Archived at the Flinders Academic Commons:
http://dspace.flinders.edu.au/dspace/
This is the publisher’s copyrighted version of this article.
The original can be found at: http://www.agu.org/journals/jc/jc0609/2006JC003592/
2006JC003592.pdf
© 2006 Journal of Geophysical Research
Published version of the paper reproduced here in accordance with the copyright policy of the
publisher. Personal use of this material is permitted. However, permission to reprint/republish
this material for advertising or promotional purposes or for creating new collective works for
resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in
other works must be obtained from Journal of Geophysical Research.
Eddy shedding and energy conversions in the East Australian Current
Mauricio M. Mata,1Susan E. Wijffels,2John A. Church,2 and Matthias Tomczak3
Received 18 March 2006; revised 29 May 2006; accepted 12 June 2006; published 26 September 2006.
[1] Mesoscale variability and eddy shedding in the Tasman Sea, particularly of the East Australian Current (EAC), is studied through the analysis of remotely sensed observations and outputs from a global ocean model. Previous observations of the western boundary current separation from the coast showed strong variability at periods ranging between 90 and 140 days. We show from satellite altimetric observations that rapid northward migration of the separation point of the EAC follows the formation of large eddies at periods of 100 days. After an eddy separation event the normally southward flowing current swiftly assumes a more zonal configuration near the separation latitude, with a cyclonic circulation developing inshore. The formation of large separation eddies is preceded by the southward propagation of sea level anomalies along the east Australian continental slope. From 25S, sea level anomalies grow as they travel south, eventually being pinched off in the form of large anticyclones at 32S, in the current
retroflection area. Energy conversion terms in a global ocean model and in altimetric data suggest both barotropic and baroclinic instability may account for the growth of these anomalies as they propagate south. East of the main EAC jet there is evidence that eddies may be feeding potential energy back to the mean flow.
Citation: Mata, M. M., S. E. Wijffels, J. A. Church, and M. Tomczak (2006), Eddy shedding and energy conversions in the East Australian Current,J. Geophys. Res.,111, C09034, doi:10.1029/2006JC003592.
1. Introduction
[2] The ocean western boundary currents of the Southern Hemisphere are less understood than their northern counter- parts despite their recognized importance to ocean heat transport and climate [e.g., Bryden et al., 1991]. In the Southwestern Pacific (Figure 1), the East Australian Current (EAC) flows southward along the east Australian continen- tal boundary. Among the several western boundary currents associated with the subtropical gyres, the EAC is recognized as the weakest, carrying on average only 22 Sv (1 Sv = 106 m3s1) at 30S [Mata et al., 2000]. Nevertheless, its flow is associated with strong eddies that impose variability levels comparable with larger western boundary currents such as the Gulf Stream, the Kuroshio and the Agulhas Current [e.g.,Gordon et al., 1983; Feron, 1995]. The EAC transport time series at 30S, obtained from the WOCE Pacific Current Meter Array 3 (PCM3) and reported byMata et al. [2000], hereinafter M2000, is dominated by extreme volume transport events occurring every 3 to 4 months.
Sea surface temperature (SST) images of the area have shown that these events are related to the presence or absence of the main jet of the EAC in the current-meter array domain and imply that there are times when the
current is separating from the continental slope north of 30S.
[3] SST composites of the Tasman Sea suggest that a sudden northward shift in the current separation point occurs after the shedding of a strong anticyclonic eddy south of 32S. Feron [1995] found similar behavior in GEOSAT altimeter data and the FRAM (Fine Resolution Antarctic Model) ocean model with quasiperiodic ring formation taking place in the EAC region roughly every 130 days. In the OCCAM (Ocean Circulation and Climate Advanced Modeling Project) global ocean model at around 30S, the EAC sheds eddies regularly at an interval of 100 days [Saunders et al., 1999].
[4] During the PCM3 deployment energetic and intermit- tent oscillations, with periods contained in the temporal mesoscale band (hereafter we define the m-band as 170 days > period > 70 days) dominated the variability of the EAC transport across the array. Using a combination of satellite altimetric and SST dataBowen et al.[2005] found mesoscale variability in the EAC with periods between 90 and 180 days propagating southward along the Australian continental boundary up to the EAC separation latitude.
After that point, the propagation direction was westward following the Tasman Front across the Tasman Sea.Bowen et al.[2005] found no evidence of remote forcing on the EAC variability, suggesting instead that self-generated barotropic instabilities can explain the observed variability.
[5] Despite recent progress, several aspects of EAC variability remain unexplained. For example, clear evidence for a strong role of barotropic/baroclinic instability in ring formation is lacking, and it is not known whether eddy dissipation can affect the mean flow. The character of EAC
Herefor
ArticleFull
1FURG – Departamento de Fı´sica, Rio Grande (RS), Brazil.
2CSIRO Marine and Atmospheric Research, Hobart, Tasmania, Australia.
3Flinders University of South Australia, SOCPES, Adelaide, South Australia, Australia.
Copyright 2006 by the American Geophysical Union.
0148-0227/06/2006JC003592$09.00
eddy-shedding events remains poorly described. Here we explore the nature of the EAC variability using a combina- tion of in situ and satellite data in addition to the outputs of a global ocean model. The next section presents the altim- etry data and a brief description of the ocean model and is followed by the analysis of the sea level variability in the study area. We then examine the eddy-mean energy inter- actions in the Tasman Sea, followed by a general discussion and summary of our main results.
2. Data and Methods
[6] The WOCE PCM3 current meter array was deployed to estimate the volume transport and variability of the EAC
in the region of its maximum strength at 30S (Figure 1). It was the first multiyear deployment in the EAC, spanning over 80 km, and was expected to cover most of the transport of the EAC, from 100 m to 4600 m water depth.
