1. Introduction
As the Loop Current flows into the Gulf of Mexico, it presses against the western side of the Yucatan Straits. The outflow from the Gulf constrains the Loop Current to form a large anticyclonic pattern, as shown in Fig. 1; eventually an anticyclonic ring separates from the main flow.
The ring shedding is quite erratic. A ring will sometimes separate quickly after the Loop Current has extended into the Gulf, but at other times several months will elapse before the ring detaches completely. In contrast, rings detach from the Gulf Stream almost immediately after they are formed (e.g., Auer 1987; Olson 1991). Leben (2005) has described as well as quantified the erratic behavior of ring separations based on sea surface height (SSH) from altimetry. Schmitz et al. (2005), although discussing the distinction between ring formation and detachment, show that rings do not separate unless there is a cyclonic ring on both sides of the flow to aid in the pinching-off process. Oey et al. (2005) have discussed ring formation from the point of view of numerical modeling. Of course, there is the possibility that the ring formations are chaotic, but Lugo-Fernandez (2007) concludes that they are not. There are many similarities between Loop Current and Gulf Stream ring separation events, although one major difference is that the Gulf Stream sheds both anticyclonic and cyclonic rings.
Figure 1a shows a typical ring separation. However, as often happens, the ring may become reattached to the main body of the Loop Current and separate later; such an event is shown in Fig. 1b.
In this paper, we have used the data and the metrics employed by Leben (2005) to determine when a ring has become detached: the abrupt decrease in area enclosed by the Loop Current, as described by the 17-cm height contour on SSH maps. We have compared these ring separations with the transport of the Florida Current as measured by the cable across the Florida Straits (see, e.g., DiNezio et al. 2009; Baringer and Larsen 2001). We have used other data as well, but the transport measurements emerge as the significant variable of interest.
When comparing these records, we noticed a remarkable and consistent feature: the transport of the Florida Current, as measured by the cable data, shows pulses of increased flow shortly before each separation event. These increases in transport are usually accompanied by increases in sea level on the offshore side of the flow; these have ∼5–10-cm amplitude and are of 20–25-days duration, suggesting that waves propagating in from the open Atlantic are instrumental in the ring separation trigger process.
We wish to emphasize that we are describing a set of observations that are associated with ring separations. Whether these events are indeed the operative mechanism requires further exploration. It is generally understood that ring separations result from a flow instability. Our study has focused on a decades-old question: Why is there often such a long and irregular delay between ring formation and the eventual detachment? An important motivating factor for our investigation is the fact that, to the best of our knowledge, modern numerical models are not able to predict when a ring will separate unless sea surface height data are assimilated. Thus, we have searched for mechanisms or processes that might be difficult to capture in a free-running model simulation. The trigger mechanism discussed here, in fact, appears to be pulses of transport that could easily be lost in the background of “ocean noise.” Also, the mechanism acts from the “downstream” direction in contrast to the potential upstream influences identified in both modeling (Murphy et al. 1999; Oey et al. 2003) and observational studies (Candela et al. 2002; Abascal et al. 2003).
Therefore, we emphasize that what we describe here as a possible trigger mechanism could also appear to a critical observer as merely noise in the data. The essential point is that these seemingly random pulses of transport occur throughout the record, and only when they occur when a ring is otherwise poised for detachment do they become an operative mechanism.
2. Data
The primary variable we use to describe the Loop Current is the sea surface height from the combined satellite altimetric record available since 1993 from the Ocean Topography Experiment (TOPEX)/Poseidon, European Remote Sensing Satellite-1 (ERS-1) and ERS-2, Geosat Follow-On, Jason-1, and Envisat satellite missions. Processing of the SSH data is based on near-real-time mesoscale analysis techniques designed to exploit the multisatellite altimetric sampling (Leben et al. 2002). All along-track data were referenced to a mean sea surface (Wang 2001) and detrended using an along-track loess filter that removes a running least squares fit of a tilt and a bias from the along-track data within a sliding 200-s window. Daily analysis maps of the detrended SSH anomaly were estimated using an objective analysis procedure (Cressman 1959) and added to a model mean to calculate the synthetic sea surface height estimates used in this study. A more complete description of the data processing as it relates to Loop Current monitoring may be found in Leben (2005).
