. World-wide anomalies associated with El Nin˜o-Southern Oscillation (ENSO) in the equatorial Paciﬁc Ocean within the period range of 3–7 years (Philander, 1981). 3. Globally occurring quasi-cycles with periods between 20 and 90 years, which may underpin observed regime shifts in marine ecosystems (Latif & Barnett, 1994).

These signals are likely to be embedded in data from the South East Atlantic, although the relative impor- tance of these and other mechanisms will vary locally.

1.1.4.1. Benguela Ninõs and tropical-origin anomalous events Four major warm events are recognised in the Benguela region, viz. 1934, 1963, 1984 and 1995 (Shannon, Boyd, Brundrit, & Taunton-Clark, 1986; Gammelsrød, Bartholomae, Boyer, Filipe, & O’Toole, 1998), although there is some evidence of similar events in 1900, 1910, mid-1920s, 1949–1950, 1972–1974 and 1987 (Shannon, 1985a; Shannon & Taunton- Clark, 1989; Shannon & Agenbag, 1990). All of the four well-documented, major events involved large intrusions of warm, saline tropical water into the Northern Benguela region (Shannon et al., 1986; Gam- melsrød et al., 1998), manifested as an extreme southward migration of the Angola–Benguela front superimposed on its normal seasonal intrusion (Boyd et al., 1987; Shannon, Crawford, Brundrit, & Underhill, 1988). The influx of tropical water from the north-west (offshore) during the 1984 event, distinguished it from the normal seasonal intrusion (Shannon & Nelson, 1996), however, influxes from offshore have not been so clearly associated with other warm events. Such southward ‘outbreaks’ of equatorial water can obstruct coastal upwelling within affected regions 190 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181–221 by depressing the thermocline (downwelling) and stabilising the water column (Shannon et al., 1988). In some years (1949, 1963, 1972, 1984) this was associated with a reduction in equatorward windstress (Shannon et al., 1986; Taunton-Clark, 1990). Extremely warm, nutrient poor waters replace the cold, nutrient-rich upwelling water with drastic consequences for productivity in the Northern Benguela ecosystem. Warmer surface water temperatures feed back to the atmosphere and enhance convection and precipitation over the whole region. Thus, well-known warm years are characterised by a reduction of upwelling in the Northern Benguela and increased rainfall over the Namibian hinterland (Shannon, 1985a). Shannon et al. (1986) highlighted a number of similarities between the 1963 and 1984 events and El Ninõ-Southern Oscillation (ENSO) events in the Pacific and suggested the term “Benguela Ninõ” to describe them. Similarities noted included apparent connections to anomalous atmospheric conditions in the tropical Atlantic in terms of relaxed westerlies along the equator and released dynamics of eastward propagating equatorial Kelvin waves, poleward propagation of tropical water along the Namibian coast, suppression of phytoplankton production, displacement of fish stocks and a reduction in fish catches (Gammelsrød et al., 1998). Long-term sea level records have shown anomalously high sea levels corresponding to these years of strong positive anomalies in sea temperature along the entire southwest African coast (Brundrit et al., 1987). These authors suggest that signals propagate polewards from the equatorial Atlantic, although the theoretically expected speed of propagation is too fast to be resolved from the monthly records used. Nonetheless, the sea level signature of these events is consistent with the theory of coastal trapped waves and is similar to descriptions of the poleward propagation of warm water events along Peruvian and Chilean coasts during ENSO events in the Pacific(Enfield & Allen, 1980). On the other hand, the interannual frequency at which these events are observed in the Benguela region appears to be much lower than in the Pacific(Shannon et al., 1986). In the eastern equatorial Atlantic, warm events have been noted in 1953–1954, 1962–1963, 1973, 1984 and 1987–1988 (Koranteng, 1998), suggesting a slightly higher frequency since the 1970s. Furthermore, other studies have shown some obser- vational evidence that there has been an increasing tendency in the total number of such events during the last 30 years in both oceans (Latif & Gro¨tzner, 2000; Hagen et al., 2001). 1.1.4.2. Agulhas Current intrusions Because coastal upwelling of the Benguela Current regime is also bounded at its southern limit by warm water originating from the Agulhas Current, anomalous warm intrusions into the Southern Benguela region have also been observed (Shannon & Agenbag, 1990; Shan- non & Nelson, 1996). For instance, a strong, warm intrusion was documented in 1986, linked to changes in winds in the southwest Indian Ocean and over southern Africa (Shannon, Agenbag, Walker, & Lutjeharms, 1990). Hydrographic data collected between 1957 and 1964 suggest similar intrusions may have taken place over this period and satellite observations of SST show warm intrusions during summer 1992/1993 and June 1994 (Shannon & Nelson, 1996). Modelling studies have indicated that local wind effects may influence the penetration of the Agulhas Current water into the South East Atlantic (de Ruijter, 1982; de Ruijter & Boudra, 1985). Although 1982 and 1983 have generally been described as cool years in the Benguela region, a minor warming was observed around the Cape Peninsula during these years and has been linked to ENSO events in the Pacific, via the Indian Ocean (Shannon, Crawford, & Duffy, 1984a; Walker, Taunton-Clark, & Pugh, 1984). Agenbag (1996) investigated the influence of ENSO events between 1982 and 1993 on local winds and sea temperatures off Cape Agulhas and concluded that during ENSO years there was a reduction in the local summer easterly wind component, but could find no consistent relationship with local SST. 1.2. Satellite remote sensing of the South East Atlantic Ocean Satellite data have been used in many studies of the South East Atlantic Ocean, particularly in the Benguela region. Initially these focused on individual images or short sequences of images to show ‘snap- N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181–221 191 shot’ patterns. Shannon (1985a) presented an infra-red satellite image to show frontal features of the Benguela region in SST and Shannon et al. (1987) showed the occurrence of double fronts in SST in the Angola–Benguela frontal zone using sequences of infra-red satellite images. A few more extensive studies have looked at several years of satellite data. Lutjeharms and Meeuwis (1987) investigated the extent and variability of upwelling in the South East Atlantic between 1982 and 1985 using satellite derived SST and were able to identify eight distinct upwelling cells. Meeuwis and Lutjeharms (1990) used the same data set to investigate surface thermal characteristics of the Angola–Benguela front, showing it to be a permanent feature but with seasonal variation in its geographic location, width, seaward extent and temperature gradient. They were also able to describe eddy formation within the frontal zone. Lutjeharms and Stockton (1987) carried out a remote sensing study of mesoscale features in the Benguela upwelling area visible in SST and were able to describe upwelling filaments, plumes and frontal eddies, as well as warm filaments from the Agulhas Current. Agenbag (1992) used sequences of Advanced Very High Resolution Radiometer (AVHRR) images to compute surface advection velocities in the Agulhas Retroflection and southern Benguela regions. As well as SST, other variables have been investigated using satellite data. The volume of Shannon (1985b) describes several studies using Coastal Zone Color Scanner (CZCS) ocean colour data to investigate optical properties of upwelled waters and develop algorithms for detection of near-surface chlorophyll and suspended solids (Brown & Henry, 1985; Walters, 1985; Walters & Schumann, 1985). Shannon, Hutchings, Bailey, and Shelton (1984b) used CZCS data in combination with in situ measurements to describe the distribution of chlorophyll throughout the Benguela region and Shannon, Walters, and Mostert (1985) combined CZCS data with satellite SST measurements to describe near-surface chlorophyll during upwelling in the Southern Benguela. Lutjeharms and Walters (1985) also used CZCS data to investigate changes in chlorophyll associated with the Agulhas Current and Subtropical Convergence, and were able to link changes in chlorophyll to specific water masses. Weeks and Shillington (1996) combined CZCS and AVHRR data to determine the relationship between phytoplankton distribution and SST in the subtropical convergence region south of Africa and found that the Agulhas front appeared to play a role in limiting the spatial distribution of phytoplankton pigment. Matano et al. (1998) described seasonal variability of the Agulhas Retroflection using Topographic Experiment (TOPEX)/Poseidon altimeter data, highlighting differences between summer and winter patterns. Thomas et al. (2001) used data from the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) to describe the seasonality of chlorophyll cycles in these regions and compared productivity with Ekman transport from coincident satellite scatterometer data, showing them to be in phase. Carr (2002) used SeaWiFS data to compare productivity from different eastern boundary upwelling areas, showing the Benguela system to be the most productive system. Strub, Shillington, James, and Weeks (1998) combined GeoSat and TOPEX altimeter data with Levitus mean sea level and AVHRR SST data to describe seasonal surface circulation in the Southern Benguela system. They were able to show that north of 32° S, winds are upwelling favourable and currents equatorward all year round but strongest in summer, whereas from Cape Columbine to the Cape Peninsula winds and currents are more seasonal. They were able to relate seasonal changes in circulation to influxes of higher sea levels from the Agulhas retroflection region and the development of lower sea surface heights due to upwelling. Several studies have attempted to apply these data to wider applications. Cole and McGlade (1998) and Cole (1999) applied principal components analysis to several years of AVHRR SST data to investigate the potential effect of environmental conditions on sardine and anchovy recruitment in the Northern Benguela. In general, sardine recruitment showed a positive relationship with the number of SST events above 19 °C, while anchovy recruitment showed a negative relationship. Roy et al. (2001) investigated two anomalous events in the Southern Benguela (one warm and one cool) during the 1999–2000 summer season using high resolution AVHRR data and considered their impact on productivity of phytoplankton and anchovy. They concluded that these two events may have contributed to a record high level of anchovy 192 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181–221 recruitment in 2000. Weeks, Currie, and Bakun (2002) attempted to describe eruptions of hydrogen sulphide gas along the Namibian coast using SeaWiFS data, although these have not been validated.

2. Data and methods

2.1. Data

2.1.1. Sea surface temperature from satellite Eighteen years of SST data were obtained from the Envifish SST data product of the European Com- missions’ Joint Research Centre (JRC), produced from Global Area Coverage (GAC) data from the Advanced Very High Resolution Radiometer (AVHRR) sensors onboard the United States’ National Oce- anic and Atmospheric Administration (NOAA) satellites. This was undertaken as a scientific collaboration between the United States’ National Aeronautics and Space Administration (NASA) and the JRC. Pro- cessing of the images included cloud identification and masking following the method of Saunders and Kriebel (1988). For cloud free areas over the ocean, SST was calculated following the algorithm proposed by Castagne´, Borgne, Le Vourch, and Olry (1986). Finally, composite images were assembled at weekly and monthly intervals. For the time period 1982–1999, approximately 21,000 images were processed. Resol- ution of the composite images was kept at the original spatial resolution of 0.04° at the equator (approx. 4.5 km). They have a relative precision of approximately 0.1 °C and an absolute precision of approximately 0.5 °C(McClain, Pichel, & Walton, 1985).

