Forcing mechanisms and eddy properties

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 117, C07003, doi:10.1029/2012JC008008, 2012
Mesoscale variability in the northeastern tropical Pacific:
Forcing mechanisms and eddy properties
Jun-Hong Liang,1 James C. McWilliams,1 Jaison Kurian,1 François Colas,1 Peng Wang,1
and Yusuke Uchiyama1
Received 26 February 2012; revised 21 May 2012; accepted 23 May 2012; published 7 July 2012.
[1] The generation mechanisms and properties of the mesoscale eddies in the northeastern
tropical Pacific (NETP) are studied using a series of numerical solutions and satellite
observations. The spatial and temporal resolution of mountain wind jets over the Gulfs
of Tehuantepec and Papagayo are essential for an accurate modeling of the regional
mesoscale variability. Three previously proposed eddy generation mechanisms are the
local transient wind forcing, the combined low-frequency wind and boundary forcing, and
the remote equatorial Kelvin wave forcing. In our model, the transient wind-forcing was
represented by wind fields with synoptic variability, and the remote equatorial Kelvin wave
forcing by 5-day open boundary conditions with interannual variability. Solutions with
and without high-frequency forcings are contrasted to evaluate the contributions of the
three mechanisms. The combined low-frequency wind and boundary forcing is the primary
eddy forcing mechanism in the NETP. In the Tehuantepec region, the high-frequency
wind-forcing contributes more to the mesoscale variability than the remote equatorial
Kelvin wave forcing, while in the Papagayo region, their contributions are comparable.
Eddies in the Tehuantepec and Papagayo regions are larger in both amplitude and size, and
are more likely to deflect equatorward than eddies in the rest of the NETP. The
maximum temperature anomaly of eddies is at around 50 m depth. Eddies with lifetimes
>6 weeks are more likely anticyclonic, while eddies with lifetimes <6 weeks are more
likely cyclonic. The anticyclone-dominance for large-amplitude long-lived eddies is the
combined consequence of wind-driven asymmetric dipole spin-up and the evolutionary
fragility of large-amplitude cyclones.
Citation: Liang, J.-H., J. C. McWilliams, J. Kurian, F. Colas, P. Wang, and Y. Uchiyama (2012), Mesoscale variability in the
northeastern tropical Pacific: Forcing mechanisms and eddy properties, J. Geophys. Res., 117, C07003,
doi:10.1029/2012JC008008.
1. Introduction
[2] The northeastern tropical Pacific (NETP), defined as the
Pacific 5 N–18 N and 120 W to the coast of Central
America, plays an important role in global climate (El Niño
Southern Oscillation (ENSO)), weather (hurricane genesis),
and local fishery industry [Wyrtki, 2006; Wang and Fiedler,
2006]. The mean state and variability of the NETP are
strongly modulated by three powerful wind jets over the Gulfs
of Tehuantepec, Papagayo, and Panama, mostly from mid-fall
to spring. These strong wind jets funnel through three narrow
mountain gaps in the Central American cordillera and are
forced by either cold surges from the North American
1
Department of Atmospheric and Oceanic Sciences, University of
California, Los Angeles, California, USA.
Corresponding author: J.-H. Liang, Department of Atmospheric and
Oceanic Sciences, University of California, Los Angeles, CA 90095, USA.
([email protected])
©2012. American Geophysical Union. All Rights Reserved.
0148-0227/12/2012JC008008
continent or by trade winds from the Caribbean Sea
[Chelton et al., 2000]. They are stronger and more frequent
in winter than in summer. Regional circulation and thermocline structure conform to the Sverdrup relation and are
strongly correlated with the strength of the mean wind jets
[Kessler, 2002, 2006; Xie et al., 2005]. The annual cycle of
sea surface temperature (SST) in the Gulfs of Tehuantepec and
Papagayo is influenced by the gap outflow over the two gulfs
[Sun and Yu, 2006]. Seasonal dynamics of sea surface salinity
at the eastern Pacific off Panama has been shown to be modulated by wind jet over the Gulf of Panama [Alory et al., 2012].
There is strong variability in the wind jets at synoptic timescales. Nearshore oceanic responses to the high-frequency
wind events include the spin-up of asymmetric eddies
[McCreary et al., 1989; Trasviña et al., 1995; Trasviña and
Barton, 2008], the decrease in SST [Clarke, 1988; Barton
et al., 1993; Trasviña et al., 2003; Barton et al., 2009], and
the enhancement of surface nutrients and biological production [McClain et al., 2002; Pennington et al., 2006].
[3] On an intraseasonal timescale, satellite altimetric data
reveal the existence of two bands of strong mesoscale
C07003
1 of 13
C07003
LIANG ET AL.: CENTRAL AMERICA EDDIES
Figure 1. (top) A Central America ROMS domain. Also
shown in the figure is a color map of 90-day high-passed
root-mean square sea level anomaly (cm) in this region from
AVISO altimetric data. Three arrows indicate, from north to
south, three mountain gaps behind the Gulfs of Tehuantepec,
Papagayo, and Panama, respectively. The black box delimitates computational domain with dashed lines delineating
open boundaries. (bottom) A schematic showing the generation mechanisms of eddies in the Northeastern Tropical
Pacific.
variability off the coast of Central America in the NETP due
to nonlinear eddies from the Gulfs of Tehuantepec and
Papagayo (Figure 1, top) [Giese et al., 1994]. One originates
from the Gulf of Tehuantepec and extends southwestward to
around 115 W 10 N; and the other one originates from the
Gulf of Papagayo and extends to the west of 95 W. The
Tehuantepec eddies first move southwestward and then join
the same track along 11 N where the Papagayo eddies
propagate westward. It is also observed that there is a preference for anticyclonic eddies over cyclonic eddies in these
two regions of high eddy activity [e.g., Palacios and
Bograd, 2005]. These eddies transport coastal nutrient-rich
water to the open ocean [Samuelsen and O’Brien, 2008] and
C07003
were recently shown to disperse deep-sea products from
hydrothermal vents in the East Pacific Rise [Adams and
Flierl, 2010]. There have been a few mechanisms proposed
for the genesis of Tehuantepec and Papagayo eddies [Willett
et al., 2006]. The most obvious possibility is that positive
wind curl on the left flank of the wind jets through the
mountain gaps spins up cyclonic eddies and negative wind
curl on the right flank of the wind jets spins up anticyclonic
eddies. On the other hand, the shallowness of the local
thermocline suppresses upwelling and the associated cyclonic eddies, which leads to asymmetric dipoles with stronger
anticyclonic eddies [McCreary et al., 1989]. There is, however, generations of eddies that are not associated with any
synoptic gap wind events [Hansen and Maul, 1991;
Zamudio et al., 2006; Liang et al., 2009]. This leads to two
other explanations of the generation mechanism. The first
are the barotropic and baroclinic instabilities of the mean
ocean currents [e.g., Willett et al., 2006; Farrar and Weller,
2006], which are part of the large-scale circulations and are
modulated by low-frequency wind. The second one is the
instability of coastal currents triggered by passing poleward
propagating coastal Kelvin waves [Zamudio et al., 2006].
Poleward propagating coastal Kelvin waves are mainly initiated when eastward-propagating Equatorial Kelvin waves
hit the America Continent. They induce strong horizontal
and vertical shear flows and trigger barotropic and baroclinic
instabilities during their passage. The eddy generation
mechanisms are schematically illustrated in Figure 1
(bottom).
