ARTICLE IN PRESS
Continental Shelf Research 30 (2010) 1241–1249
Contents lists available at ScienceDirect
Continental Shelf Research
journal homepage: www.elsevier.com/locate/csr
Upwelling-enhanced seasonal stratification in a semiarid bay
Peng Cheng a,n, Arnoldo Valle-Levinson a, Clinton D. Winant b, Aurelien L.S. Ponte b,
Guillermo Gutierrez de Velasco c, Kraig B. Winters b
a
Department of Civil and Coastal Engineering, University of Florida, Gainesville, FL, USA
Integrative Oceanography Division, Scripps Institution of Oceanography, University of California, San Diego, La Jolla, CA, USA
c
CICESE, Unidad La Paz, La Paz, Baja California, Mexico
b
a r t i c l e in fo
abstract
Article history:
Received 20 July 2009
Received in revised form
22 February 2010
Accepted 10 March 2010
Available online 9 April 2010
The role of wind-driven upwelling in stratifying a semiarid bay in the Gulf of California is demonstrated
with observations in Bahı́a Concepción, Baja California Sur, Mexico. The stratification in Bahı́a Concepción is
related to the seasonal heat transfer from the atmosphere as well as to cold water intrusions forced by
wind-driven upwelling. During winter, the water column is relatively well-mixed by atmospheric cooling
and by northwesterly, downwelling-favorable, winds that typically exceed 10 m/s. During summer, the
water column is gradually heated and becomes stratified because of the heat flux from the atmosphere.
The wind field shifts from downwelling-favorable to upwelling-favorable at the beginning of summer, i.e.,
the winds become predominantly southeasterly. The reversal of wind direction triggers a major cold water
intrusion at the beginning of the summer season that drops the temperature of the entire water column by
3–5 1C. The persistent upwelling-favorable winds during the summer provide a continuous cold water
supply that helps maintain the stratification of the bay.
Published by Elsevier Ltd.
Keywords:
Stratification
Wind-driven upwelling
Wind straining
1. Introduction
Water column stratification of the density field in coastal
environments can arise from a variety of causes. The major
mechanisms controlling vertical stratification are buoyancy
inputs, mixing produced by winds and tides, and the interaction
of vertically sheared currents with a horizontal density gradient,
i.e. tidal straining (Simpson et al., 1990). In coastal environments
devoid of freshwater influences, the main stratifying agent to the
water is atmospheric input of heat at the water’s surface, which
produces typical seasonal patterns of vertical stratification. This
study will show that coastal upwelling can provide an additional
source of buoyancy, in this case negative buoyancy, into a coastal
embayment from the adjacent ocean thus enhancing water
column stratification.
Classically, coastal upwelling has been understood as a windforced process (e.g. Gill, 1982) with alongshore wind stress
driving offshore surface Ekman transport and upwelling of deeper
waters. Therefore, the forcing by winds and the reversal of their
direction can be expected to play an important role in determining the water temperature and density stratification in coastal
embayments. For instance, in the western coast of the Iberian
Peninsula, shelf winds follow a seasonal pattern and water
n
Corresponding author. Tel.: + 1 631 816 9046.
E-mail address: [email protected]fl.edu (P. Cheng).
0278-4343/$ - see front matter Published by Elsevier Ltd.
doi:10.1016/j.csr.2010.03.015
column stratification in the Rı́a de Vigo is enhanced by the cold
water driven by upwelling winds (Gilcoto et al., 2007).
In this study, year-long observations of water column stratification were carried out in Bahı́a Concepción (Fig. 1), located in the east
coast of the Baja California peninsula in Mexico. The main objective
is to explore the relationship between water column stratification
and wind direction, and in general to demonstrate the role of the
adjacent ocean in enhancing the stratification of a bay or lagoon. In
Section 2, a description of study site and data collection is given. In
Section 3, the seasonal evolution of stratification, the estimation of
advective contribution to heat content and the dominant modes of
temperature variability are presented. In Section 4, relationships
between winds and water surface level are addressed. Also, two
mechanisms relevant to wind forcing are discussed: (a) the
generation of longitudinal density gradients by wind-driven
upwelling, thus contributing to density-driven flows; and (b) the
mixing/stratification competition. In Section 5, the main conclusions
of this study are presented.
2. Study site and observations
Bahı́a Concepción is a 40 km-long and 5–10 km-wide bay
oriented along a nearly NNW–SSE axis with an opening to the Gulf
of California, the bay entrance, to the north (Johnson and
Ledesma-Vásquez, 2001). The bay’s bathymetry is relatively
simple. The entrance to Bahı́a Concepción consists of a central
channel with a sill at about 15 m tucked against the eastern shore.
ARTICLE IN PRESS
1242
P. Cheng et al. / Continental Shelf Research 30 (2010) 1241–1249
−35
−35
−25 −5
32
28
−15
−2
5
−2 5
−2
− 15
−5
26.9
5
−5
Mexico
−5
−5
ali
−5
fC
lf o
Gu
5
−1
r
fo
24
nia
−5
26.8
− 25
−15
−113
5
−25 −2 −
−5
−1
5
5
−5
5
−2
−5
−25
26.7
−5
−15
−5
−1−55
Latitude, degree
−115
−−15
5
(http://hycom.rsmas.miami.edu/dataserver/). HYCOM uses a penetrating solar radiation scheme (Kara et al., 2005) and can
predict relatively accurate sea surface temperatures (Wallcraf
et al., 2008). Because no model grid points are available inside
Bahı́a Concepción, the data from the grid point nearest to the bay
entrance was selected as a representative of the entire bay.