2.1.2. Sea level from satellite The available 7 full years (1993–1999) of blended TOPEX/POSEIDON, ERS-1 and ERS-2 sea level difference data were obtained from the Archiving, Validation and Interpretation of Satellite Data in Ocean- ography (AVISO) Mapped Sea Level Anomaly (MSLA) product of Collecte Localisation Satellites (CLS) (Ducet, Le Traon, & Reverdin, 2000). These data have a spatial resolution of 0.25° at the equator, a temporal resolution of 10 days and 2 cm accuracy. Sea level data were corrected by AVISO for all geophysical, media, instrument and orbital effects as well as correlated noise (details of the processing can be found in Le Traon, Gaspar, Bouyssel, & Makhmara, 1995; Le Traon & Ogor, 1998 and Le Traon, Nadal, & Ducet, 1998). In this study, sea level differences were calculated relative to the mean sea level from January 1993 to December 1997, the base period used for calculation of the mean in the original MSLA data set.

2.1.3. Chlorophyll a from satellite The available three full years of chlorophyll a data (1998–2000) from SeaWiFS on board the OrbView- 2 satellite were obtained for the study region from the Goddard Earth Sciences Distributed Active Archive Centre. These data were produced by the NASA SeaWiFS project from ocean colour data using the OC4 third reprocessing algorithm, which is suitable for use in the coastal zone (O’Reilly et al., 2000), and are available as a standard mapped image level 3 product. These data have an accuracy of 35% but may still be contaminated by high sediment loadings in areas of large freshwater input. The data used were monthly mean composites with 9 km spatial resolution.

2.1.4. Surface winds from models Wind data were not available from remote sensing for a time series comparable to AVHRR SST data. Instead, 17 years of model data from the European Centre for Medium-Range Weather Forecasts (ECMWF) operational archive were used. The ECMWF model includes assimilated satellite SST and wind observations from radiosondes and other in situ sources (ECMWF, 1996). Weekly composites of zonal (u) and meridional (v) winds for the time period 1982–1999 were derived by ECMWF from daily winds (10 m N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 193 above the ocean at 1200 UTC), which were obtained by linear interpolation between the surface and the lowest model level at 1000 hPa (ECMWF, 1995). Spatial resolution varied throughout the time series, being 1.875° from August 1982 to April 1985, 1.175° from May 1985 to December 1991 and 0.5° from January 1992 to August 1999. Although many changes in the ECMWF model occurred throughout this period, the data used are sufﬁcient to show large-scale climatic changes throughout the time series (Trenberth, Large, & Olson, 1990).

2.2. Methods

2.2.1. Monthly maps To identify important oceanographic and meteorological features, and large-scale patterns of variability, maps showing the monthly mean spatial structure and overall standard deviation of surface variables in the South East Atlantic were constructed. Maps of mean values during each month of the year were calculated on a pixel-by-pixel basis for each variable (SST, surface winds, sea level, chlorophyll). For surface winds, mean vector wind speed and mean wind magnitude were calculated. Standard deviation maps were also calculated on a pixel-by-pixel basis for the whole available time series for each parameter. For mean and standard deviation maps of wind variables, only the period of highest resolution (0.5°) data was used (1992–1999). Austral seasons are used for seasonal descriptions in this paper.

2.2.2. LatitudeÐtime plots To show the temporal evolution of patterns along the South Atlantic coast of Africa, monthly mean values for the area within 50 km of the coast were plotted against latitude for the whole coast for wind, sea level and chlorophyll a. Monthly mean SST was calculated for a strip between 8 and 28 km from the coast. The use of a narrower strip for SST was due to the finer scale resolution of these data, allowing finer-scale coastal processes to be resolved. The 8 km (2 pixel) offset was used to remove any errors associated with coastline geolocation in the SST data product. Although the 8 km offset may result in the most intense upwelling inshore not being included, the seasonal coastal upwelling signal is still clearly captured. The coastal–offshore SST gradient index was calculated as the difference between the mean coastal SST and the mean SST in another 20 km wide strip 400 km offshore. Negative gradients indicate temperatures at the coast are cooler than those offshore and vice versa. For winds, the 50 km (one pixel) adjacent to the coast was used. To estimate offshore Ekman transport, we calculated the mean vector pseudo-windstress (tvp) for each month using the common approximation of the square of the monthly mean wind speed parallel to the coast, allowing for wind direction (see Eq. (1)). This approximation does not take into account the drag coefficient, which is non-linear for wind speeds less than 3 m sϪ1 (Large & Pond, 1981; Trenberth et al., 1990), however, this non-linearity is negligible compared to the variance of the data. Only pseudo-windstress leading to offshore Ekman transport (i.e. Ͼ0) is plotted Ͼ ϭ 2 if cos q 1:tvp (W cos q) ; (1) Ͻ ϭϪ 2 if cos q 1:tvp (W cos q) where W is the scalar wind speed and q is the angular difference between the wind direction and the orientation of the coast. Coastline angles were determined by visually fitting a line to the dominant trend of the coast over 0.5° latitude intervals. A wind mixing index was calculated as the mean of the cube of weekly scalar wind speed adjacent to the coast. This is proportional to the rate at which turbulent kinetic energy of the wind is added to the upper ocean and becomes available to mix the stable thermocline layers (Niller & Kraus, 1977). Weekly scalar wind speeds were cubed before averaging to account for the cumulative effect of short-scale (weekly) events. Unfortunately, events on scales shorter than 1 week are not resolved in this data set. It is important 194 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 to note that this index only estimates the wind component of mixing and does not take into account mixed- layer depth or bottom depth. This index is, however, comparable with the wind speed cubed results of Parrish et al. (1983).

2.2.3. Annual sea surface temperature anomaly maps Maps of annual anomalies were calculated for SST on a pixel-by-pixel basis. First the mean SST of each pixel for each year was calculated. Then the mean SST of each pixel for the whole time series was subtracted from the annual mean. These maps were only calculated for SST as this was the longest time series available to investigate interannual spatial variability ϭ Ϫ xijyear ¯rijyear ¯rijtime series (2) where xijyear is the anomaly at pixel (i, j) for a particular year, r¯ijyear is the mean at pixel (i, j) for a particular year and r¯ijtime series is the mean at pixel (i, j) for the whole time series.

2.2.4. Time series of anomalies Anomalies of mean values for SST, surface winds and sea level were calculated for the areas shown in Fig. 2 to show unsmoothed interannual variability. Chlorophyll data were not used in this study because the 3-year time series was considered too short. Anomalies were calculated by subtracting the climatic mean for each time step in a year from each time step in the time series (see Eq. (3)) ϭ Ϫ xjweek rj ¯rweek (3) where xjweek is the anomaly at time step j with respect to the weekly mean, rj is the value at time step j and r¯week is the mean value for the particular climatic week (1–48) corresponding to time step j.

2.2.5. Spectral analysis To highlight cycles in SST and wind anomalies, power spectra were calculated. Anomalies were ﬁrst corrected for linear trend before power spectra were generated using a Fast Fourier Transform (FFT) and smoothed using a Tukey window with three degrees of freedom. Power spectra are plotted against frequency up to the Nyquist frequency to show the distribution of interannual peaks relative to background noise, and against period in the range 1.8–9 years to emphasise interannual peaks.

Fig. 2. Map of the areas used in calculation of anomalies. N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 195

3. Results and discussion

3.1. General overview

The main large-scale oceanographic and atmospheric features of the South East Atlantic are clearly identiﬁable in remote sensing and model data. For the winds (Fig. 3) these consist of the southeasterly

Fig. 3. Monthly mean and overall standard deviation maps of surface winds (msϪ1). 196 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 trade winds, the mid-latitude westerlies (south of ~35° S) and the southwesterly monsoon (Gabon and Angola coast). Trade winds and westerlies migrate meridionally on a seasonal basis, being further north in winter and further south in summer related to the seasonal movement of the ITCZ. Areas of water with different temperature signals can be identiﬁed at the surface from SST data (Fig. 4) and correspond to the following well-known water masses: tropical surface water, both north of the Angola–Benguela front and

Fig. 4. Monthly mean and overall standard deviation maps of sea surface temperature (°C). N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 197 in the Agulhas Current to the south; subtropical surface water south of the Angola–Benguela front in the offshore region; upwelled central water (South Atlantic and Indian Ocean) mixed with surface water types in the coastal Benguela upwelling zone; and subantarctic surface water south of the subtropical front (Garzoli et al., 1996). Areas of exceptionally warm surface temperatures and higher sea levels (Fig. 5) are noted in areas where

Fig. 5. Monthly mean and overall standard deviation maps of sea level differences (mm). 198 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 near-surface currents convey tropical water from low to high latitudes, such as the Angola Current, or feed relatively warm mixed water types from the Agulhas Current towards the southern upwelling boundary. Transition zones between the cold upwelling water and these warm waters maintain frontal zones with a high variability in space and in time. These regions, therefore, show a high degree of variance. Particularly high variance in SST and sea level extends to the northwest of the Agulhas Current, an area dominated by Agulhas eddies. The permanent wind-driven parts of the Benguela Current and South Equatorial Current cannot be distinguished by SST or sea level changes sufficiently. However, they can be identified by the large central area of lower sea level variance that extends from the Benguela coast to the northwest. A classical seasonal cycle is generally seen in offshore SSTs, with temperatures warmest in summer and coolest in winter. Some coastal areas also reflect this classical seasonal cycle, however, others are more complex and these are discussed in more detail below. These seasonal patterns are also reflected in sea level changes due to the mass-deficit resulting from thermal expansion of the water column and Ekman offshore transport. The Angola–Benguela frontal zone is clearly seen in both the SST maps (Fig. 4) and the latitude–time plot of SST (Fig. 7a) from 15°–17° S, the same position as reported by Shannon et al. (1987). The gradient associated with this front is slightly weaker than that observed by these authors (2–3 °C per 1° latitude rather than 4 °C per 1° latitude), probably because our methods were based on climatological mean values.

