Downloaded 10/04/21 10:04 PM UTC TT FORECASTING: SEC CASES

The TT Problem Forecasting the Tropical Transition of Cyclones

BY CHRISTOPHER A. DAVIS AND LANCE F. BOSART

ccording to the Tropical Cyclone Reports issued Category 2 intensity, their tendency to form close to by NOAA's Tropical Prediction Center, the de- North America can create significant forecast and Avelopment of nearly half of the Atlantic tropical evacuation problems. In addition, many TT cases cyclones from 2000 to 2003 depended on an extra- become ET cases and can affect land areas from east- tropical precursor (26 out of 57). Many of these dis- ern North America to western Europe. turbances had a baroclinic origin and were initially considered cold-core systems. A fundamental dy- TT CLASSIFICATION. It is convenient to repre- namic and thermodynamic transformation of such sent TT cases with two paradigms, based on the am- disturbances was required to create a warm-core plitude and structure of the precursor disturbance: tropical cyclone. We refer to this process as tropical strong extratropical cyclone (SEC) and weak extrat- transition (TT), to be contrasted with extratropical ropical cyclone (WEC). The distinguishing factor transition (ET), which results in an extratropical dis- between these archetypes is that in SEC cases, extra- turbance given a tropical cyclone. tropical cyclogenesis produces a surface cyclone ca- Tropical cyclogenesis associated with extratropical pable of wind-induced surface heat exchange precursors often takes place in environments that are (WISHE; Emanuel 1987), whereas in WEC cases, the initially highly sheared, contrary to conditions be- baroclinic cyclone is an organizing agent for convec- lieved to allow tropical cyclone formation. The adverse tion. The convection must then undergo self-organi- effect of vertical wind shear1 exceeding 10-15 m s_1 zation to produce a disturbance capable of self-am- on the formation of low-latitude storms (equatorward plification. Because these archetypes represent end of 20°N) is well documented (DeMaria et al. 2001). points on a spectrum of precursors, we do not antici- However, a beneficial role of vertical shear, hypoth- pate the existence of a clear threshold separating one esized to organize convection, was indicated by the type from another. Once a sufficiently strong surface statistical analysis of Bracken and Bosart (2000) for vortex is formed, there is no obvious distinction of the 24 developing cases in the northern Caribbean Sea. ensuing tropical cyclone intensification in either SEC This article, focusing on the Atlantic basin, reviews or WEC cases. briefly what is known about TT and how it can be In reality, TT cases reside within what is an even anticipated. While TT storms typically do not exceed broader continuum of marine cyclogenesis, ranging from cool-season baroclinic cyclones to hurricanes 1 Throughout this article, vertical shear is expressed as a veloc- initiated from weak extratropical systems. While the ity difference through the depth of the troposphere. intensity of many marine cyclones is enhanced through storm-induced fluxes, the TT subspectrum AFFILIATIONS: DAVIS—National Center for Atmospheric of marine cyclones is dominated by such fluxes. How- Research* Boulder, Colorado; BOSART—University at Albany, State ever, the detailed pathway to TT appears principally University of New York, Albany, New York determined by baroclinity (given a sufficiently warm CORRESPONDING AUTHOR: Christopher A. Davis, NCAR, P.O. underlying ocean). Box 3000, Boulder, CO 80307 E-mail: [email protected] As we will show, SEC cases have a more consistent DOI: 10.1 175/BAMS-85-11-1657 and repeatable evolution than WEC cases. WEC cases *The National Center for Atmospheric Research is sponsored by can arise through a variety of extratropical precursors. the National Science Foundation. Because the precursor is merely an organizing agent, ©2004 American Meteorological Society its detailed structure is perhaps less important than in SEC cases. Furthermore, being of smaller ampli-