The transport of the Florida Current is monitored by a cable between the U.S. east coast and the Bahamas, and data have been made easily available on a Web page maintained by C. Meinen (available online at http://www.aoml.noaa.gov/phod/floridacurrent/FC_cable_transport_2008.dat).
Sea level data at Settlement Point, Bahamas, are available from the Hawaii Sea Level Center (available online at http://uhslc.soest.hawaii.edu/uhslc), who also have the responsibility for maintaining that tide gauge. Key West tidal data are available at the primary National Oceanic and Atmospheric Administration (NOAA) data center (available online at http://tidesandcurrents.noaa.gov).
In many of the plots shown here the data have been normalized by their individual standard deviations (std dev) for the full record. For Loop Current area, the std dev is 2.89 × 1010 km2. For transport by cable, the std dev is ∼3.5 Sv for the ∼2000 days before an approximately 3-yr cable break near the middle of the record. The cable data have extremes of ±13 Sv, but some of this is attributed to the side-to-side fluctuations of the flow that are difficult to correct for, as well as for contributions from shelf waves propagating down the coast (Johns and Schott 1987; W. Schmitz 2009, personal communication). The std dev of sea level at Settlement Point is ∼8.3 cm for the duration we use here. We have used daily means of all records.
3. Results
Figure 1a shows the position of a LC ring as it is separating. It reattaches, as is sometimes the case, and its subsequent separation is shown in Fig. 1b. Figure 1c shows the positions of the various locations described here. Figure 2a shows a set of data for the first separation interval, highlighting both the typical signal contained in the individual time series and the noise levels of the data. This ring appears to separate in March 2006, when the Loop Current area drops abruptly, as shown by the vertical red line. The arrow in mid-February indicates an abrupt pulse of increased transport in the Florida Current cable data. Cyclonic features are found on both sides of the flow as the necking-down process progresses, as discussed by Schmitz et al. (2005). The amplitude of the transport pulse is ∼5 Sv, or ∼1.5 standard deviations of the transport, which is a substantial amount.
Figure 2b shows a slightly different set of data for the separation in September 2001. The cable was down, but sea level data are available on both sides of the flow. The red curve shows sea level on the offshore side, at Settlement Point, and the green curve shows the difference between Settlement Point and Key West. Because the two signals are so similar, it is reasonable to conclude that they mimic the transport between them, as we might expect, although that is sometimes not the case with these data.
Figure 2c shows a larger dataset for a different separation event when more of the relevant variables are available. A ring separates in July 1993; prior to the separation, there are two jumps in the transport of the Florida Current, which is indicated by the two arrows. The figure also shows the sea level signal at Settlement Point and the sea level difference across the flow, as in Fig. 2b, as well as the noise level in the data.
Although there is a troublesome level of noise in these data (by which we refer mainly to real oceanic variability, not to errors in measurements), in most cases the pulses stand out clearly from the noise. For those cases where the pulses of transport are ∼2 std dev, the sea level difference is also essentially ∼2 std dev. Two std dev of transport represents ∼7 Sv (a typical value in Fig. 4), whereas the sea level difference represents an increased difference of ∼17 cm. The means of these primary variables are ∼30 Sv and ∼75 cm, so these pulses represent (as would be expected) fluctuations of ∼23% in both values. This close agreement is, of course, reassuring.
Another obvious comparison would be to examine sea level on the inshore side of the stream. Because the Miami tide gauge has many breaks during this time period, we chose the record from Key West. It turns out that Key West sea level does not look at all like the record of transport or of ring separations. This holds true both for individual events and for the mean.
