Partial decoupling of primary productivity from upwelling in the

LETTERS
PUBLISHED ONLINE: 30 MAY 2016 | DOI: 10.1038/NGEO2722
Partial decoupling of primary productivity from
upwelling in the California Current system
Lionel Renault1*, Curtis Deutsch2, James C. McWilliams1, Hartmut Frenzel2, Jun-Hong Liang3,4
and François Colas5
Coastal winds and upwelling of deep nutrient-rich water along
subtropical eastern boundaries yield some of the ocean’s
most productive ecosystems1 . Simple indices of coastal wind
strength have been extensively used to estimate the timing
and magnitude of biological productivity on seasonal and
interannual timescales2 and underlie the prediction that anthropogenic climate warming will increase the productivity by
making coastal winds stronger3–6 . The effect of wind patterns
on regional net primary productivity is not captured by such
indices and is poorly understood. Here we present evidence,
using a realistic model of the California Current system and
satellite measurements, that the observed slackening of the
winds near the coast has little effect on near-shore phytoplankton productivity despite a large reduction in upwelling
velocity. On the regional scale the wind drop-off leads to
substantially higher production even when the total upwelling
rate remains the same. This partial decoupling of productivity
from upwelling results from the impact of wind patterns on
alongshore currents and the eddies they generate. Our results
imply that productivity in eastern boundary upwelling systems
will be better predicted from indices of the coastal wind that
account for its offshore structure.
Upwelling indices are based on a large-scale pressure-gradient estimate of the wind field, but the spatial structure of the surface winds
in eastern boundary upwelling systems (EBUS) is complex, and so
is the oceanic response. Alongshore winds are typically strongest
offshore, becoming weaker towards the coast owing to orography,
surface roughness, and air–sea interaction7 . The nearshore drop-off
in winds diminishes coastal upwelling, spreading it over a broader
offshore region with slower vertical velocities (‘Ekman pumping’).
It can also modulate the mean current structure8 . The partition of
the total wind-driven upwelling between rapid coastal and slower
offshore components has been suggested to influence the upper
trophic levels of the ecosystem9,10 . However, the impact of the wind
drop-off on mesoscale activity and total net primary productivity
(NPP) at a regional scale has not yet been assessed11 .
To investigate the influence of the coastal wind drop-off on NPP,
we conducted simulations of the California Current system with an
oceanic circulation and biogeochemical model (see Supplementary
Methods). The model is forced by realistic climatological surface
and open boundary conditions in three simulations that differ only
in the cross-shore gradient of alongshore wind, the main component
of mean wind stress curl (Fig. 1a). A base case (‘uniform’) is
constructed from satellite scatterometer wind data, with a simple
extrapolation from its reliable offshore measurements across its
blind zone to the shoreline11 . Two additional simulations are
conducted with wind stress reduced by 60% at the coast, consistent
with observations for the upwelling season, that is, spring and
summer7 . The cross-shore wind tapering distance is applied over
widths of 25 km (‘sharp’) and 80 km (‘wide’) (Fig. 1a), which span
the variation of the drop-off scale7 . Neither ‘sharp’ nor ‘wide’ profiles
can be considered the most realistic because the real drop-off scale
is not uniform7 . Model solutions are analysed along the central
California coast, between 38◦ N and 43◦ N and within 100 km from
shore, where the alongshore wind, eddy kinetic energy (EKE), and
biological productivity are all relatively high (see Supplementary
Fig. 1). Analyses are carried out for the spring season because
the reversal of coastal winds during this season initiates the
phytoplankton bloom timing12,13 , and an accumulation of surface
nutrients during that season ensures that higher productivity
persists into summer.
Consistent with Ekman theory, a stronger drop-off diminishes
the horizontal transport of surface water and thus the upwelling
into the photic zone near the coast (Fig. 1b). Despite the weaker
coastal upwelling, rates of NPP integrated over the photic zone
(0–70 m, Fig. 2d) and within 20 km of the coast do not decrease
and even slightly increase (Fig. 1c). In all cases, horizontal transport
is constrained by the same wind stress at 100 km, so that offshore
Ekman pumping compensates for differences in coastal upwelling,
to maintain a similar total upwelling mass flux. Nevertheless, the
integrated NPP over the photic zone (0–70 m) significantly increases
by 30% when using a broader wind drop-off (‘wide’, Fig. 1c). The
increases in NPP are even larger for regions closer to the shore
(a 36% increase from 40 km) and in surface waters (0–10 m: a
75% increase from 0–100 km; not shown). These results imply that
despite being limited by N supply, NPP is not simply related to the
strength of wind-driven upwelling, either at the coast or on a broader
regional scale.
