A climatology of various ocean features in the Gulf of Mexico (GOM) was developed using the combination of satellite remote sensing and in situ data that spanned periods as long as 32 years. Twelve separate statistics were created, some of which described characteristics of the Loop Current, while others are involved with warm core rings that separate from the Loop Current and cold core rings. These statistics examined the frequency with which the Loop Current was found in the GOM, the orientation of the Loop Current, the frequency of intrusion of Loop Current water onto the west Florida shelf and into the GOM common water region, ring separation period from the Loop Current, ring dissipation, ring speed, ring path, frequency of ring water in the western GOM, and the frequency of warm and cold core rings (WCRs and CCRs, respectively) in the GOM. The results indicate that CCRs were principally responsible for mass and heat redistribution in the eastern GOM (EGOM) and WCRs are responsible for mass and heat redistribution in the western GOM (WGOM). The average period for WCR separation from the Loop Current was 11 months and the range from 5 to 19 months. WCRs moved through the WGOM most often using the central path (i.e., their trajectory was found between 24° and 26°N latitude) and they decreased to about 55% of their initial size when they reach the western wall of the GOM. CCRs were most often found in the EGOM, and their frequency of occurrence in the EGOM surpassed that of WCRs anywhere in the GOM.
In the late 1980s, a climatology (Vukovich and Hamilton 1990, 163–168) of a limited number of oceanographic features in the Gulf of Mexico (GOM) was created that was based on satellite remote sensing and in situ data. A climatology of other features in the GOM has also been reported in the literature (e.g., Vukovich 1988a, 1995; Herring et al. 1999; Vukovich et al. 1978, 1979a, b; Sturges 1992, 1994; Sturges and Leben 2000; Jacobs and Leben 1990). These features include the frequency of the Loop Current in the eastern GOM, Loop Current eddy separation periods, cold core rings (CCRs) on the Loop Current boundary, and warm core ring (WCR) water in the GOM. Up to 26 years of data were used in at least one of those studies, but most used datasets that covered about a 10–15-yr period. Presently, combined satellite and in situ datasets that span periods up to 32 years can be used to develop a climatology of certain features in the GOM.
The data resources used to create the climatology presented in this paper included sea surface temperature (SST) data from the Television and Infrared Observation Satellite-M (TIROS-M), Heat Capacity Mapping Mission (HCMM), Seasat, Geostationary Operational Environmental Satellite (GOES), and numerous National Oceanic and Atmospheric Administration (NOAA) satellites; ocean color data from Coastal Zone Color Scanner (CZCS), Sea-viewing Wide Field of View Sensor (SeaWiFS), and Moderate Resolution Imaging Spectroradiometer (MODIS); altimeter data from Ocean Topography Experiment (TOPEX)/Poseidon, JASON, and European Remote Sensing Satellite (ERS); analyzed in situ data from ships of opportunity and from Mineral Management Service (MMS) field programs in the GOM; and information from the GOM programs managed by the various oil and gas companies. It is expected that the information derived from this climatology may be useful in planning field programs, to marine biologists and oceanographers, as background information for environmental impact statements (EISs), for oil and gas exploration and exploitation in the GOM, and in evaluating and directing improvements for models that calculate the ocean dynamics in the GOM (Herring et al. 1999).
The principal dataset used to derive many of the statistics is monthly analyses of the fronts in the GOM, which provided the “characteristic” position of the fronts for a given month. These were created using an integral of available clear-sky satellite images and any available information on ocean features and/or analyzed in situ data available for the month in question. The Loop Current front in these analyses was the approximate average front for the month unless a WCR separated from the Loop Current during that month (Fig. 1), in which case it was the position of front after separation. WCR fronts are the approximate average front for the month. If a warm Loop Current water mass not directly associated with the Loop Current or with major WCRs was found in the eastern GOM (EGOM), then it was represented in the frontal analysis by its most widespread effect in the EGOM for that month. If an intrusion of warm Loop Current water onto the west Florida shelf (WFS) was found in the EGOM in a given month, then it was also represented by its most widespread effect on that shelf for that month.
The development of these frontal analyses was based, for the most part, on satellite remote sensing data. In the periods 1976–78 and 1986–91, only SST data were available to create the ocean front analysis. Most of the SST data were obtained from the NOAA/Advanced Very High Resolution Radiometer (AVHRR), though significant use was also made of GOES and HCMM data in that period. As a result, the frontal analyses could only be determined for as little as 5 months and as many as 8 months in a year, depending on when the mixed layer developed in the spring/summer and eroded in autumn. However, in the periods when SST data were not useful, significant use was made of “ship-of-opportunity” data to fill data gaps (e.g., position of the northern boundary of the Loop Current, the location of WCR centers, the month in which separation of a WCR from the Loop Current occurred, etc.). In the period 1979–85, CZCS data were used to supplement the SST data, which were used to detect ocean features in the warm season. After 1991, altimeter data were available and the frontal analyses for most features could be developed for all 12 months in the year. However, significant use of SST and ocean color data was also made during that period.