From a more basic point of view, the comparison between the transport as observed by cable and the sea level difference between the two tide gauges is worth examining and is shown in Figs. 2b,c. There are many cases during which the cable transport and sea level difference are quite similar, but there are a few times when there is no resemblance whatsoever. Although this is an interesting issue, it is not the point of the present work and is not pursued further here.
For completeness, plots of approximately half of the remaining separations covered by this analysis period are shown in the appendix. We examined each separation event to determine the delays between the pulses of transport and the following ring separation, knowing a priori that the ring had separated. Table 1 shows these values; the first entry in the table is for the first transport jump prior to the ring separation; the second transport jump, if it occurs within a reasonable time, is also shown. In three cases, the cable was out of commission, so sea level height at Settlement Point was used as a surrogate variable. Using the tide gauge data in those cases remains a bit speculative, but it allows us to examine all the separation events. The mean of all the first delay times is 13 ± 5 days. (If the three events using Settlement Point data were removed, the mean would change only in the first place beyond the decimal.)
There may appear to be a certain amount of subjectivity here in deciding exactly where the transport jumps occur. However, after examining all these separation events, the jumps in transport emerge as a consistent feature even if occasionally obscured by the noise in the different variables.
The distribution of the delay times is shown in Fig. 3. Although the individual pulses prior to the separations are a noisy dataset, in the mean they clearly are grouped into the bimodal pairing shown here; a pulse is seen roughly two weeks prior to each ring separation and a second one is seen roughly two weeks before that. We show, in a later section, that the variability in these delay times is significantly correlated with the length of the Loop Current before each ring separation. Because this result was encouraging, we examined the average of these events further in Fig. 4.
Figure 4a shows the transport anomaly during each of the shedding events relative to the time of separation as well as their mean; that is, the vertical line in the middle of the figure is the time of ring separation, the transport anomaly of each separation event is synchronized relative to that time, and all events are shown. A transport pulse in the mean (the blue curve) is seen ∼14 days preceding the ring separations and again at about twice that, which is consistent with the results in Table 1. Note also that a third pulse is seen shortly after the ring has separated, which is suggestive of a wave train.
Figure 4b similarly shows a set of all the data, again relative to the time of separation; the individual plots of sea level anomaly at Settlement Point, relative to the separations, make up the noisy dataset. The mean curve, however, is consistent with Fig. 4a. Sea level is high, in the mean, roughly two weeks before the ring separation events and again almost two weeks earlier than the first rise. High sea level on the offshore side of the flow of course is consistent with increased transport. Note that, in these figures, the nonnormalized values are shown so that the absolute amplitudes can be seen.
It is not surprising that the transport increases that appear here have a much reduced amplitude in the mean. We also find that third pulse of increased transport shortly after the ring separation. The three pulses are not always present in the individual separation events, in part because of noise in the data or because one or more of the individual data sources is not available.
a. Which comes first?
The increased sea levels on the offshore side can be found as a robust feature, as described in a later section, but it is a fair question to ask the following: Might the increased sea level on the right-hand side of the flow simply be a result of the increase in transport rather than a cause? To explore this question, we computed the cross-spectrum between sea level at Settlement Point and the cable transport (not shown). There is a small but consistent phase shift between the two, with sea level leading transport by roughly 10° at periods of ∼30 days. This phase shift is equivalent to roughly one day of lead and is in the correct sense to assure us that the increase in offshore sea level is a causative mechanism and is not a result of the increased transport. Although we can assume that this delay represents the response time of the flow within the straits, because this study is based on data with only daily resolution it is difficult to assign too much meaning to the one-day delay.
1) The incoming waves
Waves with a period of order one month are such ubiquitous features in the open ocean that one does not know whether to call them eddies, waves, or simply noise. In an early but profound investigation of such variability in ocean signals, Schroeder and Stommel (1969) referred to “month persistent” features, using the hydrographic and tide gauge data at Bermuda to show that they were indeed genuine features of oceanic processes and not merely errors or unexplained noise in the data.