The array consisted of 26 recording current meters and 2 upward looking ADCPs distributed along six moorings, which were deployed for the approximately 23 months from October 1991 to September 1993. Details on the individual records, data recovery, dominant space-time variability and transport estimates can be found inMata et al.[1998, 2000, 2006].
[7] We use the Mapped Sea Level Anomalies (MSLA) with respect to a long-term mean sea level from the AVISO Figure 1. The key circulation features in the Southwestern Pacific Ocean are presented in dark
schematics: the South Equatorial Current (SEC) the East Australian Current (EAC) and the Tasman Front (TF). During the shedding of large anticyclones in the retroflection area (light grey schematics), the EAC separation position can shift to the north as far as (or past) the PCM3 array location. Depth contours are displayed in 1000 m intervals. Regions shallower than 3000 m are shaded in light grey.
C09034 MATA ET AL.: EDDY SHEDDING AND ENERGY CONVERSIONS C09034
Project (Archive, Validation et Interpretation des donnes des Satellite Oceanographiques), the oceanographic satellite data bank of the French space agency CNES. In the AVISO processing the TOPEX/POSEIDON (hereafter T/P) and ERS-1 and 2 altimeter measurements are quality controlled, corrected for instrumental errors and environmental factors [AVISO, 1998]. The final product has error levels around 2 – 3 cm rms. The time-variable circulation can be determined independent of the marine geoid and thus an estimate of the mean SSH is removed from the data set, resulting in along- track sea level anomalies (SLA). MSLA are obtained using a space/time objective analysis method described in Le Traon et al. [1998], which maps the original satellite ground track data onto a regular 0.250.25 grid every 10 days (matching the T/P orbital repeat period). Note that from 16 December 1993 to 31 March 1995, no ERS data were available and thus during those times the MSLA is based on measurements from T/P only.Ducet et al.[2000]
assessed the improvements obtained by merging the T/P and the ERS-1/2 data sets, especially for the study of the ocean mesoscale phenomena [see also Hernandez et al., 1995; Le Traon and Dibarboure, 1999]. They concluded that the merged data set provide a more homogeneous data set and reduced mapping errors than either individual data set.
[8] Despite the fact that satellite altimetry only provides reliable information about the time-variable part of ocean currents, the total surface circulation (mean + variable) has commonly been inferred by adding a climatological mean surface dynamic topography to the SLA fields in order to estimate the total Sea Surface Dynamic Topog- raphy (SSDT) [e.g., Willebrand et al., 1990; Ichikawa and Imawaki, 1994]. Using SSDT one can estimate the absolute surface geostrophic velocity, and its variability, taking advantage of the robust altimetry time series.
Here we use the CSIRO (Commonwealth Scientific and Industrial Research Organisation) Atlas of Regional Seas (CARS) climatological dynamic height relative to 2000 dbar (Figure 2) to add to the MSLA fields and thus obtain the surface geostrophic currents. CARS has proven to be a robust climatology covering the Tasman and Coral Seas, which allows for high-resolution in ranging from the large-scale structure to narrow coastal features [Ridgway et al., 2002].
[9] The high-resolution global ocean model Parallel Ocean Program (POP) [Maltrud et al., 1998] is also used to investigate the energy conversions in the Tasman Sea.
We use 10-day snapshots from run 11 b which differ from the description given in Maltrud et al. (run 11) only by the incorporation of ECMWF daily-averaged wind- forcing. Maltrud et al. [1998] performed a thorough analysis of POP11 results and compared them with T/P observations. Their overall conclusion was that the model provides a good simulation of several aspects of the wind-driven circulation such as the mean mass transports of the main current systems. However, the separation point of the western boundary currents lay poleward of where observations suggest and many aspects of merid- ional overturning remain problematic in the model (it was a relatively short integration). The eddy energy, although higher than found in previous lower-resolution models, is still lower than that suggested by altimetry - only 60%
of what is observed [McClean et al., 1997]. Some of these problems (the separation point and eddy energy) were partially overcome in the recent 1/10 run [Maltrud and McClean, 2005], which suggested that an even finer spatial resolution is needed to obtain quantitative agree- ment between modeled and observed energy levels.
Nevertheless, the model reproduces the salient features of the 100-day oscillations around 30S and therefore is a useful tool in the study their dynamics.
3. Sea Level Variability in the Tasman Sea [10] Sea level in the Tasman Sea features an area of high variability west of 160E between 28and 38S where the RMS > 0.15 m (Figure 2). The feature follows the orien- tation of the East Australian continental boundary, except near the typical latitude of the Tasman Front where it extends eastward along 32S. The PCM3 array was located to the north of the region of maximum SLA variability near 30S/154E. Although the SLA variability is due to sea level oscillations on all timescales, it is essentially the result of strong eddy activity known to be present in the area [e.g., Morrow et al., 1992; Feron, 1995; Bowen et al., 2005].
Moreover, the spatial distribution of the region of high RMS variability is related to major features of the surface mean flow (Figure 2, top): high SLA variability follows the path of the mean EAC retroflection and the mean latitude of the Tasman Front. The high variability region is also very much confined to the deep Tasman Basin (>3000 m) which suggests an influence of bottom topography [Ridgway and Dunn, 2003].
[11] The spatial distribution of SLA RMS variability in POP11b generally agrees with that observed (Figure 3), although some discrepancies can be readily seen such as lower energy levels and smaller spatial extent of the high SLA variability region. The model also fails to reproduce the northern and eastern extensions of the high variability zone. While lower energy values are probably related to limitations on the spatial resolution of the model (both horizontal and vertical) another possible source of error is a poor representation of the mean flow. As for SLA variability, mean surface sea level (Figure 3, top) agrees qualitatively with observations (Figure 2, top) although the separation latitude of the observed EAC and the position of the retroflection are located further south.
