Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Numerical Simulation of a Synoptic Event in the1
Southern California Bight2
Changming Dong∗, James C. McWilliams, Alex Hall,
Department of Atmospheric and Oceanic Sciences andInstitute of Geophysics and Planetary Physics,
University of California, Los Angeles, CA 90095
Mimi Hughes
Cooperative Institute for Research in the Environmental SciencesNOAA/ESRL, Boulder, CO
3
August 4, 20104
∗Corresponding author: Dr. Changming Dong, IGPP, University of California, Los Angeles, 405 Hilgard Avenue,Los Angeles, CA, 90095-1567, email: [email protected]
1
ABSTRACT5
In the middle of March 2002 a synoptic upwelling event occurred in the Southern California6
Bight; it was marked by a precipitous cooling of at least 4◦C within 10 to 20 km of the coast.7
By the end of the month the pre-event temperatures had slowlyrecovered. The Regional Oceanic8
Model System (ROMS) is used to simulate the event with an atmospheric down-scaling reanalysis9
for surface wind and buoyancy flux forcing. Lateral boundaryconditions of temperature, salin-10
ity, velocity, and sea level are taken from a global oceanic product. Barotropic tidal fields from a11
global barotropic model are imposed along the open boundaries. The simulation reproduces well12
the upwelling process compared with observed data. The sensitivity of the simulation is exam-13
ined to wind resolution, heat flux, and tidal forcing. The oceanic response to the different wind14
resolutions converges at the level of the 6-km resolution, which is the finest scale present in the15
terrain elevation data set used in the atmospheric down-scaling. The combination of an analytical16
diurnal cycle in the solar radiation and the empirical coupling with the instantaneous ROMS sur-17
face temperature (SST) produces a similar oceanic responseto the down-scaled heat flux. Tidal18
effects are significant in the upwelling evolution due to theincrease in wind energy input through19
a quasi-resonant alignment of the wind and surface current,probably by chance.20
2
1 Introduction21
Dong et al., (2009) applies a high-resolution (dx = 1 km) Regional Oceanic Modeling System22
(ROMS) in the Southern California Bight (SCB) on mean, interannual, seasonal, and intra-seasonal23
(eddy) time scales during the period 1997-2003 (Fig. 1; alsonote the measurement sites). Ex-24
tensive assessment of the model results with observationaldata demonstrates general consistency25
between the two. In this paper, using the same model configuration, we extend the model perfor-26
mance assessment to the simulation of a strong upwelling event in the SCB. Our approach is purely27
forward-in-time modeling, without data assimilation but with some degree of optimization of the28
external forcing that includes surface momentum, heat flux,freshwater flux, and lateral boundary29
conditions including tides.30
Coastal upwelling frequently occurs along the U.S. West Coast due to persistent or recurrent31
equatorial wind stress (Strub and James (2000), especiallytheir Fig. 1). Strong upwelling is less32
common in the SCB than along the central California coast north of Point Conception. Generally33
warmer water is found to the south in the SCB, partly due to advection by the Southern California34
Counter Current and also due to sheltering from the upwelling-favorable northerly winds (Caldeira35
and Marchesiello 2002). Mostly during spring the winds episodically are favorable to upwelling36
in the SCB (Winant and Dorman, 1997). Evidence of such upwelling events is visible in the time37
series of daily-mean surface temperature (SST) anomalies at National Data Buoy Center (NDBC)38
buoy 46025 (Fig. 3), located about 50 km offshore from the coast. Nearly every spring, once or39
twice the SST dips by 2-4◦C for a week or two at a time.40
A strong upwelling episode in the SCB occurred in the middle of March 2002. This is confirmed41
by SST data from the satellite Pathfinder (Caseyet al., , 2009). Comparing the SST images on42
March 13 with those on March 18 (Fig. 2), we see a sharp drop in SST from Point Conception all43
the way south to San Diego. This is our simulation target withROMS. In Sec. 2 we describe the44
atmospheric and oceanic models and the experimental design. In Sec. 3 the baseline experiment45
3
results are presented, and the oceanic dynamics in the upwelling event are discussed. In Secs. 4-546
the sensitivities of the numerical solution to the different external forcings are investigated and47
assessed through a statistical skill estimator. Section 6 is the summary.48
2 Experimental Design49
ROMS solves the rotating primitive equations (Shchepetkinand McWilliams, 2005) and uses a50
generalized sigma-coordinate system in the vertical direction and curvilinear grid in the horizontal51
plane. It is a split-explicit, free-surface oceanic model,where short time steps are used to advance52
the surface elevation and barotropic momentum equations, with a larger time step used for tem-53
perature, salinity, and baroclinic momentum. A third-order, upstream-biased advection operator54
allows the generation of steep gradients in the solution, enhancing the effective resolution of the55
solution for a given grid size when the explicit viscosity issmall. The numerical diffusion implicit56
in the third-order, upstream-biased operator allows the explicit horizontal viscosity to be set to zero57
without excessive computational noise or instability. Thevertical viscosity is parametrized using58
a K-profile parametrization (KPP) scheme (Largeet al., , 1994, Blaaset al., , 2007). The no-slip59
lateral boundary condition is also imposed through the momentum advection operator and yields60
an implicit lateral stress (Dong and McWilliams, 2007).61
The ROMS model domain is in Fig. 1. It has a grid spacing ofdx = 1 km horizontally and 4062
levels vertically. The grid resolves all eight islands in the SCB. Mixed boundary conditions are63
used along the open boundaries,i.e., the Orlanski radiation condition in the tangential direction64
and the Flather condition with adaptive restoration of material properties to imposed data under65
inflow conditions (Marchesielloet al., , 2001). The restoring data for the lateral open-boundary66
conditions and the initial conditions are extracted from a 1996-2003 ROMS solution in a larger U.S.67
West Coast domain that has two-level online nested grids (dx = 20 km and 6.7 km) forced with68
Simple Ocean Data Assimilation (SODA; Cartonet al., 2000) monthly data along open boundaries69
4
for the 20 km grid. The same model configuration has been applied to an island wake study in the70
SCB (Dong and McWilliams, 2007) and a long-term reanalysis (Dong et al., , 2009).71
The surface flux fields (momentum, heat, and freshwater) are generated with the regional atmo-72