The pulses of increased transport are similar to simple background noise, so we investigated the spectrum of sea level at Settlement Point. Figure 5 shows the spectrum computed from two data segments, each two years long (to avoid gaps in the data). The individual raw spectra were combined and smoothed slightly.
One feature of the spectrum should be emphasized. The power in the pulses we have been describing, having periods of order 20–30 days, is ubiquitous. A more detailed examination shows that there is a great deal of variability in the amplitude of these but no obvious seasonal dependence. To explore these incoming waves, we made animations of the (high passed) SSHA data from the AVISO open ocean product. These waves come in directly from the east and do not propagate across the ocean as typical low-frequency Rossby waves. The typical Rossby waves usually increase in amplitude as they cross the ocean; by contrast, these 20–30-day waves arise abruptly in the last 400 km or so from the coast and have wavelengths of the order of 200–250 km. We assume that they are locally wind driven, but we have not explored this further.
2) One speculative mechanism
There is an old idea that sea level in the Gulf must be slightly higher than in the Caribbean because (i) the northward inflow at Yucatan has to be reversed and (ii) the increased static head in the Gulf provides the necessary momentum to overcome the turbulent frictional losses on the inshore side of the Florida Current. We explored this idea by examining the mean sea level over the entire Gulf and over the entire Caribbean; Fig. 6 shows the mean values, plotted relative to the time of each separation. Although the resolution here is in 10-day increments, we see that sea level peaks approximately 20 days before the ring separations, in the mean.
A remarkable process shows up in Figs. 2 and 6; sea level rises on both sides of the outflowing current—but not by the same amount—before a ring separates. Sea level rises roughly 4 cm at the coast (Key West) and ∼8 cm on the offshore side (Settlement Point) during the 2-month interval shown here. The peaks in sea level at Settlement Point occur ∼10 days before the separation; the difference signal looks remarkably similar to sea level at Settlement Point near the two peaks prior to the separation. The difference between the two sea levels, which is (strictly) a measure of the surface velocity, follows the trend of the two on either side, with the difference rising ∼4 cm.
In terms of the sea surface velocity, the ∼4-cm increase is only a ∼5% increase. A possibly more significant factor, however, is the pressure head between the Gulf and open Atlantic. The downstream slope is on the order of 5 cm between the Keys and Cape Hatteras. As this pressure head in the Gulf increases, it has three possible consequences: first, it could increase the transport out of the Gulf in the Florida Current; second, it could decrease the transport into the Gulf from the Caribbean; and third, it could increase the deep transport out of the Gulf back to the Caribbean. Because we do not have adequate measurements of either the upper inflow or the deep outflow, much of our conclusions about the process here have to be speculative.
In a ground-breaking result from their mooring experiment in Yucatan, Bunge et al. (2002) showed that, as the Loop Current extended into the Gulf displacing large volumes of water in the upper layers, a compensating outflow was observed in the deep layers in Yucatan. What we see here is that the compensation is not always perfect and that before a ring separation there also could be some increase in the Florida Straits outflow.
One final part of the scenario comes into play. As the Loop Current extends farther and farther into the Gulf before a ring has detached, the result is that a substantial fraction of the transport that will go out through the Florida Straits has a longer pathway within the Gulf. The detachment process can be thought of merely as the point at which none of the Florida Straits outflow goes around the nascent Loop Current ring.
Although there is a great deal of scatter in the individual cases that form Fig. 6, the results from each separation are similar to the mean in 13 of the 20 events in our data. Thus, the idea of large-scale sea level pressure forcing remains a plausible result.
