Hydrography and frontogenesis in a glacial fjord off the Strait of Magellan

Transcription

1 Ocean Dynamics DOI /s Arnoldo Valle-Levinson. José Luis Blanco. Máximo Frangópulos Hydrography and frontogenesis in a glacial fjord off the Strait of Magellan Received: 7 March 2005 / Accepted: 4 November 2005 # Springer-Verlag 2005 Abstract Current velocity and hydrographic profiles obtained for the first time in a Chilean glacial fjord were combined with under-way surface temperature and salinity measurements to describe the formation of tidal intrusion fronts and plume-like fronts. These fronts formed within several hundred meters from each other in the vicinity of a shallow sill, maximum depth of approximately 3 m, in a glacial fjord off the Strait of Magellan in the Chilean Patagonia. Measurements were obtained in mid-december of 2003 and 2004, during late austral spring, under active glacier melting and calving. The glacial fjord is approximately 18 km long from the face of the glacier to the connection with the Strait of Magellan and typically less than 1 km wide throughout the system. Between the glacier face and the 3-m sill, depths are typically less than 100 m, and seaward of the sill, depths increase to more than 200 m. Velocity and salinity data obtained during flood periods revealed that water with oceanic salinity was aspirated to near-surface levels from depths of approximately 30 m as flood flows accelerated from approximately 10 cm s 1, seaward of the sill, to approximately 60 cm s 1 at the sill crest. The upwelled water was then slightly diluted by mixing at the sill crest before plunging down to the basin Responsible Editor: Paulo Salles A. Valle-Levinson (*) Civil and Coastal Engineering Department, University of Florida, 365 Weil Hall, Gainesville, FL , USA arnoldo@ufl.edu Tel.: Fax: J. Blanco Code 614, Observational Science Branch, NASA Wallops Flight Facility, Rm. E226, Bldg. N-159, Wallops Island, VA 23337, USA M. Frangópulos Centro de Estudios del Cuaternario Fuego-Patagonia (CEQUA), Instituto de Fomento Pesquero, Punta Arenas, Chile between the glacier and the sill. The plunging of salty water over the sill created dramatic tidal intrusion fronts only a few tens of meters from the sill crest and pumping of salt with every flood period. During ebb periods, the low salinity waters derived from the glacier and a small river near the glacier converged at the sill crest. After some mixing, the buoyant waters were released within a thin layer ( 3 m deep) lead by a plume-like front that remained coherent for a few hundred meters seaward of the sill. The main findings of this study were that tidal intrusion and plume fronts were observed within 2 km from each other, and that tidal pumping was the predominant mechanism for salt fluxes into the system. Introduction The hydrography and frontogenesis of glacial fjords depend mainly on freshwater contributions arising from melting at the face of the glacier. The source of freshwater appears at the bottom of the basin and gives rise to intense mixing near the glacier terminus (or face). This freshwater source contrasts that of fjords where riverine freshwater enters at the surface. Some glacial fjords feature a very shallow sill at a previous position of the glacier s moraine. The general hydrography of glacial fjords has been relatively well documented in northern hemisphere systems (e.g., Matthews and Quinlan 1975; Cowan 1992). However, little is known about the hydrographic characteristics in Patagonian glacial fjords. The purpose of this study is to document hydrographic distributions and frontogenesis in a glacial fjord off the Strait of Magellan in southern Chile (Fig. 1) throughout complete tidal cycles. This could be the first time that any type of hydrographic measurement is carried out in a southern hemisphere glacial fjord. Special attention was paid to the effects on hydrography produced by a very shallow (<3 m deep) sill interacting with a stratified flow. Navigation constraints limited the investigation of the sill effects on the landward side of the sill to only one semidiurnal experiment. The most striking features of the experiment were the generation of tidal intrusion and plume-like fronts in close

