Transverse variability of ﬂow and density in a …arnoldo/ftp/papers/caceres...Continental Shelf...

Continental Shelf Research 22 (2002) 1683–1698

Transverse variability of flow and density in a Chilean fjord

Mario C!aceresa,*,1, Arnoldo Valle-Levinsonb, H!ector H. Sep !ulvedab,Kristine Holderiedc

aServicio Hidrografico y Oceanografico de la Armada, Errazuriz 232, Casilla 324, Valparaiso, ChilebDepartment of Ocean, Earth and Atmospheric Sciences, Center for Coastal Physical Oceanography, Old Dominion University,

Norfolk, VA 23529, USAcCenter for Coastal Monitoring and Assessment, NOAA, National Ocean Service, Silver Spring, MA 20910, USA

Received 2 February 2001; accepted 4 December 2001

Abstract

Measurements of velocity and density profiles were made to describe the transverse structure of the flow in Aysen

Fjord, Southern Chile (45.21S and 73.31W). Current profiles were made with a 307.2 kHz acoustic doppler current

profiler (ADCP) during 20 repetitions of a cross-fjord transect during one semidiurnal tidal cycle. The transect had a

B3201 orientation, 3 km length, and its bathymetry consisted of two channels, one on the southern side (230m depth)

and the other on the northern side (180m depth), separated by a bank ca. 65m depth, which was located B1 km from

the northern coast. Density measurements to a maximum depth of 50m were made at the extremes of each transect

repetition and over the bank. Also, a total of nine CTD stations that covered the surroundings of the bank was sampled

2 days prior to the ADCP sampling. During the sampling period the mean flow showed a three layer structure that was

consistent with up-fjord wind-induced exchange: a thin (o8m) weak outflow close to the surface due to river discharge;

a layer of inflow (down-wind) underneath attributed to the effect of wind-stress; and a deep compensatory outflow due

to the barotropic pressure gradient set up by the wind. The bank caused the strongest wind-induced net inflows and

outflows to be shifted to the channels and also disrupted the three-layer structure. Also, the strongest tidal current

amplitudes were located over the bank. The near-surface flow and density distributions suggested that the transverse

dynamics were ageostrophic in a layer above B50m. This was indicated by the wind-induced shifting of the location of

salt water intrusion from the northern side to the southern side of the fjord. Density measurements also suggested an

alternation to quasigeostrophic conditions in this upper layer during calm winds. Below this layer the dynamics

remained quasigeostrophic. r 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Acoustic current meters; Ageostrophic flow; South America; Chile; Aysen; Fjords; Aysen Fjord; Physical oceanography;

ADCP; Momentum balance; Physics of estuaries; Fjords oceanography

1. Introduction

Fjords are high latitude estuaries that areusually long relative to their width, relatively steepsided and deep, normally possess one or more sills,and usually feature a river discharging at their

*Corresponding author. Fax: +56-32-266-542.

E-mail address: [email protected] (M. C!aceres).1Formerly at the Center for Coastal Physical Oceanography,

ODU.

0278-4343/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.

PII: S 0 2 7 8 - 4 3 4 3 ( 0 2 ) 0 0 0 3 2 - 8

most upstream reaches (their head). As in othertypes of estuaries, studies in fjords have focusedparticularly on the understanding of the long-itudinal and vertical variability of the flow, whilethe transverse distributions have received relativelylittle attention. The works of Freeland et al. (1980)and Farmer and Freeland (1983) provide extensivereviews and abundant references about the circu-lation in fjords, confirming that the preferentialinterest has been given to the longitudinal andvertical dimensions.

The hydrodynamics of fjords are usuallyassumed to have an along-fjord momentumbalance in which the pressure gradient is balancedby friction or advective accelerations (in constric-tions), and a lateral geostrophic balance, i.e.,between Coriolis accelerations and pressure gra-dient accelerations (Dyer, 1997). The geostrophicapproximation has been demonstrated to be validin fjords of British Columbia (Cameron, 1951) andin Juan de Fuca Strait (Tully, 1958). However,wind stress may frequently become relevant to thelateral dynamics. Wind stress may also reverse theclassical mean velocity profile in estuarine systems.For instance, year-long series of current measure-ments in a coastal-plain estuary have shown that,at least during 21% of the time, inflow wasobserved on the surface and outflow near the bed(Elliot, 1974). This effect may not be very differentin fjords. Their steep coastal walls may confine thewind flow, thus enhancing the along-fjord intensityof wind and hence the wind-induced flow. Thereversal of the surface layer flow by up-fjord windshas been well documented in British Columbia(Pickard and Rodgers, 1959; Farmer, 1976; Farm-er and Osborn, 1976) and Norway (Svendsen andThompson, 1978; Svendsen, 1980). In some caseswind has been postulated to be one of the mainforcing agents for deep water renewal (Gade, 1973;Gade and Edwards, 1980). Some efforts have beenmade to model the wind-driven circulation infjords (Klinck et al., 1981; Krauss and Brugge,1991). The across-fjord variations of the flow,nevertheless, are still a matter that requires morescrutiny. The objective of this study is to documentthe lateral structure of the flow and density fieldsnear the mouth of a Chilean fjord. The objectiveis pursued with measurements of velocity and

density profiles obtained at a cross-fjord transectover a bank located near the mouth of AysenSound in southern Chile, South America.

