2007jc004144wfigs - ESSIE

advertisement
1
Density-driven exchange flow in terms of the Kelvin and Ekman numbers
2
3
Arnoldo Valle-Levinson
4
Civil and Coastal Engineering Department
5
University of Florida
6
Gainesville, FL 32611
7
Tel. 352-392-9537 ext. 1479
8
arnoldo@ufl.edu
9
10
11
Abstract
The pattern of density-induced flow influenced by basin’s width, friction and Earth’s
12
rotation is investigated as a function of the Ekman (Ek ) and Kelvin (Ke) numbers. A semi-
13
analytical solution is used to determine the conditions under which the density-induced exchange
14
flow is vertically sheared or horizontally sheared. Solutions are obtained over diverse laterally
15
varying bathymetries. It is found that the exchange flow is horizontally sheared under high
16
frictional conditions (Ek > 1) independently of the width of the basin (Ke). The horizontally
17
sheared pattern describes inflow in the channel and outflow over shoals, with the inflow
18
occupying the entire water column. The exchange flow pattern is also horizontally sheared
19
under weak friction (Ek →0) and in wide (Ke > 2) basins. In that case, however, the outflow is
20
concentrated on the left (looking into the basin in the northern hemisphere) portion of the cross-
21
section and inflow appears on the right. Also under weak friction, the exchange pattern becomes
22
more vertically sheared, with outflow at surface and inflow underneath, as the width of the basin
23
becomes small (Ke < 1). Bathymetry is not very influential in the weak friction exchange
24
patterns. Finally, under moderate friction (0.01< Ek < 0.1) the exchange pattern is both
25
horizontally and vertically sheared for all widths. The horizontally sheared pattern is best
26
defined in wide basins (high Ke) whereas the vertically sheared pattern practically dominates in
27
narrow basins (low Ke). These findings allow classification of various estuaries in the Ek - Ke
28
parameter space.
29
2
30
Introduction
31
It has been traditionally recognized that a basin’s width determines whether Earth’s
32
rotation effects on density-induced or wind-induced water exchange are appreciable or not (e.g.
33
Pritchard, 1952). The common view is that the basin should be wider than the internal Rossby
34
radius Ri for rotation to be important (e.g. Gill, 1982). In a density-induced flow, Ri is given by
35
(g’h)½/f, where g’ is the reduced gravity, h is the depth of the buoyant part of the density-induced
36
flow and f is the Coriolis parameter. In turn, g’ equals g Δρ/o, where g is the gravity
37
acceleration, o is a reference water density and Δρ is the contrast between the buoyant water
38
density and the density underneath. The importance of Ri in containing the buoyant flow may be
39
characterized by the non-dimensional Kelvin number Ke, which compares the basin’s width B to
40
Ri, i.e., Ke = B/Ri (Garvine, 1995). Earth’s rotation effects are supposed to be most prominent
41
when Ke > 1.
42
Kasai et al. (2000) and Winant (2004) pointed out that water column depth, rather than
43
basin’s width, should determine whether Earth’s rotation (or Coriolis) effects are important.
44
Their argument was that over depths greater than several Ekman layers DE (e.g. > 4 DE), Coriolis
45
effects were important regardless of the width. The value of DE is given by (2 Az / f ) ½, where Az
46
is the flow’s eddy viscosity. Earth’s rotation effects on exchange flows may then be cast in
47
terms of the Ekman number Ek (= Az / [f H 2], where H is water depth), which compares frictional
48
to Coriolis effects. Coriolis effects become negligible at high Ek (>1). The objective of this
49
paper is to reconcile these ideas with the help of semi-analytical results that portray density-
50
induced exchange flows in terms of the Ekman and Kelvin numbers. This study extends that of
51
Valle-Levinson et al. (2003) by considering the effects of basin width, i.e. the Ke dependence, on
52
density-induced exchange flows. Results show that the density-induced exchange pattern is
3
53
independent of the basin’s width (or Ke) at high Ek and depends on width (or Ke) at low and
54
moderate Ek.