spheric model MM5 (Grellet al., , 1995). Four nested grids with horizontal resolutionsdx = 54,73
18, 6, and 2 km are implemented with MM5. Two-way communication takes place with the parent74
nest at the lateral boundaries of the three innermost domains. The coarsestdx = 54 km domain75
covers the western U. S. and an equivalent-sized portion of the Pacific Ocean, while the finest76
dx = 2 km domain covers the SCB. Each nest, therefore, includes theSCB and the ROMS domain77
shown in Fig. 1. The outermost 54 km domain is forced at its lateral and surface boundaries with78
data from NCEP Eta model reanalysis for the entire month of March 2002. The lateral boundary79
conditions are available every 3 hours from this archive, and we interpolated them in time. SST is80
updated every 3 days with satellite analysis. This simulation can be thought of as a reconstruction81
of regional atmospheric conditions during March 2002, consistent with our best estimate of the82
large-scale conditions, the resolved topography, and MM5 model physics. Conil and Hall (2006)83
analyze an atmospheric configuration identical to the present one except it does not include a 2 km84
domain. They verify MM5 winds against observations for a multi-year reanalysis. Comparing the85
6 km simulated daily mean wind anomalies with the daily mean wind anomalies observed at 1686
stations over land and two buoys over the ocean (NDBC 46025 and 46053), they show they are87
highly coherent throughout the 6 km domain. For example, correlations between observed anoma-88
lies in wind direction and those simulated at the nearest model grid points are greater than 0.5 and89
are generally around 0.7. Wind speed correlation values areabove 0.4 at all 18 stations. At 1090
locations they are above 0.6, with the highest correlation reaching almost 0.8. For the two ocean91
buoys, the direction correlation values are about 0.7, and the speed correlation values are about92
0.7 and 0.5 for buoys 46025 and 46053, respectively. We perform a further verification of MM593
winds over the ocean during the March 2002 period by projecting the winds at the buoys onto their94
5
principal axis of variability and computing correlations between simulated and observed winds95
for all four MM5 resolutions (Table 1). While the coarse-resolution winds are not very realistic,96
correlation values reach 0.5-0.7 at 46025 and 46053 for the 6km and 2 km winds. However, the97
correlation at the SMB buoy is generally lower even at the highest resolution, perhaps due to the98
rather complicated wind sheltering in Santa Monica Bay. Finally, we also compared a snapshot of99
25 km resolution QuikSCAT winds with the 2 km resolution MM5 winds at a moment of intense100
wind forcing during March 2002 (Fig. 4). The magnitudes of the winds agree well, though the101
model winds tend to be rotated a few degrees clockwise of QuikSCAT. We see good agreement for102
offshore winds but larger discrepancies for nearshore winds because the QuikSCAT wind is not103
reliable within 50 km of land. In summary, MM5 does a reasonable job capturing the magnitude,104
direction, and variability of the winds, particularly at its 2 km and 6 km resolutions.105
Tides are not included in the long-term (1996-2003) ROMS reanalysis (Donget al., , 2009), but106
in this simulation tidal forcing is applied along the open boundaries. The tidal amplitudes and107
phases in both sea surface heights and barotropic velocities of eight tidal constituents are obtained108
from a global inverse barotropic tidal model (TPX0.6) (Egbert and Erofeeva, 2002) with a horizon-109
tal resolution of0.25o. The eight tidal constituents areM2, K1, O1, S2, N2, P1, K2, andQ1 ordered110
by their amplitude in the region. The barotropic transport from the TPX0.6 solution is adjusted111
based on the ROMS bathymetry because the bathymetry fields from the ROMS and TPX0.6 are112
different. Nodal correction (Foreman, 1977) is applied to the sea surface height and barotropic113
transport of the TPX0.6. Our approach follows one previously used for Monterey Bay (Wanget114
al., , 2009).115
Numerical experiments are conducted to test sensitivitiesof the model solution to various exter-116
nal forcings. MM5 wind data of different resolutions (54 km,18 km, 6 km and 2 km) are imposed117
to test the sensitivity of the oceanic response. The tidal forcing is turned on and off to examine118
the tidal effect on the sub-tidal circulation. Because it isa challenge for a regional atmospheric119
6
model to represent boundary-layer processes generating stratocumulus clouds and surface heat flux120
(Brethertonet al., , 2004; McCaaet al., , 2004), the MM5 heat fluxes are replaced by Climatolog-121
ical Oceanic and Atmospheric Data Set (COADS) heat flux and freshwater flux (DaSilvaet al., ,122
1994) in some experiments. A model-internal-step modification of the heat flux based on the MM5123
air temperature and the ROMS simulated sea surface temperature (Marchesielloet al., , 2003) is124
adopted as an alternative forcing. Table 2 lists the series of numerical experiments conducted in125
the study. The case T12 is considered as a baseline case. It isforced by MM5 momentum, heat126
flux, and freshwater flux and eight tidal constituents.127
3 Baseline Simulation Case128
The model output consists of hourly samples for the period March 1-31, 2002. The upwelling129
event is during March 14 to 24, lasting about 11 days. Figure 5plots the surface currents before130
and during the event. The current develops a strong alongshore equatorward flow during the up-131
welling period, which is significantly different from both the current before the event and the mean132
circulation pattern simulated in Donget al., (2009). An alongshore surface current is expected by133
geostrophic balance when the interior isopycnals are uplifted adjacent to the coast by upwelling.134
Sea surface elevation data at six SCB tide gauges are used (Fig. 1) to assess the modeled tides.135
The data are from the Operational Oceanographic Products and Service, NOAA (http://tidesandcurrents.noaa.gov).136
A tidal harmonic analysis is applied to extract tidal signals of the eight tidal constituents (i.e., the137
T−TIDE package from Pawlowiczet al., (2002) adapted from Foreman (1977, 1978)). The sea-138
surface height (SSH) amplitudes and phases for each of the eight tidal constituents are averaged139
over the six tide gauge locations and plotted in Fig. 6. The comparison shows that ROMS simu-140
lates the barotropic tides in the SCB rather well.141
The evolution of the upwelling event is evident in hydrographic measurements during March142
2002. There are two NDBC buoys in the SCB (46025 and 46053) andone mooring at the Santa143
7