2 Fig. 1 Study area and sampling tracks. a Study area in the Strait of Magellan, South America. b Tip of South America showing the Strait of Magellan (broken line), and the arrow points to Seno Ballena. c Seno Ballena with the tracks for under-way measurements. The white tracks denote 24-h sampling on December 11 12, 2003; red tracks represent 12-h sampling on December 16, 2004; and the orange track shows a survey on December 15, The bottom profile recorded on the survey of December 15, 2004 is shown in d proximity to the sill, recirculations produced by tidal flows colliding with the sill, and plunging of salty waters during floods. The latter gave rise to predominance of tidal pumping in the salt flux over the sill. Tidal intrusion fronts and plume-like fronts are considered to be formed by an analogous physical mechanism, namely, the density gradient enhancement in the presence of flow convergence and sinking (Pelegri 1988; O Donnell 1993). These types of fronts have been extensively studied in temperate estuaries of the northern hemisphere. For example, Marmorino and Trump (1996) found vertical velocities of order 10 cm s 1, convergence rates of order 10 1 s 1, length scales of a few meters, and propagation consistent with gravity current theory associated with tidal intrusion fronts. Similar characteristics were observed in plume fronts by Garvine (1974), O Donnell (1997) and O Donnell et al. (1998). All these studies have concentrated on either type of these fronts because of the relatively long distance that may exist between them if they appear associated to the same estuary. For instance, the Chesapeake and Delaware Bay plume fronts form outside the estuary (Boicourt et al. 1987; Sanders and Garvine 1996), most ubiquitously during ebb. During floods, tidal intrusion fronts may be appreciable inside the estuary (e.g., Marmorino et al. 1999) or in the tributaries (Brubaker and Simpson 1999), more than 10 km away from the plume fronts. Therefore, it is difficult to observe them in consecutive tidal cycles because of the long tidal excursions in those systems. In typical temperate estuaries, these frontal features could contribute to salt fluxes through the tidal pumping mechanism because they are usually tied to certain tidal phases, e.g., plume fronts to ebb and tidal intrusion fronts to floods. However, the interaction between gravitational circulation and the mean salinity field is the dominant mechanism for salt flux, with tidal pumping being of secondary but nonnegligible importance (e.g., Geyer and Nepf 1996). In this study of a glacial fjord off the Strait of Magellan, both tidal intrusion fronts and plume-like fronts are observed within a few kilometers from each other. In this system, tidal pumping is the predominant process responsible for salt flux over a shallow sill. These characteristics probably distinguish this glacial fjord relative to other estuarine systems studied to date. Study area Seno Ballena is a Patagonian glacial fjord located off the Strait of Magellan in southern Chile (Fig. 1) isolated from direct anthropogenic influence. Therefore, it represents as pristine environmental conditions as one may encounter in

3 our planet. The fjord is approximately 15 km long and typically less than 1 km wide. It is characterized by a glacier at the head of the fjord and a tall sill, less than 3 m water depth, located approximately 7 km from the glacier. Landward of the sill, typical depths are less than 100 m, although a small isolated basin shows depths more than 100 m (Fig. 1c). Seaward of the sill, depths increase rapidly and exceed 200 m over a large portion of the fjord all the way to its connection with the Strait of Magellan. The glacier calves during the spring and its ice is abundant on the basins landward of the sill. The presence of ice complicates navigation of those waters but more markedly during ebb stages, when the ice is released seaward. During spring and summer, melting of ice and snow from the nearby surrounding mountains, together with the calved ice from the glacier, provide a thin buoyant layer to the fjord. It is unclear what happens to these buoyancy sources during winter. Tides in this area are mixed with semidiurnal predominance and typical ranges of 2 m. The deep cross sections of the fjord channels and the relatively small tidal range translate into tidal currents of less than 10 cm s 1 in most parts of the fjord except over the sill, where they increase markedly. Winds in the region are predominantly from the west and frequently reach speeds more than 15 m s 1, hindering access to the study area. The data reported in this study were obtained under weak wind conditions, typically less than 5 m s 1. Until their collection, the waters from this system were uncharted, and therefore, the data represent novel information of their kind. Methodology Under-way observations of water velocity, surface temperature and salinity, and profile measurements of salinity and temperature were obtained for the first time in a Chilean glacial fjord. The purpose of the sampling plan was to explore the hydrodynamic processes associated with the high primary productivity of the system. These data provide the first glimpse at Seno Ballena, a glacial fjord off the Magellan Strait in the Chilean Patagonia (Fig. 1). Data were collected during weak wind conditions, at the transition from austral spring to summer, between December 11 and 13, 2003, and