55
56
Approach
57
Density-induced exchange flow patterns are obtained with a semi-analytical solution (see
58
Valle-Levinson et al., 2003) that compares very favorably with observations. The model solves
59
for the non-tidal or mean along-basin u and transverse v flows at one basin cross-section. The
60
flows are produced by pressure gradients and assumed to be modified only by Coriolis and
61
frictional influences. Advective effects from tidal currents are assumed to be at least one order
62
of magnitude smaller than other influences (e.g. Geyer et al., 2001) and their influence on the
63
pattern of density-induced exchange flows is insignificant (Huijts et al., 2006). In a right-handed
64
coordinate system (x, y, z), where x points seaward, y across the basin and z upward, the non-tidal
65
(or steady) momentum balance is a set of two differential equations:
66
 g 
 2u
 f v  g

z  Az 2
 x  x
z
 g 
 2v
f u  g

z  Az 2
y  y
z
67
where f, g, , , Az are the Coriolis parameter, the gravity acceleration (9.8 m/s2), water density
68
(kg/m3), surface elevation (m), and vertical eddy viscosity homogeneous in z and y (m2/s),
69
respectively. Equations (1) may be solved for a complex velocity w = u + iv, where i 2 = 1 is the
70
imaginary number:
71
w(z) = gNF1 (z) + F2 (z) .
(1)
(2)
72
In (2), N represents the sea level slope from the barotropic pressure gradient (∂η/∂x+ i∂η/∂y).
73
The functions F1 and F2 depict the vertical structure of the barotropic (from sea level slope) and
4
74
baroclinic (from density gradient) contributions to the flow, respectively:
F1 
75
iD
F2 
f
i
f

cosh  z 
1 
 cosh  H y

 





 z
cosh  z 
H y
 H y
 e  z  e
cosh  H y






.
(3)

76
In (3), D equals g/ρ(∂ρ/∂x + i∂ρ/∂y) and is independent of depth; the parameter  equals (1 +
77
i)/DE, where DE is the Ekman layer depth [2Az / f ]½. Equations (3) are obtained by assuming no
78
stress at the surface (F1 /z = F2 /z = 0 at z = 0) and no-slip at the bottom (F1 and F2 = 0 at
79
z = Hy). Solutions (2) and (3) require prescription of Hy (as any function of y), a sea level slope
80
N, an eddy viscosity Az, and a density gradient D that is dynamically consistent with N. The
81
dynamically consistent value of D may be obtained by assuming a net volume flux R (m3/s)
82
along or across a cross-section (Kasai et al., 2000), i.e.,
B 0
  w dz dy  R
83
(4)
0 H y
84
where B is the basin’s width. The value of D that satisfies a prescribed N and R is
B

R f 2  i g  N ( y ) (e
85
D
H y
 H y ) tanh( H y )  (1  e
H y

  2 H y2 2) dy
0
B


(5)
i  tanh( H y )  H y dy
0
86
and is a constant independent of y and z. As explained by Kasai et al. (2000), the solution
87
consists of a unidirectional outflow, represented by the barotropic contribution gNF1 in (2 and 3),
88
and bidirectional exchange flows given by the baroclinic contribution F2 in (2 and 3). The key to
89
the solution is the way in which N(y) is prescribed (Valle-Levinson et al., 2003). On the basis of
90
observations and numerical model results, they prescribed a slope with a value N0 at the coast
91
that decayed exponentially across the basin: N = -N0 {1+ i exp[-(y/B)2]} (Fig. 1).
5
In all of Valle-Levinson et al.’s (2003) solutions, the Kelvin number Ke was 1 because the
92
93
exponential decay of the transverse slope of sea level spanned the width B of the basin. Those
94
results are extended here by allowing a more general range of Ke values. This is done by
95
normalizing the across-basin distance with the internal radius of deformation Ri rather than with
96
B:
97
N = - N0 {1+ i exp[-(y/Ri)2]}.
(6)
98
The real part of (6) could also be prescribed as a function of y but the results are practically the
99
same (Valle-Levinson et al., 2003). Solutions (2) and (3) are obtained for a value N0 of 1×10-6, R
100
of zero and different values of Ek (function of Az) and Ke (function of Ri) over various
101
bathymetric profiles. The bathymetric variation across the domain Hy is given by:
102
Hy = H0 exp(-(y-yp)2/b12),
(7)
103
where yp is the across-basin location of the deepest part of the channel (H0) and b1 determines the
104
lateral slope of the channel. Furthermore, assuming along-basin uniformity, the vertical
105
component of motion vw may be calculated from
106
v
v
 w.
y
z
As discussed by Valle-Levinson et al. (2003), the analytical solution ignores advective
107
accelerations and also assumes uniform Az. Despite these simplifications, the patterns produced
108
by solutions (2) and (3) emulate the essence of those observed. Discrepancies exist only in
109
details such as exact area of outflows/inflows and exact slope of isotachs (lines of equal speed).