The data analyses described here are based on 15-min averages
of the original observations, except for the half-daily QuikSCAT
winds and the daily HYCOM heat flux. Short gaps are filled by linear
interpolation and periods where longer gaps exist are excluded from
the analysis. Wind stress is estimated from wind speed and direction
following Large and Pond (1981) for a neutral atmosphere. Subtidal
water level is obtained by low-pass filtering the bottom pressure
with a Lanczos filter at half-power of 34 h. Local wind stresses were
rotated along the axis of the south part of the bay while the remote
wind stresses were rotated to the alongshore direction representative of the west coast of the Gulf of California.
3. Results
−5
3.1. Seasonal evolution of stratification
temperature
26.6
wind/water level
quickscat wind
− 25
−5
−111.95
−111.85
−15
−5
−111.75
−111.65
Longitude, degree
Fig. 1. Bahı́a Concepción in the context of the Gulf of California, showing the
measurement stations. Circles indicate the position of thermistor chains, squares
show the stations where bottom pressure and winds were measured. The triangle
in the insert map shows the location for QuikSCAT winds. The bathymetric
distribution of the bay is contoured at 5 m intervals. The position of Bahı́a
Concepción is shown by the square on the top left map.
In the southern half, the bay geometry resembles a 30 m-deep
bathtub. The seasonal wind pattern is predominantly NW during
autumn and winter (cold period with well-mixed water column)
and SE during spring and summer (warm period with stratified
water column). During the summer season, the vertical temperature difference is well defined while salinity stratification is less
obvious (Lopez-Cortes et al., 2003; Palomares-Garcia et al., 2006;
Canar et al., 2008).
Moored records of water temperature were obtained from
February to October 2005 along a cross-bay transect (Fig. 1)
consisting of six thermistor (Tloggers) chains. At the central
mooring sites (stations 3 and 4), temperature sensors were
deployed at five equidistant depths ranging between near-surface
and near-bottom (i.e. 5, 10, 15, 20, and 25 m). At the four other
stations, water temperature was measured at three equidistant
depths: 5, 10, and 15 m.
Bottom pressure was measured at 15 min intervals near the
entrance and close to the head, at the south end of the bay using
CTDs with Paroscientific pressure sensors (SeaBird SBE26). Winds
were recorded near the pressure gauges using Aanderaa anemometers sampling at 1 Hz. The anemometers were deployed on
land at a height of 10 m above mean sea level. In addition,
regional winds were collected from QuikSCAT satellite open data
with 12.5 km resolution (http://airsea.jpl.nasa.gov/DATA/QUIKS
CAT/wind/). The station in the middle of the Gulf of California
(27.3751N, 111.3751W) was chosen as a representative of remote
wind conditions for the bay. Daily net heat flux at the water
surface was obtained from the Global HYCOM (HYbrid Coordinate
Ocean Model) model, which has 7 km mid-latitude resolution
Observations that spanned nearly a year are illustrated in Fig. 2.
Water temperature at the central site (station 4) was selected as
representative of the other sites (first panel). Generally, water
temperature at all five depths increased from winter to summer,
showing a seasonal warming (Fig. 2a). During the winter period
(before day 110), water temperature was nearly uniform in depth
indicating a well-mixed water column. Starting on day 110, the
water temperature at the five depths diverged as the water column
became stratified. From day 145 to 150, water temperature dropped
throughout the water column. This drop was steeper at larger depth.
After that drop, the water column restored its warming trend and
water temperature continued to increase until past the autumn
equinox, day 275, when vertical stratification was 0.5 1C/m.
During the stratified summer season (after day 150), the water
temperature near the bottom (at depths of 20 and 25 m) increased
approximately linearly (if tidal variations are neglected) by 5 1C in
125 days or 0.04 1C/day. In contrast, the temperature in the middle
of the water column (depths of 10 and 15 m) fluctuated distinctly at
higher amplitudes and frequencies than near the bottom. This
indicated that the cooling event on days 141–155 resulted from a
major cold water intrusion into the bay that dropped the entire
water column temperature. After the major cooling event,
intermittent cold water input also affected the heat content of the
water column but not as markedly as the major cold water intrusion
(this will be shown in Section 3.2).
The net heat flux increased from winter to summer showing
seasonal heat transfer from the atmosphere to the ocean (Fig. 2b).
