Water mass analysis and alongshore variation in upwelling intensity in the eastern Great Australian Bight

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 111,, doi: /2004jc002699, 2006 Water mass analysis and alongshore variation in upwelling intensity in the eastern Great Australian Bight Sam McClatchie, 1 John F. Middleton, 2 and Tim M. Ward 1 Received 3 September 2004; revised 31 January 2006; accepted 6 April 2006; published 5 August [1] A study of climatological and conductivity-temperature-depth (CTD) data for 2004 is made to provide a conceptual model of upwelling for the eastern region of the Great Australian Bight. In particular, the data and other studies provide strong evidence that shelf break upwelling is confined to the southwest Kangaroo Island region and does not occur farther to the west off the Eyre Peninsula. Rather, the upwelled water is likely to remain in a Kangaroo Island pool until subsequent upwelling events draw the water to the shallower and surface coastal regions of the eastern Bight. In this manner the surface upwelling apparent off the Bonney Coast, Kangaroo Island, and the eastern Great Australian Bight (GAB) can appear to be simultaneous. Moreover, it appears likely that the water within the Kangaroo Island pool remains nutrient rich. Support for this model comes from CTD sections collected in 2004 that show that the upwelled signal (cool, <17 C; fresher, <35.6; dense, s t >26kgm 3 ) diminishes in width and intensity with increasing distance from Kangaroo Island. The pattern of fluorescence is similar to that for temperature in the upwelled plume and indicates that the Kangaroo Island pool remains nutrient rich. Relatively low oxygen concentrations may indicate a previous bloom. The warmest water is found near the shelf break along with very low values of fluorescence and relatively higher levels of oxygen suggesting nutrient-limited growth of phytoplankton. These data also support the notion that the upwelled nutrient-rich water is supplied from the Kangaroo Island pool and not by shelf break upwelling in the eastern GAB. Anomalously salty and fresh water is identified as resulting from evaporation in coastal bays and groundwater aquifer discharge. Citation: McClatchie, S., J. F. Middleton, and T. M. Ward (2006), Water mass analysis and alongshore variation in upwelling intensity in the eastern Great Australian Bight, J. Geophys. Res., 111,, doi: /2004jc Introduction [2] The eastern Great Australian Bight (GAB) supports Australia s largest finfish fishery, based on sardines (Sardinops sagax), known locally as pilchards. The South Australian sardine fishery and the surveys required to monitor these stocks have provided us with the conductivitytemperature-depth (CTD) data that we have used to study the water masses and upwelling in the eastern GAB. [3] South Australian oceanography spans a region with shelf widths varying from 30 km to 220 km, a shelf break with some very notable canyons, and two large Gulfs, of which the Spencer Gulf is described as an inverse estuary (evaporation exceeds precipitation). The narrow southeast shelf underlies a large, summertime upwelling plume (called the Bonney upwelling) that extends to the northwest as a 1 South Australian Aquatic Sciences Centre, SARDI Aquatic Sciences, Henley Beach, South Australia, Australia. 2 School of Mathematics, University of New South Wales, Sydney, New South Wales, Australia. Copyright 2006 by the American Geophysical Union /06/2004JC series of discrete upwelling centers off western Kangaroo Island [Kitani, 1977] and along the Eyre Peninsula [CSIRO Marine and Atmospheric Research, 2001; Kampf et al., 2004] (see Figure 1). [4] Early work on upwelling off South Australia focused on the narrow shelf area off Robe (see Figure 1), southeast of the two gulfs, where upwelling of sub-antarctic water (following terminology of Newell [1974]) comes to the surface in the summer months, forming the Bonney upwelling [Rochford, 1977; Lewis, 1981; Bye, 1983; Provis and Lennon, 1981]. Rochford [1977] measured moderate nitrate enrichment in surface waters off Port MacDonnell (Figure 1) and speculated that Ekman flux driven by southeast winds (upwelling favorable in the southern hemisphere) contributed to upwelling in the area. Using a more extensive set of cross-shelf transects, and 5 years of monthly samples of temperature, salinity, nitrate and silicate at Port MacDonnell, Lewis [1981] measured nitrate concentrations 6X higher in the lower layer of sub-antarctic water compared to the upper layer of subtropical water in summer. He described three summer upwelling centers where sub- Antarctic water reaches the surface in the Bonney upwelling (southeast of Port MacDonnell, south of Southend, and 1of13