For the statistics dealing with the spatial frequency Loop Current water, ring water, warm and cold core rings, etc., a latitude–longitude grid was constructed and overlaid onto the frontal analyses. The spatial resolution of the grid was 0.5° for spatial frequencies of water masses and 1° for the spatial frequencies of warm and cold core ring centers. In the cases dealing with the frequency of water masses, if the water mass occupied more than 50% of any grid square in a given month, then it was assumed that the water mass was present in that grid square at that time. Unless otherwise specified, the spatial frequency of water masses was derived using a 28-yr database (1976–2003). In all cases, hand digitization was used, which was the simplest method to provide decisions on those cases when only a portion of the grid square was occupied by the water mass. Procedures for statistics other than spatial frequencies are discussed separately in the section discussing the statistic.
3. Loop Current and Loop Current water
a. Frequency of Loop Current water in the EGOM
The Loop Current enters the EGOM through the Yucatan Channel and exits through the Straits of Florida, penetrating northward in the EGOM until instabilities form in the current and a WCR separates. As a result, the amount of time the water mass directly associated with the Loop Current resides at locations in the EGOM varies from one location to the next. The statistic presented in this section is the spatial frequency of water directly associated with the Loop Current. This statistic may also be used to provide a crude estimate of the spatial frequency of the location of high-speed current associated with the Loop Current front in the EGOM.
The Loop Current frontal boundary (Fig. 2) seldom reached the 28° latitude line (i.e., about 5% of the time over the 28-yr period). It reached the 27° latitude line about 20% of the time. The Loop Current usually penetrates about as far north as 27°N latitude just prior to WCR separation. The water mass associated with the Loop Current occupied the region south of 24°N latitude and east of 86°W longitude the greatest amount of time. The analysis shows that there is a mean westward tilt (i.e., a westward tilt of approximately 30°) of the axis of the frequency analysis. The variation of the orientation (i.e., the westward tilt) of the Loop Current is associated with WCR separation and will be discussed in detail in the next subsection.
b. The orientation of the Loop Current
The Loop Current orientation is defined as the angle made by the intersection of a line drawn parallel to the eastern and western frontal boundaries of the Loop Current through the frontal axis and a longitude line (Fig. 3). The orientation of the Loop Current changes in association with processes involved with cycle of the Loop Current, which was described above. The Loop Current usually tilts greatest to the west at the time a WCR is about to separate. Recently, large orientation angles (i.e., the order of 60°) and considerable westward extension of the Loop Current have been observed (Fig. 4), in which case the Loop Current influenced the central GOM and parts of the western GOM (WGOM). After the ring separates, the Loop Current usually reestablishes itself around 25°N and, at that time, is usually oriented north–south (i.e., the orientation angle is 0°).
The Loop Current orientation angle was determined for each month using the monthly frontal analyses for the period 1976–2003. The data were used to determine standard statistics and a frequency distribution for the orientation angle. The frequency distribution was established by binning the data in 5° increments.
About 80% of the angles (Fig. 5) lie between 0° and 30° (i.e., the Loop Current orientation is between north–south and north-northwest–south-southeast). Orientation angles less than 30° are normally associated with a relatively stable Loop Current (i.e., a Loop Current not about to shed a WCR anytime in the near future). Orientation angles greater than 30° are found about 20% of the time. The primary mode in the histogram is in the 0°–4° band. Secondary modes are noted in the bands 10°–14° and 20°–24°. These bands comprise orientation angles for a stable Loop Current. The Loop Current, at times, persists in these orientations for extended periods. Over the 28-yr period 1976–2003, the average orientation angle was about 17° with a standard deviation of about 14°. The single most observed orientation angle (i.e., the mode of the dataset) was 0°—a north–south orientation of the Loop Current. The maximum angle observed over the period was 63°. An angle greater than 55° only occurred twice over the 28-yr period, and angles between 50° and 54° were also observed only twice.
c. Frequency of isolated warm water in the EGOM
Warm pools of water that become isolated from the Loop Current are often found in the EGOM. These can be established by minor, short-lived rings that separate from the Loop Current and dissipate in the EGOM, by transport of Loop Current water associated with the circulation of CCRs found on the boundary of the Loop Current (Vukovich 1988a; Wang et al. 2003) (Fig. 6), or by dragging surface water from the Loop Current northward through the action of a strong wind stress. These pools are an indication of the transport of Loop Current water mass and of any foreign mass (i.e., oil spilled, dinoflagellates, etc.) residing in the Loop Current, from the Loop Current into the common water area of the EGOM.
Pools of Loop Current water were found throughout the region north of 25°N in the EGOM over the period (Fig. 7). The highest frequency at which those pools were found was about 14% and the center of high frequency was located immediately west of the shelf break of the WFS at about 27.25°N, 85.25°W. This feature was created, for the most part, as a result of many cases in which there was transport of warm Loop Current water by the circulation associated with CCRs located on the northeastern boundary of the Loop Current. The 14% maximum frequency corresponds to about one event every 10 months or about six events in 5 years.
It should be noted that the association of a frequency value with an event, which is directly related to a process, in this case and cases discussed later in this paper, does not necessarily refer to separate events. This is particularly true in those cases when a frequency value is associated with more than one event per year. A feature may persist for more than a month in a given location, creating a frequency that is related to the feature’s persistence rather than to multiple occurrences of similar features.
d. Frequency of Loop Current water on the west Florida shelf
The circulation associated with CCRs on the eastern boundary of the Loop Current occasionally transports warm Loop Current water onto the WFS (Fig. 6). In time, the Loop Current water mass mixes with the shelf water so that the Loop Current water is no longer distinguishable from shelf water using satellite data. As in the previous case, the transport of Loop Current water onto the shelf also means that any foreign mass residing in the Loop Current will also be transported onto that shelf. This analysis provides specific information on the spatial frequency of warm Loop Current water that was transported onto the WFS and, in so doing, provides information on the frequency of processes that cause this specific kind of transport.