To identify the oceanic wind response that decouples NPP from
upwelling rates, we computed the nutrient budget of the photic zone
(0–70 m) within 100 km of the shore along the central California
coast (38◦ –43◦ N). Although the model accounts for multiple potentially limiting nutrients, the reservoir and flux of nitrate (NO3 − ) are
what limit overall production in the photic zone. Its budget can be
expressed (see Supplementary Methods) as:
∂N
= Fmean + Feddy − J (N )
∂t
1 Department of Atmospheric and Oceanic Sciences, UCLA, 405 Hilgard Avenue, California 90095-1565, USA. 2 University of Washington, School of
Oceanography, Box 357940 Seattle, Washington 98195-7940, USA. 3 Department of Oceanography and Coastal Sciences, Louisiana State University,
Baton Rouge, Louisiana 70803, USA. 4 Center for Computation and Technology, Louisiana State University, Baton Rouge, Louisiana 70803, USA. 5 Institut
de Recherche pour le développement (IRD), UMR LOCEAN, IRD/Sorbonne Universités (UPMC Univ Paris 06)/CNRS/MNHN, 4 Place Jussieu,
Paris Cedex 75252, France. *e-mail: [email protected]
NATURE GEOSCIENCE | VOL 9 | JULY 2016 | www.nature.com/naturegeoscience
© 2016 Macmillan Publishers Limited. All rights reserved
505
NATURE GEOSCIENCE DOI: 10.1038/NGEO2722
LETTERS
a
Wind factor
Factor
1.0
0.5
0.0
−100
Vertical velocity (m d−1)
b
8
6
4
Uniform
Sharp
Wide
−80
−60
−40
−20
Cross-shore distance, d (km)
0
20
0
20
0
20
Vertical velocities
Uniform: 9.4 × 104 m2 d−1
Sharp: 9.8 × 104 m2 d−1
Wide: 9.2 × 104 m2 d−1
2
0
−2
−100
−80
c
−60
−40
−20
Cross-shore distance, d (km)
NPP (102 mmol m−1 s−1)
NPP
15
10
Uniform: 5.54 × 107 mmol s−1
Sharp: 6.32 × 107 mmol s−1
Wide: 7.22 × 107 mmol s−1
5
−100
−80
−60
−40
−20
Cross-shore distance, d (km)
Figure 1 | Impact of wind drop-off on total upwelling and NPP. a, Coastal
wind profile factor applied to the wind product. b, Mean vertical velocity at
70 m depth between 38◦ N and 43◦ N during spring (April–June). The total
upwelling rate integrated over a distance of 100 km offshore is indicated in
the legend. c, NPP integrated over the photic zone (0–70 m depth) and
between 38◦ N and 43◦ N during spring. The total NPP (integrated over a
distance of 100 km offshore) is indicated in the legend. The shaded areas
represent the standard deviations. The means and standard deviations are
estimated using 8 years of simulations.
where N represents nitrate concentration and (∂N /∂t) is the
buildup of nitrate inventory change over time according to the
imbalance between total physical nitrate transport, due to both
the time-averaged flow (Fmean ) and its fluctuations (Feddy ), and the
total nitrate uptake by the ecosystem, J (N ). The biological uptake
of nitrate approximates the net community production of organic
nitrogen (dissolved and particulate) that is exported to depth.
During spring, when winds become upwelling favourable, the
physical nitrate supply (Fmean + Feddy ) exceeds the rate of ecosystem
production in all cases, leading to a significant buildup of the
nitrate reservoir (Fig. 2a). The larger integrated NPP in cases with
a wind drop-off (‘sharp’ and ‘wide’) is reflected in J (N ) and the
more rapid accumulation of surface nutrients (∂N /∂t) ensures
that higher productivity persists into summer. In the case of a
broader wind drop-off (‘sharp’ and ‘wide’), both the increase of
J (N ) and (∂N /∂t) is due to the higher physical nutrient delivery.
As the boundary layer is shallower than 70 m (estimated from the
K-profile parameterization, not shown), the effect of wind shape on
nutrient supply must be due to changes in advective transport rather
than mixing.
A broader wind drop-off alters significantly the mean alongshore
currents, especially the coastal undercurrent, which transports
506
high-nutrient tropical water poleward. Consistent with Sverdrup
dynamics, a stronger wind stress curl in ‘sharp’ and ‘wide’ yields
a stronger, shallower poleward flow8 that induces a weaker mean
southward flow between 0–70 m depth (Fig. 2b), and a higher
nitrate transport than in ‘uniform’. For the coastal region (0–20 km),
the reduction of the vertical velocity from ‘uniform’ to ‘wide’ (by
54%, Fig. 1b) is offset by the stronger alongshore undercurrent
that brings higher nutrient water below the photic layer. However,
the increase of nitrate transport does not continue from ‘sharp’
to ‘wide’, in spite of a further strengthening and shoaling of the
mean undercurrent. Thus, the change in the mean undercurrent
only partially explains the larger advective supply rate induced by
a broader wind drop-off. Moreover, the effect of wind shape on the
alongshore nutrient supply is offset by changes in nutrient transport
to the east, such that horizontal nutrient fluxes in the photic layer
as a whole are insensitive to the wind drop-off (Fig. 2a). Hence, the
nitrate supply by mean vertical velocities is also roughly insensitive
to the wind drop-off (Supplementary Fig. 8). Together, these results
imply a major role for nutrient transport by fluctuating components
of the circulation.