AMERICAN METEOROLOGICAL SOCIETY NOVEMBER 2004 BAPIS• I 1657 Unauthenticated | Downloaded 10/04/21 10:04 PM UTC TT FORECASTING: SEC CASES. Synoptic-scale. The composite structure for four SEC cases (Florence, Michael, Karen, and Gustav) shows a pronounced, localized anomalous trough in the upper troposphere to the west of the surface low just prior to TT (Fig. la). The lower-tropospheric temperature pattern ex- hibits warm and cold anomalies consistent with horizontal transport due to the precursor cyclone and anomalous warmth across the northeastern United States and eastern Canada (Fig. lb). This mini-composite strongly resembles the 24-case composite shown by Bracken and Bosart (2000). It also resembles the anticyclonic wave-breaking scenario (LCI) pre- sented in Thorncroft et al. (1993). In SEC cases, a low-latitude frontal cyclone devel- ops to an intensity sufficient to trigger WISHE. The vertical shear is eliminated by diabatic processes, as shown in Davis and Bosart (2003, hereafter DB03). The reduction in shear is caused by both the upper-tropospheric outflow from convection and from the diabatic redistribution of potential vorticity (PV), both of which tend to homogenize the horizontal gradients of PV di- rectly above the storm center. The process leaves an equilibrated cyclone resembling an occluded system and creates a subsynoptic "cocoon" of weak shear, within which TT occurs and the resulting tropical cyclone grows. Observations of occluded and secluded cool-season marine cyclones (e.g., Shapiro 1990) have also re- FIG. I. (a) 250-hPa height anomaly and (b) 850-hPa tem- vealed a deep column of weak shear over the cyclone perature anomaly for four-case composite prior to center. However, the occlusion prior to TT departs tropical cyclogeneis. The position of the composite from the classical model in which the surface cyclone surface low center is indicated with an "L." Dates of migrates poleward (that is, toward colder air) beneath composite are 10 Sep 2000, 15 Oct 2000, I I Oct 2001, and 9 Sep 2002. For each date, the four available analy- the upper-tropospheric jet. In TT cases, the jet itself ses (0000, 0600, 1200, and 1800 UTC) are averaged. is rearranged by diabatic processes and leads to an ap- Images were provided by the NOAA-CIRES Climate parent migration of the surface cyclone toward Diagnostics Center, Boulder, CO, from their Web site warmer mean tropospheric air (on the synoptic scale). at www.cdc.noaa.gov (also see Kalnay et al. 1996). Mesoscale. The rainfall and cloud signatures of the tude, precursors in WEC cases are more difficult to four SEC cases discussed above are shown in Fig. 2 identify in conventional data, and it is more difficult just prior to TT. In each case, there is a pronounced to definitively state the essential steps toward tropi- asymmetry associated with the rainfall, with a ten- cal cyclogenesis. Recent examples of SEC storms in- dency for a "bent back" frontal structure and heavy clude Florence (2000), Michael (2000), Erin (2001)2, rainfall on the west and even southwest side of the still Karen (2001), Noel (2001), Olga (2001), and Gustav extratropical (or perhaps subtropical) surface low. (2002). Examples of recent storms in the WEC cat- This structure is present prior to some of the stron- egory are Leslie (2000), Nadine (2000), Allison (2001), ger hurricanes resulting from TT. Gabrielle (2001), and Humberto (2001). It is possible that the bent-back structure merely indicates a stronger precursor disturbance from which it 2 Note that Erin was a tropical storm in the eastern Atlantic, but is easier to create a stronger tropical cyclone. However, it almost completely decayed in the central Atlantic. Its regen- it is also possible that such a bent-back structure, with eration took place in the presence of extratropical perturbations. enhanced rainfall upshear from the surface low (the

1658 I BAflS- NOVEMBER 2004 Unauthenticated | Downloaded 10/04/21 10:04 PM UTC FIG. 2. (a) Visible satellite and Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) 85-GHz polarized corrected temperature (PCT) imagery at I 145 and 1209 UTC 10 Sep 2000, respectively; (b) (upper right), as in (a) but for visible and TMI 85-GHz PCT at 1245 and 1335 UTC 15 Oct 2000; (c) IR and Special Sensor Microwave/Imager 85-GHz PCT at 0015 and 0219 UTC 10 Sep 2002; and (d) IR and TMI 85-GHz PCT at 2045 and 2132 UTC I I Oct 2001. In all panels, the "L" indicates the position of the surface low. Images were obtained courtesy of the Naval Research Laboratoryatwww.nrlmry.navy.mil/tc_pages/tc__home.html.