110
In the analytical solution, the effects of tidal forcing on mixing may be approximated by
111
prescribing different values of Az to emulate the observations. Obviously, the complete approach
112
to this problem would be to analyze numerical model results that allow Az to change in space and
113
in time. Nonetheless, the results presented below and those in Valle-Levinson et al. (2003)
114
exhibit essentially the same across-basin distribution as non-linear, turbulence closure numerical
6
115
results for high Ek (e.g.Valle-Levinson and O’Donnell, 1996) and for low to moderate Ek (e.g.
116
Weisberg and Zheng, 2006). Still, a comprehensive study based on numerical experiments
117
should follow the approach proposed here.
118
119
120
Results
Results are presented for the across-basin distribution of the density-induced along-basin
121
flow u and cross-basin flow v, i.e., for the real and imaginary parts of (2), respectively. All
122
representations show non-dimensional flows looking into a northern hemisphere basin where f =
123
1×10-4 s-1. The along-basin flows are normalized by the maximum inflow. First, five
124
bathymetric configurations are chosen to portray the lateral structure of flows for 3 selected
125
values of Ek and Ke. Second, the strength of the exchange flow over the deepest part of the
126
channel (H0) is examined for a wide range of Ek and Ke values. The strength of exchange flow is
127
characterized by the difference between maximum inflow and outflow and is explored for the
128
same five bathymetric distributions plus a sixth one. Third, the strength of the exchange flow
129
allows characterization of a basin as vertically sheared or horizontally sheared, or between these
130
two extremes, according to its locations in the Ek vs. Ke parameter space. Examples of several
131
estuaries are placed in such parameter space.
132
133
Lateral structure of flows
134
In the depiction of the lateral structure of flows, three bathymetric distributions feature
135
the deepest part H0 in the middle of the section. A fourth one shows H0 close to the left edge of
136
the section and a fifth one exhibits H0 close to the right edge (looking into the basin). The three
137
bathymetric distributions with H0 in the middle of the section had different lateral slopes to
7
138
explore the shape of the flows over different morphologies. The other two bathymetric
139
distributions examined the influence of the channel’s position in the section.
140
141
Flows over weak lateral slope
142
The exchange pattern (along-basin flows) over a nearly flat cross-section (contours in
143
Fig. 2) shows a strong dependence on Ke at low Ek (very weak friction). In contrast, the
144
exchange pattern remains nearly unaffected by Ke at high Ek. Under large Ke (wide basin) and
145
weak to moderate friction (Ke ≥ 1 and Ek < 0.1) the isotachs are steeply sloped (Fig. 2a, 2b, 2d,
146
2e). Most of the outflow in Figures 2a and 2b appears constrained by the internal radius of
147
deformation and the inflow occupies most of the cross-sectional area. As Ke decreases from 4 to
148
1, the outflow occupies up to one half of the cross section and the isotachs tilt less steeply (Fig.
149
2d and 2e) until they become almost flat at Ke of 0.25 (Fig. 2g and 2h). When friction becomes
150
much more dominant than rotation (Fig. 2c, 2f, 2i) then the exchange becomes that of typical
151
estuarine circulation with outflow at the surface and inflow underneath across the entire section.
152
This pattern is independent of Ke as Coriolis is not relevant anymore and the dynamics are
153
determined by the balance between pressure gradient and friction (e.g. Pritchard, 1952; Hansen
154
and Rattray, 1965). Still, slight bathymetric effects favor the development of two branches of
155
outflow over the shallower portions of the section. This becomes more evident as the
156
bathymetric slopes increase (e.g. Wong, 1994).
157
The cross-basin flows, depicted by arrows in Figures 2 to 6, show three features that are
158
consistent among all the bathymetric configurations explored. The first feature is a clockwise
159
circulation pattern, in the vertical plain, that remains qualitatively the same at low Ek regardless
160
of Ke (top row of Figures 2 through 6). This clockwise gyre is comprised of currents with similar
8
161
magnitude as the along-basin flows and is tied to those along-basin flows through Coriolis
162
effects. Thus, the lateral flow at surface moves to the right (in the northern hemisphere) of the
163
outflow and the lateral flow in the layer underneath moves to the right of the inflow. Such
164
linkage between along and cross-basin flows is best illustrated in panels d) and g). The second
165
general feature is that the cross-basin circulation has the opposite direction at high Ek (bottom
166
row of Figures 2 to 6) than at low Ek (top row of Figures 2 to 6). In general, cross-basin flows
167
are from left to right at surface (0 < y < 0.3 in the c) panels) and from right to left in the deepest
168
part of the section. This pattern contrasts that of low Ek and illustrates a region of flow
169
convergence over the channel. The lateral flow at such high Ek is rather weak (consistently < 1
170
cm/s) and is caused by the lateral pressure gradient being balanced by friction, rather than
171
Coriolis. The dynamic balance between pressure gradient force and friction results in a
172
‘sideways’ estuarine circulation (Valle-Levinson et al., 2003) even though it is weak. The third
173
general feature has to do with intermediate Ek (middle row in Figures 2 to 6). The cross-basin
174
flow pattern reverses from that at high Ke (e.g. Fig. 2b), which also resembles the flow pattern at
175
low Ek (Fig. 2a), to that at intermediate and low Ke (e.g. Fig 2e and 2h). This means that at high
176
Ke and intermediate Ek (Fig. 2b) the lateral flow is to the right in the deep part of the channel but
177
it becomes toward the left at lower Ke (Fig 2e, 2h). The lateral flow at such intermediate Ek
178
values is 2-3 times smaller than the along-basin flow so it is relevant to the transport of solutes.
