Warm-Core Formation in Tropical Storm Humberto (2001)

APRIL 2012 D O L L I N G A N D B A R N E S 1177

KLAUS DOLLING AND GARY M. BARNES University of Hawaii at Manoa, Honolulu, Hawaii

(Manuscript received 27 July 2011, in ﬁnal form 18 October 2011)

ABSTRACT

At 0600 UTC 22 September 2001, Humberto was a tropical depression with a minimum central pressure of 1010 hPa. Twelve hours later, when the ﬁrst global positioning system dropwindsondes (GPS sondes) were jettisoned, Humberto’s minimum central pressure was 1000 hPa and it had attained tropical storm strength. Thirty GPS sondes, radar from the WP-3D, and in situ aircraft measurements are utilized to observe thermodynamic structures in Humberto and their relationship to stratiform and convective elements during the early stage of the formation of an eye. The analysis of Tropical Storm Humberto offers a new view of the pre-wind-induced surface heat exchange (pre-WISHE) stage of tropical cyclone evolution. Humberto contained a mesoscale convective vortex (MCV) similar to observations of other developing tropical systems. The MCV advects the exhaust from deep convection in the form of an anvil cyclonically over the low-level circulation center. On the trailing edge of the anvil an area of mesoscale descent induces dry adiabatic warming in the lower troposphere. The nascent warm core at low levels causes the initial drop in pressure at the surface and acts to cap the boundary layer (BL). As BL air ﬂows into the nascent eye, the energy content increases until the energy is released from under the cap on the down shear side of the warm core in the form of vigorous cumulonimbi, which become the nascent eyewall. This series of events show one possible path in which a mesoscale convective system may evolve into a warm-cored structure and intensify into a hurricane.

1. Introduction and Riehl 1960; Ooyama 1964). Some have proposed that the intensification process should be split, with a pre- Wind-induced surface heat exchange (WISHE) is one WISHE phase, in which some other mechanism spins up of the leading theories explaining the intensification of a vortex to a critical point at which the WISHE theory can a strong tropical depression to a full-fledged hurricane then take over (Montgomery and Farrell 1993; Bister and (Emanuel 1986; Rotunno and Emanuel 1987). In this Emanuel 1997; Ritchie and Holland 1997; Montgomery theory the boundary layer (BL) air acquires heat and and Enagonio 1998; Davis and Bosart 2001; Hendricks water vapor from the sea, which in turn is transported et al. 2004; Molinari et al. 2004). upward in the eyewall resulting in a maximum temper- Theories about the pre-WISHE phase have recently ature anomaly in the upper troposphere. This warming been framed as either being top down or bottom up. Both hydrostatically reduces the surface pressure in the core of these approaches assume that the inner core of the cy- of the tropical cyclone (TC), and the enhanced pressure clone evolves from a mesoscale convective system (MCS). gradient between the core and the environment pro- Some MCSs develop a mesoscale convective vortex duces the destructive winds associated with a hurricane. (MCV) in the stratiform region, which trails the leading The acceptance of the WISHE theory for mature TCs is edge of convection (Gamache and Houze 1982; Leary and widespread among the scientific community, essentially Rappaport 1987; Brandes 1990; Chen and Frank 1993). formalizing the importance of sea to air energy transfer Others have hypothesized that if an MCS moves into an recognized earlier as vital for tropical cyclone develop- area favorable for genesis to occur, the stratiform rain ment (Byers 1944; Riehl 1948; Kleinschmidt 1951; Malkus region might play a role in tropical cyclogenesis (Bosart and Sanders 1981; Velasco and Fritsch 1987; Menard and Fritsch 1989; Frank and Chen 1991; Chen and Frank 1993; Corresponding author address: K. Dolling, Department of Mete- orology, University of Hawaii at Manoa, 2525 Correa Rd., Honolulu, Simpson et al. 1997; Rodgers and Fritsch 2001). HI 96822. Top-down theories propose that the MCV migrates to E-mail: [email protected] the surface at which point WISHE takes over. Simpson