upshear direction based on a synoptic-scale average), ondary circulation, rather than from quasi-horizon- is particularly efficient for eliminating the vertical shear tal advection by the swirling flow around the cyclone. over the cyclone center as shown by the conceptual model in Fig. 3. These schema are based on simula- Forecast rules. The conditions favoring development tions of Michael (DB03) and of nontransitioning cases outlined by DeMaria et al. (2001) still apply to SEC (Davis and Bosart 2002) and are intentionally simpli- cases, but only after the environment is modified by fied to illustrate the salient differences. The pathway the precursor extratropical disturbance. Because such to an occluded cyclone corresponding to Fig. 3a is modification usually occurs on time scales less than distinct from classical occlusion because it is driven 1 day, the key forecast challenge is to anticipate a fa- by diabatic heating and advection arising from its sec- vorable environmental modification.

AMERICAN METEOROLOGICAL SOCIETY NOVEMBER 2004 BAI1S* I 1659 Unauthenticated | Downloaded 10/04/21 10:04 PM UTC bances approach the surface cyclone and increase the vertical wind shear over its center before occlusion can occur or delay the occlusion until the storm is over cool water. The western Atlantic cyclones occurring on 1-2 October 2000 and 14-15 November 2001 illustrate this mode of TT failure (DB03). Because the occlusion process often requires more than a day to complete, this qualitatively represents a minimum allowable period of upper-tropospheric waves. Furthermore, if the upstream disturbance is of comparable or greater amplitude than the disturbance causing cyclogenesis, occlusion will probably not occur or will occur in response to development associated with the upstream wave.

WEC CASES. Synoptic-scale. Two types of WEC precursors considered herein are midtropospheric mesoscale vortices and baroclinic systems with a structure similar to SEC cases (i.e., an upshear tilt with height) but with smaller amplitude. The weak baroclinic systems in this category simply have insufficient amplitude to create a surface cyclone capable of amplifying by WISHE without invoking an inter- mediate process to enhance mesoscale vorticity. Hur- ricane Diana (1984) (Bosart and Bartlo 1991) and Hurricane Humberto (2001) are examples of storms that began by such a process. Their structure strongly FIG. 3. Schema showing the effect of convection (blue resembles the composite of developing depressions area) (top) downshear and (bottom) upshear, relative to a surface low ("L"). Small arrows indicate divergent shown by Bracken and Bosart (2000). motion near the tropopause. Large arrow indicates flow within upper-level jet. Solid lines are two initial PV con- Mesosco/e. Midtropospheric vortices, themselves often tours (PV2 > PV(), and red dashed lines indicate posi- formed by antecedent convection, have been observed tions of the same contours after deep convection has to initiate tropical cyclogenesis in Danny (1997) developed. (Molinari et al. 2004) and Gabrielle (2001) (K. Musgrave 2003, unpublished manuscript). Midtropospheric The primary empirical forecast rule for predicting convectively generated vortices have also been ob- the tropical transition of cases with strong extratropi- served to initiate tropical cyclogenesis in lower latitudes cal precursors may be stated, "The precursor cyclone (e.g., Simpson et al. 1997). It is through understand- must occlude and remain over warm water (>~26°C) ing the initiation of tropical cyclogenesis by mesoscale for at least a day following occlusion." The failure of vortices at higher latitudes that the link with tropical TT occurs for one of two reasons. First, transition fails cyclone formation in the deep Tropics can be made. if the occluding cyclone is embedded within a mean Just as there is a continuum between the weak and current that translates it over cool water before TT strong baroclinic precursors, there is a range of real- can occur (hence, the empirical "1-day rule" above). izations between mesoscale vortices and weak It should be noted that transition can occur, though baroclinic systems. rarely, over sea surface temperatures lower than the Numerical simulations of Diana (1984) (Powers empirical 26°C. Tropical Storm Ana (2003), occur- and Davis 2002) showed that the path to tropical cy- ring in April, was possibly one example. Another was clogenesis required a lower-middle-tropospheric vor- the South Atlantic "hurricane" of March 2004.3 Second, transition fails if the primary cyclone is 3 It is possible that the range of similar disturbances should be prevented from occluding. This happens when addi- extended to "hurricane-like" vortices that are occasionally ob- tional upper-tropospheric short wavelength distur- served in the Mediterranean Sea and in polar regions.