179
180
Flows over moderate lateral slope
181
The along-basin patterns over a moderately sloped cross-section are consistent with
182
those over weak slope but only for the low friction cases (Fig. 3). As Ke decreases, the isotachs
183
tilt less and the exchange flow changes from horizontally sheared to vertically sheared (Fig. 3a,
9
184
3d, 3g). The influence of friction allows the development of two branches of outflow. Under
185
moderate friction the exchange pattern becomes vertically sheared for all Ke examined (Fig. 3b,
186
3e, 3h). Furthermore, the left branch of outflow (looking into the basin) is more prominent
187
because of the influence of Coriolis accelerations. Such prominence of the left branch, however,
188
decreases with decreasing Ke, i.e., the asymmetry becomes less evident in narrow channels. The
189
asymmetry also diminishes under strong forcing (Fig. 3c, 3f, 3i). At high Ek the exchange
190
pattern is a) symmetric about H0, b) independent of Ke and c) horizontally sheared. This
191
exchange pattern at high Ek contrasts that over weak lateral slope, which is vertically sheared.
192
The cross-basin flow exhibits the same three general features described for the
193
bathymetry with weak lateral slope. The magnitude of the lateral flows is greater over this
194
bathymetry than over a gentler slope. This is the result of the same prescribed pressure gradient
195
acting over a smaller cross-section. The ratio of cross-basin to along-basin flows remains similar
196
to that over the gentler sloped bathymetry. This means that lateral flows are of similar
197
magnitude as the along-channel flows for low Ek, 2-3 times smaller for intermediate Ek, and
198
around 5 times smaller for large Ek.
199
200
Flows over steep slope
201
The exchange flow patterns resulting over a steep slope are very similar, qualitatively, to
202
those obtained over a moderate slope (Fig. 4). Once again, the exchange pattern shows a strong
203
dependence on Ke under very weak friction (top row of Fig. 4). For a wide channel (Fig. 4a) the
204
flows are greatly influenced by rotation and the exchange pattern is horizontally sheared,
205
whereas for a narrow channel (Fig. 4g) the exchange is vertically sheared. The high friction
206
cases (Fig. 3c, 3f and 3i) show symmetric exchange patterns, independent of Ke, with net inflow
10
207
appearing throughout the water column. The cross-basin flows again show the same three
208
features discussed above and an increase in magnitude because of the reduced cross-section.
209
Also, the ratio of the cross-basin to along-basin flow magnitudes remains similar to the other
210
bathymetries explored. The generalities of the along-basin and cross-basin flow patterns stay the
211
same regardless of the position of H0. Only a few details change with H0 located toward the left
212
or right.
213
214
Flows over moderate slope, H0 on the left and right
215
The exchange patterns arising over a channel located toward the left of the cross-section
216
(Fig. 5) are analogous to those of a channel in the middle (Fig. 3). There is a difference that
217
appears in the high Ke and moderate friction case (Fig 5b). In this case, there is a region of net
218
inflow that develops from bottom to surface. In turn, the patterns related to a channel on the
219
right of the cross-section (Fig. 6) are very similar to those of Figure 3. Therefore, the position of
220
the channel seems to have a minor effect on determining the shape of the exchange patterns.
221
In summary, under high frictional conditions (Ek >1) the exchange pattern is practically
222
invariant to Ke. In contrast, under very weak frictional conditions (Ek →0) the exchange pattern
223
depends on Ke. The exchange flow is horizontally sheared under high Ke (dynamically a wide
224
basin) and vertically sheared under low Ke (narrow basin). These findings are explored further
225
by examining the strength of exchange flows in the deepest part of the channel H0.