The modeled offshore return current seems narrower than observations suggest and the SSH field is generally smoother. Fu and Smith [1996] showed that in the POP11 run both the Gulf Stream and the Kuroshio separate to the north of the observed latitude and were followed by smaller and less energetic current extensions into the open ocean basin.
[12] Comparing the spectra of surface meridional veloc- ities between the POP11b run and the EAC observations demonstrate that there is a good general agreement between the model and the data with regard to the dominant periods of variability but, as expected, the model energies are too low [Mata, 2000]. Upon closer examination we found that the EAC jet is broader than observed and thus is thus likely more stable to barotropic instability. We suspect this is due at least in part to poor representation of the continental slope at 30S in the model - the model EAC jet is located
Figure 2. Mean Surface Dynamic Topography field (with respect to 2000 dbar) from the Climatology of the Australian Regional Seas (CARS) [Ridgway et al., 2002] (top, units are dyn. meters) and the Sea Level Anomaly (MSLA) RMS relative to the first 265 T/P cycles (bottom, units in meters). The black circles next to the Australian coast at 30S denote the positions of the PCM3 moorings. Also shown is the 3000 m isobath.
C09034 MATA ET AL.: EDDY SHEDDING AND ENERGY CONVERSIONS C09034
approximately 40 km offshore from its position observed by the PCM3 array.
4. Surface Currents at PCM3
[13] The surface meridional geostrophic velocity derived from the altimetry SSDT was calculated for the mid-PCM3
location close to the position of PCM3 mooring 5 (153560E; 30100S) and compared to the average velocity across the array estimated at two different levels (350 m and 500 m), after correcting for vertical motion (Figure 4). Most of the PCM3 top instruments (above 1000 m) returned complete records between September 1992 and November Figure 3. Sea surface height means from 5 years of POP11b output (top) and the RMS Sea Level
Anomaly (bottom). The black circles next to the Australian coast at 30S denote the positions of the PCM3 moorings. Also shown is the 3000 m isobath. Units are meters.
1993. Details on the PCM3 time series, quality control and how the data was treated to minimize vertical motion and gaps can be found in Mata et al. [1998, 2000]. The correlation coefficients between the satellite velocities and in situ values near 350 m and 500 m depth are 0.90 and 0.87, respectively. Thus SSDT clearly captures the primary variability observed during the PCM3 deployment, as these top layer velocities (350 m and 500 m) are good proxies for transport variability (M2000). The difference in amplitudes between the altimetric and the array velocity probably results from the fact that the former is at the sea surface while the latter are at lower levels (the 350 m amplitudes are approximately 55% of the altimeter’s). During strong south- ward flow events vertical shear is large, while during the northward flow events, the flow becomes nearly barotropic with surface SSDT-derived and deeper current meter veloc- ities matching (Figure 4). This tight agreement implies that CARS captures the mean EAC meridional flow remarkably well.
[14] Having established that the SSDT data capture the primary variability in the EAC for the PCM3 period, we can extend our analysis across the much longer 7 year altimetric record (Figure 5). Oscillations with a period of about 100 days dominate but are stronger in 1993 (which coincided with the second year of the PCM3 deployment) and from mid-1995 to mid-1996. These are the only times when the surface velocity actually reverses from its predominant southward direction.
Frequencies lower than annual are also present, but are much less energetic than those in the mesoscale band at this location.
[15] The spectral characteristics of the time series of velocity at 30S confirms that the dominant peak of vari- ability has a period of100 days (Figure 6), corresponding to the main signal of interest of this study. In order to enhance our estimates of the dominant spectral peaks, for this analysis we used an extended satellite data set (12 years data set, from 1992 to 2004), which was available after the majority of this study was completed. The integrated variance between periods of 70 and 170 days (the shaded area in Figure 6) contains approximately 35% of the total variance of the time series. Secondary peaks are present at 225 days (a broad peak reflecting both the seasonal and interannual oscillations) and 55 days. The latter peak could be the result of aliasing of the semi-diurnal tide wave by the altimeter sampling scheme [Schlax and Chelton, 1994].
5. Typical Eddy Shedding Event
[16] The transport reversals and vertical shear changes at the PCM3 location can be related to the migration of the separation point of the EAC to the north of the array site and the currents at the PCM3 location tend to have a northward flow whenever the EAC is not present in the array area (M2000). Satellite SST imagery suggested the eddy shed- ding behavior had repeatable aspects, suggesting a compos- iting technique might reveal the spatial pattern associated with the eddy-shedding events.
[17] An elliptic band-pass filter [e.g.,Parks and Burrus, 1987] is applied to separate the variability of interest to this Figure 4. Surface meridional geostrophic velocity across the PCM3 array estimated from MSLA +
CARS SSDT data (thick) and mean velocity from the PCM3 current meter at 350 m depth (asterisks) and 500 m depth (diamonds). The values in the bottom right box indicate the correlation coefficient between the respective current meter series and the Altimeter estimate. Negative values indicate southward flow.
C09034 MATA ET AL.: EDDY SHEDDING AND ENERGY CONVERSIONS C09034
study (100 days) from the total signal presented in Figure 5.
We chose the minima in spectral energy (Figure 6) to define the limits of the filter which removes oscillations with periods less than 70 days and greater than 170 days. The resulting filtered time series of surface velocity across the PCM3 array is defined here as the Index of the EAC at 30S (Figure 7). The Index should be a good temporal indicator to both the migration of the separation point of the EAC and the associ- ated eddy-shedding events. Following the approach thatFeng et al.[2000] applied to the variability of the Kuroshio in the Tokara Strait, we defined the extreme time periods when positive (negative) anomalies are larger than half of the time series standard deviation as the offshore (inshore) state of the EAC main jet where the anomalies are outside the limits of the dotted lines (Figure 7). Using the index we calculate composites of SLA for the inshore and offshore phases of the EAC at 30S for the entire study domain, using the band-passed altimeter data (170 >
m-band > 70 days). Thus, the EAC inshore composite is defined as the SLA temporal average of all times that the EAC index falls below the bottom half standard deviation line. Conversely, the offshore composite results from the SLA temporal average of all times the EAC index falls above the top half standard deviation line. To the SLA composites, we add the CARS mean height fields to show total sea level.