between December 14 and 16, Observations in 2003 had to concentrate on the seaward side of the sill because of the lack of knowledge on the bathymetric distribution of the area. Upon arriving at the study area, we realized that the draft of the M/V Chonos, which was used during that sampling period, was too deep to cross the shallow sill. Observations in 2004 were obtained across the shallow sill onboard the M/V Cabo Tamar I. Sampling consisted of a 4-km along-fjord transect, seaward of the sill, that was repeated for 24 h on December 11 12, 2003 (Fig. 1). A shorter along-fjord transect, approximately 0.8 km long, that crossed the sill was sampled during 12 h on December 16, This transect was part of an elaborate circuit that also investigated the lateral variability of flows on the landward side of the sill, but those data are not reported here. During these transect repetitions, under-way velocity profiles were obtained with a kHz acoustic Doppler current profiler (ADCP) mounted on a catamaran. Near-surface temperature and salinity were obtained with a conductivity-temperature recorder (Sea Bird SBE-37) attached to the catamaran. Hydrographic profiles were measured at each end of the along-fjord transects with a Sea Bird SBE-19 conductivitytemperature depth (CTD) recorder. Additionally, two alongfjord hydrographic sections consisting of five CTD casts each were carried out from near the glacier face to approximately 200 m seaward of the sill with a launch boat in December Another section consisting of eleven CTD casts was executed in December 2004 with the M/V Cabo Tamar I. Results The results are presented first in terms of the water properties indicated by the hydrographic variables obtained at the along-fjord sections and with all the near-surface temperature and salinity values obtained underway. Then, those water properties are placed in the context of sill processes and front formation with the time series of profiles obtained at the ends of each along-fjord ADCP transect. Near-surface velocity data are then represented in the space time domain, together with surface salinity data, to illustrate the timing and location of tidal intrusion and plume-like fronts in the vicinity of the sill. Finally, the vertical structure of these fronts, and of the velocity profiles at either side of the sill in general, is illustrated with the measurements obtained in the along-fjord transect that crossed the shallow sill. Along-fjord hydrography The CTD profiles measured from near the glacier face to seaward to the sill showed that the sill was effective in blocking and modifying saline waters (Fig. 2). The profiles obtained during an ebbing tide on December 12, 2003 showed that the profile on the seaward side of the sill (profile labeled as 5 on Fig. 2a) had slightly greater salinities than the other four profiles. The profiles on the landward side of the sill (profiles labeled 1 to 4 on Fig. 2a) were very similar among each other. The blocking effect was more notable below depths of approximately 6 m, where salinity values seaward of the sill were clearly different. The hydrographic differences on either side of the sill were even more appreciable during the flooding tide on December 13, In this case, the blocking effect was evident throughout the water column because there was a marked difference between the station seaward of the sill ( 5 on Fig. 2b) and those landward ( 1 to 4 on Fig. 2b). In general, salinity values below the pycnocline (>10 m depth) at any given station changed by less than 0.5 over depths of approximately 100 m. That is the reason for showing only the upper part of the water column. The along-fjord section obtained in December 2004 (Fig. 2c) also illustrated mixing in the region of the sill

4 Fig. 2 Water salinity, temperature, and density anomaly profiles at different locations along Seno Ballena on a December 12, 2003 (ebb), and b December 13, 2003 (flood). Only station 5 is seaward of the sill. c Contours of salinity temperature and density anomaly drawn from 11 CTD stations along the fjord on December 15, 2004 during flood where the upper layer of buoyant water becomes saltier and where the salty water is slightly diluted before moving landward over the sill. Note the rapid increase in salinity seaward of the sill relative to the region landward of the sill. Also note the limited 3-m thickness of the buoyant water that could easily go unnoticed, and the strong link between