2 Figure 1. Map of the survey area showing stations sampled during the March 2004 sardine survey. Locations referred to in the text are marked. A CTD profile was collected at each of the stations (marked by red dots). south of Robe, see Figure 1). Lewis [1981] made a qualitative link between upwelling intensity and the occurrence of southeast winds, and speculated that the canyons at the shelf edge might play a role in focussing the upwelling. [5] Schahinger [1997] used time series measurements of wind stress, currents from 3 current meters, and bottom temperatures combined with CTD sections to examine the dynamics of the Bonney upwelling. He calculated the internal Rossby radius to give the subsurface scale of thermocline uplift, discussed the strengthening effect of the daily sea breeze, and examined the timing and duration of two upwelling events. He related cross shelf gradients in density to vertical differences in alongshore velocity using the thermal wind equation, and estimated the thickness of the surface Ekman layer and mixed layer which was used to compare the calculated mean flows with the measured flows. Schahinger [1997] speculated that the spatial scale of the upwelling was O(20 km), and agreed with Lewis [1981] that the upwelling centers were localized, while the spatial effect of the upwelling was broadened by advection. [6] Later modeling work [Wenju et al., 1990] simulated the onset of upwelling along the Bonney coast and tested the effects of wind direction, bottom topography and interfacial stress. Their three layer and two layer models were able to simulate wind-driven upwelling, and predicted uplift of interfaces beginning a day after the onset of favorable winds. The effect of bottom topography rather than coastal curvature was shown to be significant, and the most favorable areas for upwelling were in an O(20 30 km) wide band adjacent to the coast between Portland and Cape Jaffa, near Robe (see Figure 1), as well as on the western edge of Kangaroo Island [Wenju et al., 1990]. The model also predicted a westward flowing jet of up to 70 cm s 1 adjacent to the coast. CSIRO Marine and Atmospheric Research [2001] made extensive use of remote sensing imagery of the Southeast region, clearly showing the Bonney upwelling, as well as upwelling off western Kangaroo Island and Coffin Bay Peninsula (see Figure 1) between December and March. [7] The eastern GAB along the western coast of the Eyre Peninsula (Figure 1) is also an upwelling region [Middleton and Platov, 2003; Middleton and Cirano, 2002; Baird, 2003; Kampf et al., 2004] and it is assumed that the upwelling-driven enhancement of primary and secondary production supports the large pelagic fish resource. Exactly how the enrichment occurs, its spatial and temporal scales, and its variability are poorly understood, although the seasonal shelf circulation has been modeled [Middleton and Platov, 2003; Middleton and Cirano, 2002]. Recent work showed that there are on average two to three winddriven upwelling events during the austral summer [Kampf et al., 2004]. These events are associated with southeasterly winds, that prevail between December and April [Bye, 1983]. The coastal upwelling was shown to produce surface phytoplankton patches within a week of the upwelling event [Kampf et al., 2004], that may sink and form the observed subsurface chlorophyll-a maxima [Kampf et al., 2004]. 2of13

3 [8] Two-dimensional modeling studies predict that upwelling in the eastern GAB is strongly influenced by bathymetry [Baird, 2003]. To the east, where the shelf is narrow and bathymetry is steep, modeling suggests that upwelling occurs within 10 km of the coast, raising cool water and nutrients to the surface, where phytoplankton concentrations then develop. Zooplankton patches lag the development of phytoplankton concentrations, and generally occur downstream [Baird, 2003]. In contrast, to the west where the shelf is broad and relatively shallow out to 200 km offshore, modeling predicts that upwelling of cooler water will occur more slowly and be located at the shelf break rather than inshore. Here the model suggests that nutrients are utilized by phytoplankton before reaching the surface, a subsurface phytoplankton peak develops, and the phytoplankton and zooplankton patches are spatially and temporally colocated [Baird, 2003]. For the middle to western end of the GAB, the study of Middleton and Platov [2003] indicates that downwelling and not upwelling should occur along the shelf break. The downwelling results from convergence of deep ocean and shelf slope Sverdrup transports that are both driven by the generally positive wind stress curl. [9] In the region separating the Bonney coast upwelling and the western Eyre Peninsula upwellings, Hahn [1986] conducted an extensive seasonal study