226
227
Strength of exchange flows over H0
228
The strength of the exchange flows over H0 is determined by the difference (Δu) between
229
maximum outflow (positive normalized values) and maximum inflow (negative normalized
11
230
values). The values of Δu are always positive. When the maximum inflow develops over H0 and
231
there is inflow from bottom to surface then Δu = 1, as in Figure 3c, 3f, 3i. If 0 < Δu < 1, then
232
there is only inflow over H0 but the maximum inflow in the section is found outside the channel
233
as in Figures 4a, 5a. Those cases of Δu between 0 and 1 represent horizontally, rather than
234
vertically, sheared flows. Values of Δu > 1 indicate vertically sheared exchange flows. The
235
greater Δu, the stronger the vertical shear in the exchange flows. A total of 3131 solutions of
236
equations (2) and (3) plus corresponding values of Δu were obtained for a combination of 31 Ek
237
values and 101 Ke values for each of 6 different bathymetric sections. These results are
238
portrayed in Figure 7 where the darkest shades indicate the parameter space for which
239
unidirectional inflows occupy practically the entire water column at H0 (Δu ≤ 1.1). The
240
unshaded areas represent values of Ek and Ke for which maximum exchange develops over H0
241
(Δu > 1.5). The lightly shaded areas denote the parameter space of two-layer exchange in which
242
net inflows occupy a greater portion of the water column than outflows over H0 (1 < Δu < 1.5).
243
The results portrayed in Figures 2 through 6 are also placed in the context of Figure 7 in terms of
244
where each of those solutions lies in the Ek - Ke parameter space.
245
Noteworthy of these results are three main features that should be expected in basins
246
where Coriolis accelerations and frictional effects are relevant to the exchange hydrodynamics.
247
First, for every bathymetric section and Ke considered, the largest values of Δu develop at
248
moderate frictional influences, i.e. at Ek between 0.01 and 0.1 (between -2 and -1 in the abscissa
249
of Fig. 7). Second, under any particular frictional influence and for all bathymetries, the greatest
250
values of Δu appear mostly at low Ke (< 1.6, i.e., 0.2 in the ordinate of Fig. 7) and in general Δu
251
tends to decrease with increasing Ke. Third, in cases with appreciable bathymetric lateral
252
variability (Fig. 7c through f), net inflow develops from surface to bottom at Ek > 0.3 (equivalent
12
253
to -0.5 in the abscissa of Fig. 7). A few more comments are pertinent for each of these three
254
features.
255
The development of largest Δu under moderate friction indicates that some friction is
256
required to generate vertically sheared exchange flows. Otherwise, for too little or too much
257
friction the inflow may occupy most of the water column over H0. Values of Δu > 1 will
258
develop, however, under low Ke (< 1.6, i.e., 0.2 in the ordinate of Fig. 7) and low Ek (<0.01, i.e.,
259
-2 in the abscissa of Fig. 7). This implies that under very weak friction, the exchange will be
260
vertically sheared in a narrow basin (low Ke) and horizontally sheared in a wide basin (high Ke).
261
The Δu dependence on Ke is related to the second feature, mentioned above, that for any given Ek
262
the largest Δu develops mainly at low Ke. For large Ke (wide basins), Coriolis accelerations limit
263
the outflow to the left portion (looking into the basin) of the cross-section. However, vertically
264
sheared exchange flows (Δu > 1.6) may develop under moderate friction (Ek between 0.03 and 1,
265
equivalent to -1.5 and -1 in the abscissa of Fig. 7). With too little friction the inflow occupies the
266
whole water column and the maximum inflow appears to the right of H0. The third feature is
267
associated with horizontally sheared exchange flows developing over relatively steep lateral
268
bathymetric slopes. For less steep cross-sections, tending toward a flat bottom, bathymetric
269
effects obviously do not play a role and the flow is vertically sheared even under strong frictional
270
effects. In the cross-section as a whole (figure not shown), the difference between maximum
271
outflow and inflow at the entire section (not only over H0) is greatest at intermediate Ek (same
272
values as above) and high Ke. The results depicted in all figures suggest that the combined
273
influence of Coriolis and friction is crucial for the development of vigorous exchange flows in a
274
basin.