Although the wind structure has a relatively small effect on
Fmean , it exerts a strong indirect influence on the mesoscale eddies
that also have a net nutrient transport (Feddy ). A broader wind
drop-off in ‘sharp’ and ‘wide’ weakens the vertical shear of the
alongshore current below the thermocline, flattens the isopycnal
tilt, and reduces the EKE (Fig. 2c). The rate at which EKE is
converted from eddy potential energy by baroclinic instability is
diagnosed from the integrated eddy vertical buoyancy flux, which
is reduced under the wind drop-off (Supplementary Fig. 6). In
the EBUS, eddies have been shown to reduce NPP by subducting
nutrients along isopycnal surfaces that plunge below the euphotic
layer offshore, termed ‘eddy quenching’14 . Indeed, all our cases show
a negative (downward) eddy nitrate flux during spring (Fig. 2d),
except in the first 10 km nearshore, where the upwelling prevails. By
diminishing the mesoscale eddy activity, a broader wind drop-off in
‘sharp’ and ‘wide’ weakens significantly this removal of nitrate from
the photic zone and thus allows its more complete consumption.
This inverse relationship between ‘eddy quenching’ and the wind
drop-off accounts for the minimal response of NPP to winds near
the coast, and the shallow subduction of nitrate contributes to the
higher NPP offshore under a strong wind drop-off14,15 .
Ekman theory and model simulations predict that the wind stress
magnitude is the main driver of productivity and, further, that wind
drop-off modulates NPP by being negatively correlated with EKE,
but positively correlated with phytoplankton growth and biomass.
We tested these predictions using satellite data for chlorophyll a (a
proxy for NPP, from SeaWiFS), wind stress (from QuikSCAT), wind
stress curl (a measure of the wind drop-off), and the EKE (from the
Archiving, Validation, and Interpretation of Satellite Oceanographic
data (AVISO)) (Fig. 3a–c). The scatterometer blind zone near the
coast allows only a partial sampling of the wind drop-off profile16 ;
therefore, a positive wind curl anomaly can be interpreted as a
broader wind drop-off more fully sampled by QuikSCAT, and a
negative anomaly implies a sharper wind drop-off, more of which
occurs within the blind zone. The offshore wind stress gives a reliable
metric for the total upwelling.
Over the period of overlapping satellite records (2000–2009),
interannual fluctuations in the wind stress, wind stress curl,
EKE and chlorophyll are significantly correlated. Consistent with
earlier studies17 , the mean upwelling is the main driver of the
productivity. However, years with a larger wind stress curl generally
have smaller EKE and larger chlorophyll a (Fig. 3d). The former
link implicates eddy modulation by wind-induced changes in
the unstable alongshore currents, and the latter link supports
eddy quenching of NPP. The importance of the wind drop-off in
modulating total NPP is most evident in years when anomalies
NATURE GEOSCIENCE | VOL 9 | JULY 2016 | www.nature.com/naturegeoscience
© 2016 Macmillan Publishers Limited. All rights reserved
NATURE GEOSCIENCE DOI: 10.1038/NGEO2722
a
20
Rate (106 mmol s−1)
15
LETTERS
Nutrient budget
Uniform
Sharp
Wide
Horizontal transport
Vertical transport
10
5
0
−5
−10
Storage
b
Uptake
Transport
Mean
c
Uniform
Sharp
Wide
d
Uniform
Sharp
Wide
44
40
38
42
Depth (m)
Latitude (° N)
42
40
38
36
36
−0.2
0
0.2
Alongshore current (m s−1)
Total
0 100 200 300 400
EKE (cm2 s−2)
Mean
Eddy
Sharp
Uniform
−20
−40
−60
−80
−100
−120
−140
−160
−180
−200
−100
Total
Wide
15
10
5
0
−5
(mmol m−2 s−1)
Latitude (° N)
44
Eddy
−10
−50
d (km)
0
−100
−50
d (km)
0
−100
−50
d (km)
0
−15
Figure 2 | Wind drop-off control of the NPP by modulation of the eddy physical fluxes. a, Nutrient budget during spring between 38◦ N and 43◦ N from
70 m depth to the surface from the three model experiments. Storage is ∂N/∂t. Uptake is J(N). Transport is F. b, Alongshore current during spring, averaged
over a cross-shore band 100 km wide and 70 m in depth. The dashed black lines indicate the 38◦ –43◦ N region. c, Mean surface EKE, averaged over a
cross-shore band 100 km wide. The dashed black lines indicate the 38◦ –43◦ N region. d, Mean vertical eddy nutrient supply during spring averaged
between 38◦ N to 43◦ N. The black lines indicate the corresponding mean simulated euphotic depth. d, cross-shore distance. The error bars (a) and shaded
areas (b,c) represent the standard deviations estimated using 8 years of simulation.