3.2. Seasonality

3.2.1. Angola Current and Angola coastal upwelling Both the coastal–offshore SST gradient (Fig. 7b) and the sea level difference (Fig. 7c) latitude–time plots show a bimodal seasonality with two warmer, high sea level seasons and two cooler, low sea level seasons. The signal is only present north of the Angola–Benguela frontal zone in SST. In sea level, however, it occurs along the entire coast from the equator to 30° S, although it remains stronger north of the Angola– Benguela front. The first high sea level season (February–April) appears to last slightly longer and to impact the Benguela region more than the second high season (October–November). This may be influenced by summer warming, but it also corresponds to the observed late February–March intrusion of the Angola Current into the Benguela region. The second high season, however, appears to have a slightly stronger signal in the coastal–offshore SST gradient, probably enhanced by cooler offshore temperatures at this time of year. The main low sea level season (July–September) corresponds to equatorial upwelling, but any phase shift indicating propagation southward along the coast is not detectable in monthly averages. There is also a minor low, cool season in December and January. This bimodality along the Angolan coast has previously been noted by Verstraete (1992), who described two warm, high sea level seasons in February–March and October–November and two cool, low sea level seasons in December–January and June–July. Bimodality has also previously been described in the equatorial undercurrent (Hagen & Schemainda, 1983). More recently, Pingree, Kuo, and Garcia-Soto (2002) identified this pattern in satellite sea level data in the equatorial region. The bimodal pattern matches that observed north of the equator, in the Gulf of Guinea, where there are also two cool, upwelling seasons interspersed with two warm seasons (Binet & Marchal, 1993; Roy, 1995; Hardman-Mountford & McGlade, 2002). In the Gulf of Guinea, the major upwelling season from July to September is attributable to remote forcing from the western equatorial Atlantic. An abrupt northerly shift in the ITCZ and associated intensification of the southeasterly trade winds, followed by relaxed southeast trade winds, sets up a baroclinic, equatorial Kelvin wave. It needs about 2 months to travel from west to east along the equatorial wave- guide and then becomes trapped at the African west coast, travelling poleward both north and south of the equator. In the wake of the northward travelling pulse, the thermocline shoals to or near to the surface along the central Gulf of Guinea coast allowing upwelling to take place, possibly through interaction with N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 199

Fig. 6. Monthly mean and overall standard deviation maps of surface chlorophyll a (mg mϪ3). other features such as the strong, eastward Guinea Current and the inﬂuence of an enhanced southwest monsoon. A similar explanation has been given for the upwelling seen along the African coast just south of the equator, although this is less well documented. Support for this mechanism along the Angolan coast comes from the observations by Picaut (1983) of a poleward propagating upwelling signal along the coast from 1° to 13° S, commencing at the same time and propagating at the same phase speed as the upwelling signal north of the equator (~0.7 m sϪ1). Although 200 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221

Fig. 7. Latitude by month plot of (a) coastal sea surface temperature (°C); (b) coastal–offshore sea surface temperature gradient (°C 400 kmϪ1); (c) coastal sea level differences (mm); (d) vector pseudo-windstress parallel to the coast (m2 sϪ2); (e) coastal wind mixing (m3 sϪ3), and (f) coastal surface chlorophyll a (mg mϪ3). The width of the coastal strip considered in these plots is (a) 20 km, (c)– (f) 50 km. In (b), coastal–offshore sea surface temperature gradient is calculated as the SST difference between a 20 km strip at the coast and a 20 km strip situated 400 km offshore. N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 201 less is known about the minor cool season, the bimodal signal seen in the western equatorial Atlantic (Pingree et al., 2002) suggests that remote forcing is probably responsible for this upwelling as well, both north and south of the equator. Subsequent relaxation of the westward component of southeast trade winds produces a downwelling signal along the equator as warm water flows back eastwards, leading to warm seasons and a deepening thermocline in the eastern equatorial Atlantic and along the African coast. Our results show that the bimodal seasonality of this area is also reflected in chlorophyll (Figs. 6 and 7f). During the major cool, low sea level season, chlorophyll levels are elevated along almost the entire coast north of the Angola–Benguela front. During the minor cool, low sea level season, however, elevated chlorophyll values are only seen from the equator southward to about 9° S. Despite the disruption of the bimodal signal by year-round high chlorophyll due to the Congo River discharge around 5° S, it is noteworthy that during both these seasons the signal extends south of this feature. Outside of these high chlorophyll seasons, and to the south of the high chlorophyll signal during the minor cool season, chlorophyll levels are low. This is consistent with observations of Angola Current intrusions into the Northern Benguela region being associated with low productivity (Thomas et al., 2001). Bimodality in the seasonal chlorophyll cycle along this coast, at the equator and along the Gulf of Guinea coast between Accra (Ghana) and Abidjan (Coˆte d’Ivoire) was also identified by Pingree et al. (2002) using Fourier analysis. They observed the dominant peak in chlorophyll in July–August and the secondary peak occurred between December and February. Peak chlorophyll lagged the minimum in sea level (representing upwelling) by 10 days, consistent with the lag time between nutrient enrichment of the photic zone and the start of phytoplankton production.

3.2.2. Congo River plumes Outflow from the Congo River is not detectable from SST or sea level data. In chlorophyll, however, the mouth of the Congo River shows a strong signal all year round (Fig. 7f) and large plumes extend offshore (Fig. 6). These chlorophyll signals are probably contaminated by high loadings of suspended particulate matter and dissolved organic substances. Although the plume might locally enhance chlorophyll a levels, the extent to which its turbidity affects the determination of chlorophyll a from satellite is still unknown. Nonetheless, these ocean colour signals highlight real oceanographic features. A puzzling feature of the plumes is their bimodal seasonal signal. Although this is observed in other oceanographic features along this coast, it is not seen in Congo River flow data from Brazzaville in the Democratic Republic of Congo, near to the mouth of the river (G. Mahe´, IRD, Ouagadougou, Burkina Faso, pers. comm.). Thus, this biannual signal is not likely to be related to seasonal flooding of the river and is more likely related to the biannual equatorial signal. This biannual plume was also observed by Pingree et al. (2002) and they attributed it to maximum westward flow in the South Equatorial Current drawing out the Congo Plume source region.

3.2.3. Benguela upwelling The correspondence between the wind-field and upwelling is clear in both space and time from this study. Peaks in windstress are clearly identified around Cape Frio, Lu¨deritz, Namaqualand/Hondeklip Bay and Cape Columbine/Cape Peninsula (Figs. 3 and 7d) and these locations correspond to the major upwelling cells identified from SST (Figs. 4 and 7a,b). The seasonal meridional migration of the trade winds relates to changes in the position of greatest pseudo-windstress along the Benguela coast. Strongest pseudo-windstress and upwelling are seen nearly all year round near Lu¨deritz. Upwelling occurs during summer in the seasonal cells of the Southern Benguela, coinciding with the trade winds being furthest south and pseudo- windstress in the region being strongest. These results are in agreement with those from other studies of the region (e.g. Parrish et al., 1983; Shannon, 1985a). Our results show that autumn (February–March) intrusion of warm water from the Angola Current causes a southward migration of the Angola–Benguela front into the most northern part of the Benguela region and warms water along the northern Namibian coast (Fig. 7a). This has been noted previously by several 202 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 authors (Hart & Currie, 1960; Stander, 1964; O’Toole, 1980; Badenhorst & Boyd, 1980; Le Clus & Kruger, 1982; Boyd, 1983; Kruger & Boyd, 1984; Boyd et al., 1987; Salat, Maso´, & Boyd, 1992). Intrusion of warm Angola Current water into the Northern Benguela during autumn deepens the thermocline (Salat et al., 1992), as reflected in the sea level signal, so cool water may not be brought to the surface even though upwelling may be taking place. This sea level signal is seen to extend further south than the surface temperature signal. Salat et al. (1992) showed upwelling to be coming from a subsurface core of Angola Current water between 50 and 100 m depth off Walvis Bay, further south than Angola Current water was observed at the surface. Between 16° and 22° S, wind forcing is bimodal with the first peak in October–November and the secondary peak in March–April (Fig. 7d). This was also noted by Shannon and Nelson (1996). The autumn peak in the wind field occurs just after the February–March Angola Current intrusion, whereas the spring peak occurs at the same time as the October–November warming of the Angola Current. A second warm intrusion of Angola Current water into the Benguela region is not seen in mean coastal SST (Fig. 7a) but is seen in sea level (Fig. 7c). A possible explanation is that strong equatorward windstress prevents the intrusion of Angola Current water into the Northern Benguela, but does not prevent the thermocline downwelling signal seen in sea level data, which mainly depends on remotely forced dynamics. Chlorophyll biomass is, in general, much greater in the Benguela region than along the Angolan coast (Figs. 6 and 7f). It is greatest during the summer upwelling season in the Southern Benguela, but in winter along the central Namibian coast. Wind mixing (Fig. 7e) is weakest in winter throughout the Benguela, except north of 20° S, due to the bimodal signal in this region. At Lu¨deritz, wind mixing is strong for most of the year and chlorophyll biomass is low all year, but it is enhanced downstream of this upwelling cell. Similarly, productivity is reduced during peak upwelling in the Namaqua and Cape Frio cells but elevated chlorophyll levels are observed downstream. This low chlorophyll biomass in areas of very strong wind mixing is probably because the wind mixes the phytoplankton out of the well-lit surface waters, thereby limiting photosynthesis despite the enhanced nutrient levels. Downstream from the upwelling cells, the enhanced nutrient concentrations and reduced wind mixing result in elevated chlorophyll concentrations. It is noteworthy that between the Lu¨deritz and Namaqua cells, by the mouth of the Orange River, productivity is high all year round despite periods of strong wind mixing. One factor that might explain why this area can maintain high chlorophyll levels despite strong wind mixing is that it has a relatively wide continental shelf, whereas the major upwelling cells occur over narrow shelf areas leading to greater offshore losses. Additionally, freshwater input from the Orange River may enhance stratification and, therefore, productivity in the area.

3.2.4. Agulhas Current and Western Agulhas Bank Over the Agulhas Bank, the dominant winds are the mid-latitude westerlies and these have their strongest inﬂuence on the African coastline in winter when they are furthest north (Fig. 3). During the autumn, cool waters appear on the western Agulhas Bank and by winter, these cool waters are present over the whole Agulhas Bank (Fig. 4). The cooling has been shown to be a result of deep mixing over the western Agulhas shelf (Boyd & Shillington, 1994) and is accompanied by moderately high chlorophyll biomass (Fig. 6). The cool water persists until November and then begins to warm, reaching a maximum in February. Summer winds are weak but upwelling favourable southeasterlies. The persistent nature of the Agulhas Current is reﬂected in the low temperature variability of its core. However, variations in its width and in the position of the Agulhas Return Current are noticeable as areas of highly variable SST (Fig. 4). Although the Agulhas Current is generally forced from the Indian Ocean (Gordon, 1985), it appears widest in summer, corresponding to the mid-latitude westerly winds being furthest south and, therefore, weakest in the region. The Current is not discernible in the monthly sea level maps (Fig. 5), but has a strong sea level variance signal associated with the large number of eddies in the current which pinch-off into the Atlantic. Chlorophyll data vaguely show the Agulhas Current as a low N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 203 productivity strip, surrounded by higher productivity zones over the shelf and in the subtropical front (Fig. 6). This is consistent with the current being mainly composed of low nutrient, tropical water from the Indian Ocean.