1.1. Study area

Aysen Sound (Fig. 1b) is a fjord-like inlet with alength of about 65 km measured between its mouth(station C in Fig. 1) and its head (station J inFig. 1) and an average width of about 6.5 km. Theaverage depth in the longitudinal section (stationsA–J in Fig. 1) is 217m, with maximum values of360m in the outer basin at the mouth (C) and350m in the inner basin (H–I). It has a sill of about125m depth located to the west and east-southeastof Colorada Island (station E in Figs. 1 and 3a),and banks of 60, 90 and 90m at stations F, G andH, respectively (Fig. 1). The orientation of thefjord in its first 30 km from the mouth is in theSW–NE direction. The orientation changes in thefollowing 25 km to the NW–SE direction, and tothe SW–NE direction in the last 10 km (toward thehead). As Aysen Sound is glacially carved and hasa fjord-like morphology, henceforth it will bereferred as Aysen Fjord.

Waters of oceanic origin reach the fjord chieflyfrom the north through the Meninea constriction(Fig. 1b). Modified Subantarctic waters of salinity32.0 may fill the deep basin (210m deep, letter A inFigs. 1b and c) to the south of the Menineaconstriction according to Silva et al. (1995).Because of this constriction, along with theshoalings (B50m) at Middle Passage (B) and tothe south of Elisa Peninsula in Fig. 1b, and alsobecause of the narrow channels that connect thestudy area with the ocean 50 km to the west, thecirculation in Aysen Fjord is believed to be littleaffected by remote forcing generated on thecontinental shelf (outside the Inland Sea). To thesouth of Costa Channel the system receives freshwater inputs from small rivers and glaciers and itsexchanges with the open ocean are stronglyreduced by the presence of narrow (o1 km width)and shallow (B50m) channels. Owing to thisinsulating morphology, the main forcing agents inthe vicinity of the mouth of the Aysen Fjord arepostulated to be tidal forcing from the coastalocean, local wind stress, and river discharge. The

M. C !aceres et al. / Continental Shelf Research 22 (2002) 1683–16981684

main characteristics of these forces in the regionare outlined below.

The tidal regime throughout the inland sea ismixed, predominantly semidiurnal. Outside ofAysen Fjord, four months of sea level measure-ments made during spring 1998 in the vicinity ofMeninea constriction and at the entrance to CostaChannel have shown M2 tidal amplitudes of 76.33and 81.67 cm, respectively, for these locations(Fierro et al., 1999). They have also revealed meantidal ranges during spring tides of 2.46 and 2.64mat the same locations. Inside Aysen Fjord, similarranges of 2.2m in Puerto P!erez and 2.7m at thehead of the fjord (C, Fig. 1b) also have beenreported for this region (SHOA, 1993). FollowingStigebrandt (1977), the amplitude of the tidalcurrent ðuÞ in the bank region should be about0.12m/s, using u ¼ aYw=A; where w is thefrequency of the semidiurnal tide (2p=12:42 h.), a

is the mean amplitude in water level at PuertoP!erez, Y is the horizontal area of the fjord(350 km2), and A is the cross sectional area at thebank (0.5 km2). As confirmed by this study, tidalcurrents were in the range 0.1–0.2m/s. These aresimilar to the wind-induced currents in the fjord.

According to information provided by theMeteorological Service of the Chilean Navy(SMA), the wind regime in this region of theInland Sea is dominated by southerly and south-westerly winds during spring and summer (Octo-ber–March), and northerly and northwesterlywinds during fall and winter (April–September).Westerly and southwesterly winds may favorwind-driven, up-fjord surface circulation at themouth and in the outer half of the fjord, whichwould decelerate the surface layer mean outflow.North and northwesterly winds could be impor-tant for wind-driven circulation in the inner half of

(a) (c)

(b)

Fig. 1. (a) Study area on the southern coast of Chile. (b) The measurements obtained during this experiment were made in the box

shown to the south of Elena Island. The Pacific Ocean is about 50 km to the west of the left edge of the lower panel. (c) Bottom profile

in the longitudinal transect A–J. Isobaths shown: 50, 100, 200, and 300m.

M. C !aceres et al. / Continental Shelf Research 22 (2002) 1683–1698 1685

the fjord, as the low land profile at the northernregion of Puerto Perez may favor winds comingfrom that direction.

The main source of fresh water to the fjord isprovided by the Aysen River (Fig. 1). Similar tomany other Andean and trans-Andean rivers, theAysen River exhibits two annual peaks of dis-charges at the head corresponding to the autumnincrease in precipitation (April–June) and thesummer melting of snow. Sampling for this studywas carried out in October, close to the snowmelting peak in November. The mean annual riverdischarge is estimated to be 620m3/s (Niemeyerand Cereceda, 1984) and the mean total annualprecipitation (1968–1999) for stations around thestudy area fluctuates between 1100 and 2100mm(SMA, personal communication)

In the transition zone at the mouth between Cand E (Fig. 1), a strong horizontal salinity gradientat the surface (B12.5/10 km) has been observed in

March (Silva et al., 1995), September (Mu *nozet al., 1992) and November (Silva et al., 1997).This feature suggests the presence of a tidalintrusion front in the region of confluence betweenthe brackish layer coming from the fjord and theintrusion of salty water from the west. As salinityis representative of the density field in this region,this will be used hereafter when describing densitypatterns. A satellite image obtained in February1993 (Fig. 2) also suggests the shape of the front,which might represent the tidal intrusion goinginto the fjord on the northern side and thebrackish water tending to leave the fjord on thesouth.