[4] In this study, we assess the importance of different
eddy generation mechanisms by analyzing a series of solutions from the Regional Ocean Modeling System (ROMS)
forced by surface and lateral boundary conditions of different spatial resolution and temporal variability. The necessary
forcing conditions for realistically reproducing regional
mesoscale variability are evaluated by contrasting the solutions with altimeter observations. We also seek to characterize eddy properties (size, amplitude, geographic
distribution, and vertical structures) in this region. The rest
of the paper is organized as follows: section 2 describes the
way ROMS was configured for the NETP; section 3 presents
the results, discusses the roles of different eddy generation
mechanisms, and analyzes eddy properties; and section 4 are
conclusions.
2. Data and Methods
2.1. Regional Ocean Model
[5] The numerical model used in this study is the Regional
Ocean Modeling System (ROMS), which solves the threedimensional hydrostatic primitive equations in vertical hybrid
z-sigma [Lemarie et al., 2012] and horizontal curvilinear
coordinates with innovative algorithms for advection, mixing,
pressure gradient, vertical-mode coupling, time stepping, and
parallel efficiency [Shchepetkin and McWilliams, 2005].
A configuration with a horizontal resolution of 10 km (400 620 grid points) encompasses a domain from 22 S to 35 N
and from the coast of the American continent to 140 W, with
open-boundary conditions at its western and alongshore edges
(Figure 1, top). The grid resolution (10 km) is much smaller
than the Rossby deformation radius in this region (>50 km)
[see Chelton et al., 1998, Figure 6] and is, therefore, capable
2 of 13
C07003
LIANG ET AL.: CENTRAL AMERICA EDDIES
Figure 2. (a) Wind curl in January, February, March
(JFM) from Scatterometer Climatology of Ocean Winds,
and (b) wind curl difference between SCOW and the blended
QuikSCAT/NCEP wind product. The black lines in Figure 2a
delimitate the regions where the corrected winds are imposed.
of resolving mesoscale variability in this region. The domain
extent is sufficiently large to include the currents that interconnect in the NETP: the California Current, the North
Equatorial Current, the North Equatorial Counter Current, and
the Peru Chile Current System. We used 32 vertical levels
stretched toward the surface so that roughly 20 levels cover
the first 200 m of the water column. The bottom topography
was interpolated from the ETOPO2 database [Smith and
Sandwell, 1997]. In order to reduce the pressure gradient
computation errors, the bathymetry was smoothed (analogous
to Marchesiello et al. [2003]) to satisfy the condition
dh/h < 0.18 with h the total water depth and dh the
depth difference between adjacent grid points [Beckman
and Haidvogel, 1993].
2.2. Boundary Conditions
[6] The three open boundaries of the ROMS model were
forced by temperature, salinity and velocity fields interpolated from the Simple Ocean Data Assimilation (SODA)
data between 1999 and 2008 [Carton and Giese, 2008].
Both the 5-day interannual data and the monthly climatological (long-term monthly mean) data were used. Equatorial
Kelvin waves, which can trigger eddy generation in the
NETP, are preserved in the 5-day interannual data, but are
filtered out in the monthly climatological data. The boundary
condition is a mixed radiative-relaxation parameterization
[Marchesiello et al., 2001; Mason et al., 2010]. The Flather
boundary condition was imposed on the 2D momentum
equations, and the Orlanski radiative condition was imposed
for 3D fields. A sponge layer along the open boundaries was
also set up to damp outgoing eddies and waves.
[7] Three different wind products were used. The first two
are publicly available: the QSCAT/NCEP blended wind
[Milliff et al., 2004] and the Scatterometer Climatology of
Ocean Winds (SCOW) [Risien and Chelton, 2008]. The
C07003
QSCAT/NCEP blended wind product blends QSCAT scatterometer wind and NCEP reanalysis from August 1999 to
December 2008. It is daily at a 0.5 0.5 spatial resolution.
The SCOW wind product is a monthly climatology at a
0.25 0.25 spatial resolution based on QSCAT Scatterometer wind from September 1999 to October 2009.
[8] The mountain gaps are narrow, in particular, the Chivela Pass where the Tehuantepec wind funnels through, is
only some 40 km (0.4 ) wide. Therefore, the SCOW wind
on a spatial grid of 0.25 0.25 resolves the mountain gap
wind jets better than the QSCAT/NCEP blended wind on a
spatial grid of 0.5 0.5 . Figure 2 shows the winter wind
curl in the SCOW product and the winter wind curl difference between the SCOW product and the climatology from
the QSCAT/NCEP blended wind product in winter when the
wind jets are strongest. The climatological wind curls from
the two wind products are similar except in the two wind jet
regions. The winter wind curl climatology in the QSCAT/
NCEP blended daily wind product is up to 70% weaker than
in the SCOW product over the Gulf of Tehuantepec. However, the SCOW product does not contain synoptic wind
events, which have been suggested to play an important role
in generating mesoscale eddies. To have the best spatial and
temporal resolution in the wind-forcing, we developed a
corrected daily wind product based on both the QSCAT/
NCEP daily wind product and the SCOW product. The new
wind field has the same climatology as the SCOW product
and has synoptic wind events. The wind stress field for the
Tehuantepec wind in the corrected QSCAT/NCEP blended
wind is prepared as follows:
[9] 1. The daily Tehuantepec wind index was defined as
the daily blended QSCAT/NCEP wind in the offshore
direction averaged over the nearshore region in front of the
mountain gap (from 15 N to the coast and from 96 W to
94 W).
[10] 2. The climatological monthly wind stress difference
in the Gulf of Tehuantepec between the SCOW wind product and the blended QSCAT/NCEP wind product was
defined as the typical Tehuantepec wind field for each
month. The region of the typical Tehuantepec wind field is
delimitated by the black line in Figure 2a. Outside of the
region, the typical Tehuantepec wind field decays with the
distance from the boundary following a Gaussian filter with
standard deviation of 20 km so that no artificial wind curl
was created at the boundaries of the region.
[11] 3. The typical Tehuantepec wind field was added to
the blended QSCAT/NCEP daily wind product with a
weight. The weight is proportional to the Tehuantepec wind
index and ensures that the climatological wind stress and
curl equal those in the SCOW in the region.
[12] Similar procedures were carried out for Papagayo
wind. The new daily wind product is called the corrected
blended QSCAT/NCEP wind product in this paper. The
spatial and temporal resolutions of the wind jets in the corrected blended QSCAT/NCEP wind product are the best
among the three wind products.
[13] Heat fluxes from COADS climatology [da Silva et al.,
1994] were used, and a relaxation to Pathfinder climatological SST [Casey et al., 2010] was imposed following the
parameterization of Barnier et al. [1995]. To prevent salinity
drift, we also imposed a relaxation to the COADS sea surface
salinity.
3 of 13
LIANG ET AL.: CENTRAL AMERICA EDDIES
C07003
Table 1. Abbreviated Names for Different Wind and Open
Boundary Conditionsa
Run
OBC
Wind
SCSC
S5SC
SCBC
S5BC
SCBD
S5BD
SCCC
S5CC
SCCD
S5CD
SODA clim.
SODA 5-day
SODA clim
SODA 5-day
SODA clim.
SODA 5-day
SODA clim.