275
13
276
Implications on natural systems
277
In an attempt to bring these results into a real context, Figure 7d has been recast with the
278
inclusion of several estuaries for which values of Ek and Ke are known from observations. The
279
bathymetry of Figure 7d has been chosen as a generic bathymetry for estuaries as it consists of a
280
deep channel flanked by shoals that extend to the shores. The systems chosen are only
281
illustrative of where various estuaries would lie in the Ek vs Ke parameter space and the type of
282
subtidal exchange expected. Detailed observations of the lateral structure of subtidal exchange
283
are available for each system portrayed in Figure 8. Most of those systems represented in the Ek
284
vs Ke diagram of Figure 8 were compared to analytical results depicted by equations 2 and 3 in
285
Valle-Levinson et al. (2003). These include a wide estuary with relatively weak frictional
286
influences, the Gulf of Fonseca, on the Pacific side of Central America; a wide system with
287
moderate frictional influences, the lower Chesapeake Bay; a moderately wide system with
288
moderate friction, the James River; and a narrow system with moderate friction, Guaymas Bay,
289
on the mainland side of the Gulf of California. Three other systems have been included in the
290
diagram: a narrow estuary with moderately weak friction, Saint Andrew Bay on Northern
291
Florida’s coast in the Gulf of Mexico (Valle-Levinson, unpublished data); a narrow fjord with
292
weak friction, Reloncavi fjord (Valle-Levinson et al., 2007); and a narrow estuary with moderate
293
to strong friction, the thoroughly studied Hudson River (e.g. Lerczak et al., 2006).
294
The systems that lie in the unshaded region of the Ek vs Ke diagram (Fig. 8), namely the
295
lower Chesapeake Bay, St Andrew Bay, Reloncavi Fjord and the Hudson River, exhibit a well-
296
developed vertically sheared exchange flow over their deepest channel of their cross sections.
297
Note that the Hudson River lies close to the boundary between lightly shaded and unshaded
298
regions. This indicates that an increase in frictional influences will cause the subtidal exchange
14
299
to be less vertically sheared than under reduced frictional influences. Indeed this is the transition
300
observed from neap to spring tides (Lerczak et al., 2006). A similar situation develops in the
301
James River, which also lies close to the boundary between unshaded and lightly shaded regions
302
in the Ek vs Ke diagram. The James River exhibits a fortnightly transformation from vertically
303
sheared exchange flows in neap tides to less vertically and more horizontally sheared exchange
304
flows in spring tides. It is likely that systems close to the boundaries in Figure 8 will exhibit
305
marked fortnightly transformations in exchange patterns. Guaymas Bay lies in the region
306
influenced by both vertically and horizontally sheared exchange flows as does the Gulf of
307
Fonseca, as confirmed by observations. However, Guaymas Bay’s pattern is the result of weak
308
to moderate tidal forcing and Fonseca’s pattern is the result of Coriolis effects on a wide system.
309
It also should be acknowledged that natural systems will exhibit different scales of temporal
310
variability that will position them at various locations in the Ek vs Ke diagram. Therefore, the
311
position of the symbols representing each estuary on Figure 8 is nominal. A better representation
312
would be to describe each estuary with an ellipse that circumscribes the range of possible Ek and
313
Ke values observable for the system. This might be the challenge for future investigations of a
314
given estuary. In general, any estuary could be cast in the Ek vs Ke parameter space and its
315
position would yield information on its pattern of density-induced exchange flows.
316
Noteworthy in Figure 8 is the absence of examples in the dark shaded regions. These
317
would represent either extremely strong frictional systems, for Ek > 0.3 (or -0.5 on the abscissa of
318
Fig. 8), or nearly frictionless while strongly rotating systems (upper left corner of diagram). The
319
former case of strongly frictional systems would be consistent with Wong (1994) results but
320
would require extremely large tidal currents (perhaps > 2 m/s) and very large eddy viscosities
321
(>0.05 m2/s). In systems influenced by strong tidal currents the subtidal flow will probably be
15
322
dominated by tidal rectification rather than by density gradients and the dynamics presented in
323
(1) would not apply. Similarly, the upper left corner of the diagram might have just a few
324
examples represented in nature as it denotes very wide and frictionless systems dominated by
325
density-induced flows, like the Gulf of California (Castro et al., 2006). Most estuaries in nature
326
are expected to lie in the lightly shaded and unshaded regions of the Ek vs. Ke. As seen in Figure
327
7, the limits of these regions are fairly consistent among different bathymetries, at least where
328
appreciable lateral slopes exist.