The analysis shows that intrusions of Loop Current water onto the shelf occurred most often south of 27°N between the shelf break (i.e., the 200-m isobath) and the 83°W longitude line (Fig. 8); that is, most of these intrusions occurred near the shelf break and do not penetrate deep onto the shelf. The apparent lack of identifiable deep penetration of these features on the shelf is presumably due to mixing. It also appears that the intrusions of Loop Current water onto the shelf occurred most often when a CCR is located on the southeastern portion of the Loop Current front near Dry Tortugas (i.e., the southern most portion of the WFS), where the 200-m isobath runs east–west, and the currents in the intrusion were from the south to the north, for the most part (Vukovich 1986). The maximum spatial frequency for these events is about 12%, which corresponds to a frequency of about one event every year.
The circulation on the WFS can possibly augment or deter the effects of the intrusions of Loop Current water onto the shelf. A program sponsored by MMS examined the circulation on the WFS in 1996. The shelf was seeded with hundreds of satellite-tracked surface drifters during that year. The drifter data was used to determine the characteristic circulation on the shelf for a given month by integrating all drifter locations for that month. The drifter data on the WFS for March, August, and November 1996 (Fig. 9) shows flow to the south-southeast in March and November 1996 and flow to north-northwest in August 1996. Southerly currents were found on the shelf in the fall and winter in 1996 (Fig. 10). Since the currents in the intrusions are very often from south to north, the southerly current would act opposite to the current in the intrusion, preventing widespread transport of the Loop Current water onto the shelf. Northerly currents, which would augment widespread transport of Loop Current water onto the shelf, were noted in the spring and summer. Data were not available to determine whether the 1996 data represent the characteristic cycle of the surface flow on that shelf each year. However, the shelf currents are controlled by the wind (He and Weisberg 2003), and they will vary as the year-to-year wind climatology varies.
4. Warm core rings
a. Frequency of major ring separations
The separation periods of large WCRs from the Loop Current that have been reported in the literature are highly variable: 8–9 months (Maul and Herman 1985); 11 months (Vukovich 1988a, 1995; Maul and Vukovich 1993); 12 months (Sturges and Evans 1983); bimodal periods of 8–9 months and 13–14 months (Sturges 1992, 1994); and bimodal periods 6 and 11 months with a minor mode at 9 months (Sturges and Leben 2000). Vukovich (1988a, 1995) and Maul and Vukovich (1993) noted that the 11-month period that they cited is an average value, the eddy separation periods were highly variable ranging from 6 to 17 months, and the period was seldom the same as the average period or any of the other values cited in the literature.
The studies cited above used datasets with periods as small as 10 yr and as large as 26 yr. The variable nature of the results from these studies is, in part, a testimonial to the limitations in deriving statistics using datasets that are small when the statistic deals with events that occur a very limited number of times in a year. Vukovich (1995) suggested that it may be necessary to require datasets that cover a 100-yr period in order to obtain a statistically stable estimate of the eddy-shedding period. The eddy-shedding period was reexamined using two procedures: a histogram of eddy-shedding periods using a 32-yr dataset (1972–2003) and spectral analysis of the Loop Current northern boundary oscillations using a 27-yr dataset (1977–2003).
For the histogram procedure, the eddy-separation periods were determined by documenting the month and year of the separation of each major ring. A major ring was defined as rings with diameters of about 300 km or more at the time they separated from the Loop Current, which persisted for more than five months. The month and year of separation was that time when the major eddy completely separated from the Loop Current. In some cases, a WCR separated from the Loop Current only to be reabsorbed. Under normal circumstances, one major WCR separated from the Loop Current each year. In some years, two rings separated in different months and as little as 5 months apart. In a few years, no ring separation was noted. In 2002, two major rings separated from the Loop Current in the same month. Two rings also separated from the Loop Current in the same month in 2003, but one of these dissipated quickly (i.e., in about 3 months), so it was not considered to be a major ring.
The histogram (Fig. 11) of the Loop Current’s eddy-shedding periods shows that the frequency distribution is bimodal with primary modes at 6 and 11 months, similar to that found by Sturges and Leben (2000). The primary modes have frequencies of about 14%. The range in the data is 5 to 19 months, and the average eddy-shedding period is 11 months with a standard deviation of ±4 months.
The histogram for the Loop Current’s eddy shedding as a function of the month in the year (Fig. 12) shows that the mode of the frequency distribution is at month 3 (March), the average is month 6 (June), and the standard deviation is ±3 months. These data indicated that about 45% of the major WCRs separated from the Loop Current in the late winter and spring period (i.e., March, April, May, and June). Over the 32-yr period, no WCRs separated from the Loop Current in month 12 (i.e., December) and the net frequency for eddy shedding is a minimum in months 11 and 12 (i.e., November–December).
