A ribbon of dark water: phytoplankton blooms in the meanders of the Pacific North Equatorial Countercurrent

ARTICLE IN PRESS

Deep-Sea Research II 51 (2004) 209–228

A ribbon of dark water: phytoplankton blooms in the meanders of the Pacific North Equatorial Countercurrent James R. Christiana,*, Ragu Murtuguddea, Joaquim Ballabrera-Poya, Charles R. McClainb b

a ESSIC, University of Maryland, College Park, MD, 20742, USA NASA Goddard Space Flight Center, Code 970.2, Greenbelt, MD, USA

Received 12 August 2002; received in revised form 2 June 2003; accepted 12 June 2003

Abstract In the far southwestern corner of the North Pacific Ocean, the Mindanao Current retroflects and flows northeast into the nascent North Equatorial Countercurrent (NECC). Ocean-color images show that phytoplankton blooms occur both in the nearly closed cyclonic loop of the retroflection, and in the meandering NECC, which were particularly strong during the El Nin˜o event of 1997–98. In the winter of 1997–98, the bloom persisted for about 4 months, and, at its peak, extended for several thousand kilometers along the current stream. The NECC meanders are observed in SeaWiFS and OCTS images at various times throughout the nearly 5 years of data examined, but regionally averaged chlorophyll concentrations during the 1997–98 El Nin˜o were higher than at any time since. The current meanders are sometimes visible in SeaWiFS images during winter, spring and summer, but not in autumn, consistent with the seasonal cycle of formation and decay of the Mindanao Dome (MD). Although the MD has a cyclonic circulation pattern, vertical nutrient flux appears to be maximal in the NECC, which forms its southern boundary, rather than in the MD itself. The appearance of phytoplankton blooms along the stream of the NECC is most plausibly attributed to upwelling associated with current meandering, although El Nin˜o-induced shoaling of the thermocline and seasonal Ekman pumping also may contribute. In addition, transport of nutrient-rich water from the South Pacific by the New Guinea Coastal Undercurrent (NGCUC), which is maximal in boreal winter, creates a gradient of nutrient concentration across the NECC stream with much greater concentrations to the south at a given density (J. Geophys. Res. 103 (1998) 12959). In 1997-98 a confluence of all of these factors occurred. Shoaling of the thermocline and an intensified NECC due to the El Nin˜o event coincided with the seasonal formation of the MD and increased westward transport of nutrient-rich water by the NGCUC to produce the highest chlorophyll concentrations observed in 5 years of ocean-color data. Similar though weaker blooms were observed in the spring of 2003 in association with the very weak 2002–03 El Nin˜o. Crown Copyright r 2004 Published by Elsevier Ltd. All rights reserved.

*Corresponding author. Present address: Fisheries and Oceans Canada, Canadian Centre for Climate Modelling and Analysis, University of Victoria, PO Box 1700 STN CSC, Victoria, BC, V8W 2Y2, Canada. Tel.: +1-250-363-8319; fax: +1-250-363-8247. E-mail address: [email protected] (J.R. Christian).

1. Introduction Satellite observations of ocean color provide a tool for oceanographers to identify flow patterns and hydrographic boundaries on a large scale but

0967-0645/$ - see front matter Crown Copyright r 2004 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.dsr2.2003.06.002

ARTICLE IN PRESS 210

J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228

at very high resolution (up to 1 km). Ocean-color images can reveal subsurface processes (entrainment of nutrients into the mixed layer) and dynamic features not visible in sea-surface temperature (e.g., Murtugudde et al., 1999). Similarly, waters with very different physical properties and flow characteristics may be indistinguishable in ocean-color images. Near its source region, the Pacific North Equatorial Countercurrent (NECC), meanders much like western boundary currents such as the Gulf Stream and the Kuroshio, but on a much smaller scale (Lukas et al., 1991). Upwelling associated with current meandering and the resulting enhancement of biological production have been observed in the Gulf Stream (Osgood et al., 1987; Bower and Rossby, 1989; Hitchcock et al., 1993; Ashjian et al., 1994; Mariano et al., 1996). The underlying mechanisms are not completely understood, but to first order can be understood as arising from conservation of potential vorticity (PV). If the surface layer has a thickness D, the PV of this layer is ðf þ zÞ=D where f is the planetary vorticity and z the relative vorticity (Pond and Pickard, 1983). The increase in f that occurs when the meandering current flows poleward can be compensated either by cyclonic curvature of the flow (decreasing z) or vertical motion (increasing D) so that PV is conserved (Hitchcock et al., 1993). Divergence (upwelling) occurs downstream of cyclonic meanders (troughs) and convergence (downwelling) occurs downstream of anticyclonic meanders (crests) (Bower and Rossby, 1989; Bower, 1989; Flierl and Davis, 1993; Yoshimori, 1994). In this paper we analyze ocean colour images of the nascent NECC, and use ancillary data and an ocean general circulation model (OGCM) to argue that meander-induced upwelling appears to be a proximate control on the supply of nutrients to the surface, but that the flux is also controlled by subsurface environmental conditions (thermocline depth, water mass structure) that change more gradually and are coherent with basin-scale climate fluctuations. The currents in this region were extensively studied during the Western Equatorial Pacific Ocean Circulation Study (WEPOCS) (Lindstrom et al., 1987; Lukas et al., 1991; see Fig. 1). The

NECC was observed to have a subsurface core at about 120 m in meridional transects at 143 E (Lindstrom et al., 1987). Lukas et al. (1991) concluded that Mindanao Current (MC) water contributes to the NECC via cyclonic retroflection in the Celebes Sea. The general features of the regional hydrography and circulation are illustrated in Fig. 1. (We caution that not all place names and circulation features referred to in the text are indicated on this map, and one feature is given a different name than that used by Fine et al. (1994), as discussed below). The MC flows southward along the east coast of Mindanao, retroflects as it approaches the equator, and turns to the northeast to contribute to the nascent NECC, whose meanders are indicated in Fig. 1. To the north of this current is a small cyclonic gyre called here the Mindanao Eddy, which we will refer to as the Mindanao Dome (MD) following Masumoto and Yamagata (1991). In the seasons when the MD is present (roughly the first half of the calendar year), its zonal extent is much larger than is implied in Fig. 1, extending eastward of 140 E (Masumoto and Yamagata, 1991). The New Guinea Coastal Undercurrent (NGCUC) flows northwest along the north coast of New Guinea and feeds the eastward flowing equatorial undercurrent (EUC). We will argue below that the NGCUC may be an important source of nutrients for phytoplankton in the NECC. The biogeochemistry in this region is complicated because of the confluence of many water masses (Tsuchiya et al., 1989; Fine et al., 1994). Data from WOCE section P9 show a large range of variation of nitrate and silicate concentrations, at similar densities, between the equator and 10 N (Kaneko et al., 1998). East of 140 E the NECC follows a strong salinity and potential vorticity front (Gouriou and Toole, 1993), and nitrate concentration appears to increase markedly from north to south across this front (Kaneko et al., 1998). The NGCUC transports both nutrient-rich South Pacific Tropical Water (SPTW) and nutrient-poor Antarctic Intermediate Water (AAIW) (Tsuchiya et al., 1989; Fine et al., 1994; Kashino et al., 1996; Kuroda, 2000). The latter occurs at densities exceeding 1026.8 kg m3 and is not entrained into the surface currents (Fine et al.,