From day 75 onward, the net heat flux became positive as the bay
began to gain heat from the atmosphere. The general trend of net
heat flux was responsible for the warming of the bay and
contributed to the development of summer stratification in the
water column. The major water cooling event on days 141–155
should not have been caused by atmospheric heat fluxes because
such fluxes did not decrease during this period. Interestingly, there
was a cooling from 200 to 60 W/m2, on day 150, that only
affected the water temperature at 5 m, i.e., it did not reach depths
larger than 10 m. Furthermore, the fluctuations of the thermocline
were not correlated to the net heat flux, which suggested that
atmospheric heat fluxes during the stratified period only influenced
the near-surface layer (upper 5 m) of the bay. Therefore, the water
column cooling on days 141–155 must have been caused by
advection of cold water into the bay from the adjacent Gulf of
California in response to upwelling, as explored next.
ARTICLE IN PRESS
P. Cheng et al. / Continental Shelf Research 30 (2010) 1241–1249
5m
10m
15m
20m
25m
30
degree
1243
25
20
a
400
b
W m−2
200
0
−200
speed (m s−1)
10
c
5
0
−5
−10
−15
0
50
100
150
200
250
300
day of year, 2005
Fig. 2. Time series of temperature at different depths (5, 10, 15, 20, 25 m) at station 4 (a); daily water surface net heat flux from the atmosphere, which is obtained from the
global HYCOM model (b); and QuikSCAT wind speed in the middle of the Gulf of California (c).
The wind in the Gulf of California is characterized by distinct
seasonal variations (Fig. 2c). Before day 140, during the winter and
early spring, the wind was predominantly toward the SE while
during the summer (days 140–275) it was toward the NW. After day
275, during the autumn, the wind shifted back toward SE. Winds
toward the NW during the summer drove surface water away from
the bay entrance and produced upwelling toward the coast. In
general, summer winds produce upwelling in most of the western
coast of the Gulf of California (Badan-Dangon et al., 1991). The
cooling event on days 145–150 coincided with the beginning of
northwestward winds in the gulf. Therefore, this major cooling event
should have been caused by the upwelling-driven cold water
intrusion from the adjacent ocean basin. The persistent northwestward winds throughout the stratified summer season maintained a cold water supply into the bay. During the winter and
autumn, southeastward winds dominate and cause downwelling
over the eastern coast of the Baja California peninsula. These
persistent downwelling winds should effectively reverse the
exchange pattern at the bay entrance and, together with atmospheric cooling, favor mixing throughout the bay. The water
temperature stratification in Bahı́a Concepción is thus related to
seasonal atmospheric heat fluxes as well as upwelling/downwelling
favorable winds along the eastern shore of the Baja California
peninsula. This advective effect on heat fluxes is addressed in
Section 3.2, and the relationships between winds and sea level are
explored in Section 4.1.
3.2. Estimation of the advective contribution to heat content
In order to examine the importance of horizontal advection,
the net heat flux is estimated by taking the time derivative of the
heat content of the whole water column and comparing it to the
net surface heat flux obtained from the HYCOM model. The
mismatch represents the contribution from horizontal advection.
The heat content (HC, in J/m2) of the water column can be
expressed as
HC ¼
Z0
C p r ðTTref Þdz,
ð1Þ
H
where Cp is the specific heat of seawater, r is water density, T is
water temperature, Tref is a reference temperature, which is
arbitrarily set to zero 1C, H is the depth of the water column
(taken as 30 m for station 4). The setting of Tref to zero rather than
to some other value has practically no influence on the HC
variability through r, because the density difference in the water
column Dr is small compared to the water density itself, i.e.
Dr)r. For computation of Cp, the empirical formula given by
Millero et al. (1973) has been applied. The estimation is carried
out at the center site (station 4). In the calculation of heat content,
high frequencies of the temperature data were removed by using
a Lanczos filter at half-power of 3 days. The calculated heat
content then was smoothed with a 5-point running mean (Fig. 3
top panel) in order to reduce fluctuations of the estimated
heat flux.
The heat content generally increases during the observation
period and shows, as expected, that the water column is gaining
heat from winter to summer. A noticeable decrease in heat
content around day 150 corresponds to the major cold water
intrusion revealed from the temperature evolution (Fig. 2 top
panel). Another similar cooling event occurred around day 240
but has a smaller reduction of heat content relative to the major
cooling event. From day 170 to day 220, the heat content
increased at a slower rate than that during the winter, indicating
the additional supply of cold water.
The time derivative of the heat content represents the total net
heat flux (Fig. 3 bottom panel). During winter, the estimated total
heat flux shows a similar trend as that from HYCOM data,
ARTICLE IN PRESS
1244
P. Cheng et al. / Continental Shelf Research 30 (2010) 1241–1249
a
9
10 J
2.8
2.4
2
300
b
W m –2
150
0
−150
Estimated
HYCOM data
−300
50
100
150
200
250
day of year, 2005
Fig. 3. Estimation of heat content of water column at station 4 (a) and net surface heat flux (b). Solid line represents the estimation while the dashed line represents daily
HYCOM heat flux data.
showing that the heat content in the water column is mainly
determined by atmospheric heat input. During the summer,
however, large discrepancies between the two time series of heat
flux can be found, corresponding to reductions in heat content.
The drop in the estimated heat flux indicates loss of heat and is
attributed to horizontal advection of cold water. The largest
discrepancy appears on day 150 representing the major cold
water intrusion. The other discrepancies that occurred during
summer were smaller than the major cooling event, and show
relatively weak intermittent cold water supply to the basin. This is
consistent with the temperature evolution as shown in Fig. 2a.