329
330
Conclusion
331
The main findings of this study, which extends those of Valle-Levinson et al. (2003)
332
related to density-induced exchange flows over laterally varying (channel-shoals) bathymetry,
333
are as follows. 1) The exchange pattern consisting of net inflows in the channel and outflows
334
over shoals develops only under high frictional conditions (Ek > 1) independently of the basin’s
335
width (Ke). This is a horizontally sheared exchange pattern. 2) Under weak frictional conditions
336
(Ek → 0) the exchange pattern is horizontally sheared in wide basins (Ke > 2) and vertically
337
sheared in narrow basins (Ke < 1). This response is independent of the bathymetric profile
338
because under weak friction the exchange flows are insensitive to bottom effects. 3) Under
339
moderate friction conditions (0.01< Ek < 0.1) the exchange flow is both horizontally and
340
vertically sheared in most basin widths. Only very narrow systems (Ke < 0.25) display
341
preferentially vertically sheared flow. The latter situation of moderate frictional conditions is the
342
one expected in most estuarine systems (e.g. the observational examples in Valle-Levinson et al.,
343
2003). By locating a particular system in the proposed Ek vs. Ke parameter space, one should be
344
able to infer the pattern of exchange flows to be expected in the system.
16
345
346
347
Acknowledgments
This study was funded by NSF projects 9983685 and 0551923. Conversations with R.
348
Garvine yielded the ideas developed in this manuscript, which is dedicated to him. The
349
comments from R. Chant, S. Monismith, R. Weisberg and an anonymous reviewer are gratefully
350
appreciated.
351
352
17
353
References
354
Castro R., R. Durazo, A. Mascarenhas, CA Collins, A. Trasviña, Thermohaline variability and
355
geostrophic circulation in the southern portion of the Gulf of California. Deep Sea Res. I,
356
53, 188-200, 2006.
357
358
359
360
Garvine, R.W. A dynamical system for classifying buoyant coastal discharges. Cont. Shelf Res.,
15(13), 1585-1596, 1995.
Geyer, W.R., J. Trowbridge, and M. Bowen, The dynamics of a coastal plain estuary, J. Phys.
Oceanogr., 31, 2001.
361
Gill, A.E., Atmosphere-Ocean Dynamics. Academic Press, New York, 661 pp. 1982.
362
Hansen, D.V. and M. Rattray, Jr., Gravitational circulation in straits and estuaries. J. Mar. Res.,
363
23, 104-122, 1965.
364
Huijts, K.M.H., H.M. Schuttelaars, H.E. de Swart, and A. Valle-Levinson, Lateral entrapment of
365
sediment in tidal estuaries: An idealized model study, J. Geophys. Res., 111, C12016,
366
doi:10.1029/2006JC003615, 2006.
367
Kasai, A., A.E. Hill, T. Fujiwara, and J.H. Simpson, Effect of the Earth’s rotation on the
368
circulation in regions of freshwater influence, J. Geophys. Res., 105(C7), 16,961-16,969,
369
2000.
370
371
372
373
Lerczak, J., W.R. Geyer, and R. Chant, Mechanisms driving the time-dependent salt flux in a
partially-stratified estuary, J. Phys. Oceanogr., 36(12), 2296-2311, 2006.
Pritchard, D.W. Salinity distribution and circulation in the Chesapeake Bay estuarine system. J.
Marine Res., 11, 106-123, 1952.
374
Valle-Levinson, A. and J. O'Donnell, Tidal interaction with buoyancy driven flow in a coastal
375
plain estuary, Coastal and Estuarine Studies vol. 53, Buoyancy Effects on Coastal and
18
376
Estuarine Dynamics, edited by D.G. Aubrey and C.T. Friedrichs, Am. Geophys. Union,
377
Washington, DC., pp. 265-281, 1996.
378
379
Valle-Levinson, A., C. Reyes and R. Sanay, Effects of bathymetry, friction and Earth’s rotation
on estuary/ocean exchange, J. Phys. Oceanogr., 33(11), 2375-2393, 2003.
380
Valle-Levinson, A. N. Sarkar, R. Sanay, D. Soto, and J. León. Spatial structure of hydrography
381
and flow in a Chilean Fjord, Estuario Reloncaví, Estuaries and Coasts, 30 (1), 113-126,
382
2007.
383
384
385
Winant, C.D., Three dimensional wind-driven flow in an elongated rotating basin, J. Phys.
Oceanogr., 34, 462-476, 2004.
Weisberg, R. H., and L. Y. Zheng, Circulation of Tampa Bay driven by buoyancy, tides, and
386
winds, as simulated using a finite volume coastal ocean model, J. Geophys. Res., 111,
387
C01005, doi:10.1029/2005JC003067, 2006.
388
389
Wong, K.-C., On the nature of transverse variability in a coastal plain estuary, J. Geophys. Res.,
99, 14,209-14,222, 1994.