[18] During the EAC inshore situation (Figure 8, top), a well developed anticyclonic anomaly is near the coast at 30S and the EAC is flowing strongly southwards along the coast, inshore of this anomaly. In this phase, the EAC
separation latitude is 31S, south of which a cyclonic circulation occurs next to the continental boundary. During the EAC offshore phase (maximum northward flow at PCM3; Figure 8, bottom), a very different picture arises.
South of PCM3, a large anticyclonic eddy has pinched off from the EAC main jet, while at the latitude of the array the main EAC jet is flowing south-eastward offshore of 155E and inshore the EAC has been replaced by a relatively tight cyclonic circulation near the coast. The above scenario agrees with the satellite sea surface temperature results presented in M2000. These composites suggest that the strong100 days oscillations in the transport and structure of the EAC at 30S are indeed associated with eddy shedding.
[19] The different stages of development of the large anticyclonic eddies can be followed using a lagged version of the composite analysis described above. We produced lagged composites by subtracting 10 days (the MSLA data set time increment) from all the reference times (squares in Figure 7), averaging the fields at those times to form the 10 day composite and so on until composites ranging from Day50 to +50 are obtained (Figures 9a and 9b).
[20] As the reference state was chosen during periods when the EAC is in the offshore situation at 30S, the composite for this Day 0 should be similar to the results in Figure 8 (bottom). The lagged composites for Day +50 and 50 should resemble the EAC inshore situation (Figure 8, top), as the main signal in the EAC index has a period of 100 days. Indeed, the Day50 (and Day +50) composite does show the EAC close inshore at 30S with the presence Figure 5. Surface meridional geostrophic velocity across PCM3 array estimated from the entire
MSLA + CARS SSDT time series. The dashed line shows the time mean. Negative values indicate southward flow.
of an intense anticyclonic anomaly centered at the same latitude. At this time, the EAC separation from the coast can be observed around 32S. The subsequent composites in Figures 10a and 10b show the southwestward propagation and shedding of the feature from the main EAC flow to form a large anticyclonic eddy to the south of 30S (composite Day +10), simultaneously with the shift of the EAC main flow to the east.
[21] Boland and Church[1981] noticed similar behavior analyzing the data from a series of 14 cruises in the EAC region during 1978. They observed several cyclonic anoma- lies to the east of the EAC poleward flow, and suggested that these features propagate westward eventually crossing the EAC path. This crossing leads to a large perturbation in the current path, which can move the separation point of the current far to the north of the original position. However, the sea level anomalies may instead be traveling south along the east Australian continental boundary. In the composites, significant anticyclonic anomalies are present at about 25S (indicated in Figure 8), and they travel south following the coastal boundary to form the anticyclonic features at sub- tropical latitudes, as will be discussed in the next paragraph.
Bowen et al. [2005] and Boland and Church [1981] also noted southwestward propagation of the mesoscale variabil- ity signal along the EAC using satellite measurements and situ data, respectively. Model simulations of the area also show similar perturbations that grow as they move down- stream and finally separate into anticyclonic rings [Godfrey, 1973; Marchesiello and Middleton, 2000; Roughan and
Middleton, 2004]. The eddy which is pinched off from the current at composite Day 0 appears to follow the EAC return current northward eventually turning toward the coast and finally coalescing with an anticyclone coming from the north.
[22] Composites of SLA (no mean surface height added) for the two EAC states reveal that there is a train of positive and negative sea level anomalies along the continental boundary from about 23S all the way south to the Tasma- nian coast at about 43S (Figures 10a and 10b). These intense anomalies may be the reason that several authors have described the EAC as a succession of eddies which travel poleward along the east Australian slope rather than a continuous current [e.g., Godfrey et al., 1980;Boland and Church, 1981; Cresswell and Legeckis, 1986]. More recently, taking advantage of the same data set used in this study, Mata [2000] shows the coherent propagation of those anomalies along the 2000 m isobath from around 25S to 35S.
[23] We also estimated the EAC separation index from POP11b band passed SSH fields in the same manner as for the observations to produced composites. During the EAC inshore situation (Figure 10c) there is an associated anticy- clonic SLA at about 30S, while during the offshore situation composite (Figure 10d) a depression is prominent at the same latitude, in good agreement with the observed SLA from the altimetry. The growth and decay pattern of the SLAs and their alongshore wave number, is also similar in both the model and the observations. The composites Figure 6. Variance preserving spectrum of the 12-years time series of the surface meridional
geostrophic velocity across the PCM3 array using 3 years subsegments (see text for details). The numbers above the main peaks are the approximate periods in days. The grey area highlights the periods ranging from 70 to 170 days. The thin lines are the 95% confidence interval.
C09034 MATA ET AL.: EDDY SHEDDING AND ENERGY CONVERSIONS C09034
derived from the POP11b outputs are similar to the essential features of the observed m-band variability in the EAC vicinity, and so we believe it will be useful to evaluate the model energy fluxes for the study area.