ARTICLE IN PRESS J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228

100°E

110°E

130°E

120°E

211

150°E

140°E

160°E

20°Ν

20°N

Vit iaz Str

NEC

ME NECC

Celebes Sea

NSCC

HE

SEC

ah er

New Guinea

SEC NGCUC So lo

Torres Strait

m

Lombok Strait

on

10°S Se

a

LC

GBRUC EAC

Australia

20°S 100°E

110°E

0°

SSCC

A

SJC

10°S

NGCC

a

Banda Sea

Se

Ma

a

JAV

EUC SECC

NGCUC

lm

kas sar S

Ha

trait

BORNEO RA AT M SU

0°

10°N

ait

˚

Mindanao

Solo

MUC

mon S

St. George Channel

MC

10°N

Bismarck Sea

NEC

trait

KUROSHIO

South China Sea

120°E

130°E

140°E

150°E

SEC

AAIW

20°S 160°E

Fig. 1. Map of the western tropical Pacific Ocean and Indonesian Seas showing the major geographic names and surface to intermediate depth currents, including the Kuroshio, Mindanao Current (MC), North Equatorial Current (NEC), North Equatorial Countercurrent (NECC), New Guinea Coast Current (NGCC), South Equatorial Current (SEC), South Equatorial Counter Current (SECC), East Australia Current (EAC), South Java Current (SJC) and the Leeuwin Current (LC). The subsurface currents are the New Guinea Coastal Undercurrent (NGCUC), Equatorial Undercurrent (EUC), Northern and Southern Subsurface Countercurrents (NSCC and SSCC), Mindano Undercurrent (MUC), and Great Barrier Reef Undercurrent (GBRUC). The Mindanao Eddy (ME) and Halmahera Eddy (HE) are also indicated. Solid lines indicate surface flow, thick dashed lines indicate thermocline flow, and thin dashed lines indicate the flow of Antarctic Intermediate Water. The inset shows the region of Vitiaz Strait east of New Guinea at 5 S, 148  E; after Fine et al. (1994) and reproduced courtesy of R. Lukas.

1994). The former occurs at shallower depths, and we hypothesize that it is a significant source of nutrients to the phytoplankton blooms in the NECC. This hypothesis is discussed at length below. The NGCUC also may be enriched in iron from geothermal activity or fluvial sediment (Obata et al., 1993; Mackey et al., 2002), but it is not known whether iron or major nutrients limit productivity in this region. The El Nin˜o—Southern Oscillation (ENSO) phenomenon is a dominant influence on climate throughout the tropical Pacific. In the NECC source region, sea-surface height anomalies (SSHA) are strongly correlated with ENSO phase indicators such as the Southern Oscillation Index

(SOI), and the maximal correlation occurs at very short time lags (Bray et al., 1996). Variability in SSHA reflects the canonical ‘‘horseshoe’’ pattern, with strong minima during the ENSO warm phase (Murtugudde et al., 1999; Merrifield et al., 1999). The speed and transport of the NECC are also known to intensify during El Nin˜o events (Meyers and Donguy, 1984; Qiu and Joyce, 1992; Johnson et al., 2000; Johnston and Merrifield, 2000). In this paper we show SeaWiFS images of the NECC collected during 1997-98 El Nin˜o event, when an unusual enrichment of surface chlorophyll and other absorbing substances occurred. The processes that may contribute to this enrichment of surface waters of the NECC include local

ARTICLE IN PRESS 212

J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228

upwelling associated with current meandering (e.g., Flierl and Davis, 1993), thermocline uplift associated with the seasonal varying wind stress curl (Masumoto and Yamagata, 1991; Qiu and Lukas, 1996), thermocline uplift associated with remote forcing by planetary waves (Chelton and Schlax, 1996; Boulanger and Menkes, 1999; McPhaden and Yu, 1999), presence of nutrient rich water masses on one side of the front that marks the path of the NECC (Fine et al., 1994; Kaneko et al., 1998), and advection of nutrients and/or phytoplankton from regions of topographically driven upwelling further west (e.g., off the northwest coast of Halmahera). We have used ocean-color images, various ancillary data, and an ocean general circulation model to examine these processes and attempt to explain what controls the flux of nutrients to the surface waters of the NECC, and why it increased so markedly during 1997–98.

2. Methods 2.1. Data Chlorophyll estimates were derived from SeaWiFS (McClain et al., 1998; O’Reilly et al., 1998) and Ocean Color and Temperature Scanner (OCTS) (Kishino et al., 1997; Shimada et al., 1999) satellite observations. The images shown in this paper are Level 3 (9 km) composites (monthly and daily), obtained from the Distributed Active Archive Center (DAAC, http://daac.gsfc.nasa. gov) at NASA Goddard Space Flight Center (GSFC). OCTS data (Version 4) were obtained from NASDA. Gridded (1 ) fields of SSHA were obtained from the Ocean Pathfinder project at NASA-GSFC (http://iliad.gsfc.nasa.gov/ocean. html). Additional data sets were obtained from the climate data archive at Columbia University’s Lamont-Doherty Earth Observatory. NCEP (Kalnay et al., 1996) interannual wind stress (monthly, 2.5 spatial resolution) were used to force the ocean model. Interannual sea–surface temperature (SST) fields were derived by Reynolds and Smith (1994) (monthly, 1 spatial resolution); these SST

fields are a merged product derived from ship and mooring as well as satellite observations. Values of the NINO3 index were obtained from the climate data archive, and values of the SOI from the Australian Bureau of Meteorology. These are widely used indices of the state of the ENSO cycle. The NINO3 index is an index of sea–surface temperature anomaly in the eastern and central equatorial Pacific and is positive during El Nin˜o events and negative during La Nin˜a conditions. The SOI is an index of the east–west gradient in sea-level barometric pressure and is opposite in sign to the NINO3 (negative during El Nin˜o). 2.2. Model A numerical model of the tropical Indo-Pacific ocean (Gent and Cane, 1989; Murtugudde et al., 1998) was used to estimate current fields and subsurface vertical velocities. The model domain extends from 32 E–76 W and 30 S–30 N, with horizontal resolution of 0.5 and 15 layers of variable thickness (s layers) in the vertical (Murtugudde et al., 1998, 1999). The first model layer corresponds to the mixed layer of the ocean (Chen et al., 1994), so the view of ocean-color sensors should be largely confined to this layer. The model run was forced with NCEP interannual winds starting in 1958, following a 25 year spin up from rest with climatological wind forcing. As the volume of model output is very large, we must select quantities to examine that are likely to have biogeochemical relevance. Because nutrients are nonconservative and have strong vertical gradients in the upper 300 m, the nutrient flux may be much more anisotropic than the mass flux in a region where both upwelling and downwelling occur, i.e. the net upward flux of nutrient to the surface layer may be related to the absolute amount of vertical motion rather than to vertical velocity or net vertical transport. For example, in the Sargasso Sea it has been shown that there is a net upward flux of nutrients associated with cyclonic mesoscale vortices even where the regionally averaged vertical velocity is downward (McGillicuddy et al., 1998; Siegel et al., 1999). We therefore calculated regional averages of vertical velocity (w) for the model layer immediately below