SODA 5-day
SODA clim.
SODA 5-day
SCOW
SCOW
Blended Q/N clim.
Blended Q/N clim.
Blended Q/N
Blended Q/N
Corrected Blended Q/N clim.
Corrected Blended Q/N clim.
Corrected Blended Q/N
Corrected Blended Q/N
a
Each name has four letters. The first two letters indicate the open
boundary conditions: “SC” means SODA climatology, and “S5” means
SODA interannual data updated every 5 days. The third letter indicates
the wind product used (S: SCOW; B: blended QSCAT/NCEP wind; and
C: corrected blended QSCAT/NCEP wind). The last letter indicates
whether the wind is daily (D) or climatological (C).
[14] The abbreviated names for runs forced by different
combinations of surface and lateral boundary forcings are
listed in Table 1. Simulations forced by climatological conditions were integrated for 16 years with initial conditions
interpolated from the SODA data sets. Simulations forced by
interannual surface and lateral boundary conditions were
restarted from the solutions forced by the corresponding
climatological conditions, and were integrated from August
1999 to December 2008 when both the SODA and the
blended wind data are available. The first four years of
solutions were discarded as spin-up.
2.3. Data Sets
[15] To assess the realism of the model results, satellite and
reanalysis data sets were used. Weekly sea level anomaly
C07003
(SLA) data between 1999 and 2008 on a 1/3 1/3 grid
came from the Ssalto/Duacs multimission altimeter gridded
sea level anomaly product (http://www.aviso.oceanobs.
com/en/data/product-information/duacs/index.html), which
merge data from up to four satellites [Pascual et al., 2006].
Sea surface temperature from 1999 to 2008 on a 0.25 0.25 grid originated from the Tropical Rainfall Measuring
Mission Microwave Imager (TMI) SST product (http://
www.ssmi.com). The mean seasonal sea surface height
(SSH) on a 0.25 0.25 grid was downloaded from
http://oceanwatch.pifsc.noaa.gov. The CSIRO Atlas of
Regional Seas (CARS) climatological mean ocean temperature on a 0.5 0.5 grid was downloaded from http://
www.marine.csiro.au/dunn/cars2009.
3. Results and Discussion
3.1. Annual Cycle of the NETP
[16] The regional geostrophic circulation can be inferred
from SSH maps. Figures 3a, 3d, 3g, and 3j show the observed
mean seasonal cycle of SSH. A domain mean in each season
was removed for better visual effects. This will not change
the circulation pattern, while the thermosteric effect is largely
eliminated. The SSH map agrees with the map of dynamic
height shown in Kessler [2006, Figure 7]. Two distinct features in this region are the Tehuantepec Bowl (TB) west of
the Gulf of Tehuantepec (“bowl” means a depression of the
thermocline and an elevation of the ocean surface; and
“dome” refers to the opposite situation) and the Costa Rica
Dome (CRD) west of the Gulf of Papagayo [Kessler, 2006].
The TB is the strongest in winter and almost disappears in
summer, while the strength of the CRD is less variable with
seasons. This is mainly because the Tehuantepec wind has a
stronger seasonal cycle than the Papagayo wind, and the
Figure 3. Mean-seasonal sea surface height (cm) in (a–c) January, February, and March (JFM);
(d–f) April, May, and June (AMJ); (g–i) July, August, and September (JAS); and (j–l) October, November,
and December (OND). The left column is the satellite observed mean SSH anomaly, the middle column is
solutions of run S5CD, and the right column is solutions of run SCSC. The two black dashed lines in
Figure 3a delineate 12 N and 9 N.
4 of 13
C07003
LIANG ET AL.: CENTRAL AMERICA EDDIES
C07003
Figure 4. Mean-seasonal sea surface temperature ( C) in (a–c) JFM, (d–f) AMJ, (g–i) JAS, and (j–l)
OND. The left column is the TMI satellite SST, the middle column is run S5CD, and the right column
is run SCSC.
strength of the TB and the CRD highly correlates with the
seasonal strength of the wind jets over the respective gulfs,
respectively. In winter, there are two anticyclonic cells off the
coast of the central America, one north [Brenes et al., 2008]
and one south [Chaigneau et al., 2006] of the CRD. All these
features are captured in runs S5CD and SCSC (Figure 3). The
close agreement of the two solutions indicates that the highfrequency variability in surface and lateral boundary forcing
does not influence the mean seasonal circulations. All four
solutions forced by the QSCAT/NCEP blended wind (runs
SCBC, SCBD, S5BC, and S5BD) predict weaker TB and
CRD due to the weaker wind jets (not shown here). This also
confirms the conclusion drawn by Kessler [2002, 2006] that
the mean and seasonal circulations in the NETP are determined by low-frequency wind.
[17] Figure 4 shows the maps of SST from satellite data
and from runs S5CD and SCSC. The ocean surface is substantially cooler along the two wind paths than in the surrounding areas during fall and winter. The cooling is
underestimated in the solutions probably due to the fact that
the wind jets in the surface forcing are under-resolved and
wind-driven mixing is consequently weaker. By examining
different solutions, we again see little difference in the seasonal mean SST between solutions with and without highfrequency interannual variability in surface and open
boundary conditions. Alexander et al. [2012] show that there
is a significant SST difference between El Niño and La Niña
years, and the SST difference can be reproduced in solutions
forced with 5-day interannual open boundary conditions but
not in solutions forced by climatological open boundary
conditions. The close agreement between the two solutions
shown in Figure 4 is because the SST contrast due to ENSO
is averaged out. Similar to the comparison of SSH, SST
signals produced by wind jets are further underestimated
when the wind climatology is weaker (runs SCBC, SCBD,
S5BC, and S5BD).
[18] Figure 5 displays the zonal hydrographic structures at
12 N and 9 N from run S5CD, run SCSC, and the CARS
climatology. These two latitudes, delineated by two dashed
lines in Figure 3a, are chosen since 12 N crosses the TB, and
9 N crosses the CRD. The simulated subsurface temperature
profiles between the two runs are alike. This confirms that
the mean ocean state is not influenced by the addition of
high-frequency forcing in the NETP. The surface mixed
layer depth is less than 50 m at both latitudes. The slight
thermocline shoaling at (110 W 12 N) and doming at
(90 W 9 N) are reproduced in the solutions. Compared with
observations, the simulated thermocline is slightly diffused.
3.2. Mesoscale Variability
[19] Mesoscale variability is studied using high-frequency
SLA from both altimetric data and ROMS solutions and
velocity fields from ROMS solutions. The high-frequency
fields were obtained as follows: For each grid point, a temporal mean and a raw seasonal cycle smoothed by a 30-day
running mean were first removed to obtain a non-seasonal
anomaly. After that, a high-pass filter was applied so that
any variability with periods longer than 90 days, such as
ENSO events, was essentially removed [Palacios and
Bograd, 2005].