This study concentrated on the bank near themouth located in the box marked F in Fig. 1b andhad the objective of examining the transversevariability of the flow associated with this bathy-metric feature. The working hypothesis was thatthe transverse dynamics would reflect geostrophicconditions: inflow of saltier water on the northernside and outflow of fresher water on the southernside, as suggested by Mu *noz et al. (1992) and alsoby the satellite image in Fig. 2.

2. Data collection and processing

Current velocity and density measurements wereobtained between October 14 and 17, 1998 in thevicinity of the mouth of Aysen Fjord, in order todescribe the transverse structure of density andflows in that region. Current profiles wereobtained along a B3-km-long cross-fjord transect(Fig. 3) using RD Instruments 307.2 kHz towedacoustic doppler current profiler (ADCP) during acomplete semidiurnal tidal cycle on October 16.The across-fjord transect was traversed 20 timesduring the 12 h of data collection. The samplingstarted at 13:00 h and it had a time intervalbetween repetitions of about 33min. The ADCPwas mounted looking downward on a catamaran1.2m in length, which was towed at approximately2.5m/s on the starboard side of the local boat‘‘Cordillera I’’ on October 16. The ADCP re-corded ensembles of eight pings averaged over30 s. Bin size was 4m and the first velocity bin wascentered atB8m depth. The standard deviation of

Fig. 2. ERS-1 radar satellite image from the study area

obtained on February 3, 1993. The surface roughness suggests

the position of a frontal zone (1) extending from the south of

Elena Island to the north of Traigu!en Island. A train of internal

waves (2) at the entrance of Costa Channel is also suggested in

the image.


the current measurements was then o0.03m/s.The depth limit of the ADCP profiles was 100m.Navigation was carried out using a TrimbleGlobal Positioning System (GPS). The repetitionsof the cross-fjord transect were made over a bankof ca. 65m depth placed 1 km southeast of ElenaIsland (Fig. 3). The ADCP compass data werecalibrated following Joyce (1989) and erroneous

velocity data were removed following the proce-dure explained by Valle-Levinson and Atkinson(1999). The semidiurnal tidal signal (representedby the M2 constituent with a period of 12:42 h) wasseparated from the subtidal signal of the observedflow components using sinusoidal least-squaresregression analysis (e.g. Lwiza et al., 1991). Thesubtidal signal represented the mean over the 12 hof observation.

The current measurements were complementedwith density profiles obtained with a Sea-BirdSBE-19 conductivity–temperature–depth (CTD)recorder from the Chilean Navy patrol craft‘‘Rano Kau’’ on October 14. Data processingwas followed according to the manufacturer’ssoftware. The pressure, temperature and conduc-tivity measurements were aligned, data withpressure reversals were removed and the data wereaveraged to 1m bins. Profiles were measured at theend of each transect repetition and over the bankin the center of the transect (Fig. 3b). Maximumdepth of CTD profiles was 50m owing torestrictions of the CTD’s pressure sensor. OnOctober 14, 2 days prior to the ADCP sampling,12 CTD stations (numbered 9–19, Fig. 3b) weresampled during B3 h of flood. The CTD samplingon October 14 provided a general idea of thehorizontal distribution of the density field in thevicinity of the bank. This is explained in thecontext of the velocity profiles in the next section.

The closest meteorological stations to the studyarea were Guafo Island and Raper Cape, locatedin the coastal region at 225 km to the northwestand 250 km to the southwest, respectively. Owingto this long distance and to the orography in thefjord region, they were not representative of thewind conditions in the study area. Wind data werecollected onboard during the period of sampling.Winds were calm during the first 3 h of sampling,and then they started to increase reaching max-imum values of about 7m/s in the up-fjorddirection (southwesterly winds) at the end of thesampling.

A right-handed coordinate system was adoptedfor which y was positive to the north and x waspositive to the east. It follows that the along-fjordand across-fjord components of the velocity weregiven by u and v; respectively. The data were

(a)

(b)

Fig. 3. (a) Position of the ADCP transect on October 16

(B3 km length). Locations of sills referred to in the text are

pointed out with the segmented line. Depth in meters. (b)

Position of CTD stations. Stations 4a, 6 and 5 were made in

conjunction with the ADCP sampling.


rotated counter clockwise 601 to an along- (u-flow)and an across- (v-flow) channel coordinate system.This angle was oriented to the direction of greatestvariability of the tidal currents and of weakestacross-channel tidal flows. It follows that thex- and y-axis were oriented to 0301 and 3001,respectively.

3. Results

This section presents the across-fjord distribu-tion of the subtidal and tidal flows for theobservation period. It also describes the horizontalsalinity distribution, time series of salinity profiles,

time evolution of the near-surface salinity acrossthe fjord, and the time evolution of verticalsections of salinity across the fjord. These salinityrepresentations help explain the observed subtidalflow patterns discussed below.