33
30
129 126 123 120 117
Longitude (° W)
c
150
42
100
39
36
50
Latitude (° N)
36
45
3.00
42
1.00
0.60
39
0.20
0.10
36
0.05
33
33
30
45
129 126 123 120 117
Longitude (° W)
d
30
0
Chl-a (mg m−3)
39
b
EKE (cm2 s−2)
Latitude (° N)
42
10
8
6
4
2
0
−2
−4
−6
−8
−10
Latitude (° N)
45
Wind stress curl (N m−2 per 104 km)
a
129 126 123 120 117
Longitude (° W)
Indices
Class 1: reinforcing processes
Class 2: counteracting processes
1
0
−1
2001
2002
2006
2004
2005
2007
Year
2008
2009
2000
2003
Wind index
Curl index
EKE index
Chl-a index
c(Wind, Curl) = 0.1
c(Wind, EKE) = −0.2
c(Wind, Chl-a) = 0.6
c(Curl, EKE) = −0.5
c(Curl, Chl-a) = 0.3
c(EKE, Chl-a) = −0.3
Figure 3 | An upwelling index that considers wind structure, and perhaps eddy activity, would better predict interannual NPP variations. a, Mean wind
stress curl from QuikSCAT during spring. Mean wind stress magnitude is superimposed with dashed black contours. The solid black contour indicates the
location where the indices and the budget shown in Fig. 2 are computed. b, EKE from AVISO. c, Chlorophyll a (Chl-a) from SeaWiFS. d, Indices of variability
in wind stress, wind stress curl, EKE and chlorophyll a. The indices are computed by subtracting the mean value over the area indicated by the solid black
contour in a–c from 2000–2009 in spring, and dividing the resulting anomalies by the largest magnitude over the time period. The correlations among the
indices are listed, and they all are significant at the 95% level. The years are divided into two categories: when the mean upwelling and wind drop-off effect
are reinforcing; and when they are counteracting. Similar results are found for the summer season.
NATURE GEOSCIENCE | VOL 9 | JULY 2016 | www.nature.com/naturegeoscience
© 2016 Macmillan Publishers Limited. All rights reserved
507
NATURE GEOSCIENCE DOI: 10.1038/NGEO2722
LETTERS
in wind stress curl counteract, rather than reinforce, the changes
in coastal upwelling (Fig. 3d). Such counteracting years occur in
roughly half the years in the available time series, when stronger
coastal winds are associated with weaker wind stress curl, or vice
versa. In three out of four such years in the satellite record, the
NPP anomaly is in the direction predicted by the wind drop-off, and
opposite to what would be expected by the anomaly in coastal wind.
Satellite estimates thus are consistent with the model predictions
and suggest that relatively complex indices, not solely based on
simple wind time series, will be needed to predict interannual NPP
variations in eastern boundary upwelling systems.
This consistency between satellite remote sensing and regional
modelling experiments supports a new eddy-mediated link between
the coastal wind pattern and biological productivity, at least for
the California Current system. The same mechanisms are likely
to be present in other EBUS, albeit to varying degrees that reflect
differences in wind structure, ocean stratification, and nutrient
limitation factors. For example, the steeper coastal orography of
the Andes in South America may induce a broader wind drop-off7 ,
which may explain the weak nearshore generation of EKE18 , thus
yielding less eddy quenching of nutrients and a more productive
system17 . Similarly, the density stratification has a strong influence
on baroclinic energy conversion and EKE levels, so the wind dropoff effect can only partially explain the EKE difference between
the EBUS.
As the mean upwelling is the main driver of the productivity,
indices based on large-scale winds remain useful to predict the
overall tendencies of coastal marine productivity. However, our
findings help explain residual interannual variations of NPP in the
EBUS and demonstrate the need for better predictors than indices
based on large-scale winds alone19 . Predicting how productivity
in EBUS will react to future climate change will require regional
atmospheric and/or coupled models that adequately resolve the
wind drop-off profile and the ocean–atmosphere interactions20 , as
well as changes in the oceanic state that modulate its effects on eddydriven nutrient supply.
Methods
Methods, including statements of data availability and any
associated accession codes and references, are available in the
online version of this paper.
Received 6 August 2015; accepted 25 April 2016;
published online 30 May 2016
References
1. Carr, M. E. & Kearns, E. J. Production regimes in four Eastern Boundary
Current systems. Deep-Sea Res. II 50, 3199–3221 (2003).
2. Bograd, S. J. et al. Phenology of coastal upwelling in the California Current.
Geophys. Res. Lett. 36, 1602 (2009).
3. Bakun, A. Global climate change and intensification of coastal ocean upwelling.
Science 247, 198–201 (1991).
4. Wang, D., Gouhier, T. C., Menge, B. A. & Ganguly, A. R. Intensification and
spatial homogenization of coastal upwelling under climate change. Nature 518,
390–394 (2015).
508
5. Sydeman, W. J. et al. Climate change and wind intensification in coastal
upwelling ecosystems. Science 345, 77–80 (2014).
6. Bakun, A. et al. Anticipated effects of climate change on coastal upwelling
ecosystems. Curr. Clim. Change Rep. 1, 85–93 (2015).
7. Renault, L., Hall, A. & McWilliams, J. C. Orographic shaping of US West Coast
wind profiles during the upwelling season. Clim. Dynam. 46, 273–289 (2016).
8. Song, H., Miller, A. J., Cornuelle, B. D. & Di Lorenzo, E. Changes in upwelling
and its water sources in the California Current System driven by different wind
forcing. Dyn. Atmos. Oceans 52, 170–191 (2011).
9. Rykaczewski, R. R. & Checkley, D. M. Influence of ocean winds on the pelagic
ecosystem in upwelling regions. Proc. Natl Acad. Sci. USA 105,
1965–1970 (2008).