at the mouth of the Spencer Gulf by repeating CTD transects in the axis, and across the mouth, of the Spencer Gulf combined with a moored current meter and thermistor array on the shelf. His work showed the development of strong stratification on the shelf in summer, outflow of high-salinity water from the Spencer Gulf in the autumn, and well mixed water on the shelf and in the mouth of the Gulf in the winter. The inverse estuary characteristics of the Spencer Gulf and the high-salinity autumn outflow into shelf waters were described by Lennon et al. [1987] and Nunes Vaz et al. [1990]. The interface between Spencer Gulf and shelf water creates a front in summer [Hahn, 1986] where larval fish have been shown to aggregate [Bruce and Short, 1990], possibly as a result of convergent flows, although convergence was not demonstrated. [10] The only new data currently available for examining spatial variability of upwelling at seasonal scales are bottom (60 m) temperature records, or remote sensing advanced very high resolution radiometer and ocean color imagery of surface features. Quasi-synoptic spatial scales and alongshore variability of upwelling on the other hand, can also now be determined from recent fisheries surveys that provide extensive areal and vertical coverage of CTD profiles. [11] We address the descriptive physical oceanography of the eastern Great Australian Bight (GAB) (Figure 1), which is a poorly sampled region. In section 2 we describe the satellite and CTD data. In section 3, a conceptual model is first developed on the basis of climatological data, the satellite images and numerical results. The model is then used to interpret the CTD profiles, which in turn provide support for the conceptual model. Figure 2. Sea surface temperature imagery from the MODIS-Aqua sensor showing daily average nighttime temperature for (top) 29 February and (bottom) 10 March Temperatures in C are referenced in the color bar. 2. Methods [12] Sea surface temperature (SST) imagery was obtained from the Moderate Resolution Imaging Spectroradiometer (MODIS)-Aqua sensor. Relatively cloud-free, spatially subset ( E and S) images between 15 February and 30 March spanning the survey period (16 26 March 2004) were inspected to select 2 images, separated by 10 days (29 February and 10 March 2004). [13] CTD data were collected during the annual sardine survey between 16 and 26 March 2004 [Ward et al., 2004] (Figure 1). At each station, a vertical profile was obtained by lowering a Sea-Bird 19 plus SEACAT conductivitytemperature-depth profiler fitted with sensors for dissolved oxygen, turbidity, fluorescence and irradiance. The profiler was lowered to a depth of 70 m, or to 10 m from the bottom in waters less than 80 m deep. Profiles were analyzed with Ocean Data View [Schlitzer, 2003]. The instrument and sensors were new and we applied factory calibrations to the data. Fluorescence was not converted to chlorophyll-a because subsurface water samples were not available. Calibration of fluorescence with inadequate extracted chlorophyll-a samples to resolve the effects of species composition, light acclimation and surface quenching by bright light would add additional variability to the data rather than providing reliable estimates of chlorophyll-a. Oxygen concentrations 3of13

4 Figure 3. Climatology of the CARS summer (December January) temperature as interpolated onto a bottom topography for the region. Only temperatures between 13 and 16 C have been color-contoured for clarity. Dark lines correspond to the 100 m and 200 m isobaths. Kangaroo Island pool of nutrient rich, cold water is indicated. A maximum advective scale (25 cm s 1 over 10 days) for the alongshore transport of this water is indicated. relied on factory calibration of the new sensor, and were not empirically calibrated against water samples. 3. Results and Discussion 3.1. Remote Sensing Imagery [14] The daily average surface temperature off the western Eyre Peninsula on 29 February 2004 (Figure 2) clearly shows cooler water along the coast. This upwelling signal extended from Williams Is. to the Head of the Bight (see Figure 1 for place names). The SST image shows obvious alongshore diminution of upwelled water along the western Eyre Peninsula coast from east to west. Surface water of 16 C occurred between the Coffin Bay Peninsula and Brown Point, but was not visible farther west. The crossshore width of the upwelling signal also diminished from east to west along this coast. The second SST image from 10 March 2004 is more contaminated by clouds, but is sufficient to show that the upwelling signal was still present just before the cruise (16 26 March 2004). Both images suggest the western tip of upwelling off the Bonney Coast that lies to the southeast of the Gulf