3) A testable, if speculative, hypothesis
To provide a specific mechanism for the ring separation triggers, we first review the details of a possible mechanism. A pulse of increased transport is observed. This change in transport must be propagated upstream. The path this pulse follows and the precise dynamics of the transmission are not understood and remain to be explored. The pulse follows an unspecified path until it reaches the region where the ring is about to separate. To test this possibility, we have compared the observed delays with the length of the Loop Current, as shown in Fig. 7. The scatter in this comparison is reduced dramatically if we examine only delays longer than 13 days. That is, the first 12–14 days of the delay appear to be an initiation phase, which we do not attempt to explain here; this feature requires further understanding and analysis. Nevertheless, the result of these observations is robust and thus seems worth presenting. The maximum in correlation is found in a broad region between 10 and 15 days.
The delay time, as shown in Fig. 7, is significantly correlated (0.7 ± 0.002) with the length of the Loop Current at the time of the separation. The slope of the line in Fig. 7 is 0.0084 days km−1, or 1.38 ± 0.34 m s−1. The length of the Loop Current here, from the statistics we have used previously, is actually the total length around the entire Loop Current; the relevant length for propagation of the signal is roughly half the value used for these points, because the trigger signal does not propagate all the way around the circumference of the Loop Current but propagates only to the point of separation of the ring. These values lead to an effective speed of approximately 2.8 ± 0.7 m s−1, a reasonable value for the speed of the Loop Current. This value can be most easily understood as a time that the trigger pulse is carried along with the flow of the Loop Current from Yucatan to the position of nascent ring separation. A travel time of only ∼14 days for the initial pulse to go from the Florida Straights to Yucatan, if it were to propagate as a baroclinic Kelvin wave around the perimeter, seems too short by at least a factor of 2. Note, however, that in the insightful work of Bunge et al. (2002) they found an unexplained delay of “approximately a week” between the intrusion cycle of the Loop Current and the deep outflow associated with it.
The intercept of the line in Fig. 7 is, to within one standard error, the 13-day cutoff we have chosen. Thus, the results in Fig. 7 give increased confidence that the mechanism is consistent with the observed data.
4. Discussion
For this analysis, the time when the rings separate is known a priori. The analysis cannot proceed without this knowledge.
The idea that pulses of transport could be the trigger for ring separations has been discussed for many years. Pulses of increased transport in the Agulhas Current, for example, have been shown to be associated with the separations of rings from the Agulhas Current (e.g., van Leeuwen et al. 2000, Lutjeharms et al. 2001), although many of the details there are different. This effect has been observed for nearly two decades.
It is noteworthy that Pichevin and Nof’s (1997, their Fig. 8) results show a similar effect. In their analytical model, there is a clear correlation between the pinching off of rings and the transport of the flow.
Similarly, in Leben’s (2005) analysis, his Fig. 6 shows a striking correlation between the ring separation times and the length of the Loop Current. Perhaps this correlation can be understood as a result of the greater time that has elapsed before the separation event occurs, giving the Loop Current more time to grow.
Our basic motivation here is to find a mechanism that could serve as a trigger for ring shedding but may not be apparent or computed well enough in an otherwise quite capable global circulation model. What emerges here is that pulses of transport in the Florida Current, accompanied by increased sea level on the offshore side of the stream, appear to be an operative mechanism. Although what we are describing here as signal is frequently obscured by noise, this may well be its most important feature.
It may be worth mentioning that there is a contribution of sea level variability near 14 days associated with the fortnightly tide. However, repeating the calculation for Fig. 5 with the fortnightly tide removed does not change the power in the 20–30-day band. The fact that, on average, the trigger pulses occur about 14 days before a separation suggests that the additional energy contributed by the fortnightly tide may possibly be enough to change what would otherwise be merely noise into an effective trigger mechanism; that is, although this is indeed speculative, when the essentially random ∼25-day signals coincide with the fortnightly tide, the pulses become large enough to be a trigger. Whether this is an important aspect of the process we cannot say.