1660 I BAflS- NOVEMBER 2004 Unauthenticated | Downloaded 10/04/21 10:04 PM UTC tex that formed within convection initiated by an extratropical precursor. The distinguishing character between cases such as Gabrielle (Fig. 4a) and Humberto (Fig. 4b) or Diana (1984) is that with Gabrielle, the midtropospheric vortex existed for more than two days before tropical cyclogenesis began. In such cases, the vortex must organize new convection through its interaction with vertical shear. When a weak extratropical cyclone is maintained, such as with Diana or Humberto, the organization of convection occurs through a superposition of vortex and cyclone-induced ascent. In Diana, there were multiple vortices formed within a mesoscale ascent region, and the coalescence and growth of these vortices formed the nascent tropical storm (Hendricks et al. 2004).

Forecast rules. Considerable research remains to understand how convection organizes in systems with weak precursors. Therefore, it is difficult to derive a set of forecast rules. However, it is apparent that PV "debris" extruded from the midlatitude jet is common over the warm oceans of the subtropical Atlantic, even as far south as 15°N on occasion. In September 2001 alone, we counted 34 upper-level vorticity maxima (averaged over a 3° x 3° latitude-longitude box) greater than 10~5 s_1 persisting for at least 12 h while over ocean temperatures greater than 25°C. Most of these upper-tropospheric disturbances had little effect on clouds and precipitation. Only disturbances that encountered (or persisted over) lower-tropospheric baroclinic zones helped initiate significant convection. The baroclinic zones exhibited typical FIG. 4. (a) PV on the 310 K isentropic surface (white, contrasts ranging from 0.2° to 0.5°C (100 km)"1. This contour interval 0.5 PVU) and 900-hPa relative vortic- -4 1 was insufficient for baroclinic development on time ity (red, 0.5 and 1.0 x I0 s , contoured) for 1200 UTC I I Sep 2001; (b) as in (a) but for 340 K PV and at 1200 scales of 1-2 days, but ample for focusing mesoscale UTC 21 Sep 2001. Black arrow indicates shear vector ascent. The authors believe that the lower-middle tro- orientation over low-level vorticity center. Fields are pospheric ascent is most important because it more superposed on SSM/I 85-GHz PCT as in Fig. 2 for near- effectively destabilizes the atmosphere. Overall, then, est corresponding time. favorable conditions for WEC cases of TT involve mid-upper-tropospheric cyclone PV anomalies en- countering lower-tropospheric baroclinity (and, them. Numerical simulations (K. Musgrave 2003, un- hence, vertical shear). In the cases that develop into published manuscript) of Gabrielle (2001) suggest that tropical cyclones, the shear either remains roughly following the growth of convection downshear of the 10 m s"1 or less, or it is reduced to such values follow- precursor disturbance (Fig. 4a), there is a period during convection organization (Powers and Davis 2002). ing which the shear drops off dramatically to a value less We note that at lower latitudes, convection has also than 5 ms"1. Although this period is relatively short been observed to organize when upper-tropospheric (about 6 h), the depression undergoes a full warm-core disturbances approach easterly waves, wherein sys- transformation. It is believed that the ongoing convec- tematic baroclinity is hard to identify. tion is somehow responsible for this decrease in shear. There are almost certainly other factors involved in Unlike the schematic in Fig. 3, the ambient shear in cases WEC cases of TT, but research has yet to fully clarify like Gabrielle is weak enough that the diabatic second-