10. Jacox, M. G., Moore, A. M., Edwards, C. A. & Fiechter, J. Spatially resolved
upwelling in the California Current System and its connections to climate
variability. Geophys. Res. Lett. 41, 3189–3196 (2008).
11. Capet, X. J., Marchesiello, P. & McWilliams, J. C. Upwelling response to coastal
wind profiles. Geophys. Res. Lett. 31, 13 (2004).
12. Barth, J. A. et al. Delayed upwelling alters nearshore coastal ocean ecosystems
in the northern California current. Proc. Natl Acad. Sci. USA 104,
3719–3724 (2007).
13. Thomas, A. C. & Brickley, P. Satellite measurements of chlorophyll
distribution during spring 2005 in the California Current. Geophys. Res. Lett.
33, L22S05 (2006).
14. Gruber, N. et al. Eddy-induced reduction of biological production in eastern
boundary upwelling systems. Nature Geosci. 4, 787–792 (2011).
15. Nagai, T. et al. Dominant role of eddies and filaments in the offshore transport
of carbon and nutrients in the California Current System. J. Geophys. Res. 120,
5318–5341 (2015).
16. Renault, L. et al. Impact of atmospheric coastal jet off central Chile on sea
surface temperature from satellite observations (2000–2007). J. Geophys. Res.
114, C08006 (2009).
17. Chavez, F. P. & Messié, M. A comparison of eastern boundary upwelling
ecosystems. Prog. Oceanogr. 83, 80–96 (2009).
18. Colas, F., McWilliams, J. C., Capet, X. & Kurian, J. Heat balance and eddies in
the Peru-Chile current system. Clim. Dynam. 39, 509–529 (2012).
19. García-Reyes, M., Largier, J. L. & Sydeman, W. J. Synoptic-scale upwelling
indices and predictions of phyto- and zooplankton populations. Prog.
Oceanogr. 120, 177–188 (2014).
20. Renault, L. et al. Modulation of wind-work by oceanic current interaction with
the atmosphere. J. Phys. Oceanogr. 46, 1685–1704 (2016).
Acknowledgements
We appreciate support from the Office of Naval Research (N00014-12-1-0939), National
Science Foundation (OCE-1419450 and OCE-1419323), Bureau of Ocean Energy
Management, and California Ocean Protection Council, as well as computing resources
from the Extreme Science and Engineering Discovery Environment and on the
Yellowstone cluster (ark:/85065/d7wd3xhc) provided by NCAR’s Computational and
Information Systems Laboratory, sponsored by the National Science Foundation.
Author contributions
L.R., J.C.M. and C.D. conceived and designed the experiments; L.R. performed the
experiments; L.R., C.D., J.C.M., H.F. and F.C., analysed the data; L.R., H.F. and J.-H.L.
contributed materials/analysis tools; L.R., C.D. and J.C.M. co-wrote the paper.
Additional information
Supplementary information is available in the online version of the paper. Reprints and
permissions information is available online at www.nature.com/reprints.
Correspondence and requests for materials should be addressed to L.R.
Competing financial interests
The authors declare no competing financial interests.
NATURE GEOSCIENCE | VOL 9 | JULY 2016 | www.nature.com/naturegeoscience
© 2016 Macmillan Publishers Limited. All rights reserved
NATURE GEOSCIENCE DOI: 10.1038/NGEO2722
LETTERS
Methods
Model configuration. The oceanic simulations were performed with the Regional
Oceanic Modeling System (ROMS)21 . ROMS is a free-surface, terrain-following
coordinate model with split-explicit time stepping and Boussinesq and hydrostatic
approximations. The model extends from 142.1◦ W to 114.4◦ W and from 23.9◦ N
to 50.0◦ N (see Supplementary Fig. 1). The model grid has 627 × 377 points with a
horizontal resolution of 4 km and has 42 vertical levels. The vertical grid is stretched
for increased resolution of the surface and bottom boundary layers. The bottom
topography is derived from an SRTM30 database22 . The boundary condition
algorithm consists of a modified Flather-type scheme for the barotropic mode23
and Orlanski-type scheme for the baroclinic mode (including temperature and
salinity; ref. 24). The simulation is forced at the surface by the QuikSCAT-based
daily product described in ref. 25 (based on the SCOW climatology). Heat and
freshwater atmospheric forcing are from the Comprehensive Ocean–Atmosphere
Data Set26 . The freshwater atmospheric forcing has an additional restoring term to
prevent surface salinity from drifting away from climatological values. This weak
restoring is towards climatological monthly surface salinity from the World Ocean
Atlas27 . A flux correction term is included in the atmospheric heat forcing to allow
feedback from the ocean to the atmosphere following the formulation of ref. 28. As
in ref. 29, initial and boundary information are taken from a 12 km Pacific
climatological solution and the model is run for ten years. Ref. 25 has full
information about a similar Pacific simulation at coarser resolution.