Monica Bay (SMB) available during the upwelling period. A long-term SST measurement at the144
Scripps Institute of Oceanography Pier is also used (see locations in Fig. 1). All of these stations145
measure SST, while the SMB mooring also measures vertical profiles of temperature and salinity146
in the upper 100 m. Figure 7 shows the observed time series of the SST at the four locations (blue147
lines). At the SMB mooring, a sharp drop in SST begins about March 13, 2002, the same date148
as the onset of the cooling event recorded at the NDBC buoy 46025. From then until March 18,149
SST decreases at the SMB mooring by about 4◦C, and then it slowly recovers until the end of the150
month. The event is also evident at the NDBC buoy 46053 with a drop of about 3◦C from March151
13 to 18 (Fig. 7). The temperature decreases by about2oC at Scripps Pier.152
The modeled SST is compared at these four measurement locations in Fig. 7. The correlations153
between the observational and modeling data are 0.95, 0.50,0.87, and 0.08 at the SMB mooring,154
NDBC Buoys 46053 and 46025, and Scripps Pier, respectively.The root-mean-square-errors are155
0.24oC, 0.42oC, 0.06oC, and 1.11oC. The model captures the sharp drop in the SST well, both in156
timing and magnitude at all of the four stations spanning cross the SCB coast. The correlations157
between the model solution and the observations are above0.50 except for the Scripps Pier, which158
is located 100 m from the coast, a far smaller distance than that can be resolved by the 1 km model.159
At the NDBC 46053 buoy inside the Santa Barbara Channel, the correlation is lower than those at160
the SMB mooring and NDBC Buoy 46025 located in the middle coast. The SST time series at the161
buoy in the Santa Barbara Channel (upper-right of Fig. 7) show that the upwelling onset period162
is better simulated than during the recovering period. Caution is needed in interpreting model-163
data comparisons. When eddies arise by flow instability, we expect some disagreement between164
the forward-model results and observations due to the limited predictability of intrinsic variability.165
The underlying premise of our comparisons here is that the wind forcing event is sufficiently strong166
to make unknown eddy effects relatively small.167
A comparison of vertical profiles of temperature and salinity at the SMB mooring is shown in168
8
Fig. 8. The observed data (top panels) show that from March 13to 18 colder and saltier water169
below the stratification layer is pumped up to the sea surface, destroying the thermocline and170
significantly weakening the halocline. After this interval, the ocean relaxed back to a stratified171
state. ROMS reproduces fairly well the evolution of vertical structure in temperature and salinity172
except that the water is slightly fresher than the observation at all depths and times, indicative of173
model bias from boundary-data or dynamical error.174
Coastal upwelling can be driven by either equatorward alongshore wind stressτ adjacent to the175
boundary or by a positive wind stress curl nearby. Figure 9 shows both quantities averaged over176
all the cross-shore sections shown in Fig. 1. The alongshoredirection is chosen as the one along177
the yellow line, and the wind curl is calculated from the total wind stress and mapped onto the178
red lines in Fig. 1. On March 13, the alongshore wind stress increases suddenly and lasts for179
about five days (March 13-18), and the wind stress curl picks up at the same time although it is180
not as simply persistent. The high-stress time interval is consistent with the temporal evolution of181
temperature and salinity, which indicates that this event is driven mostly by an anomaly in the wind182
stress. The equatorward alongshore wind stress drives an offshore transport in the upper layer (top183
50 m). To compensate, sea water in the lower layer flows shoreward (Fig. 10) and is upwelled to184
the surface close to the coast. These two transports are nearly balanced, at least on time scales less185
than a week, which implies that the alongshore transport divergence is not a major contributor the186
upwelled water. There are four major peaks in offshore surface-layer transport, on March 14, 16,187
18, and 24 that correspond to the four major peaks in wind stress (Fig. 9). The short dip between188
the first two peaks also matches the local minimum in the wind stress time series.189
A cross-shore section of temperature and cross-shore current u (Fig. 11) shows a temperature190
front about 40-50 km offshore during the peak upwelling period. In addition to the strongu < 0191
Ekman transport, there is au > 0 feature centered around the front. The latter is consistentwith192
the secondary circulation in frontogenesis caused by mesoscale strain (Capetet al., , 2008), and it193
9
suggests a local augmentation of the upwelling is occurringbeyond the direct effect of the wind194
stress.195
4 Sensitivities196
The baseline simulation in Sec. 3 is successful in matching the observed upwelling event in re-197
sponse to the occurrence of strong alongshore winds. In thissection the sensitivities of the event198
to various external forcings are examined, including MM5 model resolution for the wind forcing,199
several alternative heat flux fields, and the influence of tidal forcing (Table 2).200
4.1 MM5 Wind Resolution201
The MM5 configuration has four levels of nesting (Sec. 2) withdx = 56 km, 18 km, 6 km, and 2202
km. We can use the different nested fields to ask what are the consequences of finer wind scales to203
the oceanic response for this particular synoptic event.204
Alongshore wind stress and wind curl profiles averaged alongshore in the SCB and over the205
upwelling period of March 12-18 (Fig. 12) show that the negative stress and positive curl anomalies206
intensity dramatically fromdx = 54 km to 2 km, but both quantities are similar for the 6 km and 2207
km solutions. This suggests that the 6 km resolution is probably sufficient within the SCB during208
the upwelling event. This conclusion is also valid for the oceanic response in the offshore transport209
in the upper layer.210
What is the origin of the increase in alongshore wind stress with resolution, and why does it211
stop increasing once the 6 km resolution is reached? We believe it is primarily a reflection of the212
important influence of coastal terrain in Southern California. To show this we calculate orographic213
variance after applying four different spatial smoothers with filter scales corresponding to the res-214
olution of the four nesting levels in the MM5 reanalysis. We apply a second order Chebyshev filter215
with 0.1 dB of ripple and passband edge frequency that is twice the grid spacing of each domain to216
10