3.2.1. Mean and Seasonal Cycle
[20] Figure 6 shows the standard deviation of the highfrequency SLA from model runs forced by different surface
and lateral boundary conditions, and from satellite altimetric
data. The spatial patterns of the two bands of strong mesoscale variability emanating from the Gulfs of Tehuantepec
and Papagayo are reproduced in all solutions. However, the
strengths of mesoscale variability within the two bands are
under-predicted to different degrees in all solutions. Among
all runs, run SCBC under-predicts the mesoscale variability
the most (Figure 6a) because the forcing for this run has
relatively low spatial resolution and does not contain high-
5 of 13
C07003
LIANG ET AL.: CENTRAL AMERICA EDDIES
C07003
Figure 5. Zonal cross-sections of annual mean temperature at 12 N (left column) and 9 N (right column)
from (a and b) run SCSC, (c and d) run S5CD, and (e and f) the CSIRO Atlas of Regional Seas
climatology.
frequency temporal variability. With increasing spatial and
temporal resolutions in surface and lateral boundary conditions, the predicted mesoscale variability approaches to the
observed variability. The combination of a corrected daily
wind and 5-day interannual open boundary condition is the
most realistic forcing, and the solution using this forcing
(Run S5CD, Figure 6e) is consequently the closest to the
observations. Outside the two strong mesoscale variability
bands, in particular along 15 N, there is more mesoscale
variability in the solutions than in the observations. One of
the possible sources of eddies along 15 N is the Acapulco
eddies originating from the coast of Mexico [Zamudio et al.,
2001]. By contrasting Figures 6c and 6d with Figure 6b, we
can see that daily wind-forcing contributes slightly more to
the eddy activity along 15 N than 5-day interannual lateral
boundary forcing.
[21] Figure 7 displays the seasonal cycle of mesoscale
variability in the Tehuantepec and Papagayo regions, as well
Figure 6. Maps of the standard deviation of the 90-day high-passed sea level anomaly (cm) for (a) run
SCBC, (b) run SCCC, (c) run S5CC, (d) run SCCD, (e) run S5CD, and (f) satellite altimetry. The black
boxes in Figure 6f indicate the Tehuantepec and Papagayo regions.
6 of 13
C07003
LIANG ET AL.: CENTRAL AMERICA EDDIES
Figure 7. The standard deviation of the 90-day high-passed
sea level anomaly (cm) averaged over the (top) Tehuantepec
and (bottom) Papagayo regions, respectively. The Tehuantepec and Papagayo regions are delimitated with black boxes in
Figure 6f.
as quantitative comparisons between different solutions and
altimetric observations. The standard deviation of the highfrequency SLA was averaged over the regions delimitated
by the black boxes in Figure 6f, and was defined as the
strength of mesoscale variability in the regions. Mesoscale
variability exhibits a strong seasonal cycle in both regions.
In the Tehuantepec region, it is strongest in winter and
weakest in summer. In the Papagayo region, it is also
strongest in winter, and is weakest in both summer and fall.
Based on altimetric observations, mesoscale variability
during winter is 60% stronger than during summer in the
Tehuantepec region and during summer and winter in the
Papagayo region. Consistent with the results in Figure 6,
the observed mesoscale variability is best reproduced in run
S5CD. In the Tehuantepec region, the S5CD run underpredicts observed mesoscale variability by 13% in winter
and over-predicts observed mesoscale variability by 6% in
summer. In the Papagayo region, run S5CD under-predicts
observed mesoscale variability by 3% to 17% in different
seasons.
3.2.2. Forcing Mechanisms
[22] The two bands of strong mesoscale variability are
mainly the combined consequence of three different forcing mechanisms: (1) the local high-frequency wind-forcing;
(2) the combined low-frequency wind and boundary forcing;
and (3) the remote equatorial Kelvin wave forcing. The
importance of the three mechanisms can be understood by a
quantitative comparison of the standard deviation of the
high-frequency SLA in different runs shown in Figure 7.
[23] Run S5CD was forced by the most realistic wind and
open boundary conditions. Its solution is closest to observations and is therefore taken as the control run. Run SCCC
includes neither the local transient wind-forcing nor the
remote equatorial Kelvin wave forcing. It reproduces 68% to
80% of the mesoscale variability in run S5CD for Tehuantepec region and 62% to 90% for Papagayo region in different seasons. The surface and lateral forcings in runs
SCCC and S5CD have the same climatological mean. This
C07003
implies that the combined low-frequency wind and boundary forcing is the most important forcing mechanism for
mesoscale variability in the NETP. Run SCBC is also forced
by climatological wind and open boundary conditions, but
the wind forcing used in run SCBC has weaker wind jets
than in run SCCC because its spatial resolution is lower
(Figure 2). Currents driven by weaker wind jets are less
spatially intensified and less unstable. Compared with run
SCCC, mesoscale variability in run SCBC is 5% to 7%
lower in the Tehuantepec region and 8% to 19% lower in the
Papagayo region in different seasons. Because wind jets are
still under-resolved in the forcing for S5CD due to the spatial resolution, run S5CD still underestimates the observed
mesoscale variability in the two regions. Without the transient wind forcing, run S5CC under-predicts mesoscale
variability by 7% to 20% in the Tehuantepec region and by
1% to 12% in the Papagayo region in different seasons.
Without the remote equatorial Kelvin wave forcing, the
SCCD run under-predicts mesoscale variability by 6% to
12% in the Tehuantepec region and by 2% to 13% in the
Papagayo region in different seasons. In the Tehuantepec
region, the high-frequency local atmospheric forcing contributes more to eddy activity than the remote equatorial
Kelvin wave forcing. In the Papagayo region, the importance
of the two forcing mechanisms is comparable. Remote
Kelvin wave forcing not only modulates eddy activities in
Tehuantepec and Papagayo region, but also eddy activities
in other regions along the coast of America [Zamudio et al.,
2006, 2007, 2008; Echevin et al., 2011; Belmadani et al.,
2012].
[24] The respective roles of local wind-forcing and instability on eddy generation are further analyzed by examining
some terms of the eddy kinetic energy budget. The energy
conversion terms can be calculated as [Marchesiello et al.,
2003],
1
ut x þ vt y
r0
ð1Þ
1
u′t x ′ þ v′t y ′
r0
ð2Þ
FmKm ¼
FeKe ¼
∂u
∂u
∂u
∂v
∂v
∂v
KmKe ¼ u′u′ þ u′v′ þ u′w′ þ v′u′ þ v′v′ þ v′w′
∂x
∂y
∂z
∂x
∂y
∂z
ð3Þ
PeKe ¼
g
r′w′
r0
ð4Þ
where FmKm is the total wind work on ocean currents and is
largely the energy input into the mean current from lowfrequency gap winds; FeKe is the eddy kinetic energy (EKE)
generation due to transient wind, and is the eddy spin-up by
transient wind events in this study; KmKe is the energy
conversion between mean currents and eddies, and indicates
barotropic instability when it is positive; PeKe is the energy
conversion between available potential energy and EKE, and
indicates baroclinic instability when it is positive; u, v,
and w are zonal, meridional, and vertical velocities; g is
7 of 13
C07003
LIANG ET AL.: CENTRAL AMERICA EDDIES
Figure 8. The eddy kinetic energy budget terms calculated
by Equations (1) to (4) in the text for the (top) Tehuantepec
and (bottom) Papagayo regions.
gravity; t is wind stress; and ðÞ ¼ ðÞ þ ðÞ′ with ()′ the
high-frequency component of a variable ().
[25] Figure 8 shows the averaged seasonal energy budget
terms for run S5CD in the Tehuantepec and the Papagayo
regions delimitated in Figure 6f, respectively. Only the local
EKE generation terms in the EKE budget are shown in
Figure 8. Both boundary fluxes and EKE dissipation are
required to close the EKE budget, but are not shown here.