5 pycnocline and halocline. Landward of the sill, water temperature increased with depth within the pycnocline. The same sign portrayed by temperature and salinity stratification in the upper water column may give rise to double-diffusive layering. This is a topic worth investigating but was not dealt with in this particular study. The hydrographic distributions shown in Fig. 2 suggested that the main mechanism of salt transport over the sill could have been through a trickle (salt pumping) that occurred only at flood tides. This salty water should quickly sink and contribute to the near-bottom homogeneous conditions landward of the sill. The hydrography also suggested that bottom waters landward of the sill should have a long residence time because of the relatively isolated basin more than 100 m deep (Fig. 1). Those bottom waters nonetheless receive new injections of salty water with every tidal flood and do not seem to be suboxic. Near-surface hydrography The near-surface temperature and salinity values obtained throughout the sampling efforts from 2003 and 2004 illustrated two main concepts, which were derived from T S diagrams (Fig. 3). The first concept referred to the large spatial variations in hydrography, as indicated by the wide range of salinity values and the large scatter in the T Splane. These large spatial variations should have been associated to ubiquitous frontal structures. The second concept was related to the convergence of values toward the highest salinity, within a tight range of temperature and salinity values. The highest salinity value of 31 was related to waters that were usually found below the pycnocline (Fig. 2) and that tended to be insulated from air sea exchange processes. However, the values shown in Fig. 3 were measured near the Fig. 3 Temperature salinity diagram of under-way near-surface values recorded with the conductivity-temperature sensor during different surveys carried out in Seno Ballena. Dotted contour lines represent density anomalies (kilograms per cubic meter) surface, which indicated that oceanic waters could occasionally be ventilated to the surface. Because the highest near-surface salinities were similar to the salinities typically observed below the pycnocline, the ventilation probably took place with little or no mixing. The likely mechanisms that caused this upwelling of below-pycnocline waters were internal tides, as shown in Fig. 4, and Bernoulli suction (Seim and Gregg 1997), as also explored in the flow observations across the sill. Noteworthy of the convergent T S values was the temperature increase in these values from 7.5 C in 2003 to 8.0 C in This observation might be supporting the warming tendency of subglacial waters and their link to glacial retreat in other high-latitude regions (e.g., Motyka et al. 2003). Intratidal variability of salinity Intratidal variations of salinity profiles at the ends of the along-fjord transect sampled on the seaward side of the sill showed high values (>28.5) and the influence of tide-induced pycnocline oscillations (Fig. 4). These salinity profiles were measured close to the northern coastline of the fjord (Fig. 1), where sill bathymetry practically breaks the surface and blocks the direct influence of low-salinity waters. The minimum salinity observed was 28.5, which was much higher than the values of 24 observed landward of the sill (Fig. 2c). The pycnocline showed asymmetric semidiurnal oscillations during the observation period. At the sampling site closest to the sill, the pycnocline dropped rapidly between December 11.5 and 11.7 (Fig. 4a) and raised slowly between December 11.7 and These oscillations indicated ventilation of subpycnocline waters (salinity >30) around December 12.2 close to the sill. Farthest from the sill, the pycnocline showed different variability represented by an overall rise throughout the sampling period (Fig. 4b). At both stations, the 30.8 isohaline exhibited a well-defined semidiurnal oscillation. The extreme values of salinity occurred at different times at each station and did not occur at exact periods of approximately 12 h. This suggested a distortion of the tidal oscillations by the sill, which will be the topic of future studies. The semidiurnal oscillation of the halocline was also shown at the seaward end of the transect that crossed the sill (Fig. 4c). This CTD station was placed in front of the deepest part of the sill and displayed the lowest salinities that could be observed seaward of the sill. The internal oscillation of the halocline showed high-salinity water being lifted to a shallower depth (<3 m) than the maximum depth of the sill on December 16.7 (Fig. 4c). As seen later, this water was lifted even closer to the surface at the sill crest. Landward of the sill, the halocline was much shallower, and the salinity was appreciably lower than landward of the sill (Fig. 4d). Also at this location (landward of the sill), there was no clear semidiurnal oscillation of the halocline, and the salinity field showed a buoyant intrusion at the end of the sampling period associated with ebbing flow. The salinity fields depicted in Fig. 4 suggest strong sill interactions between the tidal flow and density field.