DOI: 10.1175/MWR-D-11-00183.1

Ó 2012 American Meteorological Society Unauthenticated | Downloaded 09/27/21 12:53 PM UTC 1178 MONTHLY WEATHER REVIEW VOLUME 140 et al. (1997) emphasize the importance of MCV Enagonio and Montgomery 2001; Hendricks et al. 2004; mergers during the formation of TC Oliver during the Montgomery et al. 2006). Recently, VHTs have been Tropical Ocean and Global Atmosphere Coupled Ocean– hypothesized to be part of the marsupial pouch paradigm Atmosphere Response Experiment (TOGA COARE). (Wang et al. 2010a,b). In this scenario, an area of quasi- The mergers increase the vertical extent of the resulting closed Lagrangian circulation (the cat’s eye) is moistened vortex so a surface connection can be established. The by convection and mostly protected from dry air intru- vorticity-rich MCV environment reduces the Rossby ra- sion. The parent wave is maintained and amplified by dius of deformation, which allows for the more efficient convection within the cat’s eye, defined as the intersec- warming of the troposphere. Harr et al. (1996) used tion of the wave critical latitude and the trough axis. Omega dropwindsondes (ODWs) to investigate the for- Stossmeister and Barnes (1992) documented the de- mation of Typhoon Ofelia (1993) that formed from one velopment of a new second vortex that became the persistent MCS. Although Ofelia’s formation is viewed as circulation center for Isabel (1985). This new vortex a top down, Harr et al. (1996) found warm, dry midlevel formed beneath the downwind anvil of intense convec- air near the circulation center and argues that subsidence tion in a rainband. They hypothesized that subsidence induced on the concave side of the main convective band warming beneath the anvil, similar to the mesolows might have played a role in warming the troposphere and found at the trailing edge of MCSs, could lower the in the formation of an eye. pressure and serve as the initial perturbation for a TC. In the Tropical Experiment in Mexico (TEXMEX), Heymsfield et al. (2006) investigated Tropical Storm Bister and Emanuel (1997) suggested that after an MCS Chantal in a highly sheared environment. Although has formed with a cold core in the lower atmosphere and Chantal failed to intensify into a hurricane, detailed a cyclonic circulation in the midtroposphere, the tropo- analysis of the storm provides clues to warm-core for- sphere becomes nearly saturated in the core of the vortex. mation. The low-level circulation center (LLCC) is ob- This prevents the evaporation of rain and inhibits served 80 km west-southwest of the deep convection in downdrafts that bring cool dry air to the BL. Enhanced a rain-free region. Two broad regions of warming are surface fluxes associated with increased surface winds found, one of these was argued to be due to subsidence near the core will increase the BL moist static energy. from the interaction of the storm with high wind shear, After convection occurs, the temperature of the vortex as suggested earlier by Ritchie and Elsberry (2001). core will increase, and the WISHE mechanism can then This area had a maximum temperature perturbation at act as a positive feedback to the warm-core cyclone. 500 hPa and was located partially over the low-level Bottom-up theories suggest that a low-level vorticity- circulation center. The other area of warming was found rich environment in concert with deep convection can ini- to have a maximum at the 700-hPa level. Heymsfield et al. tiate tropical cyclogenesis. Montgomery and Kallenbach (2006) hypothesized that this area of warming was due (1997), Molinari et al. (2004), and Reasor et al. (2005) have to similar mechanisms as observed by Stossmeister and hypothesized that the axisymmetrization of vorticity Barnes (1992). centers is important in spinning up a TC. Here vorticity A rare set of observations collected in Tropical Cy- is increased in localized areas by vortical hot towers clone Humberto (2001) are interpreted to gain insights (cumulonimbi with high vorticity) embedded in a well- into how a hurricane may form. The observations pro- organized, but weak low-level cyclonic circulation. vide evidence that WISHE theory, though vital for in- Axisymmetrization and the stretching of these vorticity tensification, may not explain the formation of a tropical centers intensify the initial surface vortex to a critical cyclone where deep clouds become organized and ro- point at which time WISHE becomes dominant. tational flow develops. The flights into Tropical Storm To examine the formation of TC Diana (1984), Humberto on 22 September provide observations on the Hendricks et al. (2004) used a model with a 3-km hori- pre-WISHE phase of tropical cyclone development. zontal grid that is near–cloud resolving. The simulation GPS sondes, radar from the WP-3D, and in situ aircraft shows the presence of vortical hot towers (VHTs), Cbs measurements are utilized to observe thermodynamic with high vertical vorticity. They develop from the and kinematic structures in Humberto and their rela- buoyancy-induced stretching of the local absolute vertical tionship to stratiform and convective elements. In this vorticity. At this grid spacing, VHTs are the preferred study, we examine the following questions: mode of convection. VHTs produce diabatically generated low-level PV anomalies that axisymmetrize and 1) Does the warm core form in deep convection or does merge, which is a step in the intensification of the newly it form in the stratiform area of precipitation? formed vortex (Montgomery and Lu 1997; Montgomery 2) Where does the warm core form as a function of and Enagonio 1998; Montgomery and Franklin 1998; height and do the observations verify WISHE theory?