AMERICAN METEOROLOGICAL SOCIETY NOVEMBER 2004 BAI1S* I 1661 Unauthenticated | Downloaded 10/04/21 10:04 PM UTC ary circulation can partially cancel it on the upshear side Hendricks, E. A., M. T. Montgomery, and C. A. Davis, of the convection where the incipient low resides. 2004: On the role of vortical hot towers in hurricane formation. /. Atmos. Sci., 61, 1209-1232. Kalnay, E. and Coauthors, 1996: The NCEP/NCAR Re- FOR FURTHER READING analysis 40-Year Project. Bull. Amer. Meteor. Soc., 77, Bosart, L. F., and J. Bartlo, 1991: Tropical cyclone for- 437-471. mation in a baroclinic environment. Mon. Wea. Rev., Molinari, J., D. Vollaro, and K. Corbosiero, 2004: Tropi- 119, 1979-2013. cal storm formation in a sheared environment. /. Bracken, E., and L. F. Bosart, 2000: The role of synop- Atmos. Sci., 61, 2493-2509. tic-scale flow during tropical cyclogenesis over the Powers, J. G., and C. A. Davis, 2002: A cloud-resolving, North Atlantic Ocean. Mon. Wea. Rev., 128,353-376. regional simulation of tropical cyclone formation. Davis, C. A., and L. F. Bosart, 2002: Baroclinic tropical Atmos. Sci. Lett., 3, 15-24. cyclogenesis: Developing and non-developing cases. Shapiro, M. E., 1990: Fronts, jet streams and the tropo- Preprints, 25th AMS Conf. on Hurricanes and Tropi- pause. Extratropical Cyclones. C. Newton and E. O. cal Meteorology, San Diego, CA, 395-396. Holopainen, Eds., Amer. Meteor. Soc., 167-189. , and , 2003: Baroclinically induced tropical cy- Simpson, J., E. Ritchie, G. J. Holland, J. Halverson, and clogenesis. Mon. Wea. Rev., 131, 2730-2747. S. Stewart, 1997: Mesoscale interactions in tropical DeMaria, M., J. A. Knaff, and B. H. Conell, 2001: A tropi- cyclone genesis. Mon. Wea. Rev., 125, 2643-2661. cal cyclone genesis parameter for the tropical Atlan- Thorncroft, C. D., B. J. Hoskins, and M. E. Mclntyre, tic. Wea. Forecasting, 16, 219-233. 1993: Two paradigms of baroclinic-wave life-cycle Emanuel, K. A., 1987: An air-sea interaction model of behaviour. Quart. J. Roy. Meteor. Soc., 119, 17-56. intraseasonal oscillations in the Tropics. /. Atmos. Sci., 44, 2324-2340.

UCAR Visiting Scientist Program at the Geophysical Fluid Dynamics Laboratory

The University Corporation for Atmospheric Research (UCAR) is recruiting postdoctoral scientists and short-term senior visitors to work in Princeton, New Jersey at NOAA's Geophysical Fluid Dynam- ics Laboratory (GFDL) as part of the Climate Change Research Initiative (CCRI). GFDL conducts fundamental and applied oceanic and atmospheric research on a variety of problems of importance to society and central to NOAA's mission. For the past three decades, GFDL has been a world leader in global change research, specializing in the computer modeling of the climate system. In the past several years GFDL has reorganized to develop a new generation of climate and Earth system models to support its research for the coming decade. GFDL has recently received additional funding from the CCRI to enhance its core research capability so as to enable policy related research and product generation. GFDL expects some of these new postdoctoral positions to evolve into permanent civil service hires as it expands its staff in the next several years. Placement into permanent civil service positions may require additional competition. These positions are intended to expedite the development of this new generation of climate models as well as the climate change research conducted with them. Application deadlines are May 15 and November 1. Applications are reviewed twice yearly by a steering committee and will not be reviewed unless they are complete, including letters of reference. Full text announcement located at zvzvw.vsp.ucar.edu. Send applications and letters of reference to: DB UNIVERSITY CORPORATION FOR ATMOSPHERIC RESEARCH ^^ Meg Austin, Director • Visiting Scientist Programs • P. O. Box 3000 • Boulder, CO • 80307-3000 • USA " UCAR is an Equal Opportunity/Affirmative Action Employer.