3.1. Subtidal and tidal flows

The along-fjord mean u-flow over the period ofobservations (Fig. 4) suggests a three-layer struc-ture that included a thin (o8m) surface weakoutflow close to the surface (not entirely resolvedby the ADCP measurements), a region of netinflow immediately underneath, and a compensa-tory outflow farther below. Vertical profiles of

Fig. 4. Across-fjord distribution of the mean flow (looking up-fjord). The mean flow suggested a surface (8m) outflow on the northern

side (black contour). It also showed a near-surface inflow (light tones) and an interior outflow (dark tones). Contour interval is 3 cm/s.

Vertical profiles of current velocities at selected sites of the transect are shown in the upper region of the figure. Data in the shading

region near bottom were not considered in the analysis due to side lobe effects. White region near bottom was out of the profiling range

of the ADCP.


current velocities at selected sites of the transectshowed that the average outflow within a layerextrapolated to the surface from B8m depth was0.03m/s. Below this unresolved outflow, a 30–40mthick layer showed up-fjord velocities of 0.1m/s,most likely due to the dominant southwesterlywinds during the experiment. The interior outflowlayer of 60–100m thickness had typical velocitiesof 0.05m/s and was most likely a compensatoryflow from the pressure gradient set up by the windin the along-estuary direction.

The along-fjord mean flow exhibited transversevariability related to the bathymetry. Upperlayer current magnitudes were weaker over thechannel north of the bank than to the south.This might be attributed to the fact that thenorthern channel is shallower and narrowerthan the southern channel. Over the bank,frictional effects were evident in the decreasingmagnitudes of the mean current relative to that oneither side and by the fact that most of the watercolumn showed up-fjord net flow, i.e., the near-bottom compensatory outflow was not developedin this region of the cross-section. The effectsof the steep bottom also showed that horizontalfriction might be playing a role in the dynamics,as the greatest outflow magnitudes were observedin the center of both channels at mid-depth.Throughout the section, the zero isotach atabout 50m exhibited downward slope towardthe north, which was contrary to that expectedfrom a geostrophic balance in a two-layereddensity driven flow, where the zero isotach wouldslope upward toward the north. The sign ofthe slope of the zero isotach indicated that thelateral momentum balance in the mean flow couldnot be explained by geostrophy. Below 50m,nevertheless, where the flow was westward, thecurrent was forced to the left by the Coriolis effect,resulting in a thicker layer at the south, i.e. thezero isotach was higher in the water column atthe south. The observed slope suggested then thatthe compensatory flow below 50m was consistentwith geostrophy. It appears then that the observeddistribution consisted of a frictional surface layerand a quasigeostrophic interior.

The transverse flows (Fig. 5a) were directednorthward above B50m and throughout the

northern channel, and southward only within thesouthern channel. The interaction of the flowgoing toward the north and toward the south overthe bank suggested recirculations over this feature.On both sidewalls of the bank, flow tended totravel away from the bank. Thus, the transversecirculation indicated that effects of advectioncould be important around the bank. The stron-gest northward flow was located at the surfaceover the southern channel, far from the effects ofthe frictional effects of the sidewalls. Even thoughthe velocities were in the range of the standarddeviation (70.03m/s), the distribution could notbe dismissed as it showed a coherent pattern. Themean flow is depicted by flow vectors in Fig. 5b,which shows the tendency of the vectors to followthe bathymetry around the bank. This figure alsoshows the uppermost (8m) layer going down-fjordon the northern side, and up-fjord on the southernside.

The amplitude of the semidiurnal along-fjordcurrent (Fig. 6) also showed transverse variabilitythat featured maximum tidal currents of B0.2m/sover the bank. This distribution contrasted thosein coastal plain estuaries where bottom frictioncauses the greatest magnitudes to appear in thedeepest part of the cross-section (Li and Valle-Levinson, 1999). Nonetheless, the transverse dis-tribution observed in the Aysen Fjord was con-sistent with observations of greatest flows over sillsor constrictions in response to the Bernoulli effect(e.g. Farmer and Armi, 1999), where bottomfriction plays a secondary role in shaping theflows. Therefore, the tidal flows in this area shouldbe mostly responding to the pressure gradientinduced by tidal elevations and advective accel-erations produced by the abrupt changes inmorphology.

3.2. Tidal and subtidal salinity

In general, both density and salinity profilesshowed a pycnocline in the first B8m depth withtypical salinities between 15 and 25 in the surfacelayer and 30 in the deeper layer through 50mdepth, which was the maximum depth of CTDcasts. Other observations in this area (Silva et al.,1995, 1997) showed consistent distributions with


those observed here and salinities that increasedup to 31.2 in the deepest part of the fjord.