ARTICLE IN PRESS J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228

213

the mixed layer, as well as its absolute value (|w|), which is used as an indicator of the total vertical motion in either the upward or downward direction. We also calculated heat content anomalies for each month of the simulation relative to the mean for 1986–2001, and depth- and latitudeintegrated zonal transport across 140 E. The mean zonal transport of the NECC in the model simulation was 23 Sv (1 Sv=106 m3 s1), which is similar to literature estimates such as the 27.6 Sv estimated at 142 E by Gouriou and Toole (1993), and falls in between the ‘‘El Nin˜o’’ (e.g., Fall, 1997) and ‘‘non-El Nin˜o’’ (e.g., Fall, 1998) estimates of Johnson et al. (2000).

the data contain unresolved low-frequency variability. For the time period examined here (the OCTS and SeaWiFS ocean-color missions), the variability of the western tropical Pacific Ocean is dominated by a single event (the 1997–98 El Nin˜o), and the ENSO frequency (2–7 years) cannot be resolved. We applied the methods available to test the significance of correlations in the data, but caution that a definitive estimate of the significance level is not possible given the length of the data record.

2.3. Correlation analysis

The effect of upwelling in the NECC meanders on ocean color (i.e., water leaving radiance) is illustrated by the SeaWiFS images of the far western tropical Pacific in March 1998 shown in Fig. 2. Upwelling of subsurface waters reduces the water leaving radiance in the blue region due to the presence of both colored dissolved organic matter (CDOM) and inorganic nutrients which enhance phytoplankton growth (e.g., Garver and Siegel, 1997; Nelson and Siegel, 2002). In the first image shown in Fig. 2, normalized water leaving radiance at 412 nm (Lwn(412)) indicates the presence of recently upwelled waters by strong absorption in the blue region. In this image, taken in the waning months of the 1997–98 El Nin˜o event, the meandering NECC is clearly visible as a distinct minimum in Lwn(412) over a distance of several thousand kilometres. The currents that contribute to the NECC also can be seen quite clearly in this image. The North Equatorial Current (NEC) bifurcates at the Philippine coast (B12 N, 127 E in Fig. 2), and its very oligotrophic, blue waters feed the MC and flow into the Celebes Sea and the retroflection that feeds the NECC (region between Mindanao and Halmahera). Low Lwn(412) off the north coast of Halmahera indicates upwelling on the downstream side of the retroflection. The position of the MD can be roughly estimated from the positions of the currents that form its southern, northern and western boundaries: the NECC (low Lwn), the NEC (high Lwn), and MC (high Lwn). Surface chlorophyll from the standard SeaWiFS chlorophyll algorithm (O’Reilly et al., 1998)

To gain insight into the mechanisms underlying temporal variability of nutrient flux to the surface waters of the NECC, we calculated correlation coefficients for time series of regional averages of the various data fields and the diagnostic variables calculated from the model output. Assessment of the significance of correlation coefficients for geophysical time-series data can be problematic because the data are often serially correlated (i.e. autocorrelated), so that each point is not independent as standard significance tests assume. An ‘‘effective sample size’’, measuring the number of ‘‘independent’’ observations included in the timeseries, can be estimated from the shape of the signal autocorrelation (Davis, 1976; Emery and Thomson, 2001). However, the lack of precision of these estimates, which have been found to provide ‘‘liberal’’ levels of significance (Thiebaux and Zwiers, 1984; Zwiers and von Storch, 1995; Ebisuzaki, 1997), makes the effective sample size a diagnostic quantity rather than a robust method of inferring the significance of a correlation. Ebisuzaki (1997) proposed an alternative test based on a Monte Carlo simulation, which uses surrogate time series created with power spectra identical to the original data but random phase. Because these surrogate time series have the same power spectrum as the original time series, the autocorrelation of the signal is preserved during the Monte Carlo simulation. The significance levels assigned by this test are also liberal when

3. Results

ARTICLE IN PRESS 214

J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228

Fig. 2. SeaWiFS images of (top) water-leaving radiance at 412 nm and (bottom) chlorophyll, in March 1998, showing the phytoplankton blooms off the northwest coast of Halmahera and along the meandering North Equatorial Countercurrent.

indicates the presence of phytoplankton as well as CDOM in the upwelled waters (Fig. 2, bottom panel), although the relative contributions of each can not be established. The meanders of the NECC stand out clearly in the chlorophyll image, and the bloom extends for some 2000 km.

The seasonal evolution of this event is shown in the series of monthly SeaWiFS composites for the winter of 1997–98 (Fig. 3). The bloom begins in early winter, with upwelling near the north coast of Halmahera near the point where the NECC originates in the Mindanao Current retroflection

ARTICLE IN PRESS J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228

Fig. 3. Monthly composites (Standard Mapped Image) of SeaWiFS chlorophyll from November 1997 through June 1998.

215

ARTICLE IN PRESS 216

J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228

(see Figs. 1 and 2). The bloom strengthens and extends westward during December and January, reaching its greatest zonal extent in March. In April the current and its associated upwelling appear to be weakening, and by June the event is over (Fig. 3). The penetration of blue water from the NEC between Mindanao and Halmahera is visible as a ‘‘U’’ shape in many of these images. This is most likely the nearly closed cyclonic quasivortex associated with the retroflection; it is most pronounced in the images where a strong NECC is apparent (Fig. 3). There appears to be some clear NEC water entering the Celebes Sea, particularly in December and March (Fig. 3). The following year, the bloom appears only faintly and much later in the year (Fig. 4). An enhancement of surface chlorophyll in the NECC meanders is intermittently visible in ocean-color images throughout the SeaWiFS and OCTS data sets, including April 1997, December 1999, and June 2000 (not shown). Although the highest chlorophyll concentrations were observed in boreal winter during the 1997–98 El Nin˜o, images in which the current meanders are visible occur at various times of the year, including boreal winter and summer. We have not observed any images in which the meanders are visible during boreal autumn which is consistent with the seasonal formation and decay of the Mindanao Dome (Masumoto and Yamagata, 1991). The wind-stress curl is maximal (positive) and the region of maximum curl is farthest south in boreal winter (Qiu and Lukas, 1996), coinciding with the seasonal formation of the MD. The MD remains strong through boreal summer and weakens and decays in autumn (Masumoto and Yamagata, 1991). The upwelling of nutrients in the meandering NECC appears to have been strongly enhanced during the 1997–98 El Nin˜o event, and to have been somewhat quiescent thereafter. Chlorophyll concentrations in this region were much greater in 1997–98 in than in the subsequent (La Nin˜a) years, but increased again in early 2003, a weak El Nin˜o year (Fig. 5). The seasonal evolution of the regionally averaged chlorophyll concentration appears similar in most of the years considered, with peak concentrations occurring