3.3. Interannual variability

3.3.1. Equatorial forcing and Atlantic Nin˜o-type events SST anomaly maps (Fig. 8) and time series of SST anomalies from 1982 to 1999 (Fig. 9a) show a generally increasing trend with three different periods: 1982–1986 was generally cool; 1987–1994 was intermediate and 1995–1999 was generally warm. The increasing trend is strongest at the equator with a decreasing slope further south (Table 1). Superimposed on these trends were notable anomalous warm and cool events (Table 2). In the equatorial and tropical region the extreme warm events of 1984 and 1995 were clearly observable and other warm events occurred in 1991, 1993, 1996 and 1998–1999, while cool events happened in 1982, 1983, 1991–1992 and 1997. In the Northern Benguela, some of these events were also observed: the extreme warm years of 1984 and 1995, other warm events in 1999, and cool events in 1982, 1983 and 1997. Furthermore, some of the weaker warm and cool anomalies from the equatorial area and Angolan coast were seen as strong events in the Northern Benguela, notably the warm event of 1988 and the cool events of 1987 and 1994. During the 1984, 1995, 1996 and 1999 warm events, warm SST anomalies stretched as far south as the Southern Benguela. Although satellite-derived sea level anomalies (Fig. 9d) are available for a shorter time period than SST anomalies, over the common time period they show the same increasing trend and anomalous events as the SST anomalies throughout the region, especially the March 1995 warm event. The relationship between SST and sea level was particularly strong along the Angolan Coast (r2 = 0.57, p Ͻ 0.0001, n = 87). Zonal winds at the equator also show an increasing trend in westerly wind strength (Fig. 9b, Table 1), with a switch from easterlies to westerlies from 1989 onwards (Fig. 10a), although from 1997 there is a

Fig. 8. Maps of annual mean sea surface temperature anomalies (°C). 204 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221

Fig. 9. Time series (grey line) of (a) sea surface temperature anomalies (°C); (b) surface zonal (u) wind anomalies (msϪ1); (c) surface meridional (v) wind anomalies (msϪ1), and (d) sea level anomalies (mm). The anomalies for each time series were smoothed using an annual running mean ﬁlter to emphasise interannual-scale variability without any modiﬁcations in given phase situations (black line). N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 205

Fig. 9. Continued 206 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221

Fig. 9. Continued N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 207

Fig. 9. Continued 208 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221

Table 1 Slopes of weekly SST, 10-daily sea level, and weekly u- and v-wind components (reported per decade)

Area SST (°C per decade) Sea level (mm per u-Wind (msϪ1 per v-Wind (msϪ1 per decade) decade) decade)

Equatorial 1.29∗∗∗∗ 46.58∗∗∗∗ 0.88∗∗∗∗ 0.58∗∗∗∗ Angolan Coast 1.08∗∗∗∗ 39.01∗∗∗∗ 1.11∗∗∗∗ Ϫ0.75∗∗∗∗ Northern Benguela 0.62∗∗∗∗ 33.81∗∗∗∗ 0.20∗∗∗ 0.24∗ Southern Benguela 0.35∗∗∗∗ 16.60∗∗ Ϫ0.04n.s. 0.22n.s. Agulhas 0.59∗∗∗∗ N/A Ϫ0.41∗ 0.39∗∗∗∗

∗ p Ͻ 0.05. ∗∗ p Ͻ 0.01. ∗∗∗ p Ͻ 0.001. ∗∗∗∗ p Ͻ 0.0001. n.s. Not significant. steep decline in westerly wind strength. Zonal winds in the Angolan coastal area show a similar long-term pattern, but are generally more westerly, with the shift from easterlies to westerlies occurring in 1983 and the declining phase starting earlier in 1996. Equatorward winds (Figs. 9c and 10b) show a steadily increasing trend at the equator (Table 1), where they were in a high phase during 1983, 1987, 1990–1994 and 1996–1997. These years correspond to ENSO events in the Pacific, which is consistent with the theory of a teleconnection between the Pacific and Atlantic Oceans that leads to warm ENSO events in the Pacific being related to a relaxation in the northeasterly trade winds over the North Atlantic Ocean (Roy & Reason, 2001). A relaxation in trade winds over the North Atlantic would be expected to lead to a shift in the position of the ITCZ and an intensification of southeasterly trade winds over the equatorial South Atlantic. Equatorward winds were in a low phase during 1984, 1988–1989, 1992, 1995 and 1999. These weak winds occurred during the same seasons as warm intrusions to the Northern Benguela in 1984, 1995 and 1999, but not during other years. Along the Angolan coast there was no notable linear trend up until 1992. The winds then weakened markedly until 1997 before increasing again slightly until 1999. These winds were so weak during the summer of 1996/1997 that they reversed. Spectral analysis (Fig. 11) showed the major interannual oscillations to have periods between 2 and 6 years. The approximate 2 year period was seen in the SST signal from all areas and both wind components at the equator. It probably corresponds to the QBO, noted in many environmental series (Naujokat, 1986). Three to five year cycles were observed in SST and both components of wind for the equatorial and Angolan areas, and the zonal wind component for the Northern Benguela. These cycles are of similar frequency to ENSO events in the tropical Pacific. Oscillations with decadal frequency, comparable with Benguela Ninõ events, were not detectable from these relatively short series. Warm events in the equatorial and south-eastern Atlantic have been noted by several authors (Table 2). The extreme warm event of 1984 was well documented in the eastern equatorial Atlantic and many parallels were drawn with PacificElNinõ events (Philander, 1986; Hisard, 1986), partly prompted by the proximity of the very strong 1983 El Ninõ event in the Pacific and possible teleconnections between the two basins. This event was also documented along the coasts north and south of the equator, in the Gulf of Guinea and along the coast of Angola (Verstraete, 1992). Other warm events in the Gulf of Guinea have been noted in 1953–1954, 1962–1963, 1973 and 1987–1988 (Koranteng, 1998). In the Benguela region, Benguela Ninõs were identified in 1934, 1949, 1963, 1984 (Shannon et al., 1986) and 1995 (Gammelsrød et al., 1998). From a time series of coastal sea levels from 1959 to 1985, Brundrit et al. (1987) observed high sea level events in 1963, 1974 and 1984 along the entire South Atlantic coast of Africa, from the equatorial N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 209

Table 2 Comparison of anomalous warm and cool events in the South East Atlantic from literature sources. Italics indicates warmer years and bold indicates cooler years. When both warm and cool events have been recorded for the same year, a top-bottom division signiﬁes seasonal differentiation with (1) representing the ﬁrst half of the year and (2) the second half, and a left-right division, marked &/or, indicates that no seasonal differentiation is known

Year (January) S. Benguela N. Benguela Angolan coast Gulf of Guinea

1910 Warm?i Warmi …… … … … Mid-1920s Warm?i Warmi …… … … … 1934 Warma,i Warma,i …… … … … 1949 Warm?i Warmi 1950 1951 1952 1953 Warma Warmb 1954 Warma Warmb 1955 Coola Coolb 1956 Coola Coolb 1957 Warma 1958 Coola Coola 1959 Warma 1960 Warma 1961 Warma 1962 V. warma Warmb 1963 V. warma,g,i V. warma,g,i Warmb,c 1964 Coolb–d 1965 Warmd(1) Coold Coolb,d Coold(2) 1966 1967 Coola Coold Coolc,d 1968 Warmd Coold Warmc&/or Coold 1969 Coolg Coolg 1970 Coola,g Coolg 1971 Coola,g Warmd(1) Coold,g(2) 1972 Warma,g,i Coold(1) Coold Coold Warma,d,g,i(2) 1973 Slightly warma,g,i Warma,d,g,i Warmb,c 1974 Warma,i Warma,d,i 1975 Coolb 1976 Warma Warma Coolb–d 1977 Warma Coolb,c 1978 Coola Coold Coold 1979 Coola Coold Warmc 1980 Coold Coold 1981 Coold Coold Warmc 1982 V. colda,g,h(1) Coold,g,k Coold,k Coole Warmj,l(2) 1983 V. colda,h(1) Coold Coold,k Coolb,c,e Warmj,l(2) 1984 V. warma,g–i,k V. warma,d,g,i,k V. warmd,k V. warmb,e (continued on next page) 210 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221

Table 2 (continued)

Year (January) S. Benguela N. Benguela Angolan coast Gulf of Guinea

1985 Coold Coolb,e 1986 Warmh,m Coold Coolb,c,e 1987 Coolh(1) Coold Warmd V. warmb–e Warmh(2) 1988 Warmk Warmd,k Warmb–e 1989 Warmh(1) Warmd Warme Coolh(2) 1990 Coolh 1991 Coolh Warmd(1) Warmd,k(1) Coolk(2) 1992 Coold Coold,k 1993 Warmk(2) 1994 Coold 1995 Warmk Warmf,k Warmd,f,k Warmd 1996 Warmd,k(1) Warmd,k Coold(2) 1997 Coold(1) Coolk(1) Warmd(2) Warmk(2) 1998 Warmd Warmd,k Warmd 1999 Warmk Warmd,k Warmd,k Warmd

a Shannon (1985a). b Koranteng (1998). c Koranteng and McGlade (2001). d Binet et al. (2001). e Hardman-Mountford and McGlade (2002). f Gammelsrød et al. (1998). g Walker (1987). h Shannon et al. (1992). i Shannon and Taunton-Clark (1989). j Walker et al. (1984). k This study. l The warm event in 1982/1983 was small scale and confined around the Cape Peninsula. m The 1986 warm event was due to intrusion of Agulhas Current water into the Southern Benguela. area to Cape Town. Taunton-Clark and Shannon (1988) deduced that, over an 80 year period, Benguela Ninõs had occurred with a near-decadal periodicity. The 1984 Ninõ-type event in the eastern equatorial Atlantic was characterised by an unusually warm upper ocean, a deeper than usual thermocline, a substantial reduction in the surface trade winds and a southward displacement of the ITCZ (Horel, Kousky, & Kagano, 1986; Katz et al., 1986; Philander, 1986). The displacement of the ITCZ and reduction in the trade winds were linked to heavy rainfall in Angola and Namibia as well as in north-eastern Brazil. Warm water was transported to the African coast via an unusual eastward current just south of the equator (Hisard, Heńin, Houghton, Piton, & Rual, 1986; Philan- der, 1986), identified as the South Equatorial Counter-Current (SECC) (Hisard, 1986) and by a downwelling signal in the Equatorial Under Current (EUC) (Verstraete, 1992). Furthermore, these events were tied to the normal seasonal cycle (Weisberg & Colin, 1986). All of these features are consistent with descriptions of El Ninõ events in the Pacific(Hisard, 1986). In the Benguela region, the event was seen in March/April 1984 as a single top-to-bottom intrusion of warm, saline water which inhibited upwelling along the Northern N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 211

Fig. 10. Time series of annual running means comparing interannual trends in (a) zonal (u) and (b) meridional (v) surface winds.