6. Energy Conversions in the Tasman Sea [24] Is the growth of the m-band anomalies, and the subsequent shedding of large EAC eddies, due to barotropic and/or baroclinic instabilities of the mean flow? We exam- ine two mean-eddy energy transfer terms with all variables separated into time-means and fluctuations and expressed as the volume integrals defined below:
TPE¼g
Z Z Z u00@
@xþn00@
@y
d~=dz dV ð1Þ
and
TKE¼ Z Z Z
u0u0@u
@xþu0n0 @n
@xþ@u
@y
þn0n0@n
@y
dV ð2Þ wheregis the gravity acceleration,the potential density,x the zonal and y the meridional directions, and (u, v) the horizontal velocity components. The variable e is the reference state for potential density, and is approximated as
the horizontal average of the time mean [Biastoch and Krauss, 1999;Haidvogel and Beckmann, 1999]. Here, TPE (1) represents the transfer of mean potential energy to eddy potential energy usually through baroclinic instabilities of the mean flow [e.g., Beckmann et al., 1994; Biastoch and Krauss, 1999;Haidvogel and Beckmann, 1999]. TKE (2) is the transfer rate between mean kinetic energy and eddy kinetic energy (i.e., the work of the Reynolds stresses on the mean shear) and is associated with growing barotropic instabilities.
[25] The energy conversion terms above can be calculated using the 5 years of output from POP11b. As these instability indicators are relatively ‘‘noisy’’ fields and their small-scale spatial variability is not robust [Beckmann et al., 1994], further averaging was required to extract their large- scale patterns. In addition, the TKE can be evaluated from observations at the surface only by combining the altimetry observations of the variable portion of the oceanic currents with the mean flow estimated from the CARS climatology [e.g.,Wilkin and Morrow, 1994]. TPE cannot be presently calculated from observations since the time fluctuations of potential density field (0) are not observed at the required space and timescales.
[26] There is general agreement between the observations and the model in the magnitude and shape of features in the surface TKE fields (Figure 11). Significant positive values of TKE are only found in the EAC retroflection area, between 30S and 35S. The region of weak negative TKE to the southeast of the retroflection activity is also a Figure 7. East Australia Current index at 30S. The dashed lines show the time mean and the thin lines
the value of1 =
2standard-deviation away from the time mean. The black circles mark 10-day increments of the MSLA data set, while the grey squares show the extreme events used as a reference in the time- lagged composites (see text). The mean has been added back to the whole series after filtering so that the negative values indicate southward flow.
common feature in both model and data fields. Furthermore, the relatively small and well defined area where barotropic instabilities can rapidly grow, common to both the obser- vations and the model, is an important result which provides
an explanation for the observed rapid growth of the SLAs south of 30S.
[27] Where the TKE values are weakly negative (Figure 11) the eddies lose energy to the mean flow, which is supported by the observation that, after experiencing rapid growth and Figure 8. Composites of the MSLA+CARS SSDT for the EAC inshore situation (top) and the EAC
offshore situation (bottom). Values in dyn meters relative to 0/2000 dbar. The dark circles next to the Australian coastline at 30S show the position of the PCM3 moorings. Arrows indicate the formation of anticyclones upstream from the separation region.
C09034 MATA ET AL.: EDDY SHEDDING AND ENERGY CONVERSIONS C09034
Figure 9a. Lagged composite analysis from 50 days to 0 days for the MSLA + CARS SSDT. The color scale is the same as in Figure 8. Units are dyn meters.
Figure 9b. Lagged composite analysis from +10 days to +50 days for the MSLA + CARS SSDT. The color scale is the same as in Figure 8. Units are dyn meters.
C09034 MATA ET AL.: EDDY SHEDDING AND ENERGY CONVERSIONS C09034
being shed from the current as large anticyclonic eddies, the SLAs that keep propagating southward along the coast undergo a slow decay. The smaller scale structures (‘‘dia- monds’’ in Figure 11, top) are harder to interpret, but they are probably reflect the high noise level associated with the surface TKE field derived from altimetry.
[28] The TPE model field features an elongated region of high positive TPE following the east Australian continental slope with most of the nonzero values to the south of 30S (Figure 12). Further offshore in the EAC return flow the TPE field values are distinctively negative (blue areas in Figure 12) indicating that the dissipation of eddies likely plays an important role in driving the recirculation. The
model TPE field also shows raised positive TPE values at bathymetric constriction of the Tasman Sea deep basin (22S/154E). This is consistent with the composite analysis previously discussed which revealed the presence of rela- tively strong eddies that begin to grow in that area. Hence, near to the coast baroclinic instabilities are active and thus both kinds of instability are acting to feed the growth of the SLAs that will eventually form the EAC rings.
[29] In a meridional transect (zonally averaged between 154 and 155E) of TKE there is general agreement between model and observations but the model values are generally lower (Figure 13). The large peak of positive TKE is broader in the altimetry results and, from north to south, Figure 10. SLA composite for (a) the EAC inshore situation and (b) the EAC offshore situation. Same
results as presented in Figure 8, except that the climatological mean sea level has been removed. The SLA composites from the POP model outputs for the (c) inshore and (d) offshore phases are also presented. Units are meters.
starts to differ notably from zero before 28S. In the model, positive values start to the south of 30S. One possible explanation for this discrepancy is the stabilizing effect of the model bathymetry in boundary currents. The coarse resolution of the model (1/6 average; 27 km at 30S) implies that the representation of a steep continental slope will be unrealistic, which is one recognized factor that contributes to the lower energy observed in models like POP [e.g., Wilkin and Morrow, 1994; Moore and Wilkin,
1998;Haidvogel and Beckmann, 1999]. This also helps to explain the low energy levels and weak SLAs to the north of 30S in the model simulations. Furthermore, the low levels of energy coming from the east in the form of m-band Rossby waves to the south of 25S in the model may contribute to this discrepancy. Along the EAC path a rapid jump to positive values of both terms between 30and 35S (Figure 14) shows that both barotropic and baroclinic instabilities are active, but that the TKE term is larger in Figure 11. Surface TKE field (Barotropic Instability indicator) from the MSLA + CARS data set (top)
and averaged over the upper 600 m from POP11b (bottom). Units are cm2s3.