For the second approach, a time series of the mean monthly distances between the northernmost boundary of the Loop Current and the 30°N latitude line (Fig. 13) was developed using the frontal analyses. The average distance for a given year was calculated using the monthly values and was subtracted from each corresponding monthly value to produce detrended, monthly displacements. A sign convention was developed such that positive displacements were to the north (Fig. 14). It should be noted that, when the Loop Current penetrates into the EGOM, the northern boundary generally moves north of its mean annual position. When a major eddy separates, the northern boundary moves abruptly south of its mean annual position (Fig. 15). As a result, the Loop Current’s northern boundary moves north and south of its mean annual position with a frequency, which is, for the most part, associated with the eddy-shedding period. Therefore, the dominant periods from the spectral analysis of these displacements should provide an indication of the eddy-shedding period.
Spectral analysis was applied to the time series of detrended, monthly displacements and one Hanning smoothing of the spectrum was applied. The spectrum (Fig. 16) shows that the most significant variance has a period of 12.0 months. There are secondary peaks in the spectrum at 9.5 months and 8.1 months. The histogram has a secondary mode at 9 months.
The period having the most significant variance in the spectrum varied by one month [i.e., from 12.2 months to 11.1 months and then back to 12 months] over the 10-yr period 1994–2003 as each year of data was added to the dataset of Loop Current’s northern boundary displacements (Table 1), which is probably a reflection of the variability of the eddy-shedding period over that period. On the other hand, the average eddy-shedding period from the histogram data was about 11 months each year over the same 10-yr period, suggesting that the average period may be a stable statistic.
b. Ring path
When major WCRs separate from the Loop Current, their direction of motion is generally westward until they reach the “western wall” of the WGOM (Vukovich and Waddell 1991; Elliott 1982; Hamilton et al. 1999). During their movement from the EGOM through the WGOM, some major rings (i.e., those having diameters of about 300–400 km at the time of separation) have their centers located north of 26°N most of the time so that the circulation associated with these rings impacts the Louisiana and Texas shelf and the slope region west of 90°W. Many rings have their centers located between 24° and 26°N when they move through the WGOM and travel through the deep Gulf regions (i.e., water depths ≥2000 m) for the most part, but most of these WCRs will affect the slope region west of 90°W, especially when their path is in the northern part of the 24°–26°N latitude belt. Some rings move southwestward into the WGOM and are found south of 24°N for a period of time, and then turn northward as they approach the western wall. After reaching the western wall, the WCRs generally have no prescribed direction of motion (Hamilton et al. 1999). The modeling results of Zavala et al. (1998) suggest that the rings move southward after interacting with a western boundary. Vukovich and Waddell (1991) observed that a ring moved northward, then southward after colliding with the wall. Biggs et al. (1996) noted that a ring moved into the northwest corner of the Gulf after reaching the western wall. In examining the last 10 years (1994–2003) of altimetry data, WCRs have been observed move northward, southward, and eastward after colliding with the western wall. Modeling and observational results agree on one factor: the size of the ring decreases markedly after colliding with the wall (Zavala et al. 1998; Vukovich and Waddell 1991; Vukovich and Crissman 1986).
The frequency at which major WRCs move along the three paths described above was determined using data for the 28-yr period, 1976–2003. For each month in the analysis period, the mean positions of the centers of the major WCRs were determined. Major rings that coalesced with other rings and rings that divide, producing two rings, during their motion were excluded from this analysis. It should be noted that only major rings were included in the analysis. Those rings that spent at least 75% of the time north of 26°N before reaching the western wall were considered to have the “northern path.” Those that spent at least 75% of the time between 24° and 26°N were considered to have the “central path,” and those that spent at least 75% of the time south of 24° were considered to have the “southern path.”
By far, the preferred path (i.e., with a frequency of 62%) for the WCRs through the WGOM is the central path (Fig. 17). The northern path has the next highest frequency (i.e., 24%). The least likely path for the WCRs is the southern path with a frequency of about 14%.
c. Ring speed
Previous studies have indicated that WCRs move through the WGOM at speeds on the order of 5 km day−1 (Cochrane 1972; Elliott 1982; Vukovich and Crissman 1986). This statistic has been based on studies that examined a small number of rings in most cases. Presently, a 32-yr (i.e., 1972–2003) database was used to determine the speed of major WCRs. The speed of the rings was calculated using successive locations of the center of the ring. The west-to-east and south-to-north components of the ring speed were calculated using the longitudes and latitudes, respectively, of successive positions of a ring’s center.
The average ring speed over the 32-yr period was 4.4 km day−1 with a standard deviation of ±2.9 km day−1. The median ring speed was 4.1 km day−1. The frequency distribution (Fig. 18) is bimodal. The primary mode was found in the interval 4.0–4.9 km day−1 and mode for the data was 4.1 km day−1. A secondary mode occurred at speeds on the order of 1 km day−1.
d. Decay of ring size
As major WCRs move through the WGOM, they have been observed to dissipate (Vukovich and Waddell 1991; Elliott 1982; Cochrane 1972). The size of the rings, which have diameters of about 300–400 km when they separated, decrease, resulting in diameters the order of 100–200 km when they reach the western wall. Previous studies of the decay of the ring size have used databases as small as 12 yr (Vukovich and Crissman 1986) and as large as 22 yr (Herring et al. 1999). For this study, a 28-yr (i.e., 1976–2003) database was used to determine the mean rate of decay of the WCRS. The ring dissipation was estimated in terms of the month-to-month change of the diameter of a circle whose area was identical to that determined for a specific ring that separated from the Loop Current. The area of the ring was calculated for each month using the portrayal of the ring in the monthly frontal analysis.