These budget terms show similar seasonal variability as the
strength of mesoscale variability shown in Figure 7. The total
wind work (FmKm) is the largest in fall and winter for the
Tehuantepec region and in winter for the Papagayo region.
Both barotropic (KmKe) and baroclinic (PeKe) instabilities
are also the strongest during the same seasons, because the
stronger wind jets drive more unstable currents. The magnitudes of KmKe and PeKe are comparable, which implies a
mixed baroclinic/barotropic eddy generation process. The
transient wind work (FeKe) is the largest in fall for both
Tehuantepec and Papagayo, in agreement with the strongest
enhancement of eddy variability from run SCCD to run
SCCC shown in Figure 7. The magnitude of FeKe is smaller
than the sum of KmKe and PeKe, which, from an energy
budget perspective, confirms the conclusion drawn from
Figure 7 that eddies are mostly generated by instability of the
mean seasonal circulations, not directly by high-frequency
wind events.
3.3. Eddy Properties
[26] To study the eddy properties in the NETP, we
applied an automated eddy detection and tracking method
[Kurian et al., 2011] to both satellite altimetric data and the
solutions from run S5CD as this run statistically agrees the
best with observations. An eddy is identified if a closed
contour of the 90-day high-passed SLA conforms to a set
of criteria for its shape, amplitude, and size. These criteria
are: (1) The closed contour is nearly circular; (2) Its radius
is from 15 km to 250 km for the ROMS solution and from
50 km to 250 km for the altimetric data, with the maximum
radius (250 km) considerably larger than the maximum
Rossby deformation radius in this region (150 km); (3) the
C07003
SLA magnitude monotonically increases toward the center
with a minimum 1 cm core to edge difference. Eddies are
tracked from one time step to the next by finding the nearest
eddies at consecutive solution or altimetry outputs (5 days
apart for ROMS solution and 7 days apart for altimetric data)
and comparing their properties (amplitude, size, and location)
(see Chelton et al. [2011] and Kurian et al. [2011] for details).
3.3.1. Size, Amplitude, Geographic Distribution,
and Vertical Structures
[27] Figure 9 displays eddy properties from satellite altimetric data and run S5CD. There are a comparable number of
large-size (R > 70 km) eddies in the ROMS solution and the
altimetric data. There are significantly more small-size (R <
70 km) eddies in the ROMS solution than in the altimetric data
because the ROMS solution has a better spatial resolution
(1/10 1/10 ) than the altimetric data (1/3 1/3 ). Smallsize eddies are more likely cyclonic, while large-size eddies
are more likely anticyclonic. The cyclone dominance for
small-size eddies is also found in California Current System
[Kurian et al., 2011] and the Peru Chile Current System
[Colas et al., 2011]. A distinct difference between the eddy
statistics for the NETP and for the global ocean is the
cyclone-to-anticyclone ratio when the eddy amplitude is large
(A > 10 cm). The cyclone to anticyclone ratio is significantly
larger than unity for long-lived large-amplitude eddies in the
southern hemisphere, and the cyclone to anticyclone ratio is
close to unity for eddies of similar properties in the northern
hemisphere [see Chelton et al., 2011, Figure 9]. In the NETP,
there is a strong anticyclone dominance for eddies of amplitude larger than 10 cm (Figure 9f). The mechanism for the
anticyclone dominance in the NETP will be discussed in
section 3.3.2.
[28] Because there is a general agreement in eddy statistics
between run S5CD (Figure 9) and altimetric data, we will
proceed to present results from run S5CD. Figure 10 displays the number of coherent cyclonic and anticyclonic
eddies with a lifetime of at least 4 weeks. There are more
cyclonic eddies than anticyclonic eddies in total, however,
there are more long-lived (lifetime longer than 6 weeks)
anticyclonic eddies than cyclonic eddies. Figure 11 shows
the trajectories and birth places of eddies with lifetimes
larger than 8 weeks. Eddies detected from altimetric data
also have similar geographic distribution (not shown). In
agreement with results in Figure 10, eddies of these lifetimes
are more likely anticyclonic. There are two clusters of eddies
along the two bands of strong regional mesoscale variability
originating from the Gulfs of Tehuantepec and Papagayo.
There are only one anticyclonic eddy with a lifetime larger
than 8 weeks from the Gulf of Panama in 6 years. This is
consistent with previous studies that show Panama eddies
are short-lived [e.g., Willett et al., 2006]. Outside the two
strong eddy variability bands, substantial eddy generation
and propagation also occur. It has been long recognized that
most of the open ocean is baroclinically unstable [e.g.,
Robinson and McWilliams, 1974]. Both anticyclonic and
cyclonic eddies propagate westward. 56% of anticyclonic
eddies deflect equatorward while 57% cyclonic eddies
deflect poleward. The different preferential zonal deflection
for eddies of different polarity in the NETP is consistent
with global eddy statistics [Chelton et al., 2007, 2011] and
is a combined consequence of the b effect and eddy selfadvection [McWilliams and Flierl, 1979]. Eddies in the
8 of 13
C07003
LIANG ET AL.: CENTRAL AMERICA EDDIES
C07003
Figure 9. Cumulative eddy number per snapshot with respect to (a) eddy radius R and (d) amplitude A,
eddy number distribution per snapshot with respect to (b) R and (e) A, and cyclone to anticyclone ratio with
respect to (c) R and (f) A. Altimeter data is plotted with dashed lines, and run S5CD is plotted with solid lines.
Tehuantepec and Papagayo regions are more likely to
deflect equatorward than in the rest of the NETP. 76% of
the anticyclonic eddies and 55% of the cyclonic eddies
deflect equatorward in these two regions. Strong offshore
currents with an equatorward component are driven by the
gap winds, and the advection by these currents contributes
to eddy equatorward deflection in the two regions. Maps of
shorter-lived eddies (lifetimes smaller than 8 weeks) show
similar geographical patterns, except that there are more
cyclonic eddies than anticyclonic eddies (not shown here).
Figure 10. Cumulative number of coherent eddies per year
with a life longer than 4 weeks in run S5CD.
[29] To study the vertical structure of eddies, we constructed composite vorticity and temperature maps over individual long-lived (life >8 weeks) eddies (Figures 12 and 13).
The anomaly for a field at a given time and depth level is
Figure 11. The tracks (lines) and origins (solid points) of
(top) anticyclonic and (bottom) cyclonic eddies with a life
longer than 8 weeks in 6-year solutions of run S5CD.
9 of 13
C07003
LIANG ET AL.: CENTRAL AMERICA EDDIES
C07003
Figure 12. Composite anomaly (relative to the horizontal mean of a 250 km 250 km box relative to the
eddy center) of vorticity (z/f) for (top) cyclonic and (bottom) anticyclonic eddies in the Tehuantepec
region, the Papagayo region, and the rest of the NETP.
found by removing the horizontal average of the field over a
box of 250 km 250 km centered around the eddy. Both
anticyclonic and cyclonic eddies have their maximum vorticity anomaly at the surface (Figure 12). Anticyclonic eddies are
stronger and penetrate deeper than cyclonic eddies. Eddies in
the Tehuantepec and Papagayo regions are larger in size and
relative vorticity (strength) than eddies in the rest of the
regions in the NETP. The maximum temperature anomaly
associated with the eddies is at around 50 m (Figure 13),
which is the depth of the thermocline in the NETP. The
maximum temperature anomaly at thermocline has also
been observed in cyclonic eddies in the eastern South
Pacific ocean [Chaigneau et al., 2011]. It is slightly shallower in the Papagayo region, as thermocline shoals in that
region. The depth of the maximum temperature anomaly
associated with eddies in the NETP is shallower than in the
California Current System (approximately 100 m) [Kurian
et al., 2011] and in the Peru Chile Current System (approximately 100 m for cyclonic eddies and 400 m for anticyclonic
eddies) [Chaigneau et al., 2011].