C09034 MATA ET AL.: EDDY SHEDDING AND ENERGY CONVERSIONS C09034
Figure 12. TPE field (Baroclinic Instability indicator) averaged over the upper 600 m from POP11b.
Units are cm2s3.
Figure 13. Zonally averaged (154– 155E) meridional transects of TKE: MSLA + CARS estimate in solid line and POP11b estimate in dashed line. Units are cm2s3.
the region where the rapid growth of the SLAs is observed (30– 32S).
7. Summary and Discussion
[30] The EAC sheds an eddy in a repeatable pattern roughly every 100 days. High sea level anomalies propagate poleward along the shelf slope to about 32S. They cause the EAC to strengthen, surge southward and shed an eddy, after which the separation point of the current swiftly retreats northward, the main jet moves offshore and colder and more barotropic flow encroaches on the shelf break.
While this behavior is repeatable in the main EAC jet, southwards and offshore the eddies appear to move more randomly (possibly due to weaker potential vorticity gra- dients). North and east of the EAC, anomalies in this band behave as linear first-mode Rossby waves, which have a broader spectral distribution compared to the narrow band behavior of the eddy-shedding cycle [Mata, 2000]. Along the pathway of the growing anticyclones, a fairly realistic model suggests the flow is baroclinically unstable. Just to the north and in the retroflection region both observations and model results show the flow is barotropically unstable, suggesting the local instabilities allow perturbations to grow leading up to a shedding event.
[31] The process controlling the timing of eddy shedding in the EAC remains unclear. Many studies have looked to forcing originating east of the eddy shedding region – the Tasman Front.Nilsson and Cresswell[1981] proposed that the westward propagating baroclinic Rossby waves in the
Tasman Front causes a strong poleward excursion of the EAC, constriction of the main flow and subsequent shed- ding of the EAC retroflection meander. The EAC separation latitude then migrates to the north and the process begins again.Campos and Olson[1991] evaluated this mechanism numerically for the Brazil-Malvinas Confluence and found a strong stationary pattern of meanders in the model’s confluence region similar to that ofNilsson and Cresswell [1981]. This feature, although stationary in phase, had time- varying amplitude with intermittent periods of amplitude growth and decay. Eventually, the stationary wave mode becomes strong enough to become unstable leading to a break down into a series of mesoscale eddies. The conflu- ence returns to a more zonal configuration. Although this mechanism portrays well what is observed at the EAC separation zone, it does not explain our results which show a link between eddy formation and the southward migration of the SLAs. Empirical Orthogonal Function analysis of satellite data of the area performed byBowen et al.[2005]
independently confirms our results.
[32] Recently,Marchesiello and Middleton[2000] evalu- ated the mechanism proposed by Nilsson and Cresswell [1981] in a regional version of the Princeton Ocean Model (POM). The authors were able to reproduce the formation of a large EAC warm-core eddy well, which was also associ- ated with the development of an inshore cyclonic circulation quite similar to the one in Figure 8, suggesting that Tasman Front forcing can affect the separation of the EAC from the Figure 14. TKE (dashed line) and TPE (solid line) estimated along the EAC path (roughly the 2000 m
isobath) from the POP11b model fields. The values are the result of the zonal average of all transects between the 2000 m isobath and 1to the east of that isobath. Units are cm2s3.
C09034 MATA ET AL.: EDDY SHEDDING AND ENERGY CONVERSIONS C09034
coast and the subsequent formation of the large anticyclone further south.
[33] Several authors have proposed that instabilities up- stream of the main eddy-generation region are related to the formation of the Agulhas rings [Biastoch and Krauss, 1999;
de Ruijter et al., 1999;van Leeuwen et al., 2000].Nof and Pichevin [1996] proposed an analytical model of the eddy generation mechanism in retroflecting currents. For this mechanism, the authors coined the term retroflection para- dox, which implies that a zonal retroflecting current exerts a flow force parallel to the continental wall that cannot be balanced without the generation of eddies in the current’s original direction. Pichevin et al.[1999] examined numer- ically the theoretical model proposed byNof and Pichevin [1996] by applying the model to the Agulhas Current region. Pichevin et al. [1999] showed that ‘‘transport pulses’’ (i.e., temporarily double the downstream transport next to the coast) are closely related to the production of Agulhas rings. In their study, every transport pulse produces an eddy (SLA) that propagates downstream and eventually leads to the separation of an Agulhas ring. Here, the presence of SLA upstream from the EAC separation latitude may also play an important role in the growth and formation of the EAC rings.
[34] Furthermore, we suggest that the downstream growth of anticyclones is explained in terms of local instability. The mean-eddy energy transfers in both a realistic model and in observations suggest that barotropic (and to a less extent baroclinic) instabilities are responsible for the growth of the upstream perturbations which drive a shedding event. In the Gulf Stream and Kuroshio, eddy formation processes are associated with zonal instabilities that generate meanders along the current path that grow, close upon themselves, and then pinch off to form eddies (or rings). However, the EAC eddies observed in the composite analysis are more illus- trative of retroflection eddies, associated with currents that retroflect (i.e., turn back on itself) near their separation latitude [e.g., Olson, 1991]. The retroflection eddies are generated close to the coast roughly at the same location, suggesting a geographically controlled formation model [Richardson et al., 1994;Silveira et al., 1999].