1662 I BAflS- NOVEMBER 2004 Unauthenticated | Downloaded 10/04/21 10:04 PM UTC Triple Eyewall in Hurricane Juliette

BY BRIAN D. MCNOLDY

urricane eyewalls are one of the more enigmatic phenomena in the atmosphere. Even more mys- Hterious are concentric eyewall cycles: the development of a ring of deep convection within a larger ring of deep convection, with a "moat" between them. On radar, the moat would appear as a nearly echo- free annulus, while an eyewall would appear as an annulus with radar echoes typically greater than 35 dBZ. From the first days of aircraft reconnaissance into hurricanes during the 1940s, these "double eyes" were occasionally observed in strong storms, and the advent of satellite meteorology in the 1960s has provided additional cases that we otherwise would not have known about. In particular, passive microwave imagery is the most ideal and prominent tool available for monitoring the internal precipitation structure, concentric eyewalls, and eyewall replacement cycles (Hawkins and Helveston 2004). Concentric eyewalls are ephemeral; once formed, they typically are not maintained for much longer FIG. I. Best-track positions of Juliette provided by the than 12 h. As the new outer eyewall forms, the origi- NOAA National Hurricane Center. The 0000 UTC nal inner eyewall usually lacks the necessary inflow positions are denoted by a solid square, and the 1200 to maintain itself, and it gradually dissipates. In time, UTC positions are denoted by an open circle; the date is marked at the top or right of each 0000 UTC square. dynamic processes cause the outer eyewall to contract, The track is plotted over weekly mean sea surface tem- and the process can repeat itself; this is called an perature contours, which are valid in the week from eyewall replacement cycle (Black and Willoughby 15 to 22 Sep 2001 (i.e., the prestorm period). One week 1992). Multiple eyewalls are more commonly ob- later, the SSTs cooled by 6°-8°C in the wake of Juliette served in intense tropical cyclones (i.e., Category 3, after the hurricane had passed over the area at its peak 4, and 5 on the Saffir-Simpson scale; Simpson 1974). intensity on 24-27 Sep (not shown). TRMM data are From a study of western North Pacific tropical cy- produced by Remote Sensing Systems and sponsored by NASA's Earth Science Information Partnerships and clones (TCs) during 1969-71, Willoughby et al. NASA's TRMM Science Team. (1982) estimated that approximately 53% of intense TCs (winds greater than 65 m s"1) exhibit concentric eyewalls, while only 14% of weaker TCs do. An interesting example of multiple eyewalls oc- curred during September 2001 in the eastern North Pacific basin. Tropical Depression 11 formed off the MCNOLDY—Department of Atmospheric AFFILIATIONS: Guatemalan coast at 0600 UTC on 21 September and Science, Colorado State University, Fort Collins, Colorado moved west-northwest, generally following the coast- CORRESPONDING AUTHOR: Brian D. McNoldy, Colorado State University, Fort Collins, Colorado 80523-1371 line. At 1200 UTC on 21 September, it was upgraded E-mail: [email protected] to Tropical Storm Juliette; then, at 1200 UTC on 23 DOI: 10.1175/BAMS-85- II -1663 September, it was upgraded to Hurricane Juliette.

AMERICAN METEOROLOGICAL SOCIETY NOVEMBER 2004 BAPIS• I 1663 Unauthenticated | Downloaded 10/04/21 10:04 PM UTC tember. The central pressure fell 35 hPa in the 12-h (1800 UTC on 25 September), the storm reintensified period ending at 0000 UTC on 24 September (2.9 hPa to a peak intensity of 923 hPa and 64 m s-1. Figure 1 hr-1). Juliette reached a central minimum sea level shows the track, selected intensities along the track, pressure (intensity) of 941 hPa, then weakened and weekly mean sea surface temperatures in the re- slightly over the next day. Twenty-four hours later gion (all times, pressures, and positions given are

FIG. 2. Hurricane Juliette on 26 Sep 2001 at 1638 UTC. (Upper left) GOES-10 infrared (I0.7-Aim) image showing cloud-top temperatures; (upper right) GOES-10 visible image that shows a very tightly-wound inner core; (lower left) SSM/I 85-GHz polarization-corrected temperature (PCT) image, depicting areas of heaviest precipitation; and (lower right) SSM/I 85-GHz composite microwave imagery showing upper-level heavy precipitation (red) and low-level clouds and moisture bands (green). Note that in the latter two panels, GOES-10 data are used outside the SSM/I swath. See Hawkins et al. (2001) for a detailed explanation of these products. Image courtesy of the Naval Research Laboratory in Monterey, California.