as early as January or as late as April. The spring of 1999 was the exception, with peak concentrations occurring in late spring or summer (Figs. 4 and 5). The meanders are visible in the images from January through April 2003, and the seasonal phasing appears quite similar to 1998 (not shown). Surface chlorophyll averaged from 135–150 E and 3–10 N is strongly correlated (Table 1) with the SOI and NINO3 indices, as well as the SSHA. The SSHA is known to vary strongly with the ENSO phase in this region, with minima during El Nin˜o events (indicating a shallower thermocline) and maxima (deep thermocline) during the La Nin˜a phase (Bray et al., 1996; Merrifield et al., 1999). These correlations are significant at the 0.05 or 0.01 level using the random phase test (Table 1), which is the most conservative test available, although determining the exact significance level is problematic even with this method (see Methods). A t-test with an effective sample size corrected for autocorrelation also was conducted and also indicates significance, but this test is generally more liberal (i.e. tends to overestimate significance) than the random-phase test. These correlations taken together indicate clearly that there are trends associated with ENSO that are coherent across the variables observed by oceanobserving satellites: during the warm (El Nin˜o) phase, the thermocline shoals, the sea surface cools, and the flux of nutrients and CDOM to the surface increases. The seasonal and interannual variability of the surface chlorophyll suggests some inferences about the underlying mechanisms. The wind stress is upwelling-favourable and the thermocline shoals in boreal winter (Masumoto and Yamagata, 1991; Qiu and Lukas, 1996). However, the seasonal cycle of surface chlorophyll, and therefore presumably of nutrient flux, is much weaker than the interannual variability. The apparent importance of ENSO relative to the seasonal cycle suggests that basin-scale thermocline variability and remote forcing via the effect of planetary waves are more important than local wind forcing in determining the thermocline depth. The tight coupling of the SSHA to ENSO (Bray et al., 1996; see Table 1) suggests that this is the case.

ARTICLE IN PRESS J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228

Fig. 4. Monthly composites (Standard Mapped Image) of SeaWiFS chlorophyll from November 1998 through June 1999.

217

ARTICLE IN PRESS J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228

Chlorophyll (mg m-3)

218 0.15

1997−98

0.1 1998− 99

0.05 -100

-50

0

50

100

150

200

0.15

-25

0.1

50

0.05

25

SOI

Chlorophyll (mg m-3)

Time (days from Jan. 1)

0

0

-25 1997

1998

1999

2000

2001

2002

Fig. 5. (top) Seasonal time series of mean monthly chlorophyll in the region of the NECC meanders (3–10 N, 135–150 E) for November-June of 1996–97, 1997–98, and 1988–99, 1999–2000, and 2000–2001. The El Nin˜o winter of 1997–98 is indicated by the solid symbols (); the following winter (1998–-99) is indicated by the open symbols (J); the remaining 3 years are not distinguished. (bottom) Time series of mean monthly chlorophyll (no symbols) in the same region relative to the Southern Oscillation Index (symbols).

Table 1 Correlation matrix for observations in the region of the North Equatorial Countercurrent meanders (3–10 N, 135–150 E) for the first 4 years of SeaWiFS observations CHL

SST

SSHA

SOI

NINO3 TAUX

SST 0.84 SSHA 0.62 0.56 SOI 0.72 0.60 0.71 NINO3 0.69 0.52 0.91 0.78 TAUX 0.18 TAUY 0.37

0.40 0.62

0.06 0.38

0.14 0.17

0.02 0.24

0.87

CHL=SeaWiFS chlorophyll, SST=sea surface temperature, SSHA=sea surface height anomaly, SOI = Southern Oscillation Index, NINO3=NINO3 index, TAUX=zonal wind stress, TAUY=meridional wind stress. A single  indicates significance at the 0.05 level using the random-phase test of Ebisuzaki (1997);  indicates significance at the 0.01 level.

The MD appears consistently oligotrophic in ocean-color images (Figs. 2–4), although its circulation is cyclonic. Instead, the images show enrichments of chlorophyll along the southern boundary of the Dome, in the NECC meanders. This may result from upwelling associated with

current meandering, the presence of more nutrientrich water masses to the south, advection of phytoplankton by the NECC from topographically driven upwelling centres further west, or a combination of these factors. Here, we examine each of these in turn. One possibility is that most of the chlorophyll in the NECC meanders is advected from the area between Halmahera and Mindanao. Simple scaling analysis shows the range of rates that are compatible with this hypothesis. At its greatest extent, the NECC ‘‘ribbon’’ is more than 2000 km long, giving a transport time scale of about 50 days at a surface velocity of 50 cm s1. If the chlorophyll concentration in the originating water is 2 mg m3, at a loss rate of 0.05 d1 the concentration would still be above the ‘‘background’’ of the adjacent oligotrophic waters after 50 days. At a loss rate of 0.1 d1 it would not be, but concentrations above background would still be found at lesser distances from the source (e.g., o1000 km). It is also possible that the originating water contains unutilized nutrients that would subsequently be converted to biomass.

ARTICLE IN PRESS J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228

The weakness of these scenarios is that the images show neither the monotonic decrease expected in the loss-only scenario, or the downstream peak expected in a growth-plus-loss scenario. The maximum concentration appears near the coast of Halmahera (Figs. 2 and 3) and shows a monotonic decrease for the first few hundred kilometres. In the March 1998 image there is a distinct minimum downstream of Halmahera that separates the NECC ‘‘ribbon’’ from this source. Examination of synoptic (daily) images also suggests that advection is a minor contributor, because these images indicate a minimum of surface chlorophyll northeast of the Halmahera upwelling, with enrichments in the meanders farther downstream (Fig. 6), although subsurface advection can not be entirely ruled out. The presence of steep nutrient concentration gradients, along isopycnal surfaces, may be a critical factor in determining the localization of vertical nutrient flux in the NECC meanders, as the concentration increases sharply across the front associated with the current, with higher concentrations to the south (Kaneko et al., 1998). Transport of nutrient-rich SPTW by the NGCUC is a potentially important factor contributing to the maintenance of this gradient and therefore to the variability of chlorophyll in the NECC. The NGCUC carries high-salinity SPTW into the NECC source region particularly during boreal winter (Fig. 7). It is somewhat difficult to estimate zonal transport by the NGCUC, as it is adjacent to the EUC, which generally flows in the opposite direction. The NGCUC does not appear to have a strong seasonal cycle (Cresswell, 2000; Kuroda, 2000), so the net transport is westward when the EUC is weak, particularly during boreal winter and during El Nin˜o events. We calculated the total zonal transport across 145 E between 3 S and 2 N from 75 to 250 m depth, which includes both undercurrents. This results in a seasonal cycle of net zonal transport which is westward in boreal winter and eastward in boreal summer (Fig. 8, top). Salinity at 145 E, averaged over the same depth and latitude range, shows a maximum in boreal winter (Fig. 8, bottom). The westward transport of saline nutrient-rich water is therefore at a maximum during boreal winter. The seasonal