3.3. Dominant modes of temperature variability
The spatial patterns of temperature at the observation transect
are determined using empirical orthogonal functions (EOFs)
(Fig. 4). The temporal mean of the transect shows a stratified
water column where temperature decreases approximately
linearly from surface to bottom with a top-to-bottom difference
of 6 1C or 0.2 1C/m on average. This mean temperature
distribution is practically uniform across the transect but the
isotherms slope upward (from west to east) and then downward
close to the eastern shore of the bay. This could be caused by
increased mixing effects near the edges of the bay’s channel. The
first two EOF modes explain 97% of the variance and the other
modes are negligible.
The first EOF mode accounts for 89% of the total variance and
has a similar spatial pattern to the mean, representing warming/
cooling of the entire water column. The contribution of each EOF
mode to the temperature variance is the product of the EOF mode
and the corresponding principal component (PC). Because the first
EOF mode is positive, positive PC1 represents warming while
negative PC1 indicates cooling of the water column. The PC1
increases from 25 to 22, showing a trend of warming of the
water column from winter to summer. Before day 140, the PC1 is
negative and gradually increases. This means that the water
column is cooler than the mean in that period and is gradually
heated between winter and summer. During the summer period
the PC1 is mostly positive and represents warming of the water
column. The evolution of PC1 is very similar to the heat content
(Fig. 3a). The three cooling events identified from heat content are
also found in PC1. According to the analysis just described, the
first EOF mode represents effects of both surface heating and cold
water intrusions from the adjacent gulf. The cold water intrusion
is related to the along-shelf wind, which is obtained by projecting
the remote winds to the axis of the Gulf of California (Fig. 4 lower
panel). Positive values represent northwestward winds. Because
the cold water intrusion is indirectly related to along-shelf winds,
the correlation between the wind stress and PC1 is not high.
However, the three cooling events inferred from the heat content
and PC1 can be interpreted with the evolution of the wind stress.
The cold water intrusion occurred around day 150 and is clearly
related to the major shift in wind direction. During days 170–220,
the southeastward winds are generally strong (around 10 m/s)
and account for the cooling of the water column. Around day 140,
there is a distinct pulse of southeastward winds that are
responsible for the cooling event on day 140.
The second mode explains 8% of the total variance for the
entire cross-section and is characterized by positive values near
the surface and negative values at depth. This represents a
warming at a surface layer and simultaneous cooling at a layer
underneath. Both modes 1 and 2 include the signature of the cold
water intrusions. When PC2 is positive, mode 2 tends to reinforce
the mean pattern, i.e. to increase vertical stratification in
temperature. In contrast when PC2 is negative, mode 2 tends to
reduce the vertical temperature difference because the surface
cools down and the near-bottom layer warms up. Therefore, a
positive PC2 represents increased stratification and vice versa. The
PC2 is well correlated with top-bottom temperature difference
(the correlation coefficient, r is 0.87). Because stratification is
partly driven by the remote wind, the general evolution of PC2
agrees with the trend of remote wind stress: when the wind is
toward SE (positive wind stress), PC2 is negative, indicating that
ARTICLE IN PRESS
P. Cheng et al. / Continental Shelf Research 30 (2010) 1241–1249
Depth, m
−5
0.2
0.1
0
−0.1
−0.2
25
−10
0.2
−15
−20
0.1
21
mean
−25
0
2
1245
4
EOF1
0
2
4
EOF2
0
2
4
Distance from the west coast, km
23
20
PC1
0
−20
10
PC2
0
−10
pa
0.2
remote wind stress
0.1
0
−0.1
50
100
150
day of year, 2005
200
250
Fig. 4. EOF analysis of temperature along the mooring transect. The top row shows mean of temperature and the first two EOF modes. Darker areas denote negative values.
The second and the third rows show the first and the second principal components, respectively. The low row shows along-bay component of local wind stress at the
mouth of the bay. Positive values of wind stress represent northwesterly wind.
northwesterly winds reduce stratification; when the wind is
toward NW (negative wind stress), PC2 is positive, indicating that
southeasterly winds enhance stratification.
4. Discussion
4.1. Relationships between winds and water surface elevation
Subtidal fluctuations of water level in a semienclosed basin can
be caused by remote or local forcing (e.g. Wang, 1979; Garvine,
1985; Wong and Valle-Levinson, 2002). Remote forcing produces
coastal water level fluctuations that drive unidirectional net flows
throughout the entrance to a basin. Local winds produce sea-level
slopes inside the basin and bidirectional exchange flow at the
entrance to the basin. Because the orientation of the along-basin
axis in Bahı́a Concepción is nearly parallel to that of the Gulf of
California, along an approximately northwest–southeast direction, both remote and local winds act in concert and produce
water level slopes along the bay. The subtidal water level
differences between the head and the mouth stations (Fig. 5b)
responded to the remote and local winds in the expected fashion:
northwestward winds caused sea level to rise toward the north
and east in the Gulf of California. Inside the bay, northwestward
winds caused the water level to increase at the mouth relative to
the head, while southeastward winds produced the opposite
response. The subtidal water level at the head was generally
higher than that at the mouth during the winter (positive values,
days 80–140, Fig. 5b), then dropped at the beginning of the
summer (negative value) in response to the wind reversal.