It often (but not always) appears that there is a drop in transport about the time the ring separates. Although there is no doubt that this is an accurate observation, the point of our study is to find a mechanism or event that precedes the separation event and that could possibly be a causative mechanism. The fact that the amplitudes of the transport pulses agree, via geostrophy, with the amplitudes of the sea level difference signal is a meaningful finding, and it suggests that these pulses are real oceanic signals. So, these pulses of transport might indeed appear merely to be noise in the data. And yet if the Loop Current is in a position where a ring-shedding event is poised to happen—but the separation has not yet occurred—we suggest that these pulses provide the mechanism that causes the instability to go forward and the rings to separate.
One of the puzzling differences between the events shown here and the shedding of Gulf Stream rings is that Gulf Stream rings detach almost immediately after they have formed. Because the Gulf Stream, on the western side of the ocean, is exposed to a nearly constant barrage of eddy noise, this trigger mechanism may be present all the time. The eastern Gulf of Mexico is relatively shielded from such incoming wave radiation; therefore, the Loop Current rings appear to require the sort of extra stimulus we have described.
Acknowledgments
During the course of this work, we have benefited from discussions with our colleagues A. Clarke, W. Dewar, D. Nof, and W. Schmitz. C. Meinen was most helpful in arranging for our use of the Florida Current cable data; these data are available on their Web site (available online at http://www.aoml.noaa.gov/phod/floridacurrent/) and funded by the NOAA Office of Climate Observations. R. Leben acknowledges the support of Minerals Management Service, Gulf of Mexico OCS Region Contract M08PC20043 to Science Applications International Corp. and NASA Ocean Surface Topography Mission Science Team Grant NNX08AR60G. Altimetry was obtained from ESA, NOAA, and the Physical Oceanography Distributed Active Archive Center (PO.DAAC) at the NASA Jet Propulsion Laboratory, Pasadena, California (available online at http://podaac.jpl.nasa.gov), and support as PO.DAAC Ocean Surface Topography Project Scientist is also greatly appreciated. W. Sturges is grateful for support by the MMS as well as by NSF OCE-0326233 and OCE-0925404.
REFERENCES
Abascal, A. J., J. Sheinbaum, J. Candela, J. Ochoa, and A. Badan, 2003: Analysis of flow variability in the Yucatan Channel. J. Geophys. Res., 108 , 3381. doi:10.1029/2003JC001922.
Auer, S. J., 1987: Five-year climatological survey of the Gulf Stream system and its associated rings. J. Geophys. Res., 92 , 11709–11726.
Baringer, M. O., and J. C. Larsen, 2001: Sixteen years of Florida current transport at 27°N. Geophys. Res. Lett., 28 , 3179–3182.
Bunge, L., J. Ochoa, A. Badan, J. Candela, and J. Sheinbaum, 2002: Deep flows in the Yucatan Channel and their relation to changes in the Loop Current extension. J. Geophys. Res., 107 , 3233. doi:10.1029/2001JC001256.
Candela, J., J. Sheinbaum, J. Ochoa, A. Badan, and R. Leben, 2002: The potential vorticity flux through the Yucatan Channel and the Loop Current in the Gulf of Mexico. Geophys. Res. Lett., 29 , 2059. doi:10.1029/2002GL015587.
Cressman, G. P., 1959: An operational objective analysis system. Mon. Wea. Rev., 87 , 367–374.
DiNezio, P. N., L. J. Gramer, W. E. Johns, C. S. Meinen, and M. O. Baringer, 2009: Observed interannual variability of the Florida Current: Wind forcing and the North Atlantic Oscillation. J. Phys. Oceanogr., 39 , 721–736.
Johns, W. E., and F. Schott, 1987: Meandering and transport variations of the Florida Current. J. Phys. Oceanogr., 17 , 1128–1147.
Leben, R. R., 2005: Altimeter-derived loop current metrics. Circulation in the Gulf of Mexico: Observations and Models, Geophys. Monogr., Vol. 161, Amer. Geophys. Union, 181–202.
Leben, R. R., G. H. Born, and B. R. Engelbreth, 2002: Operational altimeter data processing for mesoscale monitoring. Mar. Geod., 25 , 3–18.