of St. Vincent. Indeed, most images of surface upwelling [e.g., Kampf et al., 2004] show the simultaneous appearance of cold water plumes off the Bonney Coast, Kangaroo Island, and the eastern GAB. We discuss this further below. [15] To put these results and those below in context, we now consider the summer (December January) climatology of bottom temperature for the region (Figure 3). This climatology was obtained from the CSIRO Atlas of Regional Seas (CARS; [Ridgway et al., 2002]), and was interpolated onto the bottom topography of the region. Only water between 12 and 17 C has been color contoured so as to highlight the upwelling. As is evident, the coldest water ( C) is found to the southwest of Kangaroo Island as a plume that is largely bounded by the 100 m isobath and the shelf break (200 m). (The density of data in the CARS atlas is relatively poor, although the location of the cold pool is supported by other data collected by SARDI for the Kangaroo Is/Eyre Peninsula region). [16] The location of the C contours also indicates that this upwelled water is on average, swept to the northwest and as far as Streaky Bay in the eastern GAB. Significantly, water is not upwelled at the shelf break of the eastern GAB. Shelf break upwelling off Kangaroo Island and advection to the eastern GAB was found in the idealized ocean circulation study of Middleton and Platov [2003]. [17] Scales for this advection may be inferred from the current meter data reported by Hahn [1986] as well as from more recent modeling studies of the authors. Typical upwelling currents off Neptune Island are of order 25 cm s 1. For a relatively long upwelling event, of 10 days, this speed implies that upwelled water can be moved to the northwest by at most 225 km (Figure 2). Water upwelled off Kangaroo Island will therefore not reach Anxious Bay during a single upwelling event. 4of13

5 Figure 4. The alongshore wind stress (Pa) obtained from half-hourly data from Cape Borda, Kangaroo Island. The alongshore component resolved along 315 true is shown and has been smoothed over a 2 day period. A positive value is upwelling favorable. [18] As noted, SST imagery indicate the simultaneous appearance of surface upwelled plumes off the Bonney Coast, Kangaroo Island, and the eastern GAB. The relatively large advective scale above implies that the surface plumes off Kangaroo Island and the Bonney Coast do result directly from nearby shelf break upwelling. However, the plumes off the eastern GAB are more likely to be drawn from the pool of upwelled water that lies to the west of Kangaroo Is: this water would in turn have been upwelled by a prior upwelling event. In their examination of upwelling of the region, Kampf et al. [2004] concluded that there must be a preexisting larger-scale process that lifts cold water onto the shelf. The explanation here is that the process is simply a prior upwelling event with nutrient-rich water stored in the Kangaroo Island subsurface pool. [19] The numerical results of Middleton and Platov [2003] shed light on the likely flow field associated with the Kangaroo Island pool. As indicated by the data, deep (shelf break) upwelling is only found to the south of Kangaroo Is. At the surface, the offshore Ekman transport by the winds, leads to an alongshore current that flows to the northwest. However, the pool water is advected by deeper currents associated with the bottom boundary layer and the constraint that flow tends to follow isobaths. Indeed, the subsurface pool is found to bifurcate, with some water moving to the north of Kangaroo Island and some moving inshore and directly toward the Eyre Peninsula [Middleton and Platov, 2003, Figure 14]. [20] The existence of such a pool of enriched water is also consistent with the wind stress events and the upwelling shown in Figure 2 and below. In Figure 4 we present the alongshore component of wind stress (315 true), on the basis of half-hourly data from Neptune Is. During January the winds are generally upwelling favorable (0.03 Pa) with one very strong event exceeding 0.15 Pa. Such events are known to produce upwelling, drive surface currents to 5of13