Fig. 7 shows the horizontal salinity distributionsfrom stations 8 to 19 near the surface and 15mdepth on October 14, 2 days before the ADCPmeasurements. The time between the first and laststation was about 2.7 h. Above the pycnocline (at1m), a fresh water tongue was evident over thenorthern side, while saline water intruded from thenorthern channel into the center of the fjord.Horizontal salinity gradients across the fjordseemed to be at least as large as along-fjord

gradients in this particular instance. Below thepycnocline (at 15m), where the horizontal andvertical salinity gradients were greatly reduced, thehorizontal salinity distribution showed freshwater over the northern side and salty water onthe southwest. This distribution also suggested theinfluence of wind-induced upwelling before theADCP measurements, because the greatest sali-nities were found along the southern side. Alsoworth noting is the separation of the flowsuggested by the shape of the 29.30 isohalinearound the bank in station 15 (Fig. 7), where salty

(a)

(b)

Fig. 5. (a) Across-fjord distribution of the lateral mean flow. Darker tones denote flow to the south and lighter tones flow to the north.

Recirculation over the bank and advection away from the sidewalls of the bank are suggested in the figure. Data in the shading region

near bottom were not considered in the analysis due to side lobe effects. White region near bottom was out of the profiling range of the

ADCP. (b) Vectorial representation of the mean flow. Vectors are drawn at 10m intervals starting at 8m (black), 18 and 28m (gray),

and 38–158m (black).


water bifurcated around this feature (based on justthree points across the fjord). This providedfurther evidence of the influence of the bank inmodifying the flow. Evidence of fresher surfacewater on the northern side relative to the southernside was also suggested by time series of verticalsalinity profiles at stations 4a, 6 and 5a (Fig. 3)taken during the ADCP measurements (Fig. 8). Inthis case, the surface salinity was lower over thenorthern side than on the southern side. Thisdistribution was contrary to that expected fromthe earth’s rotation influence but could be relatedto local upwelling over the southern coastline ofthe fjord, as is also suggested by the across-fjordtilt of the zero isotach (Fig. 4) and the across-fjordflow (Fig. 5). Also worth noting is the change inthe depth of the 28 isohaline during the tidal cycle,which is associated with the along-fjord velocitychanges during this period. Below 8m, the depthresolved by the measurements, velocities may beattributed to the combined effect of tide and wind.In the southern station, the down-fjord flow is lessdominant than in stations in the center and north,probably attributed to the combined effect of up-fjord wind (Fig. 4) and low tidal amplitude shownin that side (Fig. 6). Some tidal straining effects areapparent in the three stations in the sense that thebuoyant layer thickens during ebb and thinsduring flood.

The transverse structure of the near-surfacesalinity field throughout the tidal cycle (Fig. 9)showed that, as wind started to blow from thesouthwest at about 17:30 h and persisted duringthe rest of the experiment, high salinity watersbegan to appear over the southern portion of thesection, another suggestion of upwelling. Inaccordance with the effect of the earth’s rotation,the salty water due to inflow at that depth wasexpected to be located over the northern side of thefjord as observed between 14 h and 17:30 h. Assouthwesterly winds persisted, high salinity waterscovered the whole width of the fjord at 24 h. Lowsalinities observed over the bank at about 21 hwere probably due to the effect of fresh waterconvergence at 2m depth, by the combined effectof up-fjord wind and starting of flood. The effectof the wind was also appreciable in the verticalsections of the CTD stations 4a, 6 and 5a (Fig. 10)on the first three repetitions, where all isohalinestilted down toward the south side during calmwind, as expected from a typical geostrophicbalance in the transverse direction for down-fjordflow. As winds picked up during the fourthrepetition, isohalines of 22, 24, 26 and 28 roseprogressively in repetitions 4, 5, 7 and 8, respec-tively. During repetition 8 every isohaline lowerthan 29 in the surface layer was higher on thesouthern side than on the northern side. Between

Fig. 6. Across-fjord distribution of the along-fjord M2 tidal current amplitude obtained from the ADCP measurements. Highest

amplitudes were observed over the bank. Contour interval is 2 cm/s. Data in the shading region near bottom were not considered in the

analysis due to side lobe effects. White region near bottom was out of the profiling range of the ADCP.


20 and 50m (the maximum depth of CTD casts),isohalines greater than 29 were higher in the southat every repetition (not plotted here). The slope ofthe 29 isohaline, lower in the south, could not beexplained by the measurements. It follows that,according to the salinity measurements, effects ofthe upwelling were noticeable in the first 50m.

The salinity distributions discussed in thissubsection showed a consistent message. South-westerly winds in the lower Aysen Fjord seem toproduce upwelling along the southern coast anddownwelling along the northern coast above 50mdepth, which transiently invalidates the geo-strophic approximation for the transverse dy-namics. Also during these wind episodes, thesaline intrusion shifts from the northern to thesouthern side.

4. Discussion

Wind forcing seemed to be the main drivingmechanism of the mean flow during the samplingperiod as the mean flow was up-fjord near thesurface, below 8m depth. Similar profiles to thosepresented in Fig. 3, resulting from up-fjord winddriven circulation, have been observed in the fjordsof Norway (Svendsen and Thompson, 1978;Svendsen, 1980) and British Columbia (Farmer,1976). This wind episode also may have inducedupwelling, as suggested by the zero isotach and thepycnocline both tilting downward toward thenorth, and by the presence of more saline waterin the south than in the north. Evidence of fresherwater on the northern side than on the southernside could be a transient wind-induced phenom-enon, as salinity horizontal fields observed in theregion of Colorada Island by Mu *noz et al. (1992)showed more saline water over the northern sideduring calm winds. This was also observed duringcalm winds at the beginning of our experiment.