6 Fig. 4 Intratidal variability of salinity profiles at the extremes of the white track in Fig. 1 (a Close to the sill and b farthest from the sill) and at the extremes of the red track in Fig. 1 (c seaward of the sill and d landward of the sill). Isohalines are drawn at intervals of 0.5 between 24 and 30. Above 30, isohalines are drawn at 0.1 intervals to better illustrate oscillations of salinity field Sill interactions between tidal flow and salinity The interactions of the tidal flow and density field were additionally explored with along-fjord track repetitions (white and red tracks on Fig. 1) and illustrated by nearsurface flow and salinity fields (Fig. 5, white track, and Fig. 6, red track). In the track sampled seaward of the sill, semidiurnal variations of salinity at the surface were well illustrated in the vicinity of the sill (within 2 km in Fig.5), consistent with Fig. 4a. The variations were asymmetric in time throughout most of the domain sampled. This means that the periods influenced by high-salinity waters were longer ( 0.3 day) than the periods affected by low salinities ( 0.2 day). Figure 5 illustrated high salinity values appearing close to the sill (<0.5 km) around December 11.8 and December Interestingly, soon after maximum flooding (December on Fig. 5), high-salinity water seems to have moved seaward (from left to right) over the extent of the track (4.3 km) in 0.2 day, i.e., at 0.25 m s 1. This could have represented the reflection of tidal energy that might have appeared again in the second tidal cycle (around 12.4) but was not completely resolved in the entire track. The measurements portrayed on Fig. 5 suggestedthenthatduring floods, high-salinity water was advected toward the sill, which greatly blocked the passage of this water. Accelerations produced by the passage of flow over the sill crest caused a low-pressure region that favored the aspiration of below-pycnocline waters to the surface (Seim and Gregg 1997). This mechanism, Bernoulli aspiration, should have provided a parcel of high-salinity water to the area of the sill. A portion of such aspirated salty water made it over the sill, and a portion could have been reflected seaward as suggested by Fig. 5. The along-fjord transect that crossed the sill (red track on Fig. 1) also illustrated the mechanism of Bernoulli aspiration in addition to two other phenomena: tidal intrusion fronts and the predominance of tidal pumping in the transport of salt over the sill (Fig. 6). Bernoulli aspiration was illustrated by two features: (1) the acceleration of tidal flows as they moved over the sill (between 0.2 and 0.4 km on Fig. 6a) and

7 Fig. 5 Near-surface salinity (color contours) and flow (vectors) in the space time domain at the along-fjord white track of Fig. 1 during 24 h of observations. The sill is to the left as is the source of freshwater to Seno Ballena. Salinity contours are drawn at 0.5 intervals. White dots denote location and time of observations (2) the maximum salinity was observed over the sill, centered on December 16.7 (as in Fig. 4c). The salinity values decreased seaward of the sill at the time of maximum surface salinity, which indicated the below-pycnocline origin of that high-salinity water. The aspiration of belowpycnocline waters to the surface during floods caused the sharp tidal intrusion front marked by a salinity contrast of approximately 2. This front was also centered on December Fig. 6 a Near-surface salinity (color contours) and flow (vectors) in the space time domain at the along-fjord red track of Fig. 1 during 12 h of observations. The sill is between 0.2 and 0.4 km. Salinity contours are drawn at 0.5 intervals. White dots denote location and time of observations. b Semidiurnal phase (in degrees) for alongfjord flow u (blue) and salinity (red). c Total salt flux us shown as continuous line, salt flux produced by mean flow u S as dashed line, and tidal pumping salt flux u S as dotted line