1664 I BAflS- NOVEMBER 2004 Unauthenticated | Downloaded 10/04/21 10:04 PM UTC from the NOAA National Hurricane Center Best Track dataset). Beginning at approximately this time, a second eyewall formed outside the inner ring around the "pinhole" eye. By 26 September, aircraft reconnaissance data and satellite microwave imagery indicated that a third eyewall had formed outside the inner two eyewalls. The Special Sensor Microwave Imager (SSM/I) 85-GHz channel is able to display the precipitation structure of the TC, even when other imagery is obscured by a cirrus shield (see Hawkins et al. 2001 and Spencer et al. 1989 for details on the remote-sensing aspects). Figure 2 shows an overpass on 26 September at 1638 UTC that just caught the inner core on the edge of a swath. From this image, it appears that there are three concentric eyewalls, with a moat-like feature between each of them. GOES-10 infrared and visible imagery is displayed in the top two panels, but at these frequencies, only the top cirrus shield can be seen. The lower-right panel of this figure clearly shows the inner two complete eyewalls; unfor- tunately, the swath missed about one-third of the outer eyewall, so it is difficult to conclude that it was a complete ring. However, a reconnaissance aircraft did pro- vide independent confirmation; it happened to be fly- ing in the storm just one hour after the SSM/I overpass. On 25 and 26 September, a U.S. Air Force Reserve Command WC-130 "Hurricane Hunter" aircraft from the 53rd Weather Reconnaissance Squadron flew through the intense hurricane and not only found the second-lowest pressure ever measured in the east- FIG. 3. Radial profiles of (top) tangential velocity and ern North Pacific (923 hPa), but also a very unique (bottom) relative vorticity through the northwest quad- inner-core configuration. On 25 September, the air- rant of Juliette on 25 and 26 Sep at 1819-1849 and 1722- craft was in the storm from 1745 to 2037 UTC, and 1755 UTC, respectively. The northwest quadrant is on 26 September it was in the storm from 1653 to 1933 representative of the other three but was chosen for its higher resolution and better quality data on both UTC. The highlight of the flights was on 26 Septem- days. The relevant features are the two peaks in tan- ber, when the crew found three concentric eyewalls, gential velocity on 25 Sep at 9 and 58 km (with corre- defined by three peaks in tangential wind in each of sponding peaks in relative vorticity at 7 and 55 km) and the radial legs and by three complete rings of en- the three peaks in tangential velocity on 26 Sep at I I, hanced radar reflectivity (e.g., Willoughby et al. 1982), 56, and 90 km (with corresponding peaks in relative with radii of 11, 56, and 90 km. This matches the vorticity at 9, 54, and 82 km). eyewall diameter estimates one can make using the microwave imagery in the lower two panels of Fig. 2. Data from the Hurricane Hunter aircraft are shown were noisier or of poorer resolution, so this quadrant in Fig. 3. The data shown were taken from 3 km alti- alone was chosen for containing the best data avail- tude with 30-s temporal resolution (5.5-km radial able; however, the features shown are azimuthally resolution). The top panel shows the tangential winds consistent. In this simplified radial perspective, the from radial legs in the northwest quadrant of the relative vorticity is generated solely by the radial gra- storm on 25 and 26 September, and the bottom panel dient of the tangential wind; the steepness of the vor- shows the relative vorticity from that same quadrant ticity curves may play a role in the formation of moats. on both days. Some of the data in other quadrants From the figure, one can see that the innermost