Lugo-Fernandez, A., 2007: Is the Loop Current a chaotic oscillator? J. Phys. Oceanogr., 37 , 1455–1469.
Lutjeharms, J. R. E., O. Boebel, P. C. F. van der Vaart, W. P. M. de Ruijter, T. Rossby, and H. L. Bryden, 2001: Evidence that the natal pulse controls the Agulhas Current over its full depth. Geophys. Res. Lett., 28 , 3449–3452.
Murphy, S. J., H. E. Hurlburt, and J. J. O’Brien, 1999: The connectivity of eddy variability in the Caribbean Sea, the Gulf of Mexico, and the Atlantic Ocean. J. Geophys. Res., 104 , 1431–1453.
Oey, L-Y., H-C. Lee, and W. J. Schmitz Jr., 2003: Effects of wind and Caribbean eddies on the frequency of Loop current eddy shedding: A numerical model study. J. Geophys. Res., 108 , 3324. doi:10.1029/2002JC001698.
Oey, L-Y., T. Ezer, and H-C. Lee, 2005: Loop Current, rings, and related circulation in the Gulf of Mexico: A review of numerical models and future challenges. Circulation in the Gulf of Mexico: Observations and Models, Geophys. Monogr., Vol. 161, Amer. Geophys. Union, 32–56.
Olson, D. B., 1991: Rings in the ocean. Annu. Rev. Earth Planet. Sci., 19 , 283–311. doi:10.1146/annurev.ea.19.050191.001435.
Pichevin, T., and D. Nof, 1997: The momentum imbalance paradox. Tellus, 49A , 298–319.
Schmitz Jr., W. J., D. C. Biggs, A. Lugo-Fernandez, L-Y. Oey, and W. Sturges, 2005: A synopsis of the circulation in the Gulf of Mexico and on its continental margins. Circulation in the Gulf of Mexico: Observations and Models, Geophys. Monogr., Vol. 161, Amer. Geophys. Union, 11–30.
Schroeder, E. H., and H. M. Stommel, 1969: How representative is the series of Panulirus stations of monthly mean conditions off Bermuda? Prog. Oceanogr., 5 , 31–40.
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.
Wang, Y. M., 2001: GSFC00 mean sea surface, gravity anomaly and vertical gravity gradient from satellite altimeter data. J. Geophys. Res., 106 , 31167–31174.
APPENDIX
Additional Plots Showing All Available Data for More Separation Events
Figures A1 and A2 show the observations for eight additional separation events.
a. Three perspectives of the initial detachment of Eddy Xtreme: (left) CCAR SSH data, (right) the simultaneous GOES sea surface temperature (SST) data, and (middle) both for 1, 8, and 15 Mar 2006. GOES SST data with the CCAR SSH contours are overlaid. The SSH data are in centimeters, and the SST data are in degrees Celsius. The detachment took place on 8 Mar and is marked by the breaking of the 17-cm height contour.
Citation: Journal of Physical Oceanography 40, 5; 10.1175/2009JPO4245.1
b and c. (b) As in (a), but for reattachment and final separation of Eddy Extreme for 10, 17, and 24 Apr 2006. Eddy Xtreme briefly reattached to the Loop Current on 17 April, as shown by the reconnection of the 17-cm height contours; warm Loop Current water is seen being entrained by the eddy. It separated completely from the Loop Current, based on the 17-cm contour, on 18 Apr. (c) Locations of the datasets used here: the dashed line from Florida across the Florida Straits indicates the approximate location of the undersea cable, Settlement Point is at the northwest tip of the Bahamas island at the eastern end of the cable, and Key West is at the southernmost tip of Florida. Contours show SSHA at an essentially arbitrary time to show variability in a single daily map, in 5-cm increments; dashed contours are negative. SSHA contours are not shown where depth is less than 50 m.