6 presumably lead to the cold (16 17 C) water that is evident in the satellite image of 29 February (Figure 2, upper panel). At the same time these wind events would also seem responsible for the surface upwelling that is evident in Figure 2 for the Bonney coast. The next upwelling wind events occur between 3 and 14 March (Figure 4). These events are presumably responsible for the upwelling in the satellite image of 10 March (Figure 2, lower panel) and that is described below for the CTD data collected between 16 and 23 March. Figure 5. Temperature-salinity plot for all profiles from leg 2 of 2004 survey, color coded by fluorescence or oxygen. For the embayments, (blue, low fluorescence) Anxious Bay; (green, higher fluorescence) Streaky Bay. Color bar has been scaled within total data range (0 5.0 volts for fluorescence and ml l 1 for oxygen) to emphasize regional differences. Temperature-salinity pairs associated with fluorescence or oxygen greater than the maximum for the color bar are included in the maximum color bin. the northwest [Schahinger, 1997; Kampf et al., 2004] and should, during January, supply the Kangaroo Island pool with cold, nutrient-rich water. In late February, two further upwelling wind events occur that should transport the pool of water toward the eastern GAB. Indeed, these events 3.2. Spatial Variability of Vertical Profiles in Summer Autumn 2004 [21] Analysis of CTD profiles collected in the 2004 survey showed six distinct zones with characteristic temperature, salinity, oxygen and fluorescence characteristics (Figure 5). Three of these regions were grouped along a temperature-salinity mixing line. The end-members of the mixing line were upwelling-influenced waters (cooler and fresher) and the shelf break stations (warmer and more saline). Intermediate along the mixing line were waters influenced by the Great Australian Bight (GAB) warm pool. Offset to the more saline side of this main T-S mixing line was water at the entrances of two coastal embayments (Streaky Bay and Anxious Bay). In contrast, to the more saline water from the embayments, the effects of groundwater intrusions are visible to the fresher side of the main mixing line (see below). Completely separate from the main mixing line were the warm, highly saline waters of the Spencer Gulf that result from local heating and evaporation (Figure 5). [22] The color coding by fluorescence and oxygen in Figure 5 also suggests that prior photosynthesis has taken place and a bloom may now be decaying: water that lies closer to the Kangaroo Island pool has the highest fluorescence and lowest oxygen concentrations Inshore Embayments [23] The profile data revealed significant alongshore variations in physical properties at nearshore stations, progressing from upwelling influences, to inshore embayments, to GAB warm pool influences from south to north along the western Eyre Peninsula. Streaky Bay and Anxious Bay were saltier than the other inshore stations. The profile at Streaky Bay ( E, S) showed a high-salinity layer (>35.75) in the upper 10 m (Figure 6). The high salinity was associated with low surface fluorescence values, but a broad subsurface peak in fluorescence was present between 20 and 30 m at this station. The most developed oxygen minimum layer of any of the inshore stations was also measured at Streaky Bay ( ml l 1 oxygen concentration between 8 and 22 m) (Figure 6). The inshore station at Anxious Bay ( E, S) also had a high-salinity layer in the upper 10 m, and less marked oxygen minima, but with peaks in fluorescence near the surface and below the thermocline (Figure 6). For both stations, the density structure of the upper 10 m resembled Spencer Gulf profiles (not presented here) rather than the other inshore stations (Figure 6). [24] Streaky Bay is shallower than other embayments along the western Eyre Peninsula, so that a combination of high evaporation and reduced circulation could produce the high-salinity layer. Anxious Bay ( 60 m) is deeper than 6 of 13

7 Figure 6. Vertical profiles from the nearest inshore stations ((blue) northern station, Streaky Bay and (red) southern station, Anxious Bay) marked on the map. Note the density inversions, oxygen minimum layer between 3 and 20 m, the fluorescence peaks near surface and below the thermocline in Anxious Bay, and the broad subsurface fluorescence peak in Streaky Bay. Streaky Bay (<30 m) but bounded to the south by a ridge shoaling to 30 m linking Cape Finnis to Flinders Island and Ward Island. This ridge may reduce circulation in Anxious Bay by reducing the alongshore flow from the southeast. In addition, Anxious Bay receives intermittent outflow from Venus Bay, a saline (S = 35 40) lagoon [Edyvane, 1998]. A combination of the postulated reduced flow across the ridge, and saline outflow from the lagoon, could contribute to the high-salinity surface layer. The anomalous hydrography in Streaky and Anxious Bays can be clearly seen in the T-S plots as two groups of points with salinities of and temperatures of C (Figure 5, blue (low fluorescence) is Anxious Bay, green (higher fluorescence) is Streaky Bay). The observation that the T-S relationship of Anxious Bay and Streaky Bay is off the axis of the mixing line in the T-S plot supports our speculation that the processes producing the high-salinity layer are local to the embayments, rather than due to coastal intrusion of another water mass Upwelling