The Biogeochemical Elemental Cycling30 model is coupled to ROMS. It
includes multiple limiting nutrients (N, P, Fe and Si) and three phytoplankton
functional groups (diatoms, diazotrophs, and small phytoplankton) that represent
the biogeographical variability of different oceanic biomes, for example, highly
productive coastal regimes versus the oligotrophic open ocean areas of the
subtropical gyres. It includes the dissolved iron cycle, including inputs of iron from
sediments and from atmospheric dust deposition. The degree of realism of the
simulation here is similar to the results of ref. 14.
A set of three experiments has been carried out. The only difference between
them is the coastal wind profile used to force the model (Fig. 1a). ‘uniform’ is the
control run, the QuikSCAT SCOW wind is interpolated onto the ROMS grid, and
the missing coastal values are tapered using a simple extrapolation11 . As QuikSCAT
monitors only partially the wind drop-off in this region, we consider this
experiment as without wind drop-off. ‘sharp’ and ‘wide’ add a wind drop-off using
the factors shown in Fig. 1a. As a result, ‘sharp’ and ‘wide’ have wind reduction by
60% and a wind drop-off length of 25 km and 80 km, respectively. Note, as shown
by ref. 7, the wind drop-off is not uniform and presents latitudinal variation both in
length and wind reduction. The values chosen for this study are in the range of
values found in that former study. However, for the purpose of this study, idealized
experiments using such a wind modification allow us to assess how the coastal
wind shape controls mesoscale activity and NPP.
In this study, the winter, spring, summer and autumn seasons correspond
to the months (January–March), (April– June), (July–September) and
(October–December), respectively.
Budget analysis. To analyse the role of eddies in the nutrient evolution, we
decompose the advective flux of any nutrient concentration C into time-mean and
fluctuation (eddy) components:
UC = Ū C̄ + U 0 C 0
(1)
where U denotes the three-dimensional velocity. As in ref. 31, where an analysis is
made for C equal to the buoyancy, the ‘instantaneous’ simulation outputs are 2-day
averages, and the mean is defined as a multi-year seasonal average. In the following
we denote the alongshore and cross-shore currents as v and u and the offshore
distance as d. The balance equation for any C is
∂C
= ∇ · K · ∇(C) − ∇h · uh C − ∇v · wC − J (C)
∂t
(2)
where K is the eddy kinematic diffusivity tensor, ∇ is the three-dimensional
gradient operators, and uh , and w are the horizontal and vertical velocities, and
J (C) is the biogeochemical source minus sink term.
We present the budget for inorganic nitrogen, because it is the nutrient that
ultimately limits biological productivity at a regional scale in the California
Current. We further restrict our analysis to nitrate (NO3 − ), because it is by far the
largest reservoir and physical supply of inorganic nitrogen (compared with
ammonium transport and concentration, which are both very small). Nitrate also
has the advantage of not being sensitive to the internal ecosystem nutrient
transformations. The only sinks of nitrate are from phytoplankton uptake, and its
only biological source is nitrification, which is inhibited by light in the photic zone.
The biological uptake of nitrate in the model is thus equal to ‘new production’, and
balances the portion of NPP that is exported to depth.
From equation (2) a mean-seasonal balance is estimated for the 38◦ –43◦ N ×
100 km cross-shore × 70 m depth region, which includes both the turbulent surface
boundary layer and euphotic zone in spring. By 70 m depth the mixing term
becomes negligible. Equation (2) will be first analysed in the reduced form,
dN
= Ftot − J (N )
dt
(3)
where N represents nitrate concentration, Ftot is the total physical nitrate transport,
and J (N ) the total nitrate uptake by the ecosystem. This budget analysis is then
further decomposed between mean transport (Fmean ) and eddies (Feddies ) following
equation (1),
∂N
= Fmean + Feddy − J (N )
(4)
∂t
and between horizontal (Fhor ) and vertical transport (Fver ),
∂N
= Fhor(mean) + Fhor(eddy) + Fver(mean) + Fver(eddy) − J (N )
∂t
(5)
Data availability and description. QuikSCAT wind stress. The near-surface
atmospheric circulation over the ocean is described through daily QuikSCAT zonal
and meridional wind components, obtained from Centre ERS d’Archivage et de
Traitement on a 0.25◦ × 0.25◦ resolution grid32 . This product is built from both
ascending and descending passes from discrete observations (available in
JPL/PO.DAAC Level 2B product) over each day. Standard errors are also computed
and provided as complementary gridded fields.
SeaWiFS chlorophyll a. Surface chlorophyll concentrations were estimated from
SeaWiFS data33 for the 2000–2009 period. We used Level 3 (9 km) monthly
composites obtained from the Distributed Active Archive Center at NASA
Goddard Space Flight Center.
AVISO sea level anomalies. The sea level anomalies come from the Archiving,
Validation, and Interpretation of Satellite Oceanographic data (AVISO)
multimission mapped altimetry product34 . We use the Delayed Time 2014 version,
in which data from at least two (up to four) simultaneous satellite altimeter
missions were merged and mapped onto a 0.25◦ Mercator grid at daily intervals for
the period October 1992–December 2013; the 1993–1999 mean was removed at
each grid point. The surface geostrophic currents are computed by using the sea
level anomalies and the eddy kinetic energy (EKE) is defined as 1/2(u02 + v 02 )
where u0 and v 0 are velocity perturbations relative to a seasonal time-mean (same
method is applied when estimating the EKE from the model).