The diameters of the rings were determined for each month during the period when the ring traveled from the EGOM to the western wall in the WGOM. This was done primarily because for certain portions of the period 1976–2003, only satellite SST data were available to track the ring. When the ring became small (i.e., diameters of about 100–200 km), the ring could not always be detected using the SST. When ocean color and/or altimetry data were available, the rings could be tracked for considerably longer periods. However, for consistency in the period over which the ring diameter was determined for each ring, only data in the period when the ring traveled from the EGOM to the western wall was considered.
Time was normalized relative to the separation month (i.e., the month that the ring separated from the Loop Current was set to zero and the months were in sequential order thereafter). The decay of the diameter of WCRs was also normalized relative to the average diameter at the time of separation (i.e., at month = 0, the normalized ring diameter equaled 1.0). The average diameter of the rings at the time of separation was 330 km.
Data for only an 8-month period is provided because, though data were available for longer periods in some cases, the 8-month period provided a near-uniform number of data points to calculate the average diameter for each month (Fig. 19). After the eighth month, the number of data points available decreased markedly. Also shown in the figure is a best-fit curve of the decay. The best-fit curve had an r = 0.97 and average error of about 9 km relative to the observations.
According to the observations, the ring diameter decayed to approximately 55% of its initial size at separation in about 8 months, or about 240 days. At that time, the rings were located near or at the western wall in the WGOM. There was a 2-month period starting at month = 3 and ending at month = 5 when the ring decay was very weak. After that period, the ring decay was about equivalent to or greater than that found in the first three months. The period of weaker ring decay, which was associated with momentary ring intensification, took place for many of the rings used to create this statistic.
According to the best-fit curve, the ring diameter decayed at a rate of about 17 km month−1, on average. The actual decay rate was larger than the average indicated by the best-fit curve in the first few months (i.e., the average decay rate in the first 4 months of about 18 km month−1) and the last few months (i.e., the average decay rate in the last three month was about 30 km month−1). The maximum decay rate in the data was found between months 6 and 7 and was 40 km day−1. The best-fit curve did not capture the period of weak decay of the ring noted in the observations in the period month = 3 to month = 5, and indicated that the e-folding took place in approximately 12 months.
e. Frequency of WCR water in the WGOM
When WCRs move through the WGOM, the warm water that defined the rings affect a large area in the WGOM. The area affected by a single ring is generally largest when the ring enters the WGOM immediately after it has separated from the Loop Current. Some parts of the WGOM will be considerably more affected by ring water than other parts because of the frequency of paths that WCRs follow through the WGOM. These rings play an important role in the heat and salt budget in the WGOM (Elliott 1982).
The maximum frequency (i.e., ∼24%) for WCR water is found near 26°N, 90°W in the eastern portion of the WGOM (Fig. 20). There is a zone of high frequency of ring water that stretches along a line oriented west-southwest–east-northeast from about 26°N, 90°W to about 24.75°N, 95°W that is nearly parallel to the central path of the major WCRs. The spatial frequency ring water varied along that line from 24% to 16%. There is a low probability of finding water associated with major rings in the southern portions of the WGOM (i.e., the Bay of Campeche). This is because major rings are no longer included in the frontal analysis when the diameter of the rings becomes ≤150 km. Rings of this size are generally undetectable in the satellite SST data. Rings whose diameters are usually about 150 km or smaller are the kinds of rings normally found in the Bay of Campeche. Though rings of this size are detectable in the satellite altimetry data, the SST data made up a greater part of the dataset so that rings with diameters ≤150 km were excluded when altimetry data were included as a part of the dataset to maintain consistency.
The frequency contours (Fig. 20) ridge into the northwestern corner of the WGOM. This region has been called the “graveyard for rings” (Biggs et al. 1996). Rings in this area, at times, become revitalized for a short periods of time (Vukovich and Waddell 1991). An anticyclonic circulation can be intensified in that region when the warm water is transported into the northwestern corner of the WGOM because of a ring colliding with the western wall, south of the northwest corner. Drifter data has clearly shown that in a January 1986 case an anticyclonic circulation became apparent in the northwestern corner after a ring collided with the western wall to the south of that region.
f. Frequency of the position of WCR centers in the GOM
Major WCRs separate from the Loop Current and move into the WGOM where they dissipate. However, minor WCRs also separate from the Loop Current, some of which move into the WGOM while others stay in the EGOM. The minor rings dissipate quickly (i.e., in about 2–5 months), but play a role in the heat and salt balance and in the circulation in the GOM. Warm rings are believed to be a major contributor to the transport in the WGOM and the maintenance of an anticyclonic cell along the western boundary of the GOM (Brooks 1984; Nowlin and McLellan 1967; Elliott 1982) and have a profound effect on the surface trajectories in the WGOM (Hamilton et al. 1999; Lewis and Kirwan 1985). In this section, an analysis of the spatial frequency of the location of the centers of WCRs is presented, using a 27-yr (i.e., 1977–2003) database.