This further illustrates the shifting from downwelling to
upwelling conditions at the beginning of the summer. The water
surface elevation difference of 5 cm during the major cooling
period around day 150 was caused by the wind reversal. As
southeasterly winds continued throughout the summer, the water
level slope bounced back by day 200 and fluctuated from then
onward.
The local wind field in Bahı́a Concepción (Fig. 5c) was
consistent with the remote wind outside the bay on the Gulf of
California (Fig. 4). Positive values represent southeastward wind
stress. It is clear that during the winter period local winds were
mostly southeastward (positive) and the wind stress was larger
than in the summer, exceeding 0.1 Pa at times. In the summer,
local winds were mostly northwestward and relatively weak. The
strong winds during the winter contributed to mix the water
column and the weak winds during the summer helped to
maintain the stratification. It is evident that the shifting of local
wind pattern is not as influential as that of remote wind. This is
partly because the local wind is highly influenced and steered by
local topographical features. This is because Bahı́a Concepción is
the result of land downthrow near a geological fault zone with
steep escarpments (Johnson and Ledesma-Vásquez, 2001). Thus
topography confines local winds along the Concepción fault zone.
In an elongate coastal embayment, the depth-averaged
momentum balance between pressure gradient and wind stress
can be simplified as
0 ¼ g
@Z
t
þ
,
@x r0 H
ð2Þ
where g is gravitational acceleration, H is water depth, x
represents longitudinal coordinate, r0 is reference density, and t
is the along-basin wind stress. A number of terms are omitted in
ARTICLE IN PRESS
1246
P. Cheng et al. / Continental Shelf Research 30 (2010) 1241–1249
a
0.05
m
0
−0.05
1
2
4
3
−0.1
b
m
0.05
0
−0.05
c
0.15
1
Pa
0.05
0
−0.05
−0.1
0
50
100
150
day of year, 2005
200
250
300
Fig. 5. Time series of subtidal water level at the mouth of the bay (a); the subtidal water level difference between the head and the mouth of the bay (b); numbers mark
different periods of measurement and the along-shelf wind stress taking from the QuikSCAT data. Positive values of wind stress represent northwesterly wind.
Eq. (2). Winant (2004) theoretically predicted that the winddriven flow is forced by the local wind stress and the down-wind
water surface slope. Therefore, the along-bay wind stress t can be
related to the mouth–head water level difference (Fig. 6a). During
the winter period, the data sets from day 20 to 80 were chosen for
analysis. The squared coherence between the local wind stress
and the subtidal water level difference is about 0.9 at almost all
low frequencies. This agrees with Ponte (2009), who showed high
correlation between local wind stress and water surface slope
during winter. The local wind is nearly in phase with the water
level slope (01, Fig. 6c). This implies that southeastward winds
produce higher water level at the head of the basin. During the
summer period, the data sets from days 190 to 270 were chosen
for analysis. The relation between local wind stress and water
level difference became less significant and the phase slightly
increased. This is due to the stratification of water column during
the summer period. Janzen and Wong (2002) pointed out that
stratification reduces the effective depth over which the wind
stress must act. Moreover, the linear dynamics for wind driven
circulation (e.g. Eq. (2)) might be impaired by stratification.
The cold water intrusion, in turn, generates horizontal density
gradients that enhance the contribution of baroclinic pressure
gradients (Fig. 6b). Under stratification, Earth’s rotation could be
important. During the stratified summer season, the internal
Rossby radius of deformation estimated using the stratification
inferred from the top-bottom temperature difference is about
3.5 km, which is smaller than its width indicating that Coriolis
forcing is effective. For wind-induced circulation, the water depth
is larger than twice the Ekman depth, allowing Earth’s rotation to
be influential (Winant, 2004; Sanay and Valle-Levinson, 2005).
The eddy viscosity under stratification has an order of
1 10 3 m2/s in Bahı́a Concepción (Ponte, 2009), thus Ekman
depth is about 5.5 m, much smaller than the average depth. In
addition, the relative importance of advection terms, especially
the lateral advection, increases with reduction of vertical mixing
(Cheng and Valle-Levinson, 2009). Therefore, effects of nonlinear
processes, Earth’s rotation, and baroclinic pressure should be
addressed in future theoretical considerations for the stratified
summer season.
The water level of a coastal bay is forced primarily by remote
along-shelf winds (Janzen and Wong, 2002). During the winter
season, the squared coherence between remote winds and water
elevation at the mouth is relatively high at certain frequencies
(Fig. 6). The phases are about 701 at the correlated frequencies.
This implies the coastal water level corresponds to remote winds
with a time lag. During the summer season, the correlation
between remote winds and coastal water level became less
obvious, suggesting that local dynamics could contribute to the
fluctuations of water level in the bay.