California Cooperative Oceanic Fisheries Investigation. Large-scale systematic
hydrographic sampling of the California Current system was initiated in 1949 by
the California Cooperative Oceanic Fisheries Investigations (CalCOFI)
programme. Since 1950, stations have been repeatedly occupied at varying intervals
based on a geographically fixed grid. In this study, lines 60 (off Point Reyes; ∼38◦ N)
and 67 (∼37◦ N), which have enough data to estimate a seasonal climatology of
respectively chlorophyll a and temperature, are used to validate the simulations.
Satellite analysis. Indices of wind stress, wind stress curl, EKE and chlorophyll a as
estimated over the black rectangle in Fig. 3 during spring are computed using the
following method: the long-time mean value over the black rectangle during spring
over the period 2000–2009 is calculated first. Then, the indices are calculated by
computing the anomalies of spring rectangle values with respect to the long-time
mean value and are finally normalized by the largest magnitude over the
time period.
Mean variability of chlorophyll a, wind stress and EKE. Model solutions are
analysed along the central California coast during spring, between 38◦ N and 43◦ N
and within 100 km from shore. The spring season is chosen because the reversal of
coastal winds during this season initiates the phytoplankton bloom timing and an
accumulation of surface nutrients during that season will ensure that higher
productivity persists into summer. Additionally, during spring, as illustrated in
Supplementary Fig. 1, the alongshore wind stress, EKE and biological productivity
are all relatively high.
Model evaluation and eddy buoyancy fluxes. To illustrate the realism of the
simulations, Supplementary Figs 2–6 depict some basic diagnostics of both physical
and biogeochemical fields. Supplementary Fig. 2 shows the sea surface temperature
mean from ‘uniform’ and in situ observations (World Ocean Atlas27 ). The
simulated sea surface temperature mean and variability are fairly reproduced by the
model, which clearly shows the upwelling signature. Supplementary Fig. 3
represents simulated and observed vertical distributions of temperature during
spring along CalCOFI line 67 (which starts around 37◦ N). It indicates that the
simulated vertical gradients are captured well. However, the onshore slope is
slightly overestimated, particularly in the nearshore region (Supplementary Fig. 2).
Supplementary Fig. 4a shows the EKE computed from ‘uniform’ using low-pass
NATURE GEOSCIENCE | www.nature.com/naturegeoscience
© 2016 Macmillan Publishers Limited. All rights reserved
NATURE GEOSCIENCE DOI: 10.1038/NGEO2722
LETTERS
filtered (7-day averaging and Gaussian spatial filter with 30-km half-width)
geostrophic velocities. The realism of the EKE indicates the ability of the model to
reproduce the mesoscale activity (AVISO EKE is shown in Supplementary Fig. 4b)
but also the mean current because mesoscale eddies arise from mean currents
instabilities. Recently, ref. 20 showed that current feedback to the atmosphere
dampens the EKE. Here, the offshore EKE is overestimated. The absence of current
feedback in the model (fluxed-forced) induces an overestimation of the eddy life,
allowing eddies to propagate further offshore.
Supplementary Fig. 5 depicts the mean chlorophyll a from ‘uniform’ and
SeaWiFS during spring. As expected, the coastal upwelling region is marked by
high concentrations of chlorophyll a. Model-simulated chlorophyll a and the
observations have similar agreements and disagreements as in ref. 14. There is an
overall tendency for the model to be biased low. The largest underestimation
occurs in the nearshore areas. Offshore, a low bias is found that is likely to be due to
the absence of picoplankton (which grow under oligotrophic conditions) in the
model. Finally, Supplementary Fig. 6 shows simulated (from ‘uniform’ and ‘sharp’)
and observed vertical distributions of chlorophyll a during spring along CalCOFI
line 60 (starting around 38◦ N). ‘uniform’ underestimates the coastal chlorophyll a
concentration. By using a broader wind drop-off (here ‘sharp’, a similar increase is
found in ‘wide’), the chlorophyll a concentration increases (see main paper),
becoming more realistic with respect to the observations. This also illustrates the
sensitivity of the simulated chlorophyll a to the coastal wind shapes.
A broader wind drop-off diminishes the southward surface current, strengthens
the undercurrent (Fig. 2b), and even can induce a surfacing of the undercurrent in
‘wide’ (not shown). This is consistent with Sverdrup dynamics in response to wind
drop-off: a positive wind stress curl produces a barotropic poleward flow that adds
up to the coastal undercurrent35,36 . As a result, the undercurrent strength is larger
with a broader wind drop-off. Note, the surfacing of the undercurrent in ‘wide’ is
not realistic (not shown); an overestimation of the wind drop-off length can induce
such a feature. A broader wind drop-off not only changes the undercurrent depth
and intensity but also induces a different vertical shear of the alongshore current.