the 3 arc-second United States Geological Survey national elevation dataset. Note that the effective217
cutoff spatial scale of the topography in MM5 is roughly twice the model grid resolution size;e.g.,218
the topography in the 54 km solution does not include any terrain with wavelengths less than 108219
km. The red line in Fig. 13 shows that orographic variance does increase with model resolution, but220
it approaches saturation in the step beyond the 6 km grid to the 2 km grid. Therefore, orographic221
effects on atmospheric flow, such as blocking and channeling, are not significantly stronger as the222
resolution increases belowdx = 6 km.223
4.2 Heat Flux224
Surface heat and water fluxes are often less reliable as products of meteorological downscaling225
reanalysis than is wind stress. Thus, it is important to determine how sensitive the oceanic simu-226
lation is to uncertainty in air-sea buoyancy fluxes. As an alternative we first replace the MM5 heat227
and freshwater fluxes with those contained in 1◦ by 1◦ COADS (DaSilvaet al., , 1994) and impose228
them on the ocean model. Because the COADS data are monthly averaged and have no diurnal229
cycle, a diurnal cycle of solar radiation based on an analytical model is added to the COADS fluxes230
to test the importance of the diurnal effects (Marchesielloet al., , 2003). To represent synoptic heat231
flux changes, a heat flux relaxation adjustment based on the difference between MM5 surface air232
temperature and ROMS SST is added (Marchesielloet al., , 2003; Donget al., , 2009). Numerical233
tests with all combinations of these modifications are made in experiments T14, T16, and T16n234
(Fig. 14). With both the diurnal cycle and the heat flux relaxation adjustment (case T16), the mod-235
eled SST at the SMB mooring is significantly improved compared to the simulations where only236
one modification is made. Comparing T16 with the baseline case T12 (Fig. 7, upper left panel),237
we see considerable agreement in the SST time series, which gives confidence in both alternatives.238
These results imply that synoptic SST coupling and the diurnal solar cycle both have important239
influences on the upwelling event.240
11
4.3 Tides241
A comparison of cases with (T12) and without (T11) tidal forcing (Figure 15) shows that the SST242
at the SMB buoy cools similarly during the onset of the event,but warms more slowly with tides243
in the relaxation phase afterward. The slower warming is more consistent with measurements.244
The slower-warming signal extends along the intersecting cross-shore section and into the interior245
(Figs. 16-17), as well as alongshore (not shown). A partial explanation is the stronger isopycnal246
uplift near the coast with the stronger geostrophically balanced southward surface current (cf.,247
Fig. 5). This means there is more cold water near the surface in the coastal zone with tides than248
without, hence more of a heat-content anomaly to warm once the upwelling wind forcing abates.249
The vertical structure ofT (x, z) near the coast does not support an increased vertical mixingas the250
explanation (unlike in tidal mixing fronts; Hill and Simpson, 1989; Donget al., , 2004).251
Figure 18 shows the map of wind work with and without tides. Inthe tidal case the work is nearly252
twice as large in the nearshore region all along the coast. Therefore, an alternative explanation is253
that the kinetic energy input through time-averaged wind work τ · usur is greater in the case with254
tides, hence a stronger alongshore current and more coastalisopycnal uplift, hence a larger cold255
heat content anomaly, hence a slower warming during relaxation. When more wind energy is input256
to the ocean, the surface alongshore current is enhanced, which results in more offshore transport.257
To balance the surface offshore transport, onshore transport in the lower layer also increases and258
then colder water is moved to the sea surface. Time series of the wind and current in the coastal259
zone (Fig. 19) show that the enhanced wind work occurs in oscillatory pulses in the surface current260
with an approximately inertial time scale during the periodof high winds, and evidently the tide en-261
hances the partly resonant alignment with the wind. Becausethe ocean currents have no influence262
on the wind in our model configuration, we must presume that the tidally-enhanced resonance is a263
chance occurrence in this particular synoptic event. Inertial oscillations are common with sudden264
wind changes. With the combination of tides and wind, the inertial resonance is sometimes further265
12
enhanced by the tides (Stockwellet al., , 2004).266
5 Statistical Skill Assessment267
To quantify the model performance of the numerical sensitivity experiments more objectively, we
define the mean square error (MSE) between observationso and model solutionsm:
MSEi(k) =1
Ni
Ni∑
j=1
(mi(j, k) − oi(j))2 , (1)
where the subscripti is the variable index;j is the index of observational data points;Ni is the
number of the observational data available for the variablei; k is the experiment index; andmi
andoi are the model and observational data for the variablei. Based on the MSE, a skill for each
experiment is calculated with respect to a reference experiment (Murphy, 1992), here chosen as
the baseline case (T12). The skill score (SS) is defined as
SSi(k) = 1 −
MSEi(k)
MSEr
, (2)
where the subscript “r” refers to the reference experiment.SS = 1 is a perfect score;SS = 0 is268
no improvement over the reference experiment;SS < 0 indicates a poorer performance than the269
reference; and a value ofSS between 0 and 1 indicates the experiment has a better performance270
than the reference.271
With multiple observational variables, an averagedSS is calculated over theSSi values for all
variables (Okeet al., , 2002; Wilkin, 2006; Liuet al., , 2009). Observational data are usually not
equally available for all the variables, so a weighted-average SS based on the number of available
observations is useful. Therefore, a weighted SS is also defined by
WSS(k) = 1 −
WMSE(k)
WMSEr
, (3)
whereWMSE(k) is the weighted MSE for experimentk,
WMSE(k) =
∑Ii=1
(MSEi(k) Ni)∑I
i=1(Ni)
, (4)
13
andI is the total number of variables used in the evaluation. The numerical values ofWSS have272
the same interpretation asSS.273
Table 3 listsMSE, SS, andWSS values for the numerical experiments. The best performance274
is case T12n. T12n is the same as the baseline case T12 except it has no heat and freshwater flux275
correction (Table 2). Both T12 and T12n use the MM5 heat flux. It shows the heat and freshwater276
flux correction does not improve the ROMS performance when MM5 heat flux is used. However,277
theWSS comparisons between T15 (WSS = −0.383) and T15n (WSS = −0.303), and between278