Influenced [25] Consistent with the upwelled climatology shown in Figure 3, the inshore stations southeast of Anxious Bay showed the influence of upwelling most clearly (Figure 7), with salinities in the range and temperatures in the range C, forming an end-member of the T-S mixing line (Figure 5). A group of stations clearly showing the upwelling signal were the three innermost stations on the three transects east of 134 E, with the exception of the inshore station at Anxious Bay (discussed in section 3.2.1). These stations stand out in the T-S plot by their higher fluorescence values (>2 volts) (Figure 5). They were characterized by a subsurface fluorescence maximum, indicating a maxima of phytoplankton abundance. The peak phytoplankton abundance generally occurred at the base of the thermocline (Figure 7). The phytoplankton peak was overlaid by a peak in oxygen levels and oxygen concentration, suggesting active photosynthesis. [26] The cross-shelf structure of the upwelled plume is illustrated in Figures 8 and 9 for Cape Finnis (closest to Kangaroo Is) and Point Bell (farthest from Kangaroo Is). In the former case, upwelling was evident as a tongue of cool (<17 C), less saline (<35.6), denser (>26 kg m 3 ) water advecting alongshore at depth (Figure 8). This upwelled water was present at the surface and up to 20 km offshore of Cape Finnis (134.8 E) off the western Eyre Peninsula. The results also indicate the colder temperature water and fluorescence to be positively correlated: this result is consistent with such cold (upwelled) water being nutrient rich. [27] To the north of Streaky Bay, the upwelling plume does not reach the surface, but it is still detectable on the 7of13

8 Figure 7. Vertical profiles from three inshore stations influenced by upwelling ((blue) most inshore station, (red) middle station, (green) most offshore station) marked on the map. Note broad maxima in fluorescence near base of the thermocline overlain by oxygen maxima. bottom at 45 m as close as km from the coast (Figure 9). The upwelling plume can be traced back in the sections to 70 m depths on the shelf 100 km offshore of Cape Finnis (Figure 8) and 150 km off Point Bell (Figure 9). Given that the plume extends so far offshore at depth it probably is a result of coastal advection rather than upwelling favorable wind stress. [28] The horizontal distribution of upwelling can be seen on the plots of isosurfaces for the temperature criteria (17 C) of upwelled water. Surface outcropping of the 17 C isotherm occurred in two discrete patches along the western Eyre Peninsula (Figure 10) in The northern patch was centered on Brown Point (134 E), separating Streaky Bay and Denial Bay. The second patch was centered on the southern side of Cape Finnis and Flinders Island (135 E) GAB Warm Pool Influenced [29] Stations to the north of Streaky Bay tended to have a deep isothermal warm (17 19 C) layer overlaying a steep thermocline (Figure 11). The deep warm isothermal layer due to strong summer heating may also be associated with the warm pool that develops in the shallow areas of the northwestern Bight [Herzfeld, 1997; Herzfeld and Tomczak, 1997]. The thermocline was near the bottom, associated with a low-salinity layer ( ). We interpret the cool, less saline water near the bottom at these stations as the upwelling signal discussed in section mixing with groundwater seepage (see below), and shown in the CTD section in Figure 9. In contrast to the upwelling-influenced water farther south, here there is no broad peak in fluorescence (implying lower phytoplankton concentrations) associated with the intrusion along the bottom this far west. [30] A further interesting feature of the results (Figure 11) is the presence of plumes of relatively fresh water that result from groundwater aquifers. From Figure 11 these plumes lead to 10 m inversions in density of about 0.1 kg m 3 and imply local mixing Shelf Break [31] Stations at the shelf break were characterized by the deepest isothermal layers found for the region (60 m). One reason for this is the lack of upwelled water at depth near the shelf break as illustrated by Figures 8 and 9. The lack of upwelled water implies that no abrupt change in density will exist. Surface mixing by the wind therefore extends to depth leading to the deep isothermal layers found. [32] Oxygen concentration levels were relatively higher ( ml l 1 ) from 20 m to the thermocline (Figure 12). Relatively higher oxygen levels suggest that photosynthesis is occurring at these shelf break sites, but the low fluorescence levels (<1.25 volts in the upper 60 m) suggest that growth may be nutrient limited (Figures 5 and 12). The shelf break stations unaffected by upwelling form the 8of13