For the period 1977–1991, the location of the centers of WCRs was derived from the frontal analyses discussed in section 2. The frontal analyses for that period were based on satellite SST data and CZCS ocean color data when they were available. Minor rings were included in the frontal analyses for this period when they could be detected and for as long as they were detectable. For the period 1992–2003, the database used for this analysis also included satellite altimetry data. When satellite altimetry data was included in this analysis, the data provided daily analyses of all rings (i.e., either WCRs or CCRs) in the GOM. Note that this analysis took into account major and minor rings, including major rings for periods after they reached the western wall. Minor rings were detected at most three months after they separated from the Loop Current using SST and/or ocean color data. Major rings could be detected at most four months after they reached the western wall using those data. Major and minor rings could be detected until they dissipated using altimeter data.
The analysis (Fig. 21) shows an east–west zone of high spatial frequency for the ring centers located in the 25°–26°N latitude belt from 88° to 94°W. Multiple centers are found in that zone, and the maximum frequency in that zone is about 12%, which is a little more than one event every year, on average. One event every year might be considered rather high considering that major rings do not necessarily separate from the Loop Current that often (see section 4a) nor is the path of the rings always in this area (see section 4b). There are a couple of reasons for this frequency. In periods when the movement of major rings slowed because they were either changing direction or their motion formed a loop (Vukovich and Crissman 1986), the ring, in some cases, did not move over a great distance and occupied the same grid square for more than a 1-month period. Another reason for the one event every year is the fact that minor rings, which were included in the analysis, move slowly, are often quasi-stationary, but dissipate quickly. Under these conditions, minor rings made a serious contribution in the eastern part of the analysis.
North of the eastern part of the east–west zone of high frequency is a secondary center of relatively high spatial frequency for the ring centers (i.e., located at around 27.5°N, 89.5°W). The maximum frequency in this center is about 7%, which is about one event every 17 months. In this region, both major and minor rings made serious contributions to the spatial frequency.
Immediately west of the east–west zone of high frequency is a center of relatively low spatial frequency (i.e., located at around 25.5°N, 94.5°W). The minimum frequency in that center is about 3%, which is about one event every 4 years. The reason for the minimum in this region is not clear. In the next section, it will be shown that this area is an area where CCRs are often found. The dominance of CCRs in this region may inhibit WCRs from moving into that region. WCRs are normally much smaller when they reach the western wall of the WGOM and have about the same dimensions as CCRs in this area, so a CCR is just as likely to occupy a position near the western wall as a WCR.
North and south of the minimum area discussed in the last paragraph are areas of relatively high frequency for ring centers. The area to the north is in the northwestern corner of the GOM, where ridging of the frequency contours for the WCR water mass was noted. The maximum frequency in that area is about 10%, which is about one event every year.
The zone of relatively high frequency in the area to the south of the minimum area is located near the western boundary of the GOM and the maximum frequency in that area is about 12%, which is a little greater than one event every year. This feature was created by a large number of major rings that reached the western wall and were reduced considerably in size after they reached the wall.
5. Cold core rings
a. Frequency of the position of CCR centers in the GOM
CCRs are generally small in size (i.e., diameters on the order of 100–150 km). They have been observed in the GOM along the boundary of the Loop Current (Vukovich et al. 1979a; O’Connor 1981; Vukovich and Maul 1985; Vukovich 1986, 1988a), off Dry Tortugas (Maul and Herman 1985; Vukovich 1988b), and on the boundary of WCRs (Elliott 1979; Merrill and Morrison 1981; Merrill and Vasquez 1983; Brooks 1984; Kelley and Brooks 1986; Kelley et al. 1986; Vukovich and Waddell 1991). Their circulation can influence the water mass characteristics on the WFS and in the EGOM (See Figs. 7 and 8). CCRs have been observed to develop or intensify along the northwestern portions of the Loop Current boundary when the Loop Current penetrates deep into the EGOM (i.e., when the northern boundary of the Loop Current is found as far north as 27°N) (Vukovich 1988a). They move along the Loop Current boundary in the direction of the flow in the Loop Current from the Yucatan Channel to the Straits of Florida. They have been observed to be associated with the separation of a WCR from the Loop Current (Vukovich et al. 1979a; Vukovich 1988a). The production of CCRs on the boundary of WCRs is considered as part of the dissipation process associated with WCRs (Cushman-Roisin 1987; Smith 1986). They move along the boundary of the WCRs in a clockwise manner associated with the anticyclonic flow in the WCRs. If these rings are persistent in regions of the GOM, the circulation associated with these rings can potentially influence local water characteristics and surface trajectories. An analysis of the spatial frequency of the location of the centers of CCRs was created using a 12-yr (i.e., 1992–2003) altimetry database. It is only in altimetry data that CCRs can be consistently detected.
In the EGOM (i.e., the region east of 90°W in the GOM), the analysis (Fig. 22) shows a northwest–southeast oriented zone of relatively high frequency of CCR centers, having two centers of high frequency. The center in the northwestern part of that zone (i.e., the center located at about 27.5°N, 88.5°W) has a maximum frequency of about 30%, which corresponds to about three–four events per year. It should be noted that because of the high frequency with which these CCRs are found in this region, they often interact with the shelf off the Louisiana and Alabama coast, transporting shelf water into the deep Gulf (Fig. 23). The center in the southeastern portion of that zone (i.e., the center located at about 25.5°N, 85.5°W) has a maximum frequency of about 17%, which corresponds to about two events per year.