4.2. Density-driven circulation
Bahı́a Concepción is a wind-dominated coastal embayment
without freshwater runoff. Due to the arid climate, evaporation
rates are high inside Bahı́a Concepción and cause a saltier water
body (e.g. Mendoza-Salgado et al., 2006; Winant and Gutiérrez de
Velasco, 2003). This could potentially drive an inverse estuarine
circulation. The horizontal density gradient can be derived from
the CTD measurement at the head and the mouth of bay at a
ARTICLE IN PRESS
P. Cheng et al. / Continental Shelf Research 30 (2010) 1241–1249
Winter period
Coherence squared
1
1247
Summer period
a
b
0.8
0.6
0.4
0.2
Phase, degree
0
150
τlocal ~ Δη
100
τremote ~ ηmouth
50
0
−50
−100
c
−150
0
d
0.1
0.2
0.3
0.4
0.5
0.1
Frequency, cpd
0.2
0.3
0.4
0.5
Frequency, cpd
Fig. 6. Coherence squared (a, b), and phase (c, d) between wind stresses and subtidal water levels. Solid lines show the relationship between local wind stress and subtidal
water level difference, while dashed lines show the relationship between remote wind stress and subtidal water level at the mouth of the bay.
depth of 5 m (squares shown in Fig. 1). Values of estimated
longitudinal density difference and density gradient are shown in
Fig. 7a and b. Positive values indicate higher density at the head of
the bay than at the mouth. The density at the head is generally
greater than that at the mouth. The largest difference is
approximately 0.45 kg/m3. The magnitude of the horizontal
density gradient @r/@x ranges from 0 to 14 10–6 kg/m4, i.e.,
two orders of magnitude lower than typical values in estuaries
(Fig. 7b). Nevertheless, the density-driven flow might be
important in this bay.
Officer (1976) provided analytical solutions of density-driven
(ud) and wind-driven (uw) flows, on the basis of the simplified
balance between pressure gradient and stress divergence:
gH3 @r
z
z2
ð3Þ
19 8 2 ,
ud ¼
48Av r0 @x
H
H
and
uw ¼
tH
z
z2
1 þ4 þ 3 2 ,
4Av r0
H
H
ð4Þ
where z represents the vertical coordinate, and Av is a constant
vertical eddy viscosity. Moreover, the relative importance of
density-driven flow over wind-driven flow can be evaluated using
the Wedderburn number (W), which may be defined as the ratio
between the magnitude of ud and uw. Taking z as 0 (i.e., the
magnitude of ud and uw) in Eqs. (3) and (4), W may be represented
as
u
gH2 @r
:
ð5Þ
W¼ d ¼
uw
129t9 @x Using the calculated along-bay density gradient and observed
longitudinal wind stress (Fig. 5c), W is estimated in Fig. 7c. The
absolute values of wind stress and longitudinal density gradient
are used to evaluate the relative importance of density- and
wind-driven flows. On average, the magnitude of the density-
driven circulation is about 75% of the wind-driven flow. At certain
periods, for example, around days 220 and 260 when density
gradient is large and wind stress is very weak, the density-driven
flow could be 1.5 times as important as the wind-driven flow.
Therefore, the longitudinal density gradient acts as an important
mechanism driving circulations in this arid bay.
4.3. Mixing/stratification mechanisms during the summer season
The level of stratification in the water column is a result of the
competition between stirring and stratifying processes. In coastal
environments, the major stirring agents are winds and tides while
the major stratification mechanisms are heating/cooling, baroclinic circulation, and tidal straining. Simpson et al. (1990) defined a
scalar parameter f (units Jm 3), the potential energy anomaly, to
determine the energy required to mix the water column
completely:
1
f¼
H
Z0
ðrrÞgzdz,
ð6Þ
H
where the overbar represents depth average. The time rate of
change of f indicates evolution of stratification. Simpson et al.
(1990) provided an equation for the evolution of stratification,
including the major mechanisms for mixing and stratification:
U3
U3
df
agQ 4
1 g 2 H 4 @r 2
eCd r T dCds ra W þ
¼
:
ð7Þ
dt
H
H
2Cp 3p
320 Av r @x
The first term on the right represents the increase in
stratification due to surface heating at a rate Q, while the second
and the third terms denote stirring by a tidal current of amplitude
UT and a wind of speed 10 m above surface UW, respectively. In
these terms, e (equals 0.0037) and d (equals 0.023) are the
corresponding non-dimensional efficiencies of mixing, and Cd
(equals 0.0025) and Cds (equals 6.4 10–5) are the effective drag
ARTICLE IN PRESS
1248
P. Cheng et al. / Continental Shelf Research 30 (2010) 1241–1249
kg m−3
23.6
a
head
mouth
23.2
22.8
22.4
15
10−6 kgm−4
b
10
5
0
2
c
1.5
1
0.5
0
200
220
240
260
day of year, 2005
Fig. 7. Longitudinal density difference (a), density gradient (b), and Wedderburn number (c).
coefficients for bottom and surface stresses (also non-dimensional). Furthermore, a is the thermal expansion coefficient and ra
is the density of air (1.2 kg/m3). The last term on the right
represents stratification induced by density-driven flow. Contribution from tidal straining is excluded since this study
concentrates on subtidal processes and the measurements
available have been filtered to eliminate tides.