219

maximum in salinity and westward transport was stronger in 1997–98 than in the subsequent years, but interannual variability associated with ENSO does not appear to be an important part of the total signal. Upwelling associated with current meandering is another potential reason why the biological response to thermocline shoaling manifests itself primarily along the NECC and not in the MD. The speed of the NECC tends to increase during El Nin˜o events (Meyers and Donguy, 1984; Qiu and Joyce, 1992; Johnson et al., 2000; Johnston and Merrifield, 2000), so meander-associated upwelling would be expected to increase. However, meanderinduced upwelling is likely to be accompanied by a near-equivalent amount of downwelling, and the regional mean vertical velocity may not indicate the periods of maximum nutrient flux, especially if a nonlinear rectification mechanism is operative (McGillicuddy et al., 1998; Siegel et al., 1999). Time series of the mean and variance of vertical velocity (w) from our OGCM simulation in the region from 135–150 E and 3–10 N, inside and outside the NECC (defined as surface zonal velocity (u) eastward at >30 cm s1), are shown in Fig. 9 for the model layer immediately below the mixed layer. The variance shows a strong increase in early 1998, with lesser peaks with similar seasonal phasing in other years (such as 1997, when the meanders were visible in OCTS images: see Fig. 5). The mean does not show such a peak, nor is it greater within the NECC relative to grid points within the same spatial domain where up30 cm s1. Frequency histograms of w in the NECC (as defined above) for December through April 1997–98 and 1998–99 illustrate the El Nin˜o/ La Nin˜a difference (Fig. 9 bottom). The mean vertical velocity was negative in 1997–98 and positive in 1998–99, but the variance was significantly (po0:001; variance ratio F test) greater in 1997–98 (Fig. 9). The association of simultaneous upwelling and downwelling with ENSO and nutrient fluxes is further illustrated by considering the correlation of vertical velocity with other model diagnostic fields, satellite chlorophyll, and the SOI (Table 2). During the 1997–98 El Nin˜o, the speed and transport of the NECC increased (negative correlation

ARTICLE IN PRESS 220

J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228

Fig. 6. Daily composites (Standard Mapped Image) of SeaWiFS chlorophyll for 8 consecutive days in early March 1998.

ARTICLE IN PRESS J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228

221

January mean salinity and currents at σθ = 25 10°N

10°N

35.35 35.30 35.25 35.20 35.15 35.10 35.05 35.00 34.95 34.90 34.85 34.80 34.75 34.70 34.65 34.60 34.55

10 cm/s

5°N

5°N

0°

0°N

130°E

140°E

p s u

150°E

3

Fig. 7. Mean salinity and currents during January on the 1025 kg m density surface (1986-2001), from a 15-layer OGCM. Current vectors west of 130 E are not shown. The depth of the density surface in the vicinity of the NGCUC (westward flow along the bottom of the figure) is approximately 180 m.

10 0 -10

145°E IZT (Sv)

20

-20

35.2

Salinity

35.1 35 34.9 34.8 1996

1997

1998

1999

2000

Fig. 8. Time series of (top) depth and latitude (75–250 m, 3 S–2 N) integrated zonal transport (IZT) across 145 E and (bottom) mean salinity at 145 E averaged over the same depth and latitude range, from a 15-layer OGCM.

with SOI) and the upper-ocean heat content and SST decreased (positive correlation). The effects of vertical motion are shown by the correlation of these quantities with the regionally averaged vertical velocity in the model layer immediately

below the mixed layer (w) and the regional average of its absolute value (|w|). The correlation with |w| is consistently stronger, indicating that the intensification of the NECC during the El Nin˜o event is associated with a greater amount of vertical

ARTICLE IN PRESS J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228

222

4

100

x 10

4

Within NECC Outside NECC

3 σ2(w)

w(cm d-1)

50

0

1

-50

-100 1997

2

1998

1999

2000

0 1997

2001

1998

1999

2000

2001

1998 - 99

1997 - 98

400

300

w= -16±97

250

w=22±55 300

200

200

150 100

100 50 0 -500

0

500

Vertical velocity (cm d-1)

0 -500

0

500

Vertical velocity (cm d-1)

Fig. 9. (top) Time series of mean and variance of vertical velocity (w) in the NECC from a 15-layer OGCM, in the model layer immediately below the mixed layer. The NECC is defined here as model grid points within 135–150 E, 3–10 N where surface zonal velocity is eastward at >30 cm s1. (bottom) Frequency histograms of w within the NECC during December through April of 1997–98 (n ¼ 864) and 1998–99 (n ¼ 561); the total number of data is scaled to a common value for graphical purposes. The mean and standard deviation of w are listed for each histogram.

motion but not necessarily with net upward flow. All of these correlations are quite weak and statistical significance cannot be established for such a short time series, but the overall pattern is highly consistent. The correlation with the absolute value is stronger for all quantities except the heat content, which is integrated over all 15 model layers and is therefore not affected by shallow upwelling/downwelling. The signs of the correlations with |w| indicate increased vertical motion during the El Nin˜o event (negative correlation with SOI), associated with a strengthened NECC (positive correlation with NECC speed and transport), leading to a cooling of the sea surface (negative correlation with SST) and an increased flux of nutrients to the surface layer (positive correlation with surface chlorophyll). Ocean-color data are not available over the entire period 1996–1999, but strong (B0.8) correlations

with SST and the SOI over the period where data are available are consistent with the pattern described. When a longer period (1986–2001) is considered, the correlations between |w| and SST (r ¼ 0:30), surface current velocity (r ¼ 0:48) and the transport of the NECC (r ¼ 0:54) are statistically significant using the random-phase test, which is the most conservative method available, and is best suited to longer time series. However, |w| is not correlated with the SOI over this period, even when time lags are considered. The basin-scale current and thermocline depth anomalies associated with ENSO clearly play a very important role in the regional circulation, but there is also local variability that may be uncorrelated with ENSO. During the 1997–98 El Nin˜o, the two combined to produce the most intense blooms observed over 5 years of data.

ARTICLE IN PRESS J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228 Table 2 Correlation matrix for selected model outputs and satellite chlorophyll (CHL) averaged over the region of the North Equatorial Countercurrent meanders (3–10 N, 135–150 E), and SOI values for 42 months from January 1996 through June 1999. Correlation coefficients for CHL consider only the 30 months where data are available. CHL SOI IZT U0 SST HCA |w| W

SOI

IZT

U0

SST

HCA |w|

0.79 0.25 0.34 0.26 0.37 0.81 0.80 0.64 0.36 0.20 0.46 0.55 0.16 0.38 0.37 0.30 0.23 0.48 0.35 0.31 0.13 0.04 0.10 0.05 0.21 0.01 0.18

0.18

IZT=depth- and latitude-integrated zonal transport across 140 E, U0=zonal surface current velocity, SST=sea surface temperature, HCA=heat content anomaly, |w|=absolute value of vertical velocity, w=vertical velocity. An  indicates significance at the 0.05 level using the random-phase test of Ebisuzaki (1997); chlorophyll correlations were not tested for significance due to the absence of a continuous time series.