Benguela coast (Philander, 1986; Boyd et al., 1987). For a detailed description of this event see Boyd et al. (1985), Le Clus (1986) and Shannon et al. (1986). Gammelsrød et al. (1998) described the March 1995 event along the Angolan and Namibian coasts as having positive temperature anomalies up to 8 °C subsurface and a deepening of the thermocline by 10– 30 m. The warm water covered the shelf from at least 5°–24° S, however, there is evidence of warmer temperatures as far south as Lu¨deritz (27° S) and satellite data showed them to extend offshore more than 300 km. The water off the Northern Benguela was saline, but along the Angolan coast it was fresh, probably due to the influence of the Congo River. Coastal upwelling was absent during this event. Similar conditions have been described for the 1934 and 1963 events (Walter, 1937; Stander & de Decker, 1969). The apparent decadal-scale period of these putative Ninõ-type events in the Benguela region is much longer than the period of ENSO events observed in the Pacific. It is also clear from this and other studies that Atlantic warm events can occur more frequently than this, especially but not exclusively in the equatorial region (Table 2), although it is unclear whether these are of Ninõ type. In the Northern Benguela, a non-Ninõ-type warm event occurred in 1988, according to Shannon and Agenbag (1990). These authors tracked the anomalies associated with this event and deduced that it could not be classed as a Benguela Ninõ because its origin, extent and duration were very different from those of the 1984 event. While we agree with the assessment of these authors that this event does not warrant the classification Benguela Ninõ, closer inspection of their results shows the warming not to be so different from the 1984 event. The extent, duration and intensity were less than in 1984, but the major source of the anomalies in the western 212 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221

Fig. 11. Power spectra for time series of sea surface temperature anomalies, surface zonal (u) wind anomalies, and surface meridional (v) wind anomalies. Spectra are smoothed using a Tukey window with three degrees of freedom. (a) Power spectra up to the Nyquist frequency, plotted against frequency; (b) power spectra within the period range 1.8–9 years, plotted against period. N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 213

Fig. 11. Continued 214 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 central Atlantic appears similar, just displaced to the south by 5°–10°. This displacement could be due to changes in the position of the ITCZ. Furthermore, the anomaly still appeared along the Angolan coast and in the Northern Benguela during February–March, tied to the seasonal cycle of Angola Current intrusion, and the results of our study show it to have spread along the coast from the eastern equatorial Atlantic. A major difference seen in our study between the Benguela Ninõ events in 1984 and 1995 and the unusual warming in 1988 is the status of southerly winds at the equator. During Benguela Ninõ years, extended periods of weak southerly winds occur in the equatorial area near the beginning of the year, a season of generally weaker winds, and these weak anomalies persist for several months. During 1988, weak equatorward winds at the beginning of the year did not persist (i.e. were not seen in the anomalies). In contrast, the coincidence in 1999 of strong warm anomalies and higher-than-usual sea levels spreading from the equator to the Benguela with extended weak southerly winds at the equator suggests this may have been a Benguela Ninõ year. It appears, therefore, that an interannual-scale quasi-cycle of warm events occurs in the equatorial Atlan- tic, of similar frequency to the PacificElNinõ phenomenon. This was also noted for the Gulf of Guinea region by Hardman-Mountford and McGlade (2002), although with too short a time series to be conclusive. Some exceptional events manage to penetrate strongly into the Benguela region. These appear to arise from a coincidence of anomalously warm water along the equator and in the SECC spreading along the African coast and a persistent reduction in the trade winds at the equator.

3.3.2. Agulhas warming and warm water intrusions to the Southern Benguela In the Southern Benguela and over the Agulhas Bank, variability is not dominated by interannual scale oscillations, as in the equatorial and Angolan coastal areas. Instead, high frequency, within-year variability appears more important (Figs. 9 and 11). However, a few noticeable SST events have been seen. As well as Benguela Ninõ and other northerly warm events sometimes becoming noticeable in the Southern Benguela, occasional strong intrusions of warm Agulhas water occur (Fig. 8). These are seen in 1986, 1992 and perhaps 1984, 1987, 1997, 1998 and 1999, although the strong warm anomalies of different sources during some of these years make interpretation difficult. Over the Agulhas Bank, the strongest warm anomalies occurred in 1992 and from 1997 to 1999 there was a strong warming trend (Fig. 9a). In the Benguela easterlies and the Agulhas westerlies, trend is negligible but oscillations are generally in phase, except for in 1991 (Fig. 10a). From 1992, the Agulhas westerlies weaken and the Northern Benguela easterlies strengthen. The large 1991 event saw westerly winds increase to a maximum over the Agulhas Bank while easterly winds in the Northern Benguela also increased to their strongest, i.e. an out- of-phase event. Over the Agulhas Bank, there was a strong correspondence between SST and westerly wind annual running means up until 1993, with both showing an increasing trend (r2 = 0.34, p Ͻ 0.0001, n = 505). After this, however, there appears to have been a shift to a strong negative correlation with westerly winds decreasing while SST continued to increase (r2 = 0.59, p Ͻ 0.0001, n = 264). Further- more, years of increases in westerly wind anomalies over the Agulhas Bank (Fig. 9b) appear to correspond to the years of major intrusions of Agulhas water into the Southern Benguela. The best known of these is the Agulhas intrusion during the winter of 1986, which has been described in detail by Shannon et al. (1990). This event penetrated as far north as Lu¨deritz and appears to have been triggered by anomalies in the distribution of windstress over a large part of the Indian Ocean and over the Agulhas retroflection area (Shannon & Agenbag, 1990; Shannon et al., 1990). Southerly winds in the Benguela region (Fig. 9c) showed little trend up until the decline of 1996 from which they increased until 1999. Peaks were seen in Benguela southerly winds in 1995 and 1999 and the largest trough occurred in 1996. Additionally, there was a peak during 1996 in the Northern Benguela. These anomalous years show no consistent relationship with anomalous SST and sea level events in the region. Between 1996 and 1999 an increasing trend in SST was observed over the Agulhas Bank, although N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 215 southerly winds in this area did not correspond to the pattern of warming but increased steadily throughout the series.

4. Conclusions

Many patterns of seasonal and interannual variability in the South East Atlantic have been described and discussed in this paper. The results show climate forcing of the coastal environment of the South East Atlantic to propagate from both the north and south of the region on both seasonal and interannual time scales. Equatorial signals propagate poleward along the Angolan coast while the equatorward trade winds and events in the Agulhas region dominate the Agulhas Bank and Southern Benguela. The Northern Benguela is a mixed regime, under the inﬂuence of forcing from both directions. The following conclusions can be drawn from this study:

1. In the equatorial and tropical regions, north of the Angola–Benguela front, the seasonal forcing has a bimodal signal that propagates as upwelling and downwelling signals along the equator and then poleward at the coast. This bimodal seasonal cycle is seen in surface temperatures, sea level and chlorophyll a. High productivity is associated with upwelling seasons and low productivity with downwelling seasons. Seasonality in the Northern Benguela is mainly forced by the southeasterly trade winds but is also impacted by seasonal intrusions of warm water in the Angola Current. 2. Years with extended periods of anomalously weak southerly winds at the equator during the first few months of the year correspond to years of very strong warm intrusions of warm water into the Northern Benguela. This pattern was observed during the known Benguela Ninõ years of 1984 and 1995, and also in 1999. However, the anomalous intrusion of 1988 was not accompanied by these weak winds. Thus, this study agrees with the findings of Shannon and Agenbag (1990) that 1988 was not a Benguela Ninõ year, although it may have been of Ninõ type in the tropical Atlantic. Our results also suggest that a Benguela Ninõ event may have occurred in 1999. 3. The frequency of Benguela Ninõ events has previously been found to be much less than that of Pacific El Ninõ events. However, consideration of warm events in the equatorial Atlantic from this and other studies shows a much higher frequency of warm events, comparable to that of PacificElNinõs. Thus, it appears that Ninõ-type events may occur on a similar frequency in the tropical Atlantic to the tropical Pacific, but penetrate into the Benguela region more rarely, requiring a coincidence with anomalous southerly winds at the equator for this condition to arise. Furthermore, years of anomalously strong southerly winds at the equator correspond to years of ENSO events in the Pacific.

Acknowledgements

This research formed part of the EU-funded Enviﬁsh project (contract number IC18-CT98-0329) and this programme supported Dr N.J. Hardman-Mountford and Dr A.J. Richardson. The altimeter products used have been produced by the CLS Space Oceanography Division as part of the European Union’s Environment and Climate project AGORA (ENV4-CT9560113) and DUACS (ENV4-CT96-0357) with ﬁnancial support from the CEO programme (Centre for Earth Observation) and Midi-Pyrenees regional council. CD-ROMs are produced by the AVISO/Altimetry operations centre. Ocean colour data used in this study were produced by the SeaWiFS Project at Goddard Space Flight Center. The data were obtained from Goddard Earth Sciences Distributed Active Archive Center under the auspices of the National Aero- nautics and Space Administration. Use of this data is in accord with the SeaWiFS Research Data Use Terms and Conditions Agreement. 216 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221