In applying this f equation, winds are treated as a mixing
agent. However, it is recognized that wind-driven flow can create
or destroy stratification depending on its interaction with the
horizontal density gradient (Scully et al., 2005). Wind-driven
flows enhance stratification when the wind blows with the
longitudinal density gradient and reduce stratification when the
wind blows in the opposite direction to the longitudinal density
gradient. In order to include the contribution of wind-driven flows
(wind straining) in the f equation, a parameter is developed
following the framework of Simpson et al. (1990):
@fwc
g @r
¼
@t
H @x
Z0
ðuw uw Þzdz,
ð8Þ
H
where fwc is the potential energy anomaly produced by winddriven flow. Substituting Eq. (4) into Eq. (8), and noting that
uw ¼ 0, results in
dfwc
1 gH2 t @r
¼
:
dt
48 Az r @x
ð9Þ
It is clear that when t and @r/@x have the same signs, the winddriven flow increases stratification (positive df/dt); when t and
@r/@x have opposite signs, the wind-driven flow decreases
stratification (negative df/dt). This is consistent with previous
studies (e.g. Scully et al., 2005; Li et al., 2008).
The data sets for heat flux, winds, temperature, and current in
the summer season are used to evaluate the relative importance
of the five mixing/stratification processes (Fig. 7). All data sets are
filtered with a Lanczos filter at half-power of 34 h in order to
remove tidal influences. The time dependent Av is estimated with
the closure proposed by Pacanowski and Philander (1981):
Av ¼ 0:01ð1 þ5RiÞ2 þ 104 ,
ð10Þ
where Ri is a gradient Richardson number.
Mixing is relatively weak during the summer season (Fig. 8a
and b). The potential energy anomalies induced by wind stirring
and tidal stirring are at least one order of magnitude smaller than
the other mechanisms. The evolution of tidal mixing follows the
tidal range and the weak tidal mixing results from the mircotides.
Surface heating is a major process enhancing stratification
(Fig. 8c). Density-driven circulation contributes to increase
stratification and has a similar magnitude to that of surface
heating (Fig. 8d). Wind-driven flow is another important mixing/
stratification mechanism (Fig. 8e). The potential energy anomaly
induced by wind straining has the same order of magnitude as
surface heating and density-driven flow. In particular, during
strong wind events, wind straining is the dominant mechanism
for producing stratification. Also, as a mixing agent, wind
straining acts more effectively than tidal and wind stirring. This
suggests that wind straining can play a dominant role in mixing/
stratification for coastal basins.
ARTICLE IN PRESS
P. Cheng et al. / Continental Shelf Research 30 (2010) 1241–1249
tion of this manuscript while a Gledden Fellow at the University of
Western Australia. We thank the two anonymous reviewers for their
insightful comments which helped improve this study.
0
a
−0.2
−0.4
0
b
References
10−5 J m−3 s −1
−0.1
−0.2
6
c
4
2
15
10
d
5
0
10
1249
e
0
−10
200
220
240
260
day of year, 2005
Fig. 8. Mixing/stratification mechanisms. (a) Wind stirring; (b) tidal stirring; (c)
surface heating; (d) density-driven circulation; and (e) wind straining.
5. Conclusions
Observations in Bahı́a Concepción demonstrate, for the first
time in a bay of the Gulf of California, the role of wind-driven
upwelling in stratifying a bay where freshwater input is
negligible. Negative buoyancy supplied by coastal upwelling can
contribute to seasonal stratification in coastal embayments, in
addition to the expected surface heat transfer from the atmosphere. Onset of summer upwelling winds triggered a cold water
intrusion that dropped the water temperature in the entire water
column. Persistent upwelling-favorable winds maintained the
supply of cold water into the bay, which increased the
temperature stratification as the surface water warmed up.
Therefore, the seasonal evolution of stratification is induced in
part by increased atmospheric heat input and in part by cold
water intrusions drawn by upwelling. Also, the wind-driven
upwelling maintains a longitudinal temperature difference which
produces a baroclinic exchange flow. Spatial patterns of temperature revealed by EOF analysis indicate section-wide warming/
cooling as depicted by the first two modes. The first two principal
components contain signature of seasonal heat transfer from the
atmosphere and of cold water intrusions. Strong evaporation
leads to a density difference between the bay and adjacent coastal
water resulting in an inverse gravitational circulation that has the
same order of magnitude as the wind-driven circulation. Also, it is
found that wind straining could be the dominant mixing/
stratification mechanism in wind-dominated coastal basins.
Acknowledgments
This study is funded by the US National Science Foundation
through Project OCE-0726697. AVL also acknowledges the comple-
Badan-Dangon, A., Dorman, C.D., Merrifield, M.A., Winant, C.D., 1991. The lower
atmosphere over the Gulf of California. J. Geophys. Res. 96 (C9), 16,877–16,896.
Canar, N., Margarita, A., Garcia, P., Ricardo, J., 2008. Grazing by microzooplankton
in Concepcion Bay, Gulf of California, Mexico. Hidrobiologica 18 (1), 141–151.
Cheng, P., Valle-Levinson, A., 2009. Influence of lateral advection on residual
currents in microtidal estuaries. J. Phys. Oceanogr. 39 (12), 3177–3190.