T16 (WSS = −0.081) and T16n (WSS = −0.075), which use the COADS heat and freshwater279
flux, show the correction improves their model performance.It is also noted that theWSS of T16280
is very close to that of T12. It demonstrates again that the COADS surface scalar flux can be used281
to replace MM5 fluxes as long as the relaxation and diurnal heat-flux corrections are used.282
Tidal effects on the model results are also clearly shown in theWSS. Comparison of the four283
pairs of experiments with and without tides (T11 and T12; T13and T14; T15 and T16; and T15n284
and T16n) all display betterWSS when tides are included, consistent with what we have seen in285
Sec. 4.3.286
Effects of wind resolution on model performance can be seen from the comparison among the287
values ofWSS from T12, T2, T3, and T4 with the wind resolution 2 km, 6 km, 18 km, and 54288
km. It is interesting to see the best performance is the experiment with the resolution as 6 km and289
not the one with the finest resolution, although the difference is too small to take seriously.290
6 Summary291
A synoptic upwelling event occurred in March, 2002 within the SCB. It is simulated with the292
ROMS without data assimilation. The simulation is successful in quantitatively matching the ob-293
served sharp drop in SST by4oC in three days after onset and the relaxation back to normal294
conditions over the next ten days at the edge of the continental shelf. The simulated structure of295
14
the alongshore and cross-shore circulations and pycnocline structure conform to the qualitative296
expectations of an upwelling event forced by alongshore equatorward wind.297
The necessary ingredients for successful simulation include a high-resolution reanalysis wind298
field, synoptic and diurnal heat flux anomalies, and tidal forcing. These influences are confirmed299
through an objective model skill assessment. Apart from establishing a realistic state for the re-300
gional circulation, the boundary data and initial conditions on the eddy scale are less important in301
simulating this event;i.e., the event forcing is sufficiently strong to overcome these influences.302
Perhaps the biggest surprise is the role of the tide in strengthening the oceanic circulation re-303
sponse and prolonging the relaxation period. The tide acts to increase the wind work by a quasi-304
resonant alignment in the wind stress and surface currents during the period of strong winds. The305
model configuration includes the winds and tides as independent external forcing fields arising306
from independent physical processes, therefore we do not expect other upwelling events to show307
the same enhancement of wind work seen in this event.308
Acknowledgments. CD and JCM appreciate support from the Office of Naval Research (grants309
N00014-02-1-0236 and N00014-05-10293). MH is supported byan NRC postdoctoral research310
associateship, and AH is supported by NSF ATM-0135136. We thank Xiaochun (Adam) Wang for311
help with the tidal evaluation.312
15
References313
Blaas, M., C. Dong, P. Marchesiello, J. C. McWilliams, and K.Stolzenbach, 2007: Sediment314
transport modeling on Southern California Shelves: A ROMS case study.Cont. Shelf Res.315
27, 832-853.316
Bretherton, C.S., J.R. McCaa, and H. Grenier, 2004: A new parameterization for shallow cumulus317
convection and its application to marine subtropical cloud-topped boundary layers. Part I:318
Description and 1-D results.Mon. Wea. Rev. 132, 864-882.319
Caldeira, R., and P. Marchesiello, 2002: Ocean Response to wind sheltering in the Southern Cali-320
fornia Bight.Geophys. Res. Lett. 29, 1-4.321
Capet, X., J.C. McWilliams, M.J. Molemaker, and A. Shchepetkin, 2008: Mesoscale to subme-322
soscale transition in the California Current System. II: Frontal processes.J. Phys. Ocean 38,323
44-64.324
Carton, J.A., Chepurin, G., Cao, X., Giese, B., 2000: A Simple Ocean Data Assimilation analysis325
of the global upper ocean 1950¡96¿95. I: Methodology and II:Results.J. Phys. Ocean. 30,326
294-309 and 311-326.327
Casey, K.S., T.B. Brandon, P. Cornillon, and R. Evans, 2009:The Past, Present, and Future of the328
AVHRR Pathfinder SST Program, inOceanography From Space, Again: Revisited, eds. V.329
Barale, J.F.R. Gower, and L. Alberotanza, Springer, in press.330
Conil, S., and A. Hall 2006: Local regimes of atmospheric variability: A case study of Southern331
California.J. Climate 19, 4308-4432.332
DaSilva, A., C. Young, and S. Levitus, 1994:Atlas of Surface Marine Data 1994, Vols. 1-5.333
NOAA Atlas NESDIS 6-10, U.S. Government Printing Office.334
Dong, C., E. Idica, and J. McWilliams, 2009: Circulation andmultiple-scale variability in the335
Southern California Bight.Prog. Ocean. 82, 168-190.336
16
Dong, C., and J.C. McWilliams, 2007: A numerical study of island wakes in Southern California337
Bight. Cont. Shelf Res. 27, 1233-1248.338
Dong, C., J.C. McWilliams, and A. Shchepetkin, 2006, Islandwakes in deep water.J. Phys. Ocean.339
37, 962-981.340
Dong, C., H-W Ou, D. Chen, and M. Visbeck, 2004: Tidally induced mean cross-frontal circula-341
tion: Analytical study.J. Phys. Ocean. 34, 293-305.342
Egbert, G.D., and S.Y. Erofeeva, 2002: Efficient inverse modeling of barotropic ocean tides.J.343
Atmos. Ocean. Tech. 19, 183-204.344
Foreman, M. G. G., 1977:Manual for Tidal Height Analysis and Prediction. Pacific Marine345
Science Report 77-10, Institute of Ocean Sciences, Patricia Bay, Sidney, BC, 97pp.346
Foreman, M. G. G., 1978:Manual for Tidal Height Analysis and Prediction. Pacific Marine347
Science Report 78-6, Institute of Ocean Sciences, PatriciaBay, Sidney, BC, 57pp.348
Grell, G., J. Dudhia, and D. Stauffer, 1995: A description ofthe fifth-generation Penn State/NCAR349
Mesoscale Model (MM5). Technical Report, NCAR Tech. Note NCAR/TN-398+STR350
Hill, A.E. and J.H. Simpson, 1989: On the interaction of thermal and haline fronts: the Islay front351
revisited.Estuarine Coast. Shelf Sci. 28, 495-505.352
Large, W.G., J.C. McWilliams, and S.C. Doney, 1994: Oceanicvertical mixing: A review and a353
model with nonlocal boundary-layer parametrization.Rev. Geophys. 32, 363-403.354
Liu, Y., P. MacCready, B. M. Hickey, E. P. Dever, P. M. Kosro, and N. S. Banas, 2009: Evalua-355
tion of a coastal ocean circulation model for the Columbia River Plume in summer 2004.J.356
Geophys. Res. 114, C00B04.357
Marchesiello, P., J.C. McWilliams, and A. Shchepetkin, 2001: Open boundary conditions for long-358
term integration of regional ocean models.Ocean Modelling 3, 1-20.359
Marchesiello, P., J. McWilliams, and A. Shchepetkin, 2003:Equilibrium structures and dynamics360
17
of the California Current System.J. Phys. Ocean. 33, 753-783.361
McCaa, J.R., and C.S. Bretherton, 2004: A new parametrization for shallow cumulus convection362