In the WGOM, the analysis shows four major centers of relatively high frequency of CCR centers: one at 24.5°N, 91.5°W; one at 26.5°N, 93.5°W; one at 25.5°N, 94.5°N; and one at 22.5°N, 94.5°W. The center located at 24.5°N, 91.5°W has a maximum frequency of about 18.4%; that at 26.5°N, 93.5°W about 16.3%; that at 25.5°N, 94.5°W about 17%; and that at 22.5°N, 94.5°W about 20.6%. In each case, the maximum frequency corresponds to about one–two events per year.
6. Discussion of findings
Isolated warm pools of Loop Current water were found throughout the region north of 25°N in the EGOM. Many of the isolated pools of Loop Current water were created as a result of transport of Loop Current water by the circulation associated with CCRs located on the boundary of the Loop Current. Intrusions of Loop Current water onto the WFS were, in most cases, due to transport associated with a CCR located on the southeastern boundary of the Loop Current. Most of these intrusions affected the region south of 27°N between the shelf break and 83°W on the WFS. The statistics from this study also showed that CCRs are either very persistent or numerous or both in the EGOM. These factors suggest that CCRs may play an important role in redistributing heat and mass in the EGOM.
The Loop Current front reached 28°N in the EGOM only about 5% of the time. It reached 27°N about 20% of the time. About 80% of the Loop Current orientation angles were between 0° (a north–south orientation) and 30° (a north-northwest–south-southeast orientation), which represents a stable orientation for the Loop Current, and ring separation is not expected soon. Less than 20% of the Loop Current orientation angles were greater than 30°, which represents an unstable mode for the Loop Current, and ring separation is usually expected soon. Both the north–south movement and orientation of the Loop Current are part of the cycle of Loop Current penetration into the EGOM and WCR separation. Deep penetration (i.e., penetration to 27° or 28°N latitude) of the Loop Current does not persist because ring separation usually occurs under these circumstances and the Loop Current reforms to the south. Examination of the satellite altimetry data would suggest that that the westward tilt of the Loop Current may be the result of the westward drift of a ring that has developed in the Loop Current (i.e., closed SSH contours formed in the Loop Current). This ring remains an integral part of the Loop Current as it drifts westward so that the Loop Current appears to “lean” to the west. The farther west that the ring drifts without separating from the Loop Current, the greater the orientation angle becomes.
It was found that the frequency distribution for the period for separation of major WCRs from the Loop Current had a bimodal distribution with modes at 6 and 11 months and a range of 5 to 19 months, results that were similar to those found by Sturges and Leben (2000). The average period was 11 months, which did not change substantially over the last 10-yr period (1994–2003). Since the average period and one of the primary modes in the histogram are approximately 11 months, this might suggest that an 11-month period may be a representative period for eddy separation from the Loop Current. However, the frequency distribution also showed that the 6-month period has the same frequency as the 11-month period and that periods of 9, 10, and 14 months also have reasonably high frequencies. Furthermore, the most significant period in the spectrum of the Loop Current northern boundary variations (i.e., 12 months) did not correspond to any of the important periods in the histogram of ring separation periods, and was not stable property since it varied by one month over the last 10 years (1994–2003), which is presumably a result of the year-to-year variability in the eddy-shedding period. It is doubtful that the spectrum will converge to some single value for the most significant period even with 100 years of data since it seems to be very susceptible to the year-to-year variability in the eddy-shedding period. Convergence might be accomplished, however, with considerable smoothing of the spectrum. These results would suggest that it is more likely that the eddy-shedding period is highly variable, as suggested before (Vukovich 1995), and that no one period can be used to represent the Loop Current’s eddy-shedding period. The precise cause for eddy shedding in the GOM is still in question (Oey et al. 2003; Candela et al. 2003; Ezer et al. 2003). Among the processes that are considered to be the cause of Loop Current eddy shedding are transport, wind-induced fluctuations through the Greater Antilles Passage, and Caribbean eddies. Any one of these factors could induce a shorter or longer eddy-shedding period.
A high frequency of ring separation was found in March, and no rings were observed to separate from the Loop Current in December. In March, the Straits of Florida transport is usually increasing in strength or has reached a maximum and, in December, the Straits of Florida transport is normally a minimum (Maul and Vukovich 1993). This coincidence of factors lends support to the notion that transport and its variations play a role, and probably a significant role, in the eddy-shedding process.
WCRs normally travel through the WGOM using the central path (62% frequency of occurrence). The northern path had the next highest frequency (24%), and the least likely trajectory for the WCRs was the southern path (18%). It should be noted that, since there were few ring paths associated with either the northern or southern paths over the 28-yr dataset used to create this statistic, the addition of just one ring for either one of those paths will change those statistics markedly.