References

Adamec, D., & O’Brien, J. J. (1978). The seasonal upwelling in the Gulf of Guinea due to remote forcing. Journal of Physical Oceanography, 8, 1050–1060. Agenbag, J. J. (1992). A procedure for the computation of sea surface advection velocities from satellite thermal band imagery, with applications to the South East Atlantic Ocean. Doctoral Thesis, University of Cape Town, Cape Town, South Africa, unpublished. Agenbag, J. J. (1996). Pacific ENSO events reflected in meteorological and oceanographic perturbations in the southern Benguela system. South African Journal of Science, 92, 243–247. Agenbag, J. J., & Shannon, L. V. (1988). A suggested physical explanation for the existence of a biological boundary at 24° 30Ј S in the Benguela system. South African Journal of Marine Science, 6, 119–132. Andrews, W. R. H., & Hutchings, L. (1980). Upwelling in the Southern Benguela Current. Progress in Oceanography, 9,1–81. Badenhorst, A., & Boyd, A. J. (1980). Distributional ecology of the larvae and juveniles of the anchovy Engraulis capensis Gilchrist in relation to the hydrological environment off South West Africa, 1978–79. Fisheries Bulletin. Sea Fisheries Institute, Republic of South Africa, 13,83–106. Bang, N. D. (1971). The southern Benguela Current region in February 1966. 2. Bathythermography and air–sea interactions. Deep- Sea Research, 18, 209–224. Bakun, A. (1978). Guinea Current upwelling. Nature, 271, 147–150. Bakun, A. (1993). The California Current, Benguela Current and Southwestern Atlantic shelf ecosystems: a comparative approach to identifying factors regulating biomass yields. In K. Sherman, L. M. Alexander, & B. Gold (Eds.), Large marine ecosystems— stress mitigation and sustainability (pp. 199–221). Washington, DC: American Association for the Advancement of Science Chapter 18. Bakun, A., & Nelson, C. S. (1991). Wind stress curl in subtropical eastern boundary current regions. Journal of Physical Oceanogra- phy, 21, 1815–1834. Berrit, G. R. (1976). Les eaux froides cotieres du Gabon a l’Angola sont-elles dues a un upwelling d’Ekman? Cahier O.R.S.T.O.M. Se«ries Oce«anographie, 14, 273–278. Binet, D., & Marchal, E. (1993). The large marine ecosystem of shelf areas in the Gulf of Guinea: long-term variability induced by climatic changes. In K. Sherman, L. M. Alexander, & B. Gold (Eds.), Large marine ecosystems—stress mitigation and sustainability (pp. 104–118). Washington, DC: American Association for the Advancement of Science. Binet, D., Gobert, B., & Maloueki, L. (2001). El Ninõ-like warm events in the Eastern Atlantic (6° N, 20° S) and fish availability from Congo to Angola (1964–1999). Aquatic Living Resources, 14,99–113. Boebel, O., Lutjeharms, J. R. E., Schmid, C., Zenk, W., Rossby, T., & Barron, C. (2003). The Cape Cauldron: a regime of turbulent inter-ocean exchange. Deep-Sea Research II, 50,57–86. Boyd, A. J. (1983). Seasonal distributions of temperature, salinity, oxygen, and nitrate off southwestern Africa in 1981/82 and their possible influence on anchovy catches. Collection of Scientific Papers—International Commission for the Southeast Atlantic Fish- eries, 10,41–73. Boyd, A. J., & Shillington, F. A. (1994). The Agulhas Bank: a review of the physical processes. South African Journal of Marine Science, 90, 114–122. Boyd, A. J. (1987). The oceanography of the Namibian shelf. Doctoral Thesis, University of Cape Town, South Africa, unpublished. Boyd, A. J., & Nelson, G. (1998). Variability of the Benguela Current off the Cape Peninsula, South Africa. South African Journal of Marine Science, 19,27–40. Boyd, A. J., Hewitson, J. D., Kruger, I., & Le Clus, F. (1985). Temperature and salinity trends off Namibia from August 1982 to August 1984, and their relation to the spawning success of pelagic fish. Collection of Scientific Papers—International Commission for the Southeast Atlantic Fisheries, 12,53–58. Boyd, A. J., Salat, J., & Maso´, M. (1987). The seasonal intrusion of relatively saline water on the shelf off northern and central Namibia. South African Journal of Marine Science, 5, 107–120. Brown, P. C., & Henry, J. L. (1985). Phytoplankton production, chlorophyll-a and light penetration in the southern Benguela region during the period between 1977 and 1980. In L. V. Shannon (Ed.), South African ocean colour and upwelling experiment (pp. 211–218). Cape Town: Sea Fisheries Research Institute. Brundrit, G. B., de Cuevas, B. A., & Shipley, A. M. (1987). Long-term sea-level variability in the eastern South Atlantic and a comparison with that in the eastern Pacific. South African Journal of Marine Science, 5,73–78. Busalacchi, A. J., & Picaut, J. (1983). Seasonal variability from a model of the tropical Atlantic-Ocean. Journal of Physical Ocean- ography, 13, 1564–1588. Carr, M. -E. (2002). Estimation of potential productivity in Eastern Boundary Currents using remote sensing. Deep-Sea Research II, 49,59–80. Carton, J. A. (1994). Tropical Atlantic circulation. In K. Majumdar, E. W. Miller, G. S. Forbes, R. F. Schmalz, & A. A. Panah (Eds.), The Oceans: physicalÐchemical dynamics and human impact (pp. 74–87). Pennsylvania Academy of Science 20, Easton: Typehouse of Easton. N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 217

Castagne´, N., Borgne, P., Le Vourch, J., & Olry, J. P. (1986). Operational measurements of sea surface temperatures at CMS Lannion from NOAA-7 AVHRR data. International Journal of Remote Sensing, 7, 953–984. Citeau, J., Berge´s, J. C., Demarcq, H., & Mahe´, G. (1988). The watch of ITCZ migrations over the tropical Atlantic Ocean as an indicator in drought forecast over the Sahelian area. Tropical Ocean-Atmosphere Newsletter, 45,1–3. Cole, J. F. T. (1999). Environmental conditions, satellite imagery and clupeoid recruitment in the northern Benguela upwelling system. Fisheries Oceanography, 8,25–38. Cole, J. F. T., & McGlade, J. (1998). Temporal and spatial patterning of sea surface temperature in the northern Benguela upwelling system: possible environmental indicators of clupeoid production. South African Journal of Marine Science, 19, 143–158. Copenhagen, W. J. (1953). The periodic mortality of fish in the Walvis Region; a phenomenon within the Benguela Current. Investi- gational Report—Division of Fisheries, South Africa, 14, 35. de Ruijter, W. P. M. (1982). Asymptotic analysis of the Agulhas and Brazilian Current systems. Journal of Physical Oceanography, 12, 361–373. de Ruijter, W. P. M., & Boudra, D. B. (1985). The wind-driven circulation in the South Atlantic–Indian Ocean—I. Numerical experiments in a one layer model. Deep-Sea Research, 32, 557–574. de Ruijter, W. P. M., Biastoch, A., Drijfhout, S. S., Lutjeharms, J. R. E., Matano, R. P., Pichevin, T., van Leeuwen, P. J., & Weijer, W. (1999). Indian–Atlantic interocean exchange: dynamics, estimation and impact. Journal of Geophysical Research—Oceans, 104, 20885–20910. Dias, C. A. (1983a). Note on the evidence of a permanent southward flow of the upper oceanic tropospheric waters off Angola at 12° S. Collection of Scientific Papers—International Commission for the Southeast Atlantic Fisheries, 10,99–102. Dias, C. A. (1983b). Preliminary report on the physical oceanography off southern Angola, March and July 1971. Collection of Scientific Papers—International Commission for the Southeast Atlantic Fisheries, 10, 103–116. Ducet, N., Le Traon, P. Y., & Reverdin, G. (2000). Global high-resolution mapping of ocean circulation from TOPEX/Poseidon and ERS-1 and -2. Journal of Geophysical Research—Oceans, 105, 19477–19498. ECMWF (1995). User guide to ECMWF products, Version 2.1, December 1995. Reading, UK: European Centre for Medium Range Weather Forecasts. ECMWF (1996). The ECMWF Re-Analysis (ERA) Project. ECMWF Newsletter, 73, 7-16. Enfield, D. B., & Allen, J. S. (1980). On the structure and dynamics of monthly mean sea level anomalies along the Pacific coast of North and South America. Journal of Physical Oceanography, 10, 557–578. Feistel, R., Hagen, E., & Grant, K. (2004). Climatic changes in the subtropical Southeast Atlantic: The St Helena climate index (1893–1999). Progress in Oceanography, this issue, doi:10.1016/j.pocean.2003.07.002. Gammelsrød, T., Bartholomae, C. H., Boyer, D. C., Filipe, V. L. L., & O’Toole, M. J. (1998). Intrusion of warm surface water along the Angolan–Namibian coast in February–March 1995: The 1995 Benguela Nino. South African Journal of Marine Science, 19, 41–56. Garzoli, S. L., Gordon, A. L., Kamenkovich, V., Pillsbury, D., & Duncombe-Rae, C. (1996). Variability and sources of the south eastern Atlantic circulation. Journal of Marine Research, 54, 1039–1071. Gordon, A. L. (1985). Indian–Atlantic transfer of thermocline water at the Aghulhas Retroflection. Science, 227, 1030–1033. Grundlingh, M. L., & Lutjeharms, J. R. E. (1979). Large-scale flow patterns of the Agulhas Current system. South African Journal of Science, 75, 269–270. Hagen, E. (2001). Northwest African upwelling scenario. Oceanologica Acta, 24(Suppl.), S113–S128. Hagen, E., Feistel, R., Agenbag, J. J., & Ohde, T. (2001). Seasonal and interannual changes in intense Benguela upwelling. Oceanolog- ica Acta, 24, 557–568. Hagen, E., & Schemainda, R. (1983). Semi-annual variations within the Atlantic Equatorial Undercurrent. Tropical Ocean-Atmosphere Newsletter, 17,4–5. Hardman-Mountford, N. J., & McGlade, J. (2002). Variability of physical environmental processes in the Gulf of Guinea: an investigation using remotely sensed sea surface temperature. In J. McGlade, P. Cury, K. A. Koranteng, & N. J. Hardman-Mountford (Eds.), The Gulf of Guinea large marine ecosystem: environmental forcing and sustainable development of marine resources (pp. 49–66). Amsterdam: Elsevier. Hart, T. J., & Currie, R. I. (1960). The Benguela Current. “Discovery” Report, 31, 123–298. Hill, A. E., Hickey, B. M., Shillington, F. A., Strub, P. T., Brink, K. H., Barton, E. D., & Thomas, A. C. (1998). Eastern boundary currents: a pan-regional review. In A. R. Robinson, & K. H. Brink (Eds.), The Sea. Volume 11. The global coastal ocean. Regional studies and syntheses (pp. 29–68). New York: John Wiley and Sons Ltd. Hisard, P. (1986). El Ninõ response of the tropical Atlantic Ocean during the 1984 year. In T. Wyatt, & M. G. Larran˜eta (Eds.), International Symposium on Long Term Changes in Marine Fish Populations, Vigo, Spain, 18Ð21 November 1986 (pp. 18–21). Vigo: Instituto de Investigaciones Marinas de Vigo. Hisard, P., Heńin, C., Houghton, R., Piton, B., & Rual, P. (1986). Oceanic conditions in the tropical Atlantic during 1983 and 1984. Nature, 322, 243–245. 218 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221