9 Figure 8. Section constructed from CTD profiles along transect off Cape Finnis marked by the red box on the map. Upwelled cool (<17 C), relatively fresh (<35.6) water reaches the surface up to 20 km from the coast. The dense upwelling plume is associated with high fluorescence. Vertical lines indicate profile positions (n1 to the left, n12 to the right). 9 of 13

10 Figure 9. Section constructed from CTD profiles along transect off Point Bell marked by the red box on the map. Upwelled cool (<17 C), relatively fresh (<35.6) water does not reach the surface but is detectable on the bottom at 20 km from the coast. Vertical lines indicate profile positions (t1 to the left, t16 to the right). 10 of 13

11 Figure 10. Spatial distribution of upwelling off the western Eyre Peninsula, based on the depth of the 17 C isotherm. Stations where CTD data were used for the plots are shown as dots. higher-temperature higher-salinity end-member of the T-S mixing line (Figure 5). 4. Conclusions [33] Our study of climatological and CTD data for 2004 provides for the first time, a conceptual model of upwelling for the eastern region of the Great Australian Bight. In particular, both numerical model results [Middleton and Platov, 2003], and the CARS climatology of bottom temperature (December January) provide strong evidence that shelf break upwelling is confined to the Kangaroo Island region and does not occur farther to the west off the Eyre Figure 11. Vertical profiles from three nearshore stations influenced by the GAB warm pool water ((blue) eastern station, (red) northern station, (green) western station) marked on the map. Note isothermal warm water in upper m and conspicuous low-salinity layer at the thermocline. 11 of 13

12 Figure 12. Vertical profiles from three shelf break stations ((blue) eastern station, (red) middle station, (green) western station) marked on the map. Note the deep (60 m) thermocline and the low-salinity layer associated with the thermocline. Peninsula. Rather, the upwelled water is likely to remain in the Kangaroo Island subsurface pool until subsequent upwelling events draw the water to the shallower and surface coastal regions of the eastern GAB. In this manner, the surface upwelling apparent off the Bonney Coast, Kangaroo Island, and the eastern GAB can, to within a day or two, appear to be simultaneous. Moreover, it appears likely that the water within the Kangaroo Island pool remains nutrient rich. [34] Support for this model comes from the CTD sections collected during March 2004 for the eastern GAB. In particular, the data shows that the upwelled signal (cool; <17 C), fresher; <35.6), dense; s t >26kgm 3 ) diminishes in width and intensity with increasing distance from Kangaroo Is. The pattern of fluorescence is similar to that for temperature and indicates that the Kangaroo Island pool remains nutrient rich. [35] The warmest water is found near the shelf break along with very low values of fluorescence and relatively higher levels of oxygen, suggesting nutrient-limited growth of phytoplankton. These data support the notion that the upwelled nutrient-rich water is supplied from the Kangaroo Island pool and not by shelf break upwelling in the eastern GAB. [36] Diagrams of temperature versus salinity summarize the conceptual model. The coldest water lies nearest the Kangaroo Island pool and has high/low values of fluorescence/oxygen. The data also indicates anomalously fresh water due to groundwater discharge (aquifers) at depths of 40 m or more. In addition, the coastal bays are sources of anomalously salty water, a likely result of evaporation. [37] The eastern GAB upwelling signal near the coast is also most evident as two distinct patches of cool water centered on Brown Point (separating Streaky Bay and Denial Bay), and the southern side of Cape Finnis and Flinders Island. Upwelling around headlands and bays is known to occur elsewhere due to local wind effects, [e.g., Roughan et al., 2005] as well as advection by headland eddies [e.g., Leth and Middleton, 2004; Oey, 1996; Penven et al., 2000]. Unfortunately, without further ocean current data, we cannot determine the mechanisms for the localized upwelling. [38] Acknowledgments. We gratefully acknowledge the SARDI staff who worked aboard the RV Ngerin to collect the survey data. Wetjens Dimmlich helpfully contributed graphics. MODIS data used in this study were acquired as part of NASA s earth science enterprise. The algorithms were developed by the MODIS science teams. The data were processed by the MODIS adaptive processing system (MODAPS) and Goddard distributed active archive center (DAAC) and are archived and distributed by the Goddard DAAC. We also thank Ken Ridgway (CSIRO) for making the CARS atlas available to us. The comments of two anonymous reviewers substantially improved the manuscript. References Baird, M. E. (2003), Effect of cross-shelf topography on a pelagic ecosystem response to upwelling favourable winds, unpublished report, Univ. of New South Wales, Sydney, Australia. 12 of 13