3.3.2. Polarity Asymmetry
[30] Figure 14 shows the mean strength (averaged vorticity) and number of long-lived eddies (life >8 weeks) for
different geographic regions defined in Figure 6f and for
different seasons. The number of long-lived Tehuantepec
anticyclonic eddies is twice that of cyclonic eddies in fall
and winter. In spring and summer, the numbers of long-lived
Tehuantepec anticyclonic eddies and cyclonic eddies are
comparable. The average strength of Tehuantepec eddies in
fall and winter is higher than in spring and summer. The
number of long-lived Papagayo anticyclonic eddies is more
than twice that of cyclonic eddies in all seasons and the
average strength of long-lived Papagayo eddies is higher
than long-lived eddies in Tehuantepec region and other
regions. Outside the two regions, the anticyclone dominance
is not evident.
[31] Two mechanisms have been previously proposed for
the anticyclone dominance (see Willett et al. [2006] for a
review). Both of these mechanisms refer to wind-driven
spin-up of asymmetric dipoles with a stronger anticyclone
due to the shallowness of the thermocline in the Gulfs (the
thermocline in the Gulfs of Tehuantepec and Papagayo is
around 30 m). Using a 1.5-layer model, McCreary et al.
[1989] showed that vertical mixing entrains thermocline
water, which is shallow in this region, to the surface, and
inhibits the strengthening of cyclonic eddies. Thomas and
Rhines [2002] proposed that Ekman transport at finite
Rossby number (Ro = z/f with z the relative vorticity and f
the planetary vorticity) enhances/weakens Ekman pumping
on the anticyclonic/cyclonic side of a dipole. This effect
forms a positive feedback to the preferential spin-up of
anticyclones. The wind-driven spin-up of asymmetric
10 of 13
C07003
LIANG ET AL.: CENTRAL AMERICA EDDIES
C07003
Figure 13. Composite anomaly (relative to the horizontal mean of a 250 km 250 km box relative to the
eddy center) of temperature ( C) for (top) cyclonic and (bottom) anticyclonic eddies in the Tehuantepec
region, the Papagayo region, and the rest of the NETP.
circulation has been previously observed [e.g., Trasviña
et al., 1995] and the two mechanisms are relevant. Because
considerable number of eddies in the Tehuantepec and Papagayo regions are generated by current instability, not by winddriven eddy spin-up (Figure 8) and both previously stated
mechanisms are relevant during wind-driven eddy spin-up, we
propose the theory given by Graves et al. [2006], which
applies during eddy propagation, as likely important. Graves
et al. [2006] show that cyclonic eddies with finite Rossby
number and small Burger number (Bu = Rd/R with Rd the first
baroclinic Rossby deformation radius and R the eddy radius)
are more easily dissipated under an ambient strain field by
neighboring eddies and mean flows. The preferential
destruction of strong cyclonic eddies and the consequent
anticyclone dominance have been observed in laboratory
flows [e.g., Perret et al., 2006]. In order to assess the relevance of this theory for eddies in the Tehuantepec and
Papagayo regions, we examined short-lived large-amplitude
eddies that have a life shorter than 10 days, a radius larger
than 50 km, and a relative vorticity z larger than 0.8f in the
two regions. The corresponding Rossby number (Ro) of these
eddies is larger than 0.8, and the corresponding Burger
number (Bu) is smaller than one because the first baroclinic
Rossby deformation radius Rd in the NETP is between 50 km
to 100 km [see Chelton et al., 1998, Figure 6]. The finite
Rossby number (Ro > 0.8) and small Burger number (Bu ≤ 1)
Figure 14. Comparison of the eddy strength (averaged vorticity: z avg/f ) within long-lived (life >8 weeks) eddies (top)
between cyclonic and anticyclonic eddies and (bottom) of
cyclone and anticyclone population number in different seasons and different regions. “T” indicates the Tehuantepec
region defined in Figure 6f, “P” indicates the Papagayo
region, and “O” indicates regions other than “T” and “P”.
11 of 13
C07003
LIANG ET AL.: CENTRAL AMERICA EDDIES
falls in the regime for preferential destruction of cyclonic
eddies according to the theory by Graves et al. [2006].
There is a strong polarity asymmetry for short-lived largeamplitude eddies in the two regions. On average, there are
9.6 and 4.6 short-lived large-amplitude eddies per year in
the two regions, respectively. There are only 0.5 shortlived large-amplitude anticyclonic eddies per year in the
Tehuantepec region, and no such anticyclonic eddies in the
Papagayo region. The existence of considerable amount of
short-lived large-amplitude eddies indicates that evolutionary fragility of cyclonic eddies plays an important role
in the anticyclone dominance for large-amplitude longlived eddies in the NETP.
4. Conclusions
[32] We present regional ocean modeling solutions of the
northeastern tropical Pacific (NETP) with the objective of
clarifying the respective contributions of three different eddy
generation mechanisms, i.e., the combined low-frequency
wind and boundary forcing, the high-frequency wind-forcing,
and the remote equatorial Kelvin wave forcing. The latter two
forcing mechanisms can be represented by surface and open
boundary conditions with high-frequency interannual variability, respectively. For the modeling of NETP, added difficulty is the narrow mountain gaps in the Central America that
strong wind jets funnel through. A daily wind product with
corrected gap winds is developed based on both the blended
QSCAT/NCEP daily wind at a 0.5 0.5 grid and scatterometer climatology of ocean wind at a 0.25 0.25 grid.
[33] Through a systematic analysis of 10 solutions forced
by different combinations of surface and lateral boundary
conditions, we show that the mean seasonal ocean state is not
sensitive to high-frequency surface and open boundary conditions. The regional mesoscale variability depends on the
spatial and temporal variability of wind and open boundary.
By comparing numerical solutions with AVISO sea level
anomaly data, we find that the run forced by corrected daily
wind and 5-day interannual open boundary condition is
closest to the real mean state and variability of the ocean. The
combined low-frequency wind and boundary forcing is the
most important eddy generation mechanism, and accounts
for more than 68% to 80% mesoscale variability in the
Tehuantepec region and 62% to 90% mesoscale variability in
the Papagayo region. Without the local high-frequency windforcing, mesoscale variability is under-estimated by 7% to
20% in the Tehuantepec region, and by 1% to 12% in the
Papagayo region in different seasons. Without the remote
equatorial Kelvin wave forcing, mesoscale variability is
under-estimated by 6% to 12% in the Tehuantepec region,
and by 2% to 13% in the Papagayo region in different seasons. These results also stress the importance of high-resolution wind and open boundary conditions in a numerical
model in reproducing the realistic mesoscale variability of
the NETP.