and its application to marine subtropical cloud-topped boundary layers. Part II: Regional363
simulations of marine boundary layer clouds.Mon. Wea. Rev. 132, 883-896.364
Murphy, A. H., 1992: Climatology, persistence, and their linear combination as standards of refer-365
ence in skill scores.Wea. Forecasting 7, 692-698.366
Oke, P.R., J.S. Allen, R.N. Miller, G.D. Egbert, J.A. Austin, J.A. Barth, T.J. Boyd, P.M. Kosro, and367
M.D. Levine, 2002: A modeling study of the three-dimensional continental shelf circulation368
off Oregon. Part I: Model-data comparisons.J. Phys. Ocean. 32, 1360-1382.369
Pawlowicz, R., R. Beardsley, and S. Lentz, 2002: Classical tidal harmonic analysis including error370
estimates in MATLAB using T−TIDE. Computers and Geosciences 28, 929-937.371
Shchepetkin, A.F., and J.C. McWilliams, 2005: The RegionalOceanic Modeling System: A split-372
explicit, free-surface, topography-following-coordinate oceanic model.Ocean Modelling 9,373
347-404.374
Stockwell, R.G., W.G. Large and R.F. Milliff, 2004: Resonant inertial oscillations in moored buoy375
ocean surface winds.Tellus A 56, 536-547.376
Strub, P.T., and C. James, 2000. Altimeter-derived variability of surface velocities in the California377
Current System: 2. Seasonal circulation and eddy statistics. Deep Sea Res.II, 47, 831-870.378
Wang, X., Y. Chao, C. Dong, J. Farraraa, Z. Li, J.C. McWilliams, J.D. Paduan, and L.K. Rosenfeld,379
2009: Modeling Tides in Monterey Bay.Deep-Sea Research 56, 219-231.380
Wilkin, J.L., 2006: The summertime heat budget and circulation of southeast New England Shelf381
waters.J. Phys. Ocean. 36, 1997-2011.382
Winant, C.D. and C. Dorman, 1997: Seasonal patterns of surface wind stress and heat flux over383
the Southern California Bight.J. Geophys. Res. 102, 5641-5653.384
18
TABLE CAPTIONS385
Table. 1 Principal-Axis Correlation between MM5 and Buoy Winds in March, 2002386
Table. 2 Numerical Experiments with Different External Forcings387
Table. 3 Skill Scores (SS), MSE, and Weighted Skill Scores (WSS)388
19
FIGURE CAPTIONS389
Fig. 1 Bathymetry [m] of the Southern California Bight ROMS domain. The red solid circles390
indicate the locations of three buoys (SMB, NDBC-46025, NDBC-46053) and Scripps Pier. Tide391
gauges are at the red-star locations plus one more near the Scripps Pier. Twenty-eight red lines and392
one yellow line indicate cross-shore sections and one along-shore section used in the simulation393
analysis.394
Fig. 2 Satellite SST [oC] distribution at the beginning (upper) and ending (lower) of the up-395
welling event. The data are AVHRR/Pathfinder (ftp://data.nodc.noaa.gov/pub/data.nodc/pathfinder/Version5.0/)396
with 4 km spatial resolution.397
Fig. 3 Time series of SST anomaly relative to the individual-year means at NDBC buoy 46025398
for 10 years (1995-2004). See Fig. 1 for location.399
Fig. 4 Comparison between QuikSCAT wind (red) and MM5 reanalysis wind with dx = 2 km400
(blue) at 10 AM, March 16 (GMT). The QuikSCAT wind resolutionis 25 km (http:www.jpl.nasa.gov).401
Fig. 5 Oceanic currents at 10 m depth the baseline case T12 before the upwelling (March 1-13;402
upper panel) and during the upwelling event (March 14-16; lower panel).403
Fig. 6 Comparison of ROMS sea-level amplitude and phase with tide-gauge data averaged over404
the six locations shown in Fig. 1 and for 8 tidal frequencies.405
Fig. 7 SST time series from the baseline case T12 and observations during March 2002 at the406
SMB mooring (upper-left), NDBC Buoys 46053 (upper-right) and 46025 (lower-left), and Scripps407
Pier (lower-right) (see locations in Fig. 1). All data are three-hourly except the daily-averaged408
Scripps-Pier data.409
Fig. 8 Vertical profiles of temperature (left) and salinity (right) at the SMB buoy: observations410
(upper panels) and case T12 results (lower panels)411
Fig. 9 Time series of alongshore wind stress [N m−2] (upper panel) and curl [Pa / 100 km] on the412
20
twenty-eight cross-shore sections from the MM5 model withdx = 2 km. Dashed lines indicate413
the dates of March 13, 18 and 24,i.e., the starting and ending dates of the first upwelling event and414
second minor upwelling event, respectively.415
Fig. 10 Comparison of onshore transports with (Case T12; upper panel) and without tidal forcing416
(Case T11; lower panel). Red indicates the transport in the layer above 50 m and blue the layer417
below 50 m.418
Fig. 11 Cross-shore sections of temperature [oC] and cross-shore current [m s−1] near the SMB419
buoy averaged over March 14-18 (i.e., the 17th section from the south in Fig. 1). They show an420
upwelling front with a cross-frontal secondary circulation.421
Fig. 12 Cross-shore wind stress and curl alongshore-averaged overthe cross-shore sections in422
Fig. 1 during the upwelling period (March 13 to 18).423
Fig. 13 Wind stress, wind-stress curl, cross-shore transport, andspatial variance of the SCB424
terrain for four different wind resolutions, normalized bythe data withdx = 2 km. The temporal425
and spatial averages are over the upwelling period (March 13-18) and over the cross-shore sections426
in Fig. 1.427
Fig. 14 SST time series from three simulation cases with different combinations of the heat flux428
functions (COADS, diurnal cycle, and empirical coupling with the ROMS SST; see Table 2) and429
the measurements at the SMB mooring.430
Fig. 15 The simulated tidal effect on SST evolution at the SMB mooring: case T12 with tides431
(thick blue) and case T11 without tides (thin blue). The observed SST is plotted in black.432
Fig. 16 SST evolution along the cross-shore section near the SMB buoy (the 17th from the south;433
Fig. 1): without tides (Case T11; left) and with tides (Case T12; right).434
Fig. 17 Time-averaged cross-section during the relaxation phase (March 20-26) near the SMB435
buoy (the 17th from the south; Fig. 1):σ0 (upper panels) andu [m s−1] (lower panels), without436
(case T11; left column) and with tides (case T12; right column).437
21
Fig. 18 Map of wind work [10−3 N m−1 s−1 averaged over the upwelling period from March 13438
to 26 with (case T12; lower panel) and without tides (case T11; upper panel). The four red stars439
are the locations used for Fig. 19.440
Fig. 19 Time series averaged over four near-shore locations markedon Fig. 18: (a) wind stress441
(amplitude and direction); (b) and (c) surface current (amplitude and direction) for cases T11 and442
T12 without and with tides, respectively; and (d) wind work for cases T11 (blue) and T12 (red).443
22
Table 1: Principal-axis Correlation between MM5 and Buoy Winds in March, 2002
Buoys 54 km 18 km 6 km 2 km
46025 0.43 0.61 0.50 0.6546053 -0.01 0.46 0.68 0.68SMB -0.18 0.24 0.26 0.26
23
Table 2: Numerical Experiments with Different External Forcings
Exp. Wind Res. HF and FF Tides? Diurnal HF? HF Relaxation?