The bank affected the across-channel distribu-tion of subtidal flow as it shifted the location ofboth the strongest near-surface inflow and thestrongest mid-depth outflow to the channels. Italso masked the three-layer mean flow that wasobserved in the channels. Evidences of flowbifurcation induced by the bank were observed inthe horizontal salinity distribution below thepycnoline. The effect of the bank in shaping theflows was also observed in the secondary circula-tion patterns, as flow tended to travel away of thesidewalls and suggested recirculations over thesummit. These effects induced by the bank suggestthat advective accelerations and friction could be

Fig. 7. Horizontal salinity distribution from stations 8 to 19 at

surface and 15m depth on October 14. Isohalines are drawn in

white and station positions are printed in black. Positions of

CTD stations made on October 16 (4a, 6, 5a) are also drawn for

reference. The area around Colorada Island has been masked

because of lack of measurements there.


important for the across-fjord momentum balanceduring strong wind episodes.

In order to establish the relative contribution ofadvection and friction to the across momentumbalance, the Coriolis term and the advective andfriction (vertical and horizontal) terms werecompared. In the across-channel direction, thesubtidal momentum balance for transects 1–4 canbe then given by

/vðqv=qyÞSþ/fuS ¼ �/1=rðqp=qyÞS

þ/Azðq2v=qz2ÞSþ/Ahðq

2v=qy2ÞS; ð1Þ

where vðqv=qyÞ represents the advective terms(the other two could not be evaluated fromthe sampling strategy), fu is the Coriolis term(f ¼ 1:03� 10�4), 1=rðqp=qyÞ is the pressuregradient term, where r is the seawater density,Azðq

2v=qz2Þ is the vertical frictional term,Ahðq

2v=qy2Þ is the horizontal frictional terms and

brackets (/S) denote tidal averages. The coeffi-cients Az and Ah denote the vertical and horizontaleddy viscosities. These were the terms of thecomplete momentum balance that could be re-liably approximated, with exception of the pres-sure gradient term.

For the vertical eddy viscosity, Az; we firstconsider wind stresses of the order of 0.03 Pa, foran average wind of 2.5m/s during the tidal cycle. Ift ¼ rAzqu=qz then Az may be given by Az ¼t=ðrqu=qzÞ: Considering r ¼ 1024 kg/m3 (the aver-age of the full set of density data) andqu=qz ¼ 3:5� 10�3 s�1 (averaged from distribu-tion of Fig. 4), an estimate of Az from windstresses will be 0.0027m2/s. The relationshipproposed by Csanady (1975), Az ¼ u2

n=200f for

estuaries of large depth, where un is the frictionalvelocity ðt=rÞ1=2; provides an estimate of Az

produced by bottom friction. This solution givestypical values in the order of Az ¼ 0:0004m2/s.

Fig. 8. Temporal variability of the salinity (contours) and flow profiles (vectors) from stations 4a, 6 and 5a on October 16. Along-fjord

ADCP measurements for those stations are overplotted. Broken line corresponds to the zero ADCP velocity contour.


Subtracting this from 0.0027 (as shown by Wong,1994; Csanady, 1975) gives Az ¼ 0:0023m2/s.Therefore, we used a constant eddy viscositycoefficient of 0.002m2/s. As expected, bottomfriction played a secondary role in the dynamicsand most of the vertical friction was attributed tothe wind stress.

Similar empirically derived forms of the magni-tude of the horizontal eddy viscosity in tidalchannels are less widely used. Tee (1976) useda value of Ah ¼ 100m2/s for an energetic tidalchannel (currents of B5m/s), Ianniello (1979)used estimates ranging from 1 to 10m2/s for tidalchannels, and Jones and Elliot (1996) usedestimates of 0.5–10m2/s in the parametrization offriction in rivers. Following the method proposedby Ianniello (1979) to estimate the width of thechannel that makes horizontal friction importantfor the momentum balance, we used an estimate ofAh ¼ 1m2/s, according to the velocities andhorizontal lengths observed in this system.

The absolute value of the ratios/vðqv=qyÞS=/fuS; /Azðq

2v=qz2ÞS=/fuS and/Ahðq

2v=qy2ÞS=/fuS is shown in Fig. 11. Theseratios showed that the non-linear accelerations,vertical friction term and horizontal friction termwere comparable to Coriolis accelerations duringthe sampling period, as most of the magnitudes ofthe sectional means were about 1. The distribu-tions showed that values >1 were located aroundthe bank and close to the sidewalls and bottom,where advective and frictional terms were expectedto be larger than the Coriolis term. There was alsoa thin layer of values >1 at about 50m depthacross the entire fjord, which may be explained bythe position of the zero velocity in Fig. 4, wherethe Coriolis term tends to zero.