[35] The presence of intra-annual Rossby waves propa- gating toward Australia has been observed byMata[2000]
andBowen et al. [2005]. The latter authors could not find any evidence of coherent propagation of those waves from the offshore region and then along the east Australian coast and, thus, suggest that they are not among the main players in controlling the formation of EAC eddies. Also, very narrow-band robust EAC eddy shedding behavior was observed in a global model forced only by mean climato- logical winds [Saunders et al., 1999] and thus lacking wind- driven Rossby waves in the m-band. These results suggest internal current dynamics can account for the bulk of the observed behavior. Nevertheless, a small amount m-band variability should be present in the model due to the remote growth of instabilities on the density gradients [Qiu and Chen, 2004] present further to the east, and may contribute to modulating the frequency of the EAC eddy formation.
[36] Acknowledgments. M. M. Mata acknowledges the support by the Brazilian research council (CNPq, grant 200167/96-0) and by the Fundac¸a˜o Universidade Federal do Rio Grande (FURG) - Brazil.
S. E. Wijffels and J. A. Church were supported by CSIRO’s Climate Change Research Project and were funded in part by the Australian Climate Change Science Program. The POP Model development group at Los Alamos National Laboratory and the AVISO Project are acknowl- edged for providing the model outputs and altimetry data used in this study. We also thank M. Bowen, D. Griffin, K. Ridgway, S. Godfrey and G. Cresswell for early discussions that helped to improve the manuscript. This study is a contribution to the World Ocean Circulation Experiment.
References
AVISO (1998), Users’ handbook for sea level anomalies (SLAs),CLS Rep.
AVI-NT-011-312-CN, Toulouse, France.
Beckmann, A., C. W. Bo¨ning, B. Bru¨gge, and D. Stammer (1994), On the generation and role of eddy variability in the central North Atlantic Ocean,J. Geophys. Res.,99(10), 20,381 – 20,391.
Biastoch, A., and W. Krauss (1999), The Role of mesoscale eddies in the source regions of the Agulhas Current,J. Phys. Oceanogr.,29, 2303 – 2317.
Boland, F., and J. A. Church (1981), The east Australian current,Deep Sea Res., Part A,28(9), 937 – 957.
Bowen, M. M., J. L. Wilkin, and W. J. Emery (2005), Variability and forcing of the East Australian Current,J. Geophys. Res.,110, C03019, doi:10.1029/2004JC002533.
Bryden, H. L., D. H. Roemmich, and J. A. Church (1991), Ocean heat transport across 24N in the Pacific,Deep Sea Res., Part A,38(3), 297 – 324.
Campos, E. J. D., and D. B. Olson (1991), Stationary Rossby waves in western boundary currents extensions,J. Phys. Oceanogr.,21, 1202 – 1224.
Cresswell, G. R., and R. Legeckis (1986), Eddies off southeastern Austra- lia,Deep Sea Res., Part A,33, 1527 – 1562.
de Ruijter, W. P. M., P. J. van Leeuwen, and J. R. E. Lutjeharms (1999), Generation and evolution of natal pulses: Solitary meanders in the Agul- has Current,J. Phys. Oceanogr.,29, 3043 – 3055.
Ducet, N., P. Y. Le Traon, and G. Riverdin (2000), Global high-resolution mapping of ocean circulation from TOPEX/Poseidon and ERS-1 and2, J. Geophys. Res.,105(C8), 19,477 – 19,498.
Feng, M., H. Mitsudera, and Y. Yoshikawa (2000), Structure and variability of the Kuroshio Current in Tokara Strait,J. Phys. Oceanogr.,30, 2257 – 2276.
Feron, R. C. V. (1995), The southern ocean western boundary currents:
Comparison of fine resolution Antarctic model results with Geosat alti- meter data,J. Geophys. Res.,100(C3), 4959 – 4975.
Fu, L.-L., and R. D. Smith (1996), Global ocean circulation from satellite altimetry and high-resolution computer simulation,Bull. Am. Meteorol.
Soc.,77(11), 2625 – 2636.
Godfrey, J. S. (1973), Comparison of the east Australian current with the western boundary flow in BRYAN and COX’s (1968) numerical model ocean,Deep Sea Res. Oceanogr. Abstr.,20, 1059 – 1076.
Godfrey, J. S., G. R. Cresswell, T. J. Golding, and A. F. Pearce (1980), The separation of the east Australian current,J. Phys. Oceanogr.,10, 430 – 440.
Gordon, A. L., K. I. Horai, and M. Donn (1983), Southern hemisphere western boundary current variability revealed by GEOS 3 altimeter, J. Geophys. Res.,88(C1), 755 – 762.
Haidvogel, D. B., and A. Beckmann (1999),Numerical Ocean Circulation Modeling, 320 pp., Imperial Coll. Press, London.
Hernandez, F., P. Y. Le Traon, and R. Morrow (1995), Mapping mesoscale variability of the Azores current using TOPEX/POSEIDON and ERS 1 altimetry, together with hydrographic and Lagrangian measurements, J. Geophys. Res., 100(C12), 24,995 – 25,006.
Ichikawa, K., and S. Imawaki (1994), Life history of a cyclonic ring detached from the Kuroshio Extension as seen by the Geosat altimeter, J. Geophys. Res.,99(C8), 15,953 – 15,966.
Le Traon, P. Y., and G. Dibarboure (1999), Mesoscale mapping capabilities of mutiple-satellite altimeter missions,J. Atmos. Oceanic Technol.,16, 1208 – 1223.
Le Traon, P. Y., F. Nadal, and N. Ducet (1998), An improved mapping method of multi-satellite altimeter data,J. Atmos. Oceanic Technol.,15, 522 – 534.
Maltrud, M. E., and J. L. McClean (2005), An eddy resolving global 1/10 degrees ocean simulation,Ocean Modell.,8(1 – 2), 31 – 54.
Maltrud, M. E., R. D. Smith, A. J. Semtner, and R. C. Malone (1998), Global eddy-resolving ocean simulations driven by 1985 – 1995 atmo- spheric winds,J. Geophys. Res.,103(C13), 30,825 – 30,853.