[34] An eddy identification and tracking algorithm was
applied to both AVISO sea level anomaly data and ROMS
solutions. Eddies in the Tehuantepec and Papagayo regions
are more likely to deflect equatorward than eddies in the rest
region of the NETP because eddies in these two regions are
advected by offshore currents that have a strong equatorward
component. Eddies in the two regions are also larger in
C07003
amplitude and size than eddies in the rest of the NETP.
Temperature anomaly associated with eddies in the NETP is
at around 50 m, shallower than that with eddies off the
coasts of California and Peru/Chile. Eddies with lives shorter
than 6 weeks are more likely cyclonic, while eddies with
lives longer than 6 weeks are more likely anticyclonic.
Besides previously proposed preferential anticyclone generation mechanisms due to the shallow thermocline, the
preferential destruction of strong cyclonic eddies at finite
Rossby number and small Burger number under ambient
deformation is also important for the anticyclone dominance
among long-lived large-amplitude eddies in the NETP.
[35] Acknowledgments. JHL was supported by NASA headquarters
under the Earth System Science Fellowship Grant NNX08AU98H. JCM,
JK, FC, PW, and YU acknowledge support from the Office of Naval
Research through contract N00014-08-1-0597. Computation in this study
was performed on supercomputers Thresher and Trestles at the San Diego
Supercomputing Center. We thank Alexis Chaigneau and two anonymous
reviewers for their constructive comments and useful suggestions.
References
Adams, D. K., and G. R. Flierl (2010), Modeled interactions of mesoscale
eddies with the East Pacific Rise: Implication for larval dispersal, Deep
Sea Res., Part I, 57, 1163–1176.
Alexander, M. A., H. Seo, S.-P. Xie, and J. D. Scott (2012), ENSO’s impact
on the gap wind regions of the eastern tropical Pacific Ocean, J. Clim., 25,
3549–3565.
Alory, G., C. Maes, T. Delcroix, N. Reul, and S. Illig (2012), Seasonal
dynamics of sea surface salinity off Panama: The far eastern Pacific fresh
pool, J. Geophys. Res., 117, C04028, doi:10.1029/2011JC007802.
Barnier, B., L. Siefried, and P. Marchesiello (1995), Thermal forcing for a
global ocean circulation model using a three-year climatology of
ECMWF analyses, J. Mar. Syst., 6, 363–380.
Barton, E. D., M. L. Argote, J. Brown, P. M. Kosro, M. Lavín, J. M. Robles,
R. L. Smith, A. Trasviña, and H. S. Velez (1993), Supersquirt: Dynamics
of the Gulf of Tehuantepec, Mexico, Oceanography, 6, 23–30.
Barton, E. D., M. Lavín, and A. Trasviña (2009), Coastal circulation and
hydrography in the Gulf of Tehuantepec, Mexico, during winter, Cont.
Shelf Res., 29, 23–30.
Beckman, A., and D. Haidvogel (1993), Numerical simulation of flow
around a tall isolated seamount, J. Phys. Oceanogr., 23, 1736–1753.
Belmadani, A., V. Echevin, B. Dewitte, and F. Colas (2012), Equatorially
forced intraseasonal propagations along the Peru-Chile coast and their
relation with the nearshore eddy activity in 1992–2000: A modeling
study, J. Geophys. Res., 117, C04025, doi:10.1029/2011JC007848.
Brenes, C. L., M. F. Lavin, and A. S. Mascarenhas (2008), Geostrophic circulation between the Costa Rica Dome and Central America, Deep Sea
Res., Part I, 55, 608–629, doi:10.1016/j.dsr.2008.02.005.
Carton, J., and B. Giese (2008), A reanalysis of ocean climate using Simple
Ocean Data Assimilation (SODA), Mon. Weather Rev., 136, 2999–3017.
Casey, K. S., T. B. Brandon, P. Cornillon, and R. Evans (2010), The past,
present, and future of the AVHRR Pathfinder SST Program, in Oceanography From Space: Revisited, edited by V. Barale, J. F. R. Gower, and
L. Alberotanza, pp. 273–288, Springer, New York.
Chaigneau, A., R. Abarca del Rio, and F. Colas (2006), Lagrangian study of
the Panama Bight and surrounding regions, J. Geophys. Res., 111,
C09013, doi:10.1029/2006JC003530.
Chaigneau, A., M. Le Texier, G. Eldin, C. Grados, and O. Pizarro (2011),
Vertical structure of mesoscale eddies in the eastern South Pacific Ocean:
A composite analysis from altimetry and Argo profiling floats, J. Geophys.
Res., 116, C11025, doi:10.1029/2011JC007134.
Chelton, D. B., R. A. deSzoeke, M. G. Schlax, K. El Naggar, and N. Siwertz
(1998), Geographical variability of the first baroclinic Rossby radius of
deformation, J. Phys. Oceanogr., 28, 433–460.
Chelton, D. B., M. H. Freilich, and S. K. Esbensen (2000), Satellite observations of the wind jets off the Pacific coast of Central America. Part I:
Case studies and statistical characteristics, Mon. Weather Rev., 128,
1993–2018.
Chelton, D. B., M. G. Schlax, R. M. Samelson, and R. A. de Szoeke (2007),
Global observations of large oceanic eddies, Geophys. Res. Lett., 34,
L15606, doi:10.1029/2007GL030812.
Chelton, D. B., M. G. Schlax, and R. M. Samelson (2011), Global observations of nonlinear mesoscale eddies, Prog. Oceanogr., 91, 167–216.
12 of 13
C07003
LIANG ET AL.: CENTRAL AMERICA EDDIES
Clarke, A. J. (1988), Inertial wind path and sea surface temperature patterns
near the Gulf of Tehuantepec and Gulf of Papagayo, J. Geophys. Res., 93,
15,491–15,501.
Colas, F., J. C. McWilliams, X. Capet, and J. Kurian (2011), Heat balance
and eddies in the Peru-Chile current system, Clim. Dyn., doi:10.1007/
s00382-011-1170-6, in press.
da Silva, A., C. Young, and S. Levitus (Eds.) (1994), Atlas of Surface
Marine Data 1994, vol. 1, Algorithms and Procedures, NOAA Atlas
NESDIS, vol. 6, 83 pp., NOAA, Silver Spring, Md.
Echevin, V., F. Colas, A. Chaigneau, and P. Penven (2011), Sensitivity of
the Northern Humboldt Current System nearshore modeled circulation
to initial and boundary conditions, J. Geophys. Res., 116, C07002,
doi:10.1029/2010JC006684.
Farrar, J. T., and R. A. Weller (2006), Intraseasonal variability near 10 N in
the eastern tropical Pacific Ocean, J. Geophys. Res., 111, C05015,
doi:10.1029/2005JC002989.
Giese, B. S., J. A. Carton, and L. J. Holl (1994), Sea level variability in the
eastern tropical Pacific as observed by TOPEX and Tropical OceanGlobal Atmosphere Tropical Atmosphere-Ocean Experiment, J. Geophys.
Res., 99(C12), 24,739–24,748.
Graves, L., J. McWilliams, and M. Montgomery (2006), Vortex evolution
due to straining: A mechanism for dominance of strong, interior anticyclones, Geophys. Astrophys. Fluid Dyn., 100, 151–183.
Hansen, D. V., and G. A. Maul (1991), Anticyclonic current rings in the
eastern Pacific Ocean, J. Geophy. Res., 96(C4), 192–194.