Analogously, the along-fjord momentum bal-ance appears to undergo transition from a balancebetween pressure gradient and advection duringcalm conditions, reminiscent of a Bernoulli-typeflow, to a balance that also includes friction during

17

1618

Fig. 9. Horizontal across-fjord salinity distribution from stations 4a, 6 and 5a plotted in time. Dotted lines correspond to the ship

track. Wind started blowing steadily at 17:30 h and caused upwelling on the southern side.


wind events. The terms in the along-fjord momen-tum balance may be given by

/vðqu=qyÞS�/fvS ¼ �/1=rðqp=qxÞS

þ/Azðq2u=qz2ÞSþ/Ahðq

2u=qy2ÞS: ð2Þ

Again, with the exception of the pressure gradientterm, these were the terms that could be reliablyestimated from the measurements. The ratios ofthe advective term, vertical friction term andhorizontal friction term to Coriolis term (notplotted here) showed absolute values of the

Fig. 10. Across-fjord vertical salinity sections at different times (looking into the fjord). Persistent southwesterly winds starting after

repetition 3 induced upwelling, as reflected by the progressive rise of the isohalines on the southern side. Time corresponds to the

beginning of each repetition. Maximum depth plotted is 20m, which was where the most appreciable changes appeared.


sectional means of 1.4, 2.6, and 1.3, respectively.This revealed that, in the along-fjord momentumbalance, vertical and horizontal friction were alsoas important as advection.

The contribution of advection to the momentumbalance is supported by the distribution of the

tidal current amplitudes. The greatest values wereobserved over the shallow areas as expected from aBernoulli-type balance. In other systems, wherebottom friction is influential, the transversedistribution of tidal current amplitudes followsthe bathymetry with greatest values appearing in

Fig. 11. Ratios of absolute values of the tidally averaged advective term, vertical friction term and horizontal friction term to the

tidally averaged Coriolis acceleration. Darker tones denote values below 1 and white denotes values over 1. The absolute value of the

sectional mean is shown at the upper right corner of each panel. The advective, vertical friction and horizontal friction terms were

comparable to Coriolis accelerations (contours shown: 0.5, 1, 5, 10, 15, etc.). Data in the shading region near bottom were not

considered in the analysis due to side lobe effects. White region near bottom was out of the profiling range of the ADCP.


the region where the depths are greatest (Li andValle-Levinson, 1999).

With respect to the frictional influences, furtherinvestigations are required to reveal the mainfrequencies of the wind stress in the region. If astrong diurnal (24 h) period is found, as observedby Pickard and Rodgers (1959), Farmer (1976)and Farmer and Osborn (1976) in fjords of BritishColumbia, the system could be affected by high-frequency (>1 cycle/day) fluctuations in theacross-fjord momentum balance.

In summary, wind forcing seemed to be themain driving mechanism of the mean flow duringthe sampling period as the mean flow was up-fjordnear the surface (below 8m depth). The bankhad salient effects on the across-channel distribu-tion of subtidal flow by shifting the location ofstrongest near-surface inflow and strongest mid-depth outflow to the channels, by masking thethree layer vertical structure, by shifting the flowdirection away from the bank, by inducingrecirculation at its top, and by causing a bifurca-tion of the flow below the pycnocline. Evidencesof wind-induced upwelling observed during thesampling period indicated that the across-fjordmomentum balance likely switches from quasi-geostrophic during calm winds to a balance that inthe surface is influenced by non-linear effects(from wind stress and advection) during windepisodes, and remains quasigeostrophic in theinterior.

Acknowledgements

We thank the Comit!e Oceanogr!afico Nacional,CONA (National Oceanographic Committee) ofChile for supporting this project within the multi-disciplinary research program Cimar-Fiordo 4 tostudy fjords of southern Chile. We also thankthe Servicio Hidrogr!afico y Oceanogr!afico de laArmada (Hydrographic and Oceanographic Ser-vice of the Chilean Navy) for coordinating thelogistics of the field work and for renting the‘‘Cordillera I’’ boat. The Gobernaci !on Mar!ıtimade Ays!en (Maritime Office of Aysen) provided theChilean Navy patrol craft ‘‘Rano Kau’’ to makethe field measurements. They allowed us the use of

three ship days onboard this boat. We are in debtto Lt. Nelson Saavedra and the crew of this boatfor supporting the experiment. The ServicioMeteorol !ogico de la Armada (MeteorologicalService of the Chilean Navy) provided themeteorological data for the general description ofthe region. The European Research Satellite (ERS)image was taken from a ground station based onthe Antarctic continent and provided by theEuropean Space Agency under a contract withthe Instituto Ant!artico Chileno (Chilean AntarcticInstitute).

References

Cameron, W.M., 1951. On the transverse forces in a British

Columbia Inlet. Transactions of the Royal Society of

Canada 45, 1–9.

Csanady, G.T., 1975. Lateral momentum flux in boundary

currents. Journal of Physical Oceanography 5, 705–717.

Dyer, K., 1997. Estuaries, A Physical Introduction, 2nd

Edition. Wiley, New York, 195pp.

Elliot, A., 1974. Observations of the meteorologically induced

circulation in the Potomac Estuary. Estuarine and Coastal

Marine Sciences 6, 285–299.

Farmer, D.M., 1976. The influence of wind on the surface layer

of a stratified inlet: Part II. Analysis. Journal of Physical

Oceanography 6, 941–952.