The average speed of the WCRs through the WGOM was 4.4 km day−1 with a standard deviation of ±2.9 km day−1. The frequency distribution of WCR speeds had a bimodal distribution with modes in the increments of 4.0–4.9 and 1.0–1.9 km day−1. Ring motion is characterized with periods of “stalls and sprints” (Vukovich and Crissman 1986; Hamilton et al. 1999). Vukovich and Crissman (1986) noted that the stalls are often associated with periods when the rings make major changes in their direction or when their motion is characterized by a tight loop. The relatively high speed motion of the rings, the sprints, occurred when the path of the ring is characterized by quasi-linear motion. Hamilton et al. (1999) suggested that the stalls may, in some cases, be attributed to the presence of lower continental slope cyclones situated to the northwest of the WCRs, which affect the motion of and the circulation in the rings. The secondary mode in the frequency distribution is most likely produced by those periods when the path of the rings was characterized by the stalls.
WCRs decayed, on average, to about 55% of their initial size in about 8 months, about the time the rings reached the western wall. There was a period of about two months (i.e., around the fourth month after separation of the ring) when the decay rate decreased significantly. Examination of the data used to create this statistic indicated that some of the rings seem to intensify (i.e., the number of SSH contours that define the ring increased) and to grow larger during this period. No explanation could be ascertained about the cause of this behavior from the data available. After reaching the western wall, the rings normally expanded to the north and south, redistributing mass, heat, and momentum (Zavala et al. 1998; Vukovich and Waddell 1991; Vukovich and Crissman 1986). Afterward, the size of the ring decreases markedly. However, the intensification discussed above occurred before the rings reached the western wall.
The frequency distribution of WCR water in the WGOM indicated that that water mass associated with major WCRs has its most significant effect north of 24°N. The frequency distribution also showed that the frequency contours for WCR water ridged into the northwest corner of the WGOM where WCRs were often observed in our dataset as indicated by the fact that a region of relatively high frequency of WCR centers was found in the northwest corner of the WGOM. In the WGOM, the water mass associated with WCRs plays an important role in the heat and salt balance (Elliott 1982) and that water mass appears to have its most significant effect on the heat and salt balance north of 24°N.
The spatial frequency distribution for WCR centers had a number of regions with relatively high frequencies. A couple of these deserve further discussion. A center of relatively high frequency was found around 27.4°N, 89.6°W for which minor rings played a significant role in developing this center. These rings could have an important role in the transport heat, mass, and momentum in the Mississippi Delta region (e.g., the transport of Mississippi River water off the shelf into the deep gulf) as well as on the Louisiana–Texas shelf. Another region of relatively high frequency of WCRs was found near the western wall between 22° and 25°N. This region was developed by the persistence of WCRs in that area of the WGOM. The existence of this region supports the notion that WCRs may be major contributors to transport in the WGOM and the maintenance of an anticyclonic cell along the GOM’s western boundary (Brooks 1984; Nowlin and McLellan 1967; Elliott 1982).
High frequency for CCRs was found in the EGOM. One of the two centers found in that region, the northwestern center, had a maximum frequency of about 30% (the highest frequency noted for any kind of ring—warm or cold). This center is found in an area of the EGOM where CCRs have been noted to either develop or intensify on the boundary of the Loop Current (Vukovich 1988a). When CCRs develop or intensify on the boundary, they become quasi stationary for a period of time (e.g., 1–2 months) before they move downstream. If a WCR separates after the CCR develops or intensifies in this area, then the CCR may also contribute to this center of maximum frequency for a number of months as it moves clockwise along the boundary of the WCR, while the WCR moves westward into the WGOM. Another center was located at around 25.5°N, 85.5°W in the Dry Tortugas region (i.e., off the southwestern coast of Florida) where CCRs have been previously noted (Maul and Herman 1985; Vukovich 1988a, b) where they have been observed to be associated with the separation of a WCR from the Loop Current (Vukovich et al. 1979a; Vukovich 1988a), and where they develop intrusions of Loop Current water onto the WFS, at times (Vukovich 1986). These CCR have been observed to persist in that location for up to two months.
In the WGOM, four region of relatively high frequency for CCR centers were found. These centers are produced by CCRs associated with a number of process events that take place in the WGOM. The CCRs are most often associated with major WCRs that separate from the Loop Current and are found on the boundary of the WCRs (Elliott 1979; Merrill and Morrison 1981; Merrill and Vasquez 1983; Brooks 1984; Kelley and Brooks 1986; Kelley et al. 1986; Vukovich and Waddell 1991). The major WCRs that move through the WGOM generally enter the WGOM north of the center of maximum frequency for the CCRs located at 24.5°N, 91.5°W in about 80% of the cases, which suggests that the CCRs that contribute to that center would be located on the southern boundary of a WCR, if they are associated with a WCR. The CCRs that contribute to the centers of maximum frequency for the CCRs that are located near the western wall of the WGOM (i.e., those located at 26.5°N, 93.5°W; 25.5°N, 94.5°N; and 22.5°N, 94.5°W) are associated with the CCRs that are often observed with WCRs after they collide with the western wall (Vukovich and Waddell 1991; Cushman-Roisin 1987; Smith 1986).
This climatology was developed through resources provided by the Mineral Management Service under Contract 0103PO74302. The author thanks Dr. Carole Current, the MMS contracting officer’s technical representative, and Dr. Alexis Lugo-Fernandez for their continuing support of this project. I also thank the people, too numerous to cite, who contributed information and data that helped to create the continuous datasets from which many of the statistics were derived.
Corresponding author address: Fred M. Vukovich, SAIC, 615 Oberlin Road, Suite 100, Raleigh, NC 27605. Email: firstname.lastname@example.org