Marchesiello, P., and J. H. Middleton (2000), Modeling the east Aus- tralian current in the Western Tasman Sea, J. Phys. Oceanogr., 30, 2956 – 2971.
Mata, M. M. (2000), On the mesoscale variability of the east Australian current at subtropical latitude, 253 pp., Ph.D. thesis, Flinders Univ. of South Aust., Adelaide.
Mata, M. M., S. Wijffels, M. Tomczak, and J. A. Church (1998), Direct measurements of the transport of the east Australian current: A data report from the WOCE Pacific current meter array 3,Data Rep. 234, CSIRO Mar. Lab., Hobart.
Mata, M. M., M. Tomczak, S. E. Wijffels, and J. A. Church (2000), East Australian current volume transports at 30S: Estimates from the World Ocean Circulation Experiment hydrographic sections PR11/P6 and the PCM3 current meter array,J. Geophys. Res.,105(C12), 28,509 – 28,526.
Mata, M. M., S. E. Wijffels, J. A. Church, and M. Tomczak (2006), Sta- tistical description of the east Australian current low-frequency variability from the WOCE PCM3 Current Meter Array,Mar. Freshwater Res., 57(3), 273 – 290.
McClean, J. L., A. J. Semtner, and V. Zlotnicki (1997), Comparisons of mesoscale variability in the Semtner-Chervin 1/4 model, the Los Alamos Parallel Ocean Program 1/6model, and TOPEX/POSEIDON data,J. Geophys. Res.,102(C11), 25,203 – 25,226.
Moore, M. I., and J. L. Wilkin (1998), Variability in the South Pacific Deep Western Boundary Current from current meter observations and a high- resolution global model,J. Geophys. Res.,103(C3), 5439 – 5457.
Morrow, R., J. A. Church, R. Coleman, D. Chelton, and N. White (1992), Eddy momentum flux and its contribution to the southern ocean momen- tum balance,Nature,357(6378), 482 – 484.
Nilsson, C. S., and R. G. Cresswell (1981), The formation and evolution of east Australian current warm-core eddies,Prog. Oceanogr.,9, 133 – 183.
Nof, D., and T. Pichevin (1996), The retroflection paradox,J. Phys. Ocea- nogr.,26, 2344 – 2358.
Olson, D. B. (1991), Rings in the ocean,Annu. Rev. Earth Planet. Sci.,19, 283 – 311.
Parks, T. W., and C. S. Burrus (1987),Digital Filter Design, 368 pp., John Wiley, Hoboken, N. J.
Pichevin, T., D. Nof, and J. Lutjeharms (1999), Why are there agulhas rings?,J. Phys. Oceanogr.,29, 693 – 707.
Qiu, B., and S. Chen (2004), Seasonal modulations in the eddy field of the South Pacific Ocean,J. Phys. Oceanogr.,34, 1515 – 1527.
Richardson, P. L., G. E. Hufford, R. Limeburner, and W. S. Brown (1994), North Brazil current retroflection eddies,J. Geophys. Res.,99(C3), 5081 – 5093.
Ridgway, K. R., and J. R. Dunn (2003), Mesoscale structure of the mean East Australian Current System and its relationship with topography, Prog. Oceanogr.,56, 189 – 222.
Ridgway, K. R., J. R. Dunn, and J. L. Wilkin (2002), Ocean interpolation by four-dimensional weighted least squares—Application to the waters around Australia,J. Atmos. Oceanic Technol.,19, 1357 – 1375.
Roughan, M., and J. H. Middleton (2004), On the East Australian Current:
Variability, encroachment, and upwelling,J. Geophys. Res.,109, C07003, doi:10.1029/2003JC001833.
Saunders, P. M., A. C. Coward, and B. A. de Cuevas (1999), Circulation of the Pacific Ocean seen in a global ocean model: Ocean Circulation and Climate Advanced Modelling project (OCCAM), J. Geophys. Res., 104(C8), 18,281 – 18,299.
Schlax, M. G., and D. B. Chelton (1994), Aliased tidal errors in TOPEX/
POSEIDON sea-surface height data,J. Geophys. Res.,99(C12), 24,761 – 24,775.
Silveira, I. C. A., G. R. Flierl, and W. S. Brown (1999), Dynamics of separating western boundary currents,J. Phys. Oceanogr.,29, 119 – 144.
van Leeuwen, P. J., W. P. M. de Ruijter, and J. R. E. Lutjeharms (2000), Natal pulses and the formation of Agulhas Rings, J. Geophys. Res., 105(C3), 6425 – 6436.
Wilkin, J. L., and R. A. Morrow (1994), Eddy kinetic energy and momen- tum flux in the southern ocean: Comparison of a global eddy-resolving model with altimeter, drifter, and current-meter data,J. Geophys. Res., 99(C4), 7903 – 7916.
Willebrand, J., R. H. Kaese, D. Stammer, H. H. Hinrichsen, and W. Krauss (1990), Verification of GEOSAT sea surface topography in the Gulf Stream extension with surface drifting buoys and hydrographic measure- ments,J. Geophys. Res.,95(3), 3007 – 3014.
J. A. Church and S. E. Wijffels, CSIRO Marine and Atmospheric Research, GPO Box 1538, Hobart, Tas 7001, Australia.
M. M. Mata, FURG – Departamento de Fı´sica, R. Alfredo Huch, 475, Rio Grande (RS) 96201-900, Brazil. ([email protected])
M. Tomczak, Flinders University of South Australia, SOCPES, P.O. Box 2100, Adelaide, SA 5001, Australia.
C09034 MATA ET AL.: EDDY SHEDDING AND ENERGY CONVERSIONS C09034