Unauthenticated | Downloaded 09/27/21 12:53 PM UTC APRIL 2012 D O L L I N G A N D B A R N E S 1179

3) Is there an MCV present and what role does it play in the formation of the tropical cyclone?

2. Data and methodology a. GPS sonde quality control The global positioning system dropwindsondes (GPS sondes) sample at 2 Hz providing data every 6–7 m in the lower atmosphere. Wind errors from the sondes are 0.5 m s21 while temperature and pressure errors are 0.28C and 1.0 hPa, respectively (Hock and Franklin 1999). Relative humidity errors average less than 5%; however, sensor wetting can still be a problem. Bogner et al. (2000) and Barnes (2008) offer schemes to mitigate this issue. The data are initially processed using the At- mospheric Sounding Processing Environment (ASPEN) program. ASPEN, developed at the National Center for Atmospheric Research (NCAR), uses an extensive series of quality control algorithms to systematically identify spurious data and either corrects or removes these from the dataset. Details of these quality control algorithms FIG. 1. The location of the GPS dropsondes, marked by the time can be found in the ASPEN user manual (Martin 2007). (UTC) they were jettisoned. Aircraft are identified with ‘‘P’’ for the NOAA WP-3D, ‘‘D’’ for the NASA DC-8, and ‘‘E’’ for the NASA b. GPS sonde deployment ER-2. Mean position of the center of the storm at the central analysis time marked by the black dot on the latitude–longitude grid. Ninety-five percent of the GPS sondes deployed on 22 September recorded data successfully. A few GPS sondes were not used because of their distance from the behind the aircraft and rotates in a vertical plane. The storms circulation center. The WP-3D dropped 21 GPS WP-3D reflectivity observations from both radars are sondes from approximately 600 hPa. The ER-2 dropped discussed by Marks (1985). The radar scans determine 2 GPS sondes, sampling from the stratosphere to the the convective and stratiform regions and they are used surface, while the DC-8 flew at 300 hPa and jettisoned to place the GPS sonde data with respect to the persis- 15 GPS sondes. The horizontal spacing of GPS sondes tent mesoscale features found in the TS. In situ aircraft is about 20 km near the circulation center and about sensors are described by Jorgensen (1984) and Aberson 40 km by 100 km from the center. Twenty-three GPS et al. (2006). sondes were used for the analysis of the low-level fields d. Analysis scheme (Fig. 1). These GPS sondes were jettisoned between 1843 and 2131 UTC, resulting in a composite time of 2 h, Determining storm track is the first step of the anal- 48 min. This is approximately one sonde launch every ysis. The technique of Willoughby and Chelmow (1982) 7–8 min. The upper-level analysis used 10 GPS sondes. is employed to establish the circulation center based on They were jettisoned between 1819 and 2236 UTC, a aircraft fixes. Best-track satellite centers for periods that composite time of 3 h, 17 min. This is an average of one extend on either side of the flight time are used to sup- sonde launch every 25–26 min. A brief summary of the plement the aircraft fixes. A few of the fixes were not Convection and Moisture Experiment (CAMEX) and used because they would cause unusual accelerations in the coordination between the aircraft is presented by the storm’s motion. On 22 September the fixes fit a linear Kakar et al. (2006). regression with an r2 5 0.98. After acquiring the storm track, the position of each c. Aircraft instrumentation sonde relative to the circulation center is determined. As Two research radars are on the WP-3Ds. The lower the sonde falls its position was corrected for its motion fuselage (LF) radar operates at a wavelength of 5.5 cm relative to the moving circulation center. These data (C band). This radar scans horizontally at all azimuth were then all composited to a central time. Here we are angles and provides plan views of the TC. The tail radar assuming that the evolution of the TS produced negli- (X band, 3.2-cm wavelength) scans from 258 ahead to 258 gible variations in structure compared to the spatial