Kessler, W. S. (2002), Mean three-dimensional circulation in the northeast
tropical Pacific, J. Phys. Oceanogr., 32, 2457–2471.
Kessler, W. S. (2006), The circulation of the eastern tropical Pacific:
A review, Prog. Oceanogr., 69, 181–217.
Kurian, J., F. Colas, X. Capet, J. C. McWilliams, and D. B. Chelton (2011),
Eddy properties in the California Current System, J. Geophys. Res., 116,
C08027, doi:10.1029/2010JC006895.
Lemarie, F., J. Kurian, A. F. Shchepetkin, M. J. Molemaker, F. Colas, and
J. C. McWilliams (2012), Are there inescapable issues prohibiting the use
of terrain-following coordinates in climate models?, Ocean Modell., 42,
57–79.
Liang, J.-H., J. C. McWilliams, and N. Gruber (2009), High-frequency
response of the ocean to mountain gap winds in the northeastern tropical
Pacific, J. Geophys. Res., 114, C12005, doi:10.1029/2009JC005370.
Marchesiello, P., J. McWilliams, and A. Shchepetkin (2001), Open boundary conditions for long-term integration of regional oceanic models,
Ocean Modell., 3, 1–20.
Marchesiello, P., J. McWilliams, and A. Shchepetkin (2003), Equilibrium structure and dynamics of the California Current System, J. Phys. Oceanogr., 33,
753–783.
Mason, E., J. Molemaker, A. F. Shchepetkin, F. Colas, J. C. McWilliams,
and P. Sangra (2010), Procedures for offline grid nesting in regional
ocean models, Ocean Modell., 35, 1–15.
McClain, C. R., J. R. Christian, S. R. Signorini, M. R. Lewis, I. Asanuma,
D. Turk, and C. Dupouy-Douchement (2002), Satellite ocean-color
observations of the tropical Pacific Ocean, Deep Sea Res., Part II, 49,
2533–2560.
McCreary, J. P., Jr., H. S. Lee, and D. B. Enfield (1989), The response of
the coastal ocean to strong offshore winds: With application to the Gulfs
of Tehuantepec and Papagayo, J. Mar. Res., 47, 81–109.
McWilliams, J. C., and G. R. Flierl (1979), On the evolution of isolated,
nonlinear vortices, J. Phys. Oceanogr., 9, 1155–1182.
Milliff, R. F., J. Morzel, D. B. Chelton, and M. H. Freilich (2004), Wind
stress curl and divergence biases from rain effects on QuikSCAT surface
wind retrievals, J. Atmos. Oceanic Technol., 21, 1216–1231.
C07003
Palacios, D. M., and S. J. Bograd (2005), A census of Tehuantepec and
Papagayo eddies in the northeastern tropical Pacific, Geophys. Res. Lett.,
32, L23606, doi:10.1029/2005GL024324.
Pascual, A., Y. Faugère, G. Larnicol, and P.-Y. Le Traon (2006), Improved
description of the ocean mesoscale variability by combining four satellite
altimeters, Geophys. Res. Lett., 33, L02611, doi:10.1029/2005GL024633.
Pennington, J. T., K. L. Mahoney, V. S. Kuwahara, D. D. Kolber,
R. Calienes, and F. P. Chavez (2006), Primary production in the eastern
tropical Pacific: A review, Prog. Oceanogr., 69(2–4), 285–317.
Perret, G., A. Stegner, M. Farge, and T. Pichon (2006), Cyclone-anticyclone
asymmetry of large-scale wakes in the laboratory, Phys. Fluids, 18,
036603, doi:10.1063/1.2179387.
Risien, C. M., and D. B. Chelton (2008), A global climatology of surface
wind and wind stress fields from 8 years of QuikSCAT scatterometer
data, J. Phys. Oceanogr., 38, 2379–2413.
Robinson, A. R., and J. C. McWilliams (1974), The baroclinic instability of
the open ocean, J. Phys. Oceanogr., 4, 281–294.
Samuelsen, A., and J. J. O’Brien (2008), Wind-induced cross-shelf flux of
water masses and organic matter at the Gulf of Tehuantepec, Deep Sea
Res., Part I, 55, 221–246, doi:10.1016/j.dsr.2007.11.007.
Shchepetkin, A., and J. C. McWilliams (2005), The regional oceanic modeling system (ROMS): A split-explicit, free-surface, topography-followingcoordinate ocean model, Ocean Modell., 9, 347–404.
Smith, W. H. F., and D. T. Sandwell (1997), Global seafloor topography from satellite altimetry and ship depth sounding, Science, 277,
1957–1962.
Sun, F., and J.-Y. Yu (2006), Impacts of Central America gap winds on the
SST annual cycle in the eastern Pacific warm pool, Geophys. Res. Lett.,
33, L06710, doi:10.1029/2005GL024700.
Thomas, L. N., and P. B. Rhines (2002), Nonlinear stratified spin-up,
J. Fluid Mech., 473, 211–244.
Trasviña, A., and E. D. Barton (2008), Summer circulation in the Mexican
tropical Pacific, Deep Sea Res., Part I, 55, 587–607.
Trasviña, A., E. D. Barton, J. Brown, H. S. Vélez, P. M. Kosro, and R. L.
Smith (1995), Offshore wind forcing in the Gulf of Tehuantepec, Mexico:
The asymmetric circulation, J. Geophys. Res., 100, 20,649–20,663.
Trasviña, A., E. D. Barton, H. S. Vélez, and J. Brown (2003), Frontal subduction of a cool surface water mass in the Gulf of Tehuantepec, Mexico,
Geofis. Int., 42(1), 101–114.
Wang, C., and P. C. Fiedler (2006), ENSO variability and the eastern tropical Pacific: A review, Prog. Oceanogr., 69, 239–266.
Willett, C. S., R. R. Leben, and M. F. Lavín (2006), Eddies and tropical instability waves in the eastern tropical Pacific: A review, Prog. Oceanogr., 69,
218–238, doi:10.1016/j.pocean.2006.03.010.
Wyrtki, K. (2006), Preface, Prog. Oceanogr., 69, 91–93.
Xie, S. P., H. Xu, W. S. Kessler, and M. Nonaka (2005), Air-sea interaction
over the eastern Pacific warm pool: Gap winds, thermocline dome, and
atmospheric convection, J. Clim., 18(1), 5–20.
Zamudio, L., A. P. Leonardi, S. D. Meyers, and J. J. O’Brien (2001), ENSO
and eddies on the southwest coast of Mexico, Geophys. Res. Lett., 28(1),
13–16, doi:10.1029/2000GL011814.
Zamudio, L., H. E. Hurlburt, E. J. Metzger, S. L. Morey, J. J. O’Brien, C. E.
Tilburg, and J. Zavala-Hidalgo (2006), Interannual variability of Tehuantepec eddies, J. Geophys. Res., 111, C05001, doi:10.1029/2005JC003182.
Zamudio, L., H. E. Hurlburt, E. J. Metzger, and C. E. Tilburg (2007), Tropical
wave-induced oceanic eddies at Cabo Corrientes and the Maria Islands,
Mexico, J. Geophys. Res., 112, C05048, doi:10.1029/2006JC004018.
Zamudio, L., P. Hogan, and E. J. Metzger (2008), Summer generation of the
southern Gulf of California eddy train, J. Geophys. Res., 113, C06020,
doi:10.1029/2007JC004467.
13 of 13