Farmer, D.M., Armi, L., 1999. The generation and trapping of

solitary waves over topography. Sience 283, 188–190.

Farmer, D.M., Freeland, H.J., 1983. The physical oceanogra-

phy of fjords. Progress in Oceanography 12, 147–219.

Farmer, D.M., Osborn, T.R., 1976. The influence of wind on

the surface layer of a stratified inlet: Part I. Observations.

Journal of Physical Oceanography 6, 931–940.

Fierro, J., Belmar, J., Rosales, S., Blanco, J., 1999.

Caracterizaci !on del r!egimen de mareas en canales interiores

en las vecindades de la constricci !on de Meninea. Res !umenes

Ampliados Crucero Cimar-Fiordo 4. Comit!e Oceanogr!afico

Nacional, Chile, 76pp.

Freeland, H., Farmer, D., Levings, C., 1980. Fjord Oceano-

graphy. NATO Conference Series, Vol. 4. Plenum Press,

New York, 115pp.

Gade, H.G., 1973. Deep water exchanges in a sill fjord: a

stochastic process. Journal of Physical Oceanography 3,

213–219.

Gade, H.G., Edwards, A., 1980. Deep-water renewal in fjords.

In: Freeland, H., Farmer, D., Levings, C. (Eds.), Fjord

Oceanography. NATO Conference Series, Vol. 4, Plenum

Press, New York, pp. 453–489.

Ianniello, J., 1979. Tidally induced residual currents in estuaries

of variable breadth and depth. Journal of Physical



Jones, B., Elliot, A.J., 1996. The parameterization of lateral and

bottom friction in depth integrated numerical models.

Journal of Marine Environmental Engineering 3, 1–13.

Joyce, T., 1989. On in situ ‘‘calibration’’ of shipboard ADCPs.

Journal of Atmopheric and Oceanic Technology 6, 169–172.

Klinck, J., O’Brien, J., Svendsen, H., 1981. A simple model of

fjord and coastal circulation interaction. Journal of Physical


Krauss, W., Brugge, B., 1991. Wind-produced water exchange

between the deep basins of the Baltic Sea. Journal of

Physical Oceanography 21, 373–384.

Li, C., Valle-Levinson, A., 1999. A two dimensional analytic

model for a narrow estuary of arbitrary lateral depth

variation: the intratidal motion. Journal of Geophysical

Research 104 (C10), 23525–23543.

Lwiza, K.M.M., Bowers, D.G., Simpson, J.H., 1991. Residual

and tidal flow at a tidal mixing front in the North Sea.

Continental Shelf Research 11 (11), 1379–1395.

Mu *noz, P., Avaria, S., Sievers, H., 1992. Presencia de

dinoflagelados t !oxicos del g!enero Dinophysis en el Seno

Ays!en, Chile. Revista de Biolog!ıa Marina 27 (2), 187–212.

Niemeyer, H., Cereceda, P., 1984. Hidrograf!ıa. In: Geograf!ıa de

Chile. Vol. 8. Instituto Geogr!afico Militar, 320pp.

Pickard, G., Rodgers, K., 1959. Current measurements in

Knight Inlet, British Columbia. Journal of the Fisheries

Research Board of Canada 16, 635–678.

SHOA, 1993. Nautical chart 8290. Hydrographic and Oceano-

graphic Service of Chilean Navy, SHOA.

Silva, N., Sievers, H., Prado, R., 1995. Caracter!ısticas

oceanogr!aficas y una proposici!on de circulaci!on para

algunos canales de la costa de Chile entre 411200S y

461400S. Revista de Biolog!ıa Marina (Valpara!ıso) 30 (2),

207–254.

Silva, N., Calvete, C., Sievers, H., 1997. Caracter!ısticas

oceanogr!aficas f!ısicas y qu!ımicas de canales australes

chilenos entre Puerto Montt y Laguna San Rafael, (Crucero

Cimar-Fiordo 1). Ciencia y Tecnolog!ıa del Mar 20, 23–106.

Stigebrandt, A., 1977. On the effect of barotropic current

fluctuations on the two-layer transport capacity of a

constriction. Journal of Physical Oceanography 7, 118–122.

Svendsen, H., 1980. Exchange processes above sill level between

fjords and coastal water. In: Freeland, H.J., Farmer, D.M.,

Levings, C.D. (Eds.), Fjord oceanography. NATO Con-

ference Series 4: Marine Science. Plenum Press, New York,

pp. 355–361.

Svendsen, H., Thompson, R., 1978. Wind driven circulation in

a fjord. Journal of Physical Oceanography 8, 703–712.

Tee, K.T., 1976. Tide-induced residual current, a 2-D nonlinear

numerical tidal model. Journal of Marine Research 34,

603–628.

Tully, J.P., 1958. On structure, entrainment and transport in

estuarine embayments. Journal of Marine Research 17,

523–535.

Valle-Levinson, A., Atkinson, L., 1999. Spatial gradients

in the flow over an estuarine channel. Estuaries 8 (2a),

179–193.


Transverse variability of ﬂow and density in a …arnoldo/ftp/papers/caceres...Continental Shelf...

Documents

Transcript of Transverse variability of ﬂow and density in a …arnoldo/ftp/papers/caceres...Continental Shelf...