Horel, J. D., Kousky, V. E., & Kagano, M. T. (1986). Atmospheric conditions in the Atlantic sector during 1983 and 1984. Nature, 322, 248–251. Houghton, R. W. (1983). Seasonal variations of the subsurface thermal structure in the Gulf of Guinea. Journal of Physical Oceanogra- phy, 13, 2070–2081. Houghton, R. W., & Colin, C. (1986). Thermal structure along 4° W in the Gulf of Guinea during 1983–1984. Journal of Geophysical Research—Oceans, 91, 11727–11739. Hutchings, L., Pitcher, G. C., Probyn, T. A., & Bailey, G. W. (1995). The chemical and biological consequences of coastal upwelling. In C. P. Summerhayes, K. -C. Emeis, M. V. Angel, R. L. Smith, & B. Zeitzschel (Eds.), Upwelling in the ocean: modern processes and ancient records (pp. 65–82). Dahlem Workshop Reports, Environmental Sciences Research Report, Vol. 18. Chichester, UK: John Wiley and Sons Ltd. Ikeda, M., & Emery, W. J. (1984). Satellite observations and modeling of meanders in the California Current system off Oregon and northern California. Journal of Physical Oceanography, 14, 1434–1450. Katz, E. J., Hisard, P., Verstraete, J. M., & Garzoli, S. L. (1986). Annual change of sea-surface slope along the equator of the Atlantic-Ocean in 1983 and 1984. Nature, 322, 245–247. Koranteng, K. A. (1998). The impacts of environmental forcing on the dynamics of demersal fishery resources of Ghana. Doctoral Thesis, University of Warwick, Coventry, UK, unpublished. Koranteng, K. A., & McGlade, J. M. (2001). Climatic trends in continental shelf waters off Ghana and in the Gulf of Guinea, 1963– 1992. Oceanologica Acta, 24, 187–198. Kruger, I., & Boyd, A. J. (1984). Investigation into the hydrology and plankton of the surface waters off southwestern Africa in ICSEAF Divisions 1.3, 1.4 and 1.5 in 1982/83. Collection of Scientific Papers—International Commission for the Southeast Atlantic Fisheries, 11, 109–133. Large, W. G., & Pond, S. (1981). Open ocean momentum flux measurements in moderate to strong winds. Journal of Physical Oceanography, 11, 324–336. Largier, J. L., & Swart, V. P. (1987). East-west variation in thermocline breakdown on the Agulhas Bank. South African Journal of Marine Science, 5, 263–272. Lass, H. U., Schmidt, M., Mohrholz, V., & Nausch, G. (2000). Hydrographic and current measurements in the area of the Angola– Benguela front. Journal of Physical Oceanography, 30, 2589–2609. Latif, M., & Barnett, T. P. (1994). Causes of decadal climate variability over the North Pacific and North America. Science, 266, 634–637. Latif, M., & Gro¨tzner, A. (2000). The equatorial Atlantic oscillation and its response to ENSO. Climate Dynamics, 16, 213–218. Le Clus, F. (1986). Aftermath of environmental perturbations off Namibia relative to the anchovy stock, and comparative pilchard spawning. Collection of Scientific Papers—International Commission for the Southeast Atlantic Fisheries, 13,19–26. Le Clus, F., & Kruger, I. (1982). Time and space distribution of temperature, salinity, plankton, and fish eggs off South West Africa in 1980/81—a preliminary data report. Collection of Scientific Papers—International Commission for the Southeast Atlantic Fisheries, 9, 121–145. Le Traon, P. Y., & Ogor, F. (1998). ERS-1/2 orbit improvement using TOPEX/ POSEIDON: the 2 cm challenge. Journal of Geophysi- cal Research—Oceans, 103, 8045–8057. Le Traon, P. Y., Gaspar, P., Bouyssel, F., & Makhmara, H. (1995). Using TOPEX/POSEIDON data to enhance ERS-1 data. Journal of Atmospheric and Oceanic Technology, 12, 161–170. Le Traon, P. Y., Nadal, F., & Ducet, N. (1998). An improved mapping method of multi-satellite altimeter data. Journal of Atmospheric and Oceanic Technology, 25, 522–534. Lutjeharms, J. R. E. (1977). Time variations in the spatial scales and intensities of ocean circulation in the vicinity of the Cape of Good Hope. Tellus, 29, 375–381. Lutjeharms, J. R. E. (1996). The exchange of water between the South Indian and the South Atlantic. In G. Wefer, W. H. Berger, G. Siedler, & D. Webb (Eds.), The South Atlantic: present and past circulation (pp. 125–162). Berlin: Springer. Lutjeharms, J. R. E., & Meeuwis, J. M. (1987). The extent and variability of South-East Atlantic upwelling. South African Journal of Marine Science, 5,51–62. Lutjeharms, J. R. E., & Stockton, P. L. (1987). Kinematics of the upwelling front off southern Africa. South African Journal of Marine Science, 5,35–50. Lutjeharms, J. R. E., & van Ballegooyen, R. C. (1988). The retroflection of the Agulhas Current. Journal of Physical Oceanography, 18, 1570–1583. Lutjeharms, J. R. E., & Walters, N. M. (1985). Ocean colour and thermal fronts south of Africa. In L. V. Shannon (Ed.), South African ocean colour and upwelling experiment (pp. 227–238). Cape Town: Sea Fisheries Research Institute. Lutjeharms, J. R. E., Cooper, J., & Roberts, M. (2000). Upwelling at the inshore edge of the Agulhas Current. Continental Shelf Research, 20, 737–761. Lutjeharms, J. R. E., Shillington, F. A., & Duncombe-Rae, C. M. (1991). Observations of extreme upwelling filaments in the southeast Atlantic Ocean. Science, 253, 774–776. N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 219

Matano, R. P., Simionato, C. G., Ruijter, W. P., van Leeuween, P. J., Strub, P. T., Chelton, D. B., & Schlax, M. G. (1998). Seasonal variability in the Agulhas Retroflection region. Geophysical Research Letters, 25, 4361–4364. McClain, E. P., Pichel, W. G., & Walton, C. C. (1985). Comparative performance of AVHRR-based multichannel sea surface temperatures. Journal of Geophysical Research, 90, 11587–11601. McDonagh, E. L., Heywood, K. J., & Meredith, M. P. (1999). On the structure paths and fluxes associated with Agulhas rings. Journal of Geophysical Research—Oceans, 104, 21007–21020. Meeuwis, J. M., & Lujeharms, J. R. E. (1990). Surface thermal characteristics of the Angola–Benguela front. South African Journal of Marine Science, 9, 261–280. Merle, J. (1978). Atlas Hydrologique saisonnier de l’oceán Atlantique intertropicale. Travaux et Documents de l’ORSTOM, 82. Paris: ORSTOM. Moore, D. W., McCreary, J. P., Hisard, P., Merle, J., O’Brien, J. J., Picaut, J., Verstraete, J. -M., & Wunsch, C. (1978). Equatorial adjustment in the eastern Atlantic Ocean. Geophysical Research Letters, 5, 637–640. Moroshkin, K. V., Bunov, V. A., & Bulatov, R. P. (1970). Water circulation in the eastern South Atlantic Ocean. Oceanology, 10, 27–34. Naujokat, B. (1986). An update of the observed quasi-biennial oscillation of the stratospheric winds over the tropics. Journal of Atmospheric Science, 43, 1873–1877. Nelson, G., & Hutchings, L. (1983). The Benguela upwelling area. Progress in Oceanography, 12, 333–356. Niller, P. P., & Kraus, E. B. (1977). One-dimensional models of the upper ocean. In E. B. Kraus (Ed.), Modelling and prediction of the upper layers of the ocean (pp. 143–172). New York: Pergamon Press. O’Brien, J. J., Adamec, D., & Moore, D. W. (1978). A simple model of upwelling in the Gulf of Guinea. Geophysical Research Letters, 5, 641–644. O’Reilly, J. E., Maritorena, S., O’Brien, M. C., Siegel, D. A., Toole, D., Menzies, D., et al. (2000). SeaWiFS postlaunch calibration and validation analyses, Part 3. In: S.B. Hooker & E.R. Firestone (Eds.), NASA Technical Memorandum 2000-206892: Vol. 11 (49 p). Greenbelt, Maryland: NASA Goddard Space Flight Center. O’Toole, M. J. (1980). Seasonal distribution of temperature and salinity in the surface waters off South West Africa, 1972–1974. Investigational Report South Africa Sea Fisheries Institute, 121,1–25. Parrish, R. H., Bakun, A., Husby, D. M., & Nelson, C. S. (1983). Comparative climatology of selected environmental processes in relation to eastern boundary current pelagic fish reproduction. In: G. D. Sharp & J. Csirke, Proceedings of the Expert Consultation to Examine Changes in Abundance and Species Composition of Neritic Fish Resources, San Jose, Costa Rica, April, 1982. F.A.O. Fisheries Reports, 291, 731–777. Peterson, R. G., & Stramma, L. (1991). Upper-level circulation in the South Atlantic Ocean. Progress in Oceanography, 26,1–73. Philander, S. G. H. (1986). Unusual conditions in the tropical Atlantic-Ocean in 1984. Nature, 322, 236–238. Philander, S. G. H. (1981). The response of equatorial oceans to a relaxation of the trade winds. Journal of Physical Oceanography, 11, 176–189. Philander, S. G. H., & Pacanowski, R. C. (1980). The generation of equatorial currents. Journal of Geophysical Research—Oceans, 85, 1123–1136. Philander, S. G. H., & Pacanowski, R. C. (1986). A model of the seasonal cycle in the tropical Atlantic-Ocean. Journal of Geophysical Research—Oceans, 91, 14192–14206. Picaut, J. (1983). Propagation of the seasonal upwelling in the eastern tropical Atlantic. Journal of Physical Oceanography, 13,18–37. Pitcher, G. C., Brown, P. C., & Mitchell-Innes, B. A. (1992). Spatio-temporal variability of phytoplankton in the southern Benguela upwelling system. South African Journal of Marine Science, 12, 439–456. Pingree, R., Kuo, Y. -H., & Garcia-Soto, C. (2002). Can the subtropical North Atlantic permanent thermocline be observed from space? Journal of the Marine Biological Association of the United Kingdom, 82, 707–728. Probyn, T. A., Mitchell-Innes, B. A., Brown, P. C., Hutchings, L., & Carter, R. A. (1994). A review of primary production and related processes on the Agulhas Bank. South African Journal of Science, 90, 166–173. Quartly, G. D., & Srokosz, M. A. (1993). Seasonal variations in the region of the Agulhas Retroflection: studies with Geosat and FRAM. Journal of Physical Oceanography, 23, 2107–2124. Roy, C. (1995). The Coˆte d’Ivoire and Ghana coastal upwellings: dynamics and changes. In X. Bard, & K. A. Koranteng (Eds.), Dynamics and use of Sardinella resources from upwelling off Ghana and Ivory Coast. Paris: ORSTOM Editions. Roy, C., & Reason, C. (2001). ENSO related modulation of coastal upwelling in the eastern Atlantic. Progress in Oceanography, 49, 245–255. Roy, C., Weeks, S., Rouault, M., Nelson, G., Barlow, R., & van der Lingen, C. (2001). Extreme oceanographic events recorded in the Southern Benguela during the 1999–2000 summer season. South African Journal of Science, 97, 465–471. Salat, J., Maso´, M., & Boyd, A. J. (1992). Water mass distribution and geostrophic circulation off Namibia during April 1986. Continental Shelf Research, 12, 355–366. Saunders, R. W., & Kriebel, K. T. (1988). An improved method for detecting clear sky and cloudy radiances over Western Europe. International Journal of Remote Sensing, 9, 123–150. 220 N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221