AMERICAN METEOROLOGICAL SOCIETY NOVEMBER 2004 BAI1S* I 1665 Unauthenticated | Downloaded 10/04/21 10:04 PM UTC eyewall produced the sharpest drop-off in vorticity, free to organize outside of the moat. This convection and the second (and third) eyewalls produced peaks can be axisymmetrized (ignoring external influences in vorticity that are not as dramatic, but still above the such as vertical wind shear), and a new ring of deep noise of minor fluctuations. convection (i.e., a concentric eyewall) is born. Some- So why do moats form? The answer is unclear, but times in just a few hours the new outer eyewall will it is believed that both the 3-D coupled thermodynam- dominate, the inner eyewall dissipates, and an eye- ics/dynamics as well as 2-D vorticity dynamics play wall replacement cycle is completed. Further re- critical roles. Rozoff et al. (2004) show a significant search is needed on this topic, investigating the 3-D dewpoint depression of 8°-10°C in the moat between structure of filamentation times and the relative role concentric eyewalls in Hurricane Gilbert in 1988, in- of subsidence. dicating subsidence, which inhibits convection. How- ever, the focus of their paper is on a difference in relative magnitude between vorticity and deformation as FOR FURTHER READING being another mechanism responsible for moat for- Black, M. L., and H. E. Willoughby, 1992: The concen- mation. Rotation-dominated regions, which are found tric eyewall cycle of Hurricane Gilbert. Mon. Wea. inside the radius of maximum wind and further out- Rev., 120, 947-957. side of steep negative vorticity gradients, allow coher- Hawkins, J. D., and M. Helveston, 2004: Tropical cyclone ent structures such as clouds and mesovortices to ex- multiple eyewall characteristics. Preprints, 26th Conf. ist (see Kossin et al. 2002 for a collection of hurricane on Hurricanes and Tropical Meteorology, Miami mesovortex observations, and references therein). Beach, FL, Amer. Met. Soc., 276-277. Strain-dominated regions, found just outside the ra- , T. F. Lee, J. Turk, C. Sampson, J. Kent, and K. dius of maximum wind in the steep negative vortic- Richardson, 2001: Real-time internet distribution of ity gradient, are basically filled with vorticity filaments satellite products for tropical cyclone reconnaissance. and cloud debris. In fact, one could consider a Bull. Amer. Met. Soc., 82, 567-578. "filamentation time scale" for strain-dominated re- Kossin, J. P., B. D. McNoldy, and W. H. Schubert, 2002: gions. If deformation is sufficiently greater than vor- Vortical swirls in hurricane eye clouds. Mon. Wea. ticity, then the filamentation time scale (defined as a Rev., 130, 3144-3149. temporal measure of the relative difference between Rozoff, C. M., W. H. Schubert, B. D. McNoldy, and J. P. deformation and vorticity) is shorter than the convec- Kossin, 2004: Rapid filamentation zones in intense tive time scale (defined to be the approximate time tropical cyclones. /. Atmos. Sci., in press. required to develop a thunderstorm in the Tropics— Simpson, R. H., 1974: The hurricane disaster potential about 30 min.), and no deep cloud can form. The scale. Weatherwise, 27, 169-186. greater the difference in magnitudes, the shorter the Spencer, R. W., H. M. Goodman, and R. E. Hood, 1989: filamentation time scale. Conversely, if the filamen- Precipitation retrieval over land and ocean with the tation time scale is longer than the convective time SSM/I: Identification and characteristics of the scat- scale, deep clouds have time to form and possibly or- tering signal. /. Atmos. Oceanic Technol., 6, 254-273. ganize into spiral bands or another eyewall. Willoughby, H. E., J. A. Clos, and M. G. Shoreibah, 1982: This conveniently leads to the case of double—or Concentric eyewalls, secondary wind maxima, and even triple—eyewalls. Once the initial eyewall forms the evolution of the hurricane vortex. /. Atmos. Sci., and perhaps establishes a moat, deep convection is 39, 395-411.

1666 I BAflS- NOVEMBER 2004 Unauthenticated | Downloaded 10/04/21 10:04 PM UTC