From ‘uniform’ to ‘wide’, the vertical shear diminishes below the thermocline,
stabilizing the water column. This is confirmed by Supplementary Fig. 7, which
depicts the mean vertical buoyancy flux w 0 b0 from all of the experiments during
spring. Negative values are important as they indicate regions where eddies act
locally contrary to the baroclinic instability theoretical expectation of positive
conversion of available potential to kinetic energy. In the stratified interior, eddy
buoyancy flux acts to balance the effect of upwelling-favourable winds, that is, to
flatten the tilted upper thermocline31 . As a result, the offshore eddy restratification
flux weakens in the progression from ‘uniform’ to ‘wide’, and there is a similar
weakening in the eddy destratification flux near the coast. By inducing a weaker
vertical shear of the alongshore current, a broader drop-off weakens the energy
flux associated with upper ocean baroclinic instability and then reduces the
EKE (Fig. 2c). A broader wind drop-off, by reducing the coastal wind
stress, also weakens the coastal wind work35 , acting again towards a reduction of
the EKE.
Supplementary Fig. 8 depicts a dissolved inorganic nitrogen budget of the
photic zone (0–70 m) within 20 km of the shore along the central California coast
◦
(38 –43◦ N) during spring (N , nitrate concentration). The effect of the reduction of
the vertical velocities from ‘uniform’ to ‘wide’ (by 54%, Fig. 1b) is damped by the
effect of changes of alongshore current on the nitrate reservoir below the photic
layer. ‘wide’ leads to a more effective coastal mean upwelling because the mean
nitrate supply by mean vertical and horizontal velocities is roughly insensitive to the
wind drop-off (Supplementary Fig. 8), whereas the opposing eddy flux is decreased.
Code availability. We have opted not to make the computer code associated with
this paper available because we use in-house versions of ROMS at UCLA and UW;
similar ROMS versions are available through Rutgers (www.myroms.org) and
ROMS-AGRIF (http://www.romsagrif.org).
Data availability. The simulation outputs that support the findings of this study
are available on request from the corresponding author (L.R.). The data are not
publicly available owing to the large size of the model output files.
References
21. Shchepetkin, A. F. & McWilliams, J. C. The regional oceanic modeling system
(ROMS): a split-explicit, free-surface, topography-following-coordinate
oceanic model. Ocean Model. 9, 347–404 (2005).
22. Becker, J. J. et al. Global bathymetry and elevation data at 30 arc seconds
resolution: SRTM30_PLUS. Mar. Geod. 32, 355–371 (2009).
23. Mason, E. et al. Procedures for offline grid nesting in regional ocean models.
Ocean Model. 35, 1–15 (2010).
24. Marchesiello, P., McWilliams, J. C. & Shchepetkin, A. Open boundary
conditions for long-term integration of regional oceanic models. Ocean Model.
3, 1–20 (2001).
25. Lemarié, F. et al. Are there inescapable issues prohibiting the use of
terrain-following coordinates in climate models? Ocean Model. 42,
57–79 (2012).
26. da Silva, A. M., Young, C. C. & Levitus, S. Atlas of Surface Marine Data Vol. 4
(NOAA Atlas NESDIS 9, US Government Printing Office, 1994).
27. Conkright, M. E. et al. World Ocean Atlas 2001: Objective Analyses, Data
Statistics, and Figures: CD-ROM Documentation (US Department of
Commerce, National Oceanic and Atmospheric Administration, National
Oceanographic Data Center, Ocean Climate Laboratory, 2002).
28. Barnier, B., Siefridt, L. & Marchesiello, P. Thermal forcing for a global ocean
circulation model using a three-year climatology of ECMWF analyses. J. Mar.
Syst. 6, 363–380 (1995).
29. Molemaker, M. J., McWilliams, J. C. & Dewar, W. K. Submesoscale instability
and generation of mesoscale anticyclones near a separation of the California
Undercurrent. J. Phys. Oceanogr. 45, 613–629 (2015).
30. Moore, J. K., Doney, S. C. & Lindsay, K. Upper ocean ecosystem dynamics and
iron cycling in a global three-dimensional model. Glob. Biogeochem. Cycles
18, GB4028 (2004).
31. Colas, F., Capet, X., McWilliams, J. C. & Li, Z. Mesoscale eddy buoyancy flux
and eddy-induced circulation in Eastern Boundary Currents. J. Phys. Oceanogr.
43, 1073–1095 (2013).
32. Bentamy, A. & Fillon, D. Gridded surface wind fields from Metop/ASCAT
measurements. Int. J. Remote Sensing 33, 1729–1754 (2012).
33. Hooker, S. B. & McClain, C. R. The calibration and validation of SeaWiFS data.
Prog. Oceanogr. 45, 427–465 (2000).
34. Ducet, N., Le Traon, P. Y. & Reverdin, G. Global high-resolution mapping of
ocean circulation from TOPEX/Poseidon and ERS-1 and -2. J. Geophys. Res.
105, 19477–19498 (2000).
35. McCreary, J. P. & Chao, S. Y. Three-dimensional shelf circulation along an
eastern ocean boundary. J. Mar. Res. 43, 13–36 (1985).
36. Marchesiello, P., McWilliams, J. C. & Shchepetkin, A. Equilibrium structure
and dynamics of the California Current System. J. Phys. Oceanogr. 33,
753–783 (2003).
NATURE GEOSCIENCE | www.nature.com/naturegeoscience
© 2016 Macmillan Publishers Limited. All rights reserved