8 16.7, spanning from to 16.75, and extended from the sill crest at approximately 0.3 km to a distance of approximately 0.65 km. This tidal intrusion front triggered oscillations with wavelengths of order 100 m as seen in the wake of the front on December The oscillations were denoted by flood flow peaks associated with highsalinity pulses. In addition to the tidal intrusion front, Fig. 6a also illustrated a plume-like ebb front where salinity decreased by approximately 2 between 0 and 0.4 km on December These fronts were favored by large-phase shifts in the flow and semidiurnal salinity signals (Fig. 6b). The along-fjord flow u had approximately the same semidiurnal phase on either side of the sill, but the phase changed by close to 90, or 6 h, at the sill crest between 0.3 and 0.35 km. There was also a marked phase lag of in the semidiurnal salinity S signal at the sill crest. These phase lags gave rise to convergent flows and sharp salinity gradients in the vicinity of the sill. A notable aspect of this data set was the formation of both types of fronts, tidal intrusion and plume-like, in close proximity to each other. This was allowed by the weak tidal currents on either side of the sill, which translate into short tidal excursions. Finally, this particular data set illustrated the predominance of tidal pumping in effecting the salt transport over the sill. Figure 6c shows the mean salt transport us along the track (continuous line), where the angled brackets represent tidal averages. It also shows the salt transport carried out by the mean flow and salinity u S (dashed line), and the salt transport affected by tidal pumping u S, where primes denote tidal variations or deviations from temporal mean. To determine these fluxes, the mean salinity throughout space and time was subtracted from all salinity data. Figure 6c shows then that tidal pumping dominated the total transport in the region of the sill between 0.2 and 0.4 km. Away from the sill, the salinity transport was dominated by the mean flow, as is customary in most estuarine systems. Therefore, this is an example of a system where salinity transport is mainly dominated by tidal pumping, at least locally. Fig. 7 Plume-like fronts appearing during ebb stages of the tidal cycle, looking toward the southern coastline. a Frontogenesis within 100 m of the sill crest. The source of freshwater is to the right. b Front propagating seaward Frontogenesis The near surface properties of the track sampled seaward of the sill illustrated mostly plume-like (O Donnell 1993) fronts during ebb. These fronts should be expected to behave as gravity currents because of the large water-column depth, the weak tidal forcing, and the relatively large buoyancy input (Pelegri 1988). The plume-like fronts were released from the sill crest as the ebb flow became supercritical (>0.3 m s 1 ). Given the depth of the buoyant outflow (H 3 m) and the density contrast (Δρ/ρ ), the surface flow was critical when it equaled (g Δρ/ρ H) ½ or approximately 0.3 m s 1, where g is 9.8 m s 2. As the flow remained subcritical, low-salinity water, ice debris and other floatable organic matter accumulated at a line of large convergence parallel to the sill crest (Fig. 7a). With the evolution of ebb, the convergence line moved over the sill crest and was released seaward in a coherent structure that crossed the fjord (Fig. 7b). These plume-like features were appreciable between December 11.9 and December 12.0, within 0 and 1 km on Fig. 5, and, as mentioned above, around December on Fig. 6a. Analogously, tidal intrusion fronts were formed during floods in the vicinity of the sill as illustrated in the track that crossed the sill (Fig. 6a). During floods, the below-pycnocline high-salinity water that was lifted through Bernoulli aspiration collided against the buoyant water on the landward side of the sill and sank underneath. The tidal intrusion front featured typical cusps and a very sharp contrast in water color and roughness (Fig. 8a). Figure 6 showed that the front propagated landward at a rate of approximately 400 m in 0.1 of a day (December ) or 0.05 m s 1, and that it could cause internal oscillations behind it. These oscillations were not only observed in the surface salinity and flows but