T11 2 km MM5 NO NO YEST12 2 km MM5 YES NO YEST12n 2 km MM5 YES NO NOT13 2 km COADS NO NO YEST14 2 km COADS YES NO YEST15 2 km COADS NO YES YEST15n 2 km COADS NO YES NOT16 2 km COADS YES YES YEST16n 2 km COADS YES YES NOT2 6 km MM5 YES NO YEST3 18 km MM5 YES NO YEST4 54 km MM5 YES NO YES
HF denotes heat flux and FF denotes freshwater flux. The baseline experiment is T12.
24
Table 3: Skill Scores (SS), MSE, and Weighted Skill Score (WSS) for the Experiments
Exp. SS: T MSE: T SS: S MSE: S SS: U MSE: U SS: V MSE: V WSS
T11 -0.705 0.733 -2.133 0.018 0.497 0.009 -1.818 0.007 -0.666T12 0.000 0.430 0.000 0.006 0.000 0.018 0.000 0.003 0.000T12n 0.073 0.399 -0.125 0.007 0.177 0.015 -0.204 0.003 0.074T13 -0.392 0.598 -1.423 0.014 0.363 0.012 -1.279 0.006 -0.371T14 -0.021 0.439 -0.145 0.007 0.142 0.016 0.205 0.002 -0.012T15 -0.400 0.602 -1.485 0.014 0.284 0.013 -1.432 0.006 -0.383T15n -0.319 0.567 -1.236 0.013 0.308 0.013 -1.204 0.006 -0.303T16 -0.081 0.395 0.040 0.006 0.149 0.015 -0.395 0.004 -0.081T16n -0.088 0.468 -0.031 0.006 0.139 0.016 0.024 0.003 -0.075T2 0.024 0.420 -0.658 0.010 0.356 0.012 -2.841 0.010 0.013T3 -0.064 0.457 -0.850 0.011 0.494 0.009 -1.574 0.007 -0.054T4 -0.248 0.537 -0.358 0.008 0.041 0.017 -0.566 0.004 -0.237
The number of the data pointsNi is 2310 at the SMB, 496 at the NDBC, and 31 at the Scripps Pier,respectively, for temperature; 2310 for salinity at SMB; and 3691 for horizontal velocity at SMB. Since thediurnal and semi-diurnal cycles are presented in the data, the independent data points in terms of statisticalview could be much less than the number listed, which is not discusse in the analysis.
25
SMB
46025
46053
SIO Pier
Santa RosaRidge
121oW 120oW 119oW 118oW 117oW
30’
33oN
30’
34oN
30’
Los Angeles
SanDiego
Point Conception
Palos Verdes Peninsula
0 500 1000 1500 2000 2500 3000 3500 4000
Figure 1: Bathymetry [m] of the Southern California Bight ROMS domain. The red solid circlesindicate the locations of three buoys (SMB, NDBC-46025, NDBC-46053) and Scripps Pier. Tidegauges are at the red-star locations plus one more near the Scripps Pier. Twenty-eight red lines andone yellow line indicate cross-shore sections and one along-shore section used in the simulationanalysis.
26
Figure 2: Satellite SST [oC] distribution at the beginning (upper) and end-ing (lower) of the upwelling event. The data are AVHRR/Pathfinder(ftp://data.nodc.noaa.gov/pub/data.nodc/pathfinder/Version5.0/) with 4 km spatial resolution.
27
1 2 3 4 5 6 7 8 9 10 11 12−5−4−3−2−1
012345
1995
1 2 3 4 5 6 7 8 9 10 11 12−5−4−3−2−1
012345
1996
1 2 3 4 5 6 7 8 9 10 11 12−5−4−3−2−1
012345
1997
1 2 3 4 5 6 7 8 9 10 11 12−5−4−3−2−1
012345
1998
1 2 3 4 5 6 7 8 9 10 11 12−5−4−3−2−1
012345
1999
Tem
pera
ture
Ano
mal
y (o C
)
1 2 3 4 5 6 7 8 9 10 11 12−5−4−3−2−1
012345
2000
1 2 3 4 5 6 7 8 9 10 11 12−5−4−3−2−1
012345
2001
1 2 3 4 5 6 7 8 9 10 11 12−5−4−3−2−1
012345
2002
1 2 3 4 5 6 7 8 9 10 11 12−5−4−3−2−1
012345
2003
Month1 2 3 4 5 6 7 8 9 10 11 12
−5−4−3−2−1
012345
2004
Month
Figure 3: Time series of SST anomaly relative to the individual-year means at NDBC buoy 46025for 10 years (1995-2004). See Fig. 1 for location.
28
121oW 120oW 119oW 118oW 117oW 30’
33oN
30’
34oN
30’
Los Angeles
10 m/s
Figure 4: Comparison between QuikSCAT wind (red) and MM5 reanalysis wind withdx = 2 km(blue) at 10 AM, March 16 (GMT). The QuikSCAT wind resolutionis 25 km (www.jpl.nasa.gov).
29
33oN
20’
40’
34oN
20’
0.2 ms−1
Before Upwelling
120oW 30’ 119oW 30’ 118oW 30’ 117oW 33oN
20’
40’
34oN
20’
0.2 ms−1
During Upwelling
Figure 5: Oceanic currents at 10 m depth for the baseline caseT12 before the upwelling (March1-13; upper panel) and during the upwelling event (March 14-16; lower panel).
30
Q1 O1 P1 K1 N2 M2 S2 K20
0.1
0.2
0.3
0.4
0.5
0.6
Am
plitu
de (
m)
Tidal Amplitude Comparison: model(r),obs(b)
Q1 O1 P1 K1 N2 M2 S2 K20
50
100
150
200
250
Tidal Constituents
Pha
se (
Deg
ree)
Tidal Phase Comparison: model(r),obs(b)
Figure 6: Comparison of ROMS sea-level amplitude and phase with tide-gauge data averaged overthe six locations shown in Fig. 1 and for 8 tidal frequencies.