Citation: Journal of Physical Oceanography 40, 5; 10.1175/2009JPO4245.1
(a) Loop Current area and Florida Current transport at the time of a Loop Current ring separation. The Florida Current transport is from cable data. Daily values are used for all variables. The vertical dashed red line indicates the time of ring separation. The vertical black arrow points to a pulse of increased transport, as shown in Table 1. (b) As in (a), also showing sea level at Settlement Point, Bahamas, but for a time when the cable transport data are not known. The similarity between the Settlement Point sea level and the sea level difference across the flow suggests that the transport signal would be similar. (c) As in (b), but at the time of the Loop Current ring separation of July 1993, also showing the sea level difference between Settlement Point and Key West. The vertical black arrows point to pulses of increased transport, as shown in Table 1.
Citation: Journal of Physical Oceanography 40, 5; 10.1175/2009JPO4245.1
A histogram of the delay times given in Table 1, as derived from the individual separation events.
Citation: Journal of Physical Oceanography 40, 5; 10.1175/2009JPO4245.1
(a) The transport anomaly of the Florida Current from cable data (black lines) for each of the Loop Current shedding events, relative to the time of shedding. The blue curve shows the mean over all separations. Note that the cable transport values are not normalized here. (b) Sea level height at Settlement Point (black lines) for each of the Loop Current shedding events, relative to the time of shedding. The blue curve shows the mean over all separations. Note that the sea level values are not normalized here. The mean value for this record is 3064 mm.
Citation: Journal of Physical Oceanography 40, 5; 10.1175/2009JPO4245.1
A variance-preserving spectrum of sea level at Settlement Point using two segments of 2 yr each. The 12.0- and 6.0-month terms were removed before computing the spectrum.
Citation: Journal of Physical Oceanography 40, 5; 10.1175/2009JPO4245.1
Mean sea level (from SSH) over the Gulf of Mexico, the Caribbean Sea, and their difference, relative to the time of separation of all rings in this discussion. This dataset has only 10-day resolution. The mean values (mm) for the Gulf and Caribbean are shown.
Citation: Journal of Physical Oceanography 40, 5; 10.1175/2009JPO4245.1
Correlation plot of delay time (y axis) as a function of the length of the Loop Current (x axis). The line shows the least squares fit, with a correlation coefficient of −0.7.
Citation: Journal of Physical Oceanography 40, 5; 10.1175/2009JPO4245.1
Fig. A1. (a) Data for the September 1993 Loop Current ring separation. The “Sett Pt” curve (red) is sea level anomaly at Settlement Point, Bahamas. All data are normalized by their own std dev. The vertical arrow shows the position of the pulse of transport prior to ring separation as entered into Table 1 of the text. For this separation, there is little or no agreement between the transport pulse and the Settlement Point sea level. (b) The August 1994 separation. (c) Data for the March 1996 ring separation. The sea level difference curve (green) is for Settlement Point minus Key West. (d) Data for the March 1998 ring separation.
Citation: Journal of Physical Oceanography 40, 5; 10.1175/2009JPO4245.1
Fig. A2. (a) Data for the October 1999 ring separation. The cable was down for this separation, so the delay times shown in Table 1 were determined from the two pulses indicated at the Settlement Point tide gauge. (b) Data for the March 2002 Ring separation. No data are available at Settlement Point. The flat spot at the beginning of the plot is when the cable was out. (c) Data for the August 2004 Ring separation. No data are available at Settlement Point. The flat spot at the end of this plot is for a time when the cable was out. (d) Data for the February 2006 Ring separation. No data are available at Settlement Point.
Citation: Journal of Physical Oceanography 40, 5; 10.1175/2009JPO4245.1
Delays between the increased pulses of Florida Current transport and the subsequent separation of a ring, as shown in Fig. 2. The “previous” column shows the delay for the previous pulse of increased transport for those occasions when one was apparent. When the transport values were not available because the cable was down, Settlement Point sea level was used as a surrogate for separations 10–12.