Unauthenticated | Downloaded 09/27/21 12:53 PM UTC 1180 MONTHLY WEATHER REVIEW VOLUME 140 variations inherent to the TS at any given time. Hum- berto deepened only a few hectopascal (hPa) during the flight resulting in our ability to link our derived vortex- scale wind and state variable fields to the long-lived mesoscale features. In the third step the storm motion u and y winds were subtracted from the earth-relative u and y winds to obtain the storm relative winds. These were then placed into cylindrical coordinates to obtain relative tangential andradialcomponents.Thefourthstepistoapplya linear interpolation to fill gaps of 300 m or less and in- terpolate values to 10-m height intervals. In the final step a piecewise cubic Hermite interpolation (Fritsch and Carlson 1980) was performed to create FIG. 2. The pressure (hPa) and streamlines at 20-m altitude on 22 a field for any chosen variable for each level. This cubic Sep 2001. Composite time is 2 h, 48 min. Latitude and longitude spline method preserved the monotonicity and the shape delineated every 18. The color bar to the right denotes pressure of the data and reproduced the observed value at its values, the white dot marks the circulation center, and the four digit observed location. The plan view maps constructed with numbers correspond to GPS sonde locations (Dolling and Barnes 2012). the cubic spline matched well with subjective analyses. A three-dimensional matrix is created by stacking the parameters for genesis (Gray 1968, 1979). At the time of plan view maps at every 10 m and keeping the grid the analysis there was still strong divergence aloft asso- boundaries the same. Once a three-dimensional matrix ciated with the TUTT to the southwest of Humberto, fa- is constructed, one may slice through it at any point and cilitating the deep convection. When the first GPS sondes in any direction to get a vertical cross section. For some were jettisoned late on 22 September Humberto’s mini- of the meteorological variables this was possible through mum central pressure was 1000 hPa and it had attained a large depth of the troposphere because the DC-8 and tropical storm strength. Twenty-four hours later, late on the ER-2 were dropping GPS sondes at high altitudes. 23 September, it would intensify to a category-2 storm. Convective available potential energy (CAPE) is calculated for each vertical column in the matrix. This is likely an underestimate as CAPE in TCs may be calculated on 3. Results lines of constant angular momentum, which tilt out from a. GPS sonde analysis and reflectivity structure the vortex core with height. More details of the analysis scheme can be found in Dolling (2010). The pressure field and streamlines at 20-m altitude (Fig. 2) reveal an area of confluence about 50 km to the e. Humberto (2001) north-northeast of the circulation center. The reflectivity Humberto had its origin in a trough that extended field (Fig. 3) exhibits an arc of deep convection, oriented southwest from the circulation of Hurricane Gabrielle north to east of the low pressure center, located 40–50 km (Beven et al. 2003). On 18 September, a westward-moving from the nascent circulation center. The arc of deep tropical upper-tropospheric trough (TUTT) enhanced convection is collocated with the area of confluence in the the deep convection and a weak low formed in the area. streamline analysis provided in Fig. 2. The system became a tropical depression near 1200 UTC Vertical cross sections of the reflectivity structure (i.e., 21 September, when deep convection developed near the RHI), taken from the tail radar of the WP-3D appear the center of a broad cyclonic circulation. At about in the next two figures. Figure 4a is a north–south ver- 1800 UTC 21 September Humberto’s pressure began to tical cross section along the line marked ‘‘a’’ in Fig. 3. drop while under the east side of the TUTT. Early on 22 This scan cuts through the arc of convection located to September, Humberto was a tropical depression with the north of the circulation center. Echo tops reach to a minimum central pressure of 1010 hPa. The storm 15–16 km, the highest echo tops in the convective arc. was located in the Atlantic basin near 298N, 668W. The Echo tops become progressively lower farther to the movement was to the north-northwest at approximately east and south in the arc. Another RHI scan (Fig. 4b) 4ms21. Humberto would continue moving to the north runs from northwest to southeast and is shown as a line and eventually recurve to the northeast. The sea surface marked ‘‘b’’ in Fig. 3. To the northwest there is an anvil temperatures (SSTs) ranged from about 288 to 28.58Cand with its base near 6 km and tops around 10 km. The area the vertical wind shear was 6–7 m s21, which are within of stratiform clouds (anvil) is the exhaust from the

Unauthenticated | Downloaded 09/27/21 12:53 PM UTC APRIL 2012 D O L L I N G A N D B A R N E S 1181