Garvine, R.W., 1985. A simple model of estuarine subtidal fluctuations forced by
local and remote wind stress. J. Geophys. Res. 90 (C6), 11945–11948.
Gill, A.E., 1982. Atmosphere–Ocean Dynamics. Academic Press, San Diego 662pp.
Gilcoto, M., Pardo, P.C., Álvarez-Salgado, X.A., Pérez, F.F., 2007. Exchange fluxes
between the Rı́a de Vigo and the shelf: a bidirectional flow forced by remote
wind. J. Geophys. Res. 112, C06001, doi:10.1029/2005JC003140.
Janzen, C.D., Wong, K.-C., 2002. Wind-forced dynamics at the estuary–shelf
interface of a large coastal plain estuary. J. Geophys. Res. 107 (C10), 3138,
doi:10.1029/2001JC000959.
Johnson, M., Ledesma-Vásquez, J., 2001. Pliocene-Pleistocene rocky shorelines
trace coastal development of Bahı́a Concepción, gulf coast of Baja California
Sur (Mexico). Palaeogeogr., Palaeoclimatol., Palaeoecol. 166, 65–88.
Kara, A.B., Wallcraft, A.J., Hurlburt, H.E., 2005. A new solar radiation penetration
scheme for use in ocean mixed layer studies: an application to the Black Sea
using a fine-resolution Hybrid Coordinate Ocean Model (HYCOM). J. Phys.
Oceanogr. 35, 13–32.
Large, W.G., Pond, S., 1981. Open ocean momentum flux measurements in
moderate to strong winds. J. Phys. Oceanogr. 11, 324–336.
Li, C., Walker, N., Hou, A., Georgiou, I., Roberts, H., Laws, E., McCorquodale, J.A.,
Weeks, E., Li, X., Crochet, J., 2008. Circular plumes in Lake Pontchartrain
estuary under wind straining. Estuar. Coast. Shelf Sci. 80 (1), 161–172.
Lopez-Cortes, D., Garate-Lizarraga, J.B.-G.I., Hernandez-Sandoval, F., MurilloMurillo, I., 2003. Phytoplankton biomasses and hydrographic conditions
during El Niño 1997–1998 in Bahı́a Concepción, Gulf of California, Mexico.
Geofı́s. Int. 42, 495–504.
Mendoza-Salgado, R., Lechuga-Deveze, C., Ortega-Rubio, A., 2006. Influence of
rainfall on a subtropical arid zone coastal system. J. Arid Environ. 66, 247–256.
Millero, F.J., Perron, G., Desnoyers, J.F., 1973. Heat capacity of seawater solutions
from 5 to 35 1C and 0.05 to 22% chlorinity. J. Geophys. Res. 78 (21),
4499–4506.
Officer, C.B., 1976. Physical Oceanography of Estuaries (and Associated Coastal
Waters). John Wiley, New York 465 pp.
Pacanowski, R.C., Philander, S.G.H., 1981. Parameterization of vertical mixing in
numerical models of tropical oceans. J. Phys. Oceanogr. 11, 1443–1451.
Palomares-Garcia, R., Bustillos-Guzman, J.J., Lopez-Cortes, D., 2006. Pigmentspecific rates of phytoplankton growth and microzooplankton grazing in a
subtropical lagoon. J. Plankton Res. 28 (12), 1217–1232.
Ponte A.L.S., 2009. Wind and tidal response of a semi-enclosed bay, Bahı́a
Concepción, Baja California. University of California, San Diego, Ph.D.
Dissertation, 140pp.
Simpson, J.H., Brown, J., Matthews, J., Allen, G., 1990. Tidal straining, density
currents, and stirring in the control of Estuarine stratification. Estuaries 13 (2),
125–132.
Sanay, R., Valle-Levinson, A., 2005. Wind-induced circulation in semienclosed
homogeneous, rotating basins. J. Phys. Oceanogr. 35, 2520–2531.
Scully, A.E., Friderichs, C., Brubaker, J., 2005. Control of estuarine stratification and
mixing by wind-induced straining of the estuarine density field. Estuaries 28
(3), 321–326.
Wang, D.-P., 1979. Wind-driven circulation in the Chesapeake Bay, winter 1975.
J. Phys. Oceanogr. 9, 564–572.
Winant, C.D., 2004. Three dimensional wind-driven flow in an elongated, rotating
basin. J. Phys. Oceanogr. 34, 462–467.
Winant, C.D., Gutiérrez de Velasco, G., 2003. Tidal dynamics and residual
circulation in a well mixed inverse estuary. J. Phys. Oceanogr. 33, 1365–1379.
Wallcraft, A.J., Kara, A.B., Hurlburt, H.E., Chassignet, E.P., Halliwell, G.H., 2008.
Value of bulk heat flux parameterizations for ocean SST prediction. J. Mar. Sys.
74, 241–258, doi:10.1016/j.jmarsys.2008.01.009.
Wong, K.-C., Valle-Levinson, A., 2002. On the relative importance of the remote and
local wind effects on the subtidal exchange at the entrance to the Chesapeake
Bay. J. Mar. Res. 60, 477–498.