9 Fig. 8 Tidal intrusion front appearing during flood flows. a Looking toward the southern shoreline with the source of freshwater to the right. b Internal oscillations manifested at the surface, in the lee of the tidal intrusion front. Looking seaward were also clearly seen in the sea surface roughness (Fig. 8b), where lines of different texture were delineated. It was clear that the phenomenon of salty water spilling over the sill crest with every flood, associated with the tidal intrusion fronts, was crucial for the landward transport of salt. This was further indicated by the residual flows measured seaward of the sill (not shown) that illustrated near-surface outflow throughout an approximately 20-m-thick layer and inflow underneath. This net inflow appeared well below the maximum depth of the sill and could not have transported salt over it. The vertical structure of the plume-like and tidal intrusion fronts was explored with the sections of flow and acoustic backscatter signal measured along the transect that crossed the sill (red track on Fig. 1). During ebb, the buoyant outflow intensified dramatically seaward of the sill, reaching supercritical values more than 0.3 m s 1, and remained in a surface layer less than 10 m deep (Fig. 9). The shape of that buoyant outflow seems to have been represented by the ADCP backscatter that showed a region with highest scatter emanating from the sill in the direction of the tidal flow. During floods, the flow once again accelerated at the sill crest and plunged downward along the bottom, emulating a gravity current (Fig. 10). The measurements did not fully resolve the gravity current but suggested that the floodrelated salty water plunged down to approximately m deep, where the salty waters must have arrived at their neutral buoyancy (Fig. 10). Fig. 9 Flows and acoustic backscatter during ebb stages of the tide illustrating plume-like fronts. Vertical flows have been exaggerated 40 times. Backscatter units are relative, but greater values indicate greater concentration of material in the water column

10 Fig. 10 Same as Fig. 9 but for flood stages illustrating the spilling of salty water as it flows landward over the sill Discussion and conclusion The first set of observations of hydrography and flow profiles in a glacial fjord off the Strait of Magellan illustrated intratidal variations that resulted from the interactions between a stratified flow and a very shallow sill. The sill practically blocked the mean landward flow of high salinity, which was pumped over the sill and toward the glacier only during flood flows. The mean inflow was too deep to be able to transport salt over the sill. Nonetheless, the acceleration of flood tidal flows in the area of the sill produced the conditions needed for pumping of below-pycnocline salt over the sill (Fig. 6b) by means of Bernoulli aspiration. These salty waters must have been pumped vertically from an aspiration depth, h a, of Seim and Gregg (1997), ( u/ x= w/ z) and assuming a vertical excursion z of 30 m associated with the aspiration of subpycnocline waters toward the surface, yields reasonable upward velocities (w) of 0.01 m s 1. The mechanism of Bernoulli suction is h a ¼ 0:5h s þð0:25h 2 s þ U 2 N 2 Þ 1=2 (1) where h s is the sill depth (3 m), N is the buoyancy frequency, and U is the flow speed at the sill (0.6 m s 1 ). The buoyancy frequency squared equals gδρ /(ρh), where H is the total water depth (60 m), g equals 9.8 m s 2, Δρ is 3 kg m 3,andρ is 1,020 kg m 3. Using these values yield an aspiration depth, h a, of 29 m, which was consistent with the below-pycnocline salinity values that were observed at the surface on December 16.7 (Fig. 6a). The horizontal divergence u/ x associated with the observed flow, which typically accelerated ( u) by approximately 0.3 m s 1 over 100 m ( x) as it moved over the sill, was s 1. Using two-dimensional continuity of mass Fig. 11 Schematic representation of the flow and density field during a tidal intrusion fronts that develop during flood, together with Bernoulli aspiration, and b plume-like fronts that develop during ebb. Both fronts cause large accumulations of suspended material

11 supposed to arise from the advective accelerations balancing the pressure gradient. Such mechanism can be verified by comparing the order of magnitude of the advective acceleration term associated with flow divergence, i.e., u u/ x, against the frictional term arising from bottom stresses, i.e., C d u 2 /H. Taking the nondimensional bottom drag coefficient C d of ,thedepthH over the sill of 3 m and the flow over the sill u of 0.3 m s 1 yields a bottom friction of ms 2. In comparison, advective accelerations were typically ms 2 or one order of magnitude larger than friction. This scaling indicated the likelihood for Bernoulli dynamics causing the flow of salty water over the sill. Friction, however, may contribute to the dynamical balance during the periods of strongest flows ( m s 1 )overthe sill. Also, during flood flows, some of the salty water pumped to the surface by Bernoulli aspiration should have