Schumann, E. H., Perrins, L. -A., & Hunter, I. T. (1982). Upwelling along the south coast of the Cape Province, South Africa. South African Journal of Science, 78, 238–242. Servain, J., Picaut, J., & Merle, J. (1982). Evidence of remote forcing in the equatorial Atlantic ocean. Journal of Physical Oceanogra- phy, 12, 457–463. Shannon, L. V. (1985a). The Benguela Ecosystem, Part I. Evolution of the Benguela, physical features and processes. Oceanography and Marine Biology: An Annual Review, 23, 105–182. Shannon, L. V. (1985b). South African ocean colour and upwelling experiment. Cape Town: Sea Fisheries Research Institute. Shannon, L. V., & Agenbag, J. J. (1990). A large-scale perspective on interannual variability in the environment in the South-East Atlantic. South African Journal of Marine Science, 9, 161–168. Shannon, L. V., & Nelson, G. (1996). The Benguela: large scale features and processes and system variability. In G. Wefer, W. H. Berger, G. Siedler, & D. J. Web (Eds.), The South Atlantic: present and past circulation (pp. 163–210). Berlin, Heidelberg: Springer-Verlag. Shannon, L. V., & Taunton-Clark, J. (1989). Long-term environmental indices for the ICSEAF area. Selected Papers—International Commission for the Southeast Atlantic Fisheries, 1,5–15. Shannon, L. V., Crawford, R. J. M., & Duffy, D. C. (1984a). Pelagic fisheries and warm events: a comparative study. South African Journal of Science, 80,51–60. Shannon, L. V., Hutchings, L., Bailey, G. W., & Shelton, P. A. (1984b). Spatial and temporal distribution of chlorophyll in southern African waters as deduced from ship and satellite measurements and their implications for pelagic fisheries. South African Journal of Marine Science, 2, 109–130. Shannon, L. V., Nelson, G., & Jury, M. R. (1981). Hydrological and meteorological aspects of upwelling in the southern Benguela Current. In F. A. Richards (Ed.), Coastal and estuarine sciences. 1. Coastal upwelling (pp. 146–159). Washington, DC: American Geophysical Union. Shannon, L. V., Agenbag, J. J., & Buys, M. E. L. (1987). Large- and mesoscale features of the Angola–Benguela front. South African Journal of Marine Science, 5,11–34. Shannon, L. V., Agenbag, J. J., Walker, N. D., & Lutjeharms, J. R. E. (1990). A major perturbation in the Agulhas retroflection area in 1986. Deep-Sea Research, 37, 493–512. Shannon, L. V., Walters, N. M., & Mostert, S. A. (1985). Satellite observations of the surface temperature and near-surface chlorophyll in the southern Benguela region. In L. V. Shannon (Ed.), South African ocean colour and upwelling experiment (pp. 183–210). Cape Town: Sea Fisheries Research Institute. Shannon, L. V., Boyd, A. J., Brundrit, G. B., & Taunton-Clark, J. (1986). On the existence of an El Ninõ type phenomenon in the Benguela system. Journal of Marine Research, 44, 495–520. Shannon, L. V., Crawford, R. J. M., Brundrit, G. B., & Underhill, L. G. (1988). Responses of fish populations in the Benguela ecosystem to environmental change. Journal du Conseil pour l’Exploration de la Mer, 45,5–12. Shannon, L. V., Crawford, R. J. M., Pollock, D. E., Hutchings, L., Boyd, A. J., Taunton-Clark, J., Badenhorst, A., Melville-Smith, R., Augustyn, C. J., Cochrane, K. L., Hampton, I., Nelson, G., Japp, D. W., & Tarr, R. J. Q. (1992). The 1980s—a decade of change in the Benguela ecosystem. South African Journal of Marine Science, 12, 271–296. Shelton, P. A., & Hutchings, L. (1982). Transport of anchovy, Engraulis capensis Gilchrist, eggs and early larvae by a frontal jet current. Journal du Conseil Permanent International pour l’Exploration de la Mer, 40, 185–198. Shelton, P. A., & Hutchings, L. (1990). Ocean stability and anchovy spawning in the southern Benguela Current region. Fishery Bulletin, 88, 323–338. Shillington, F. A. (1998). The Benguela Upwelling System off southwestern Africa. Coastal segment (16,E). In A. R. Robinson, & K. H. Brink (Eds.), The Sea. Volume 11. The global coastal ocean. Regional studies and syntheses (pp. 583–604). New York: John Wiley and Sons Ltd. Shillington, F. A., Hutchings, L., Peterson, W. T., Probyn, T. A., Waldron, H. N., & Agenbag, J. J. (1990). A cool upwelling filament off Namibia, southwest Africa: preliminary measurements of physical and biological features. Deep-Sea Research A, 37, 1753–1772. Shillington, F. A., Hutchings, L., Probyn, T. A., Waldron, H. N., & Peterson, W. T. (1992). Filaments and the Benguela frontal zone: offshore advection or recirculation loops? South African Journal of Marine Science, 12, 207–218. Stander, G. H. (1964). The pilchard of South West Africa (Sardinops ocellata). The Benguela Current off South West Africa. Investigational reports—Marine Research Laboratory of South West Africa, 11,1–42. Stander, G. H., & de Decker, A. H. B. (1969). Some physical and biological aspects of an oceanographic anomaly off South West Africa in 1963. Investigational Report of the Division of Sea Fisheries South Africa, 81, 46. Strub, P. T., Shillington, F. A., James, C., & Weeks, S. (1998). Satellite comparison of the seasonal circulation in the Benguela and California current systems. South African Journal of Marine Science, 19,99–112. Swart, V. P., & Largier, J. L. (1987). Thermal structure of Agulhas Bank water. South African Journal of Marine Science, 5, 243–254. Taunton-Clark, J. (1990). Environmental events within the South-East Atlantic (1906–1985) identified by analysis of sea surface temperature and wind data. South African Journal of Science, 86, 470–472. N.J. Hardman-Mountford et al. / Progress in Oceanography 59 (2003) 181Ð221 221

Taunton-Clark, J., & Shannon, L. V. (1988). Annual and interannual variability in the South-East Atlantic during the 20th century. South African Journal of Marine Science, 6,97–106. Thomas, A. C., Carr, M. -E., & Strub, P. T. (2001). Chlorophyll variability in eastern boundary currents. Geophysical Research Letters, 28, 3421–3424. Trenberth, K. E., Large, W. G., & Olson, J. G. (1990). The mean annual cycle in global ocean wind stress. Journal of Physical Oceanography, 20, 1742–1760. Verstraete, J.-M. (1987). Seasonal heat content of the eastern tropical Atlantic. Proceedings of the International Symposium on equatorial vertical motions, Paris, 6-10 May 1985 (Number). Oceanologica Acta, 5(6), 85-90. Verstraete, J. -M. (1992). The seasonal upwellings in the Gulf of Guinea. Progress in Oceanography, 29,1–60. Walker, N. D. (1987). Interannual sea surface temperature variability and associated atmospheric forcing within the Benguela system. South African Journal of Marine Science, 5, 121–132. Walker, N., Taunton-Clark, J., & Pugh, J. (1984). Sea temperatures off the South African west coast as indicators of Benguela warm events. South African Journal of Science, 80,72–77. Walter, H. (1937). Die o¨kologischen Verha¨ltnisse in der Namibnebalwu¨ste (Su¨dwestafrika) unter Auswertung de Aufzeichnungen des Dr G Boss (Swakopmund). In N. Pringsheim (Ed.), Jahrbu¬cher fu¬r Wissenshaffliche Botanik (pp. 58–222). Leipzig: Gebru¨der Borntraeger. Walters, N. M. (1985). Algorithms for the determination of near-surface chlorophyll and semi-quantitative total suspended solids in South African Coastal Waters from Nimbus-7 CZCS data. In L. V. Shannon (Ed.), South African ocean colour and upwelling experiment (pp. 175–182). Cape Town: Sea Fisheries Research Institute. Walters, N. M., & Schumann, E. H. (1985). Detection of silt in coastal waters by Nimbus-7 CZCS. In L. V. Shannon (Ed.), South African ocean colour and upwelling experiment (pp. 219–226). Cape Town: Sea Fisheries Research Institute. Wedepohl, P. M., Lutjeharms, J. R. E., & Meeuwis, J. M. (2000). Surface drift in the South-East Atlantic Ocean. South African Journal of Marine Science, 22,71–79. Weeks, S. J., & Shillington, F. A. (1996). Phytoplankton pigment distribution and frontal structure in the subtropical convergence region south of Africa. Deep-Sea Research I, 43, 739–741. Weeks, S. J., Currie, B., & Bakun, A. (2002). Satellite imaging: massive emissions of toxic gas in the Atlantic. Nature, 415, 493–494. Weeks, S. J., Shillington, F. A., & Brundrit, G. B. (1998). Seasonal and spatial variability in the Agulhas Retroflection and Agulhas return current. Deep-Sea Research I, 45, 1611–1625. Weisberg, R. H., & Colin, C. (1986). Equatorial Atlantic-Ocean temperature and current variations during 1983 and 1984. Nature, 322, 240–243. Wyrtki, K. (1982). The Southern Oscillation, ocean-atmosphere interaction and El Ninõ. Marine Technology Society Journal, 16,3–10.