The effect of the 1997–98 El Nin˜o event on the modelled subsurface circulation is shown in Figs. 10 and 11. Upwelling and downwelling associated with current meandering is strong in January 1998 and weak in May, and the upwelled water is much cooler water due to an elevated thermocline in January (Fig. 10). These model ‘‘snapshots’’ are intended to illustrate graphically the hypothesized mechanism of vertical nutrient flux discussed above, and the dramatic differences that occurred in 1997–98 between El Nin˜o and post-El Nin˜o conditions. The above analysis illustrates how such simultaneous upwelling and downwelling may be associated with intensification of the NECC during El Nin˜o events. Attribution of the phytoplankton blooms shown in Figs. 2 and 3 to this mechanism is the most plausible explanation given the information available but can not be conclusively demonstrated. The strength of the Mindanao Dome shows strong ENSO variability as shown by the subsurface temperature structure of the ocean model simulation. The temperature at 100 m averaged from 140–145 E is shown in a latitude-time section in Fig. 11 along with the current vectors at the same depth and longitude. The shoaling of the

223

thermocline associated with the cyclonic circulation of the MD is stronger during 1997–98 than at any other time during the simulation. It is also enhanced during 1992–94, a period of persistent El Nin˜o-like conditions (Trenberth and Hoar, 1996). The NECC forms the southern boundary of the MD, and migrates latitudinally with the seasonal and interannual variability of the strength of the MD (Fig. 11). During the 1992–94 El Nin˜o the MD was strong and its position quite far south. The 1997–98 El Nin˜o appears as a briefer but more intense event where the position of the MD does not seem to have been as far south as in 1992– 1994, except for a brief period that coincided with the formation of the most intense NECC phytoplankton bloom (Figs. 2 and 3). At this time, the temperature at 100 m in the MD was the coldest (B15 C) in 15 years of simulation examined (Fig. 11). In the years since 1998 the upwelling of cold water by the MD seems to have been quite weak. Note that while the NECC shifts position along with the MD, maximal surface chlorophyll concentrations are always found in association with the NECC rather than the MD (Figs. 3 and 4). We attribute this to a combination of meanderinduced upwelling and the presence of nutrientrich waters on the southern side of the NECC.

4. Discussion Lukas et al. (1991) observed the NECC meanders in the tracks of drifters released during the WEPOCS-III expedition. They estimated the distance between meander peaks to be 700– 800 km and the ‘‘amplitude’’ of the meanders as 200-400 km. The scales observed in the ocean-color images are similar to these estimates (Figs. 2–4). Lukas et al. (1991) further noted that there was ‘‘no zonal displacement of the meander pattern with time which would be characteristic of phase propagation’’ (p. 7093). As the meanders are visible in monthly composite images, the oceancolor data appear to confirm that the meanders do not propagate, or do so very slowly. The analyses presented above show the importance of large-scale climatic phenomena in forcing relatively small-scale oceanographic variability in

ARTICLE IN PRESS J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228

224

January 1998 (El Niño) 0 -50 -100 -150 -200 -250 -300

135

145

155

Longitude (°E) 50 cm/s May 1998 (post El Niño) 0 -50 -100 -150 -200 -250 -300

135

145

155

Longitude (°E)

7

9

11

13

15

17

19

21

23

25

27

29

31

Temperature (°C) Fig. 10. Temperature and zonal and vertical current velocities from a 15-layer OGCM, averaged from 3–7 N, in January and May 1998. Vertical exaggeration is 10000 times.

this region, and the utility of ocean-color observations for identifying and documenting such events. The seasonal phasing of the bloom in 1997–98 is consistent with the seasonal cycle of the thermocline depth and upwelling described by Masumoto and Yamagata (1991), with the seasonal variation of wind-stress curl (Qiu and Lukas, 1996), and with the seasonal variation of westward transport of SPTW by the NGCUC in our ocean model

simulation. The weakness or absence of the phytoplankton bloom in subsequent years can most plausibly be attributed to a deep thermocline and a weaker NECC associated with the ongoing La Nin˜a conditions, and weak westward transport of nutrient-rich waters by the NGCUC (Figs. 8 and 11). The NGCUC transports both nutrient-rich (SPTW) and nutrient-poor (AAIW) water masses

ARTICLE IN PRESS J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228 18

225 29 28

50 cm/s 16

27 26

14 24

12

23 10

22 21

8

20

Temperature (˚C)

Latitude (˚N)

25

19 6 18 17

4

16 2

15 1992

1994

1996

1998

2000

Fig. 11. Latitude-time section of temperature and current velocities at a depth of 100 m averaged from 140–145 E, from a 15-layer OGCM forced with NCEP interannual winds. The NECC forms the southern boundary of the MD (region of low T from B6–12 N). Note thermocline shoaling during 1992–94 and 1997–98.

(Tsuchiya et al., 1989; Fine et al., 1994; Kashino et al., 1996; Kuroda, 2000). The AAIW occurs at too great depths to be of interest here, and is not resolved by our reduced-gravity model. We therefore consider whether the NGCUC may be a source of nutrient to the nascent NECC. The ocean model predicts salinities of 35.2–35.5 and temperatures of 18–20 C in the vicinity of the salinity maximum associated with the NGCUC, consistent with observations (Fine et al., 1994). Using the NODC climatologies (Conkright et al., 1994), we estimated the nitrate concentration at this location by regressing nitrate against temperature in the central South Pacific where the SPTW originates. Within a box bounded by 160 E, 160 W, 5 S, and 20 S, at depths of 75–300 m and salinities between 35.0 and 35.5, there is a strong correlation between temperature and nitrate (r ¼ 0:86; Po0:001). Based on this linear relationship and temperatures from the OGCM, we estimate that the intrusion of SPTW along the New Guinea coast carries a nitrate concentration of 10–12 mM. This is highly consistent with the results of Kaneko et al. (1998), who observed temperatures of 19–21 C and nitrate concentrations of 8–12 mM in the vicinity of the salinity maximum. This saline water also may strengthen

the salinity front associated with the NECC, leading to increased upwelling. This front is well defined and associated with a steep gradient of nitrate concentration in the P9 section (Kaneko et al., 1998). It is difficult to quantify the variability in the upwelling velocities and the vertical fluxes of nutrients, but the OGCM results suggest vigorous vertical motions associated with current meandering (Figs. 9 and 10). The mechanisms by which a meandering current induces upwelling are complex and not entirely understood (Bower, 1989; Flierl and Davis, 1993; Yoshimori, 1994), but the basic mechanism can be explained in relatively simple terms as vertical motion arising from imbalances in the horizontal (planetary, curvature, shear) components of vorticity (Bower, 1989; Hitchcock et al., 1993). The models of Flierl and Davis (1993) and Yoshimori (1994) are incomplete in not considering shear-induced vorticity, which Bower (1989) indicates is an important contributor to the vertical motion, but they correctly locate the along-stream position of the upwelling. Our attribution of the phytoplankton blooms to meander-induced upwelling is somewhat speculative and can not be conclusively demonstrated with the data available, but it explains several