390
19
391
List of Figures
392
Figure 1. Cross-channel distribution of the lateral slope prescribed in equation (5) as compared
393
394
to numerical and observational results (as in Valle-Levinson et al., 2003).
Figure 2. Along-estuary (normalized by maximum inflow) and cross-estuary (scale appearing
395
above upper right corner in cm/s) flows in a cross-section with very weak lateral slopes.
396
Bathymetry is drawn from equation (7) for yp in the middle (y/B = 0.5) and and b1 of 1.9.
397
Darker areas denote regions of inflows. Contours are drawn at 0.2 intervals. Views are
398
looking into the estuary. For the upper row, Az = 1×10-4 m2/s; for the middle row, Az =
399
1×10-2 m2/s; and for the lower row Az = 1×10-1 m2/s.
400
Figure 3. Same as figure 1 but with b1 in equation (7) of 0.5.
401
Figure 4. Same as figure 1 but with b1 of 0.2.
402
Figure 5. Same as figure 1 but with b1 of 0.3 and yp at y/B = 0.3.
403
Figure 6. Same as figure 1 but with b1 of 0.3 and yp at y/B = 0.7.
404
Figure 7. Values of Δu (maximum normalized outflow - positive - minus maximum normalized
405
inflow -negative) at H0 for a range of Ke and Ek values and bathymetries. The subpanels on
406
top of each panel show the bathymetry related to each set of results and the location of H0.
407
Contour interval is 0.1. Unshaded areas indicate the parameter space for which vertically
408
sheared along-estuary exchange develops. Dark shades denote the parameter space for
409
horizontally sheared exchange flow. Lightly shaded areas represent horizontally and
410
vertically sheared exchange flows. Numbers accompanied by letters (e.g. ‘2a’) indicate the
411
Ke -Ek flows illustrated in the corresponding figure (e.g. Fig. 2a).
412
413
Figure 8. Same as Figure 7d but including some examples of natural systems: F is Gulf of
Fonseca, Central America; C is lower Chesapeake Bay, Virginia; J is James River,
20
414
Virginia; G is Guaymas Bay, Mexico; S is St Andrew Bay, Florida; H is Hudson River,
415
New York-New Jersey; R is Reloncavi Fjord, Chile.
416
21
417
418
419
420
421
Figure 1. Cross-channel distribution of the lateral slope prescribed in equation (5) as compared
to numerical and observational results (as in Valle-Levinson et al., 2003).
22
422
423
424
425
426
427
Figure 2. Along-estuary (normalized by maximum inflow) and cross-estuary (scale appearing above upper right corner in cm/s)
flows in a cross-section with very weak lateral slopes. Bathymetry is drawn from equation (7) for yp in the middle (y/B = 0.5) and
and b1 of 1.9. Darker areas denote regions of inflows. Contours are drawn at 0.2 intervals. Views are looking into the estuary.
For the upper row, Az = 2×10-4 m2/s; for the middle row, Az = 1×10-2 m2/s; and for the lower row Az = 1×10-1 m2/s.
23
428
429
Figure 3. Same as figure 1 but with b1 in equation (7) of 0.5.
24
430
431
Figure 4. Same as figure 1 but with b1 of 0.2.
25
432
433
Figure 5. Same as figure 1 but with b1 of 0.3 and yp at y/B = 0.3.
26
434
435
Figure 6. Same as figure 1 but with b1 of 0.3 and yp at y/B = 0.7.
27
436
437
438
439
440
441
442
Figure 7. Values of Δu (maximum normalized outflow - positive - minus maximum normalized inflow -negative) at H0 for a
range of Ke and Ek values and bathymetries. The subpanels on top of each panel show the bathymetry related to each set of
results and the location of H0. Contour interval is 0.1. Unshaded areas indicate the parameter space for which vertically sheared
along-estuary exchange develops. Dark shades denote the parameter space for horizontally sheared exchange flow. Lightly
shaded areas represent horizontally and vertically sheared exchange flows. Numbers accompanied by letters (e.g. ‘2a’) indicate
the Ke -Ek flows illustrated in the corresponding figure (e.g. Fig. 2a).
28
443
444
445
446
447
448
Figure 8. Same as Figure 7d but including some examples of natural systems: F is Gulf of
Fonseca, Central America; C is lower Chesapeake Bay, Virginia; J is James River, Virginia; G is
Guaymas Bay, Mexico; S is St Andrew Bay, Florida; H is Hudson River, New York-New Jersey;
R is Reloncavi Fjord, Chile.
29
Download