13 Bruce, B. D., and D. A. Short (1990), Observations on the distribution of larval fish in relation to a frontal system at the mouth of the Spencer Gulf, South Australia, Bur. Rural Res. Proc., 15, Bye, J. (1983), Physical oceanography, in Natural History of the South East, edited by M. J. Tyler et al., pp , Occas. Publ., R. Soc. of S. Aust., Adelaide, S. Aust., Australia. CSIRO Marine and Atmospheric Research (2001), SEF (South East Fishery) ocean movies, technical report, CD-ROM version 1.4, CSIRO Mar. and Atmos. Res., Hobart, Tasmania, Australia. Edyvane, K. S. (1998), Conserving marine biodiversity in South Australia, part 2: Identification of areas of high conservation value in South Australia, Primary Ind. and Resour. S. Aust., ISBN: , S. Aust. Res. and Dev. Inst. Aquat. Sci., Henley Beach, S. Aust., Australia. Hahn, S. D. (1986), Physical structure of the waters of the South Australian continental shelf, Ph.D. thesis, Sch. of Earth Sci., Flinders Univ. of S. Aust., Bedford Park, S. Aust. (available in S. Aust. Res. and Dev. Inst. library) Herzfeld, M. (1997), The annual cycle of sea surface temperature in the Great Australian Bight, Prog. Oceanogr., 39, Herzfeld, M., and M. Tomczak (1997), Numerical modelling of sea surface temperature and circulation in the Great Australian Bight, Prog. Oceanogr., 39, Kampf, J., M. Doubell, D. Griffin, R. L. Matthews, and T. M. Ward (2004), Evidence of a large seasonal coastal upwelling system along the southern shelf of Australia, Geophys. Res. Lett., 31, L09310, doi: / 2003GL Kitani, K. (1977), Some observations on the temperature inversion off Kangaroo Island, southern Australia, Bull. Far Seas Res. Lab. Jpn., 15, Lennon, G. W., et al. (1987), Gravity currents and the release of salt from an inverse estuary, Nature, 327, Leth, O., and J. F. Middleton (2004), A mechanism for enhanced upwelling off central Chile: Eddy advection, J. Geophys. Res., 109, C12020, doi: /2003jc Lewis, R. K. (1981), Seasonal upwelling along the south-eastern coastline of South Australia, Aust. J. Mar. Freshwater Res., 32, Middleton, J. F., and M. Cirano (2002), A northern boundary current along Australia s southern shelves, J. Geophys. Res., 107(C9), 3129, doi: /2000jc Middleton, J. F., and G. Platov (2003), The mean summertime circulation along Australia s southern shelves: A numerical study, J. Phys. Oceanogr., 33(11), Newell, B. S. (1974), Distribution of Oceanic Water Types off South- Eastern Tasmania, Rep. 59, CSIRO Aust. Div. of Fish. and Oceanogr. Nunes Vaz, R. A., G. W. Lennon, and D. G. Bowers (1990), Physical behaviour of a large, negative or inverse estuary, Cont. Shelf Res., 10, Oey, L. Y. (1996), Flow around a coastal bend: A model of the Santa Barbara Channel eddy, J. Geophys. Res., 101, 16,667 16,682. Penven, P., C. Roy, A. de Verdiere, and J. Largier (2000), Simulation of a coastal jet retention process using a barotropic model, Oceanol. Acta, 23(5), Provis, D. G., and G. W. Lennon (1981), Some oceanographic measurements in the Great Australian Bight, in Fifth Australian Conference on Coastal and Ocean Engineering 1981, pp , Natl. Comm. on Coastal and Ocean Eng. of the Inst. of Engineers, Australia. Ridgway, K. R., J. R. Dunn, and J. L. Wilkin (2002), Ocean interpolation by four-dimensional least squares: Application to the waters around Australia, J. Atmos. Oceanic Technol., 19(9), Rochford, D. J. (1977), A review of a possible upwelling situation off Port Macdonnell, Div. Fish. Oceanogr., 81, CSIRO. Roughan, M., E. J. Terrill, J. L. Largier, and M. P. Otero (2005), Observations of divergence and upwelling around Point Loma, California, J. Geophys. Res., 110, C04011, doi: /2004jc Schahinger, R. B. (1997), Structure of coastal upwelling events observed off the south-east coast of South Australia during February 1983 April 1984, Aust. J. Mar. Freshwater Res., 38, Schlitzer, R. (2003), Ocean Data View. (available at Ward, T. M., L. J. McLeay, and S. McClatchie (2004), Spawning biomass of sardine (Sardinops sagax) in South Australia in 2004, in Report to PIRSA Fisheries RD/04/0153, SARDI Aquatic Sciences. Wenju, C., R. B. Schahinger, and G. W. Lennon (1990), Layered models of coastal upwelling: A case study of the South Australian region, in Modelling Marine Systems, edited by A. M. Davies, pp , CRC Press, Boca Raton, Fla. S. McClatchie, J. F. Middleton, and T. M. Ward, South Australian Aquatic Sciences Centre, SARDI Aquatic Sciences, P.O. Box 120, Henley Beach, SA 5022, Australia. (mcclatchie.sam@saugov.sa.gov.au; john. middleton@unsw.edu.au; ward.tim@saugov.sa.gov.au) 13 of 13