ARTICLE IN PRESS 226

J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228

important aspects of the ocean-color observations. Firstly, the shoaling of the thermocline during an El Nin˜o event occurs over a wide area, but the biological response is strongly localized within the NECC meanders. Similarly, seasonal Ekman upwelling explains neither the spatial nor the temporal distribution of surface chlorophyll, although it probably makes a small but important contribution. Finally, examination of synoptic (daily SMI) SeaWiFS images (Fig. 6) indicates that the along-stream position of the color signal is consistent with the theory and observations that suggest that upwelling should occur primarily on the poleward-flowing arm of the meanders (Bower and Rossby; 1989; Bower; 1989; Flierl and Davis, 1993; Yoshimori, 1994). Furthermore, such meander-induced upwelling occurs locally even in nonEl Nino years (SeaWiFS daily SMI, not shown), but without thermocline shoaling the nutrient flux is too small to make the meanders visible in monthly composite images. These images must be interpreted somewhat cautiously, however, because along-stream advection rates (B50 km/day) are likely to be substantial relative to phytoplankton growth and loss time scales, i.e. high surface chlorophyll is not necessarily an indicator of local upwelling even in synoptic images (Flierl and Davis, 1993; Ashjian et al., 1994; Anderson and Robinson, 2001). The interactions of upwelling, growth, grazing, sinking, and advection could be examined more quantitatively with along-stream models like that utilized by Flierl and Davis (1993) or fully four-dimensional high-resolution models as employed by Anderson and Robinson (2001) but is beyond the scope of this paper. Both observations and model results show ENSO related variability that is consistent with increased zonal flow, upwelling and nutrient flux to the mixed layer in the NECC during the ENSO warm phase (Figs. 5, 10, and 11, Tables 1 and 2). This results from a confluence of several factors with both seasonal and interannual components. Firstly, the wind-stress becomes upwelling favorable (positive curl) in boreal winter (predominantly seasonal). This alone explains little of the total variability, but is generally consistent with the seasonal phasing of the blooms. Secondly, the thermocline shoals and the NECC intensifies,

bringing the nutricline closer to the surface and increasing meander-induced upwelling (predominantly interannual). Meander-induced upwelling probably occurs at other times, but nutrient fluxes would be small due to both a weaker current and a deeper thermocline. Finally, the transport of saline, nutrient-rich SPTW via the NGCUC increases, strengthening the salinity front and increasing nutrient concentrations on its southern side (predominantly seasonal, but somewhat weakened in the years following the 1997–98 El Nin˜o). The importance of this last process is relatively uncertain, but the data of Kaneko et al. (1998) suggest that, were it not present, much deeper upwelling would be required to produce an equivalent nutrient flux. During the 1997–98 El Nin˜o a confluence of all of these factors produced the ‘‘ribbon of dark water’’ shown in Fig. 2. Blooms of similar character have no doubt occurred in this region before, but have not been observed by oceanographers for lack of spacebased sensors. Dandonneau (1992) calculated empirical orthogonal functions (EOF’s) from chlorophyll data collected by ships of opportunity, and found that in the western Pacific one of his EOF modes was associated with an enhancement at the latitude of the NECC. The time series (principal component) of this mode appeared to be related to ENSO, with an enhancement of surface chlorophyll evident in the NECC during the 1982– 83 El Nin˜o. The bloom in 1997-98 was strongest in the late stages of the El Nin˜o event (Figs. 3 and 5), so future blooms in this region may be predictable months in advance; this could be advantageous both to oceanographers wishing to study the event more closely and to fishers and fisheries managers. When the original manuscript of this paper was written there had not yet been another El Nin˜o subsequent to 1998. In the revision, we incorporated data from 2002–03, and note that the blooms recurred at this time as we predicted that they would (Fig. 5), although the event was a very weak one (the highest value of the NINO3 SST anomaly index was 1.65 C vs. 3.92 C in 1997–98). While we have attempted to determine the mechanisms controlling the spatial and temporal evolution of these blooms, we acknowledge that our interpretations remain somewhat speculative. Detailed field

ARTICLE IN PRESS J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228

data will be required to determine the mechanisms with certainty. The images show significant variability over quite small space and time scales, suggesting that ocean-color imagery will need to be employed in real time for such observations to be successfully acquired.

Acknowledgements This research was supported the NASA Ocean Biogeochemistry program. We thank R. Lukas for providing the schematic diagram of western Pacific currents, and R. Lukas and D. Siegel for constructive comments on an earlier draft of this manuscript.

References Anderson, L.A., Robinson, A.R., 2001. Physical and biological modeling in the Gulf Stream region. Part II. Physical and biological processes. Deep-Sea Research I 48, 1139–1168. Ashjian, C.J., Smith, S.L., Flagg, C.N., Mariano, A.J., Behrens, W.J., Lane, P.V.Z., 1994. The influence of a Gulf Stream meander on the distribution of zooplankton biomass in the Slope Water, the Gulf Stream, and the Sargasso Sea, described using a shipboard acoustic Doppler current profiler. Deep-Sea Research I 41, 23–50. Boulanger, J.-P., Menkes, C., 1999. Long equatorial wave reflection in the Pacific Ocean from TOPEX/POSEIDON data during the 1992–1998 period. Climate Dynamics 15, 205–225. Bower, A.S., 1989. Potential vorticity balances and horizontal divergence along particle trajectories in Gulf Stream meanders east of Cape Hatteras. Journal of Physical Oceanography 19, 1669–1681. Bower, A.S., Rossby, T., 1989. Evidence of cross-frontal exchange processes in the Gulf Stream based on isopycnal RAFOS float data. Journal of Physical Oceanography 19, 1177–1190. Bray, N.A., Hautala, S., Chong, J., Pariwono, J., 1996. Largescale sea level, thermocline, and wind variations in the Indonesian throughflow region. Journal of Geophysical Research 101, 12239–12254. Chelton, D.B., Schlax, M.G., 1996. Global observations of oceanic Rossby waves. Science 272, 234–238. Chen, D., Rothstein, L.M., Busalacchi, A.J., 1994. A hybrid vertical mixing scheme and its application to tropical ocean models. Journal of Physical Oceanography 24, 2156–2179. Conkright, M., Levitus, S., Boyer, T., 1994. World Ocean Atlas 1994, Vol. 1: Nutrients, US Department of Commerce, Washington, 150pp.

227

Cresswell, G.R., 2000. Coastal currents of northern Papua New Guinea, and the Sepik River outflow. Marine and Freshwater Research 51, 553–564. Dandonneau, Y., 1992. Surface chlorophyll concentration in the tropical Pacific Ocean: an analysis of data collected by merchant ships from 1978 to 1989. Journal of Geophysical Research 97, 3581–3591. Davis, R.E., 1976. Predictability of sea surface temperature and sea level pressure anomalies over the North Pacific Ocean. Journal of Physical Oceanography 6, 249–266. Ebisuzaki, W., 1997. A method to estimate the statistical significance of a correlation when the data are serially correlated. Journal of Climate 10, 2147–2153. Emery W. J., Thomson, R.E., 2001. Data analysis methods in physical oceanography, 2nd ed., Elsevier, Amsterdam, 638pp. Fine, R.A., Lukas, R., Bingham, F.M., Warner, M.J., Gammon, R.H., 1994. The western equatorial Pacific: a water mass crossroads. Journal of Geophysical Research 99, 25063–25080. Flierl, G.R., Davis, C.S., 1993. Biological effects of Gulf Steam meandering. Journal of Marine Research 51, 529–560. Garver, S.A., Siegel, D.A., 1997. Inherent optical property inversion of ocean color spectra and its biogeochemical interpretation: 1. Time series from the Sargasso Sea. Journal of Geophysical Research 102, 18607–18625. Gent, P.R., Cane, M.A., 1989. A reduced gravity, primitive equation model of the upper equatorial ocean. Journal of Computational Physics 81, 444–480. Gouriou, Y., Toole, J., 1993. Mean circulation of the upper layers of the western equatorial Pacific Ocean. Journal of Geophysical Research 98, 22495–22520. Hitchcock, G.L., Mariano, A.J., Rossby, T., 1993. Mesoscale pigment fields in the Gulf Stream: observations in a meander crest and trough. Journal of Geophysical Research 98, 8425–8445. Johnson, G.C., McPhaden, M.J., Rowe, G.D., McTaggart, K.E., 2000. Upper equatorial Pacific Ocean current and salinity variability during the 1996–98 El Nin˜o–La Nin˜a cycle. Journal of Geophysical Research 105, 1037–1053. Johnston, T.M.S., Merrifield, M.A., 2000. Interannual geostrophic current anomalies in the near-equatorial western Pacific. Journal of Physical Oceanography 30, 3–14. Kalnay, E., et al., 1996. The NCEP/NCAR 40-year reanalysis project. Bulletin of the American Meteorological Society 77, 437–471. Kaneko, I., Takatsuki, Y., Kamiya, H., Kawae, S., 1998. Water property and current distributions along the WHP-P9 section (137-142 E) in the western North Pacific. Journal of Geophysical Research 103, 12959–12984. Kashino, Y., Aoyama, M., Kawano, T., Hendiarti, N., Syaefudin, Y., Anantasena, K., Muneyama, H., Watanabe, 1996. The water masses between Mindanao and New Guinea. Journal of Geophysical Research 101, 12391–12400. Kishino, M., Ishizaka, J., Saitoh, S., Senga, Y., Utashima, M., 1997. Verification plan of ocean color and temperature