been recirculated because not all of the volume could make it through the sill gap. This was illustrated as a reflected salty signal in Fig. 5 and might represent one of the few observational examples of tidal reflection at a fjord sill. Recirculations of flow over a sill have already been documented in other systems (e.g., Klymak and Gregg 2001; Cáceres and Valle-Levinson 2004). The flood processes described above are illustrated schematically in Fig. 11. Noteworthy was the fact that both tidal intrusion fronts and plume-like fronts were observed within a close distance from each other (Fig. 11). This was possible because the large-phase lags in the semidiurnal tidal signals (Fig. 6b), in both salinity and flow, at the sill were illustrative of flow convergences and sharp salinity gradients. This was also possible because of the short tidal excursions at both sides of the sill, where semidiurnal tidal currents had amplitudes of approximately 0.05 m s 1. The tidal excursion associated with a current amplitude of u 0 and period T is given by u 0 T/π or 0.7 km for the study area. It was clear that sill processes such as divergences, Bernoulli aspiration, and tidal distortions were crucial for the frontogenesis during both ebb and flood tidal stages. In conclusion, the two most relevant findings of this study are the (1) identification of the tidal pumping mechanism as predominant for salt transport over the sill and water renewal landward of the sill, and (2) description of tidal intrusion fronts and plume-like fronts in the close vicinity of the sill (Fig. 11). It is clear that the interaction among the very shallow sill, the tidal flow, and the very shallow stratification induced by glacier waters was fundamental in favoring the development of fronts and the tidal pumping mechanism. Acknowledgements This study was funded by the Centro de Estudios del Cuaternario Fuego-Patagonia (CEQUA), financed by the Chilean Government. A.V.L. acknowledges support from National Science Foundation (NSF) project no References Boicourt WC, Chao S-Y, Ducklow HW, Glibert PM, Malone TC, Roman M, Sanford LP, Fuhrman J, Garside C, Garvine R (1987) Physics and microbial ecology of a buoyant estuarine plume on the continental shelf. EOS 69: Brubaker JM, Simpson JH (1999) Flow convergence and stability at a tidal estuarine front: acoustic Doppler current observations. J Geophys Res 104(C8): Cáceres M, Valle-Levinson A (2004) Transverse variability of flow on both sides of a sill/contraction combination in a fjord-like inlet of southern Chile. Estuar Coast Shelf Sci 60: Cowan EA (1992) Meltwater and tidal currents: controls on circulation in a small glacial fjord. Estuar Coast Shelf Sci 34: Garvine RW (1974) Physical features of the Connecticut River outflow during high discharge. J Geophys Res 79: Geyer WR, Nepf HM (1996) Tidal pumping of salt in a moderately stratified estuary. In: Aubrey DG, Friedrichs CT (eds) Coastal and estuarine studies, vol 53. Buoyancy effects on coastal and estuarine dynamics. American Geophysical Union, Washington, DC, pp Klymak JM, Gregg MC (2001) Three-dimensional nature of flow near a sill. J Geophys Res 106(C10): Marmorino GO, Trump CL (1996) High-resolution measurements made across a tidal intrusion front. J Geophys Res 101(C11): Marmorino GO, Trump CL, Trizna DB (1999) Preliminary observation of a tidal intrusion front inside the mouth of the Chesapeake Bay. Estuaries 22(1): Matthews JV, Quinlan AV (1975) Seasonal characteristics of water masses in Muir Inlet, a fjord with tidewater glaciers. J Fish Res Board Can 32: Motyka RJ, Hunter L, Echelmeyer K, Connor C (2003) Submarine melting at the terminus of a temperate tidewater glacier, LeConte Glacier, Alaska. Ann Glaciol 36:57 65 O Donnell J (1993) Surface fronts in estuaries: a review. Estuaries 16:12 39 O Donnell J (1997) Observations of near-surface currents and hydrography in the Connecticut River plume with the surface current and density array. J Geophys Res 102(C11): O Donnell J, Marmorino GO, Trump C (1998) Convergence and downwelling at a river plume front. J Phys Oceanogr 28: Pelegri JL (1988) Tidal fronts in estuaries. Estuar Coast Shelf Sci 27:45 60 Sanders TM, Garvine RW (1996) Frontal observations of the Delaware coastal current source region. Cont Shelf Res 16(8): Seim HE, Gregg MC (1997) The importance of aspiration and channel curvature in producing strong vertical mixing over a sill. J Geophys Res 102(C2):