31
0 10 20 3010
11
12
13
14
15
16
0 10 20 3012
13
14
15
16
DAY in March, 2002
SS
T (
o C)
0 10 20 3011
12
13
14
15
16
17
DAY in March, 2002
0 10 20 3010
11
12
13
14
15
16
SS
T (
o C)
Figure 7: SST time series from the baseline case T12 and observations during March 2002 at theSMB mooring (upper-left), NDBC Buoys 46053 (upper-right) and 46025 (lower-left), and ScrippsPier (lower-right) (see locations in Fig. 1). All data are three-hourly except the daily-averagedScripps-Pier data.
32
Figure 8: Vertical profiles of temperature (left) and salinity (right) at the SMB buoy: observations(upper panels) and case T12 results (lower panels).
33
Figure 9: Time series of alongshore wind stress [N m−2] (upper panel) and curl [Pa / 100 km] onthe twenty-eight cross-shore sections from the MM5 model with dx = 2 km. Dashed lines indicatethe dates of March 13, 18 and 24,i.e., the starting and ending dates of the first upwelling event andsecond minor upwelling event, respectively.
34
5 10 15 20 25 30−0.6
−0.4
−0.2
0
0.2
0.4
0.6
Ons
hore
Tra
nspo
rt (
Sv)
Upper LayerLower Layer
5 10 15 20 25 30−0.6
−0.4
−0.2
0
0.2
0.4
0.6
DAY in March, 2002
Ons
hore
Tra
nspo
rt (
Sv)
Upper LayerLower Layer
Figure 10: Comparison of onshore transports with tidal forcing (Case T12; upper panel) and with-out it (Case T11; lower panel) integrated alongshore over all the cross-shore sections in Fig. 1.Red indicates the transport in the layer above 50 m and blue the layer below 50 m.
35
Dep
th (
m)
T
−100 −90 −80 −70 −60 −50 −40 −30 −20 −10 0−200
−150
−100
−50
0
4
6
8
10
12
14
16
−100 −90 −80 −70 −60 −50 −40 −30 −20 −10 0−200
−150
−100
−50
0
Offshore Distance (km)
Dep
th (
m)
U −0.3
−0.2
−0.1
0
0.1
0.2
0.3
Figure 11: Cross-shore sections of temperature [oC] and cross-shore current [m s−1] near the SMBbuoy averaged over March 14-18 (i.e., the 17th section from the south in Fig. 1). They show anupwelling front with a cross-frontal secondary circulation.
36
−100 −80 −60 −40 −20 0−0.18
−0.16
−0.14
−0.12
−0.1
−0.08
−0.06
Offshore Distance (km)
Alo
ngsh
ore
Win
d S
tres
s (N
m−
2 )
54km18km6km2km
−100 −80 −60 −40 −20 0−0.05
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
Offshore Distance (km)
Win
d C
url (
Pa/
100K
m)
54km18km6km2km
Figure 12: Cross-shore wind stress and curl alongshore-averaged over the cross-shore sections inFig. 1 during the upwelling period (March 13 to 18).
37
100
101
102
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
1
1.05
MM5 Resolution (km)
Alongshore Wind StressWind Stress CurlOffshore TransportTerrain Variance
Figure 13: Wind stress, wind-stress curl, cross-shore transport, and spatial variance of the SCBterrain for four different wind resolutions, normalized bythe data withdx = 2 km. The temporaland spatial averages are over the upwelling period (March 13-18) and over the cross-shore sectionsin Fig. 1.
38
0 5 10 15 20 25 3010
11
12
13
14
15
16
T14: No Analytical HFT16n: No HF CorrectionT16: AllOBS
Figure 14: SST time series from three simulation cases with different combinations of the heat fluxfunctions (COADS, diurnal cycle, and empirical coupling with the ROMS SST; see Table 2) andthe measurements at the SMB mooring.
39
0 5 10 15 20 25 3010
11
12
13
14
15
16
T11: Without TidesT12: With TidesOBS
Figure 15: The simulated tidal effect on SST evolution at theSMB mooring: case T12 with tides(thick blue) and case T11 without tides (thin blue). The observed SST is plotted in black.
40
Offshore Distance (km)
Day
in M
arch
−100 −50 0
14
16
18
20
22
24
26
28
30
11
11.5
12
12.5
13
13.5
14
14.5
15
15.5
16
Offshore Distance (km)
Day
in M
arch
−100 −50 0
14
16
18
20
22
24
26
28
30
11
11.5
12
12.5
13
13.5
14
14.5
15
15.5
16
Figure 16: SST evolution along the cross-shore section nearthe SMB buoy (the 17th from thesouth; Fig. 1): without tides (Case T11; left) and with tides(Case T12; right).
41
Dep
th (
m)
−400
−350
−300
−250
−200
−150
−100
−50
0
25
25.5
26
26.5
Dep
th (
m)
Offshore Distance (km)0 20 40 60 80 100
−400
−350
−300
−250
−200
−150
−100
−50
0
Offshore Distance (km)0 20 40 60 80 100
−0.4
−0.3
−0.2
−0.1
0
0.1
0.2
0.3
0.4
Figure 17: Time-averaged cross-sections during the relaxation phase (March 20-26) near the SMBbuoy (the 17th from the south; Fig. 1):σ0 (upper panels) andu [m s−1] (lower panels), withouttides (case T11; left panels) and with tides (case T12; rightpanels).
42
Figure 18: Map of wind work [10−3 N m−1 s−1] averaged over the upwelling period from March13 to 26 without tides (case T11; upper panel) with tides (case T12; lower panel). The four redstars are the locations used for Fig. 19.
43
0 5 10 15 20 25 300
0.2
0.4(a)
N/m
2
0 5 10 15 20 25 30−180
−90
0
90
180
Deg
ree
0 5 10 15 20 25 300
0.2
0.4
0.6
0.8
m/s
(b)
0 5 10 15 20 25 30−180
−90
0
90
180
Deg
ree
0 5 10 15 20 25 300
0.2
0.4
0.6
0.8
m/s
(c)
0 5 10 15 20 25 30−180
−90
0
90
180
Deg
ree
0 5 10 15 20 25 300
0.05
0.1(d)
Date in March, 2002
Win
d W
ork
(N/m
/s)
Figure 19: Time series averaged over four near-shore locations marked on Fig. 18: (a) wind stress(amplitude and direction); (b) and (c) surface current (amplitude and direction) for cases T11 andT12 without and with tides, respectively; and (d) wind work for cases T11 (blue) and T12 (red).
44