ARTICLE IN PRESS 228

J.R. Christian et al. / Deep-Sea Research II 51 (2004) 209–228

scanner atmospheric correction and phytoplankton pigment by moored optical buoy system. Journal of Geophysical Research 102, 17197–17207. Kuroda, Y., 2000. Variability of currents off the northern coast of New Guinea. Journal of Oceanography 56, 103–116. Lindstrom, E., Lukas, R., Fine, R., Firing, E., Godfrey, S., Meyers, G., Tsuchiya, M., 1987. The Western Equatorial Pacific Ocean Circulation Study. Nature 330, 533–537. Lukas, R., Firing, E., Hacker, P., Richardson, P.L., Collins, C.A., Fine, R., Gammon, R., 1991. Observations of the Mindanao Current during the western equatorial Pacific ocean circulation study. Journal of Geophysical Research 96, 7089–7104. Mackey, D.J., O’Sullivan, J.E., Watson, R.J., 2002. Iron in the western Pacific: a riverine or hydrothermal source for iron in equatorial undercurrent. Deep-Sea Research I 49, 877–893. Mariano, A.J., Hitchcock, G.L., Ashjian, C.J., Olson, D.B., Rossby, T., Ryan, E., Smith, S.L., 1996. Principal component analysis of biological and physical variability in a Gulf Stream meander crest. Deep-Sea Research I 43, 1531–1565. Masumoto, Y., Yamagata, T., 1991. Response of the western tropical Pacific to the Asian winter monsoon: the generation of the Mindanao Dome. Journal of Physical Oceanography 21, 1386–1398. McClain, C.R., Cleave, M.L., Feldman, G.C., Gregg, W.W., Hooker, S.B., Kuring, N., 1998. Science quality SeaWiFS data for global biosphere research. Sea Technology 39 (9), 10–16. McGillicuddy, D.J., Robinson, A.R., Siegel, D.A., Jannasch, H.W., Johnson, R., Dickey, T.D., McNeil, J., Michaels, A.F., Knap, A.H., 1998. Influence of mesoscale eddies on new production in the Sargasso Sea. Nature 394, 263–266. McPhaden, M.J., Yu, X., 1999. Equatorial waves and the 1997– 98 El Nin˜o. Geophysical Research Letters 26, 2961–2964. Merrifield, M., Kilonsky, B., Nakahara, S., 1999. Interannual sea level changes in the tropical Pacific associated with ENSO. Geophysical Research Letters 26, 3317–3320. Meyers, G., Donguy, J.R., 1984. The North Equatorial Countercurrent and heat storage in the western Pacific Ocean during 1982-83. Nature 312, 258–260. Murtugudde, R., Busalacchi, A.J., Beauchamp, J., 1998. Seasonal-to-interannual effects of the Indonesian throughflow in the tropical Indo-Pacific Basin. Journal of Geophysical Research 103, 21425–21441. Murtugudde, R., Signorini, S.R., Christian, J.R., Busalacchi, A.J., McClain, C.R., Picaut, J., 1999. Ocean color variability of the tropical Indo-Pacific basin observed by SeaWiFS during 1997–98. Journal of Geophysical Research 104, 18351–18366.

Nelson, N.B., Siegel, D.A., 2002. Chromophoric DOM in the open ocean. In: Hansell, D.A., Carlson, C.A. (Eds.), Biogeochemistry of marine dissolved organic matter. Academic, New York, pp. 547–578. Obata, H., Karatani, H., Nakayama, E., 1993. Automated determination of iron in seawater by chelating resin concentration and chemiluminescence detection. Analytical Chemistry 65, 1524–1528. O’Reilly, J.E., Maritorena, S., Mitchell, B.G., Siegel, D.A., Carder, K.L., Garver, S.A., Kahru, M., McClain, C.R., 1998. Ocean color algorithms for SeaWiFS. Journal of Geophysical Research 103, 24937–24953. Osgood, K.E., Bane, J.M., Dewar, W.K., 1987. Vertical velocities and dynamical balances in Gulf Stream meanders. Journal of Geophysical Research 92, 13029–13040. Pond, S., Pickard, G.L., 1983. Introductory dynamical oceanography, 2nd edition, Pergamon, Oxford, 329pp. Qiu, B., Joyce, T.M., 1992. Interannual variability in the midand low-latitude western North Pacific. Journal of Physical Oceanography 22, 1062–1079. Qiu, B., Lukas, R., 1996. Seasonal and interannual variability of the North Equatorial Current, the Mindanao Current, and the Kuroshio along the Pacific western boundary. Journal of Geophysical Research 101, 12315–12330. Reynolds, R.W., Smith, T.M., 1994. Improved global sea surface temperature analyses using optimum interpolation. Journal of Climate 7, 929–948. Shimada, M., Oaku, H., Mitomi, Y., Murakami, H., Kawamura, H., 1999. Calibration of the ocean color and temperature scanner. IEEE Transactions on Geoscience and Remote Sensing 37, 1484–1495. Siegel, D.A., McGillicuddy, D.J., Fields, E.A., 1999. Mesoscale eddies, satellite altimetry, and new production in the Sargasso Sea. Journal of Geophysical Research 104, 13359–13379. Thiebaux, H.J., Zwiers, F.W., 1984. The interpretation and estimation of effective sample size. Journal of Climate and Applied Meteorology 23, 800–811. Trenberth, K.E., Hoar, T.J., 1996. The 1990–1995 El Nin˜oSouthern Oscillation event: Longest on record. Geophysical Research Letters 23, 57–60. Tsuchiya, M., Lukas, R., Fine, R.A., Firing, E., Lindstrom, E., 1989. Source waters of the Pacific Equatorial Undercurrent. Progress in Oceanography 23, 101–147. Yoshimori, A., 1994. Horizontal divergence caused by meanders of a thin jet. Journal of Physical Oceanography 24, 345–352. Zwiers, F.W., von Storch, H., 1995. Taking serial correlation into account in tests of the mean. Journal of Climate 8, 336–351.