FIG. 3. Plan view reflectivity image at 1900 UTC 22 Sep 2001. The image is 120 km 3 120 km. The white dot shows the LLCC. Streamlines at 5 km are superimposed on the reflectivity image. Lines ‘‘a’’ and ‘‘b’’ are RHI reflectivity images shown in the next two figures. Two black dots show the position of the WP-3D at the FIG. 4. (a) RHI reflectivity image at 1853 UTC 22 Sep 2001. The time of the RHI scans. Red arrow shows position of aircraft as it image is 120 km wide and 20 km high and corresponds to line ‘‘a’’ encounters a mesoscale downdraft. The WP-3D is moving east- in Fig. 3. WP-3D is flying at approximately 4 km and its position northeast. The white ‘‘plus’’ sign in the center of the image is the is at 28.838N. (b) As in (a), but the time is 1854 UTC and corre- aircraft position at the time of scan. sponds to line ‘‘b’’ in Fig. 3. WP-3D is at center of image at 28.88N. The high reflectivities adjacent to the sea are due to surface contamination effects, not rain. convective arc that lies to the north of the LLCC. The WP-3D in Fig. 4b is located about 10 km northwest of the LLCC at the time of the RHI scan. To the southeast (1977) and Houze (1977) looked at the structure of strong of this scan, the reflectivity rates are about 20–30 dBZ. tropical squall lines. The leading edge of the squall line This is equivalent to a rain rate of about 1–4 mm h21 is composed of a gust front and tilted updrafts followed (Jorgensen and Willis 1982). Notice that the area of light by an area of heavy precipitation where cool saturated rain is located over the LLCC. downdrafts are collocated with a mesohigh. In the rear The plan view reflectivity with a streamline analysis of the squall is an area of lighter stratiform precipitation at 5 km superimposed on it (Fig. 3) reveals that an MCV where a mesolow is maintained. This mesolow is charac- is observed with its center located about 50 km north- terized by a specific type of sounding that is nearly satu- northeast of the LLCC and just south of the convective rated and moist adiabatic from the tropopause to around arc. The MCV advects the exhaust from the arc of deep 600 hPa. Between 600 hPa and the top of the BL, the air convection in a cyclonic direction over the LLCC. This is is warmer than the surrounding environment with high in contrast to prior work mentioned in the introduction. dewpoint depressions and a lapse rate that is nearly dry Bister and Emanuel (1997) hypothesize that the vor- adiabatic. This structure indicates subsidence; Gamache ticity associated with the MCV migrates to the surface. and Houze (1982) calculated descent rates on the order In Humberto, the center of the MCV and the LLCC are of 20.5 m s21 under the anvil of one of the stronger offset by 50 km; the MCV remains at midlevels and ad- squall lines that traveled through the GATE array. vects the anvil over the LLCC where the warm core is Leary (1980) used a one-dimensional, steady-state, observed under the trailing edge of the anvil. hydrostatic model in an attempt to reproduce this area of dry adiabatic descent on the trailing edge of the anvil. b. The role of the stratiform precipitation in forming The red dashed line (Fig. 5) exhibits a sounding for a me- the warm core soscale downdraft originating at 600 hPa with a tempera- During the Global Atmospheric Research Program ture of 08C and relative humidity of 100%. The profiles of (GARP) Atlantic Tropical Experiment (GATE), Zipser temperature and humidity in these soundings are located

Unauthenticated | Downloaded 09/27/21 12:53 PM UTC 1182 MONTHLY WEATHER REVIEW VOLUME 140

FIG.5.SkewT–logp diagram of calculated proﬁles of temperature and dewpoint for mesoscale downdrafts originating at 600 hPa with a temperature of 08C, a relative humidity of 100%, and vertical velocities of 20.2 m s21, resulting in the heavy solid green lines for temperature and dewpoint, and 20.5 m s21 (heavy dashed red lines) superposed on Zipser’s (1977, his Fig. 8) collection of soundings beneath anvil clouds. The vertical coordinate is pressure and all temperatures are in 8C (Leary 1980). under the trailing edge of a precipitating anvil cloud (Zipser temperature proﬁle is close to dry adiabatic (Fig. 6). 1977). Leary calculates that with a realistic drop size dis- Dewpoint depressions reach about 138C at 820 hPa. The tribution, the model reproduces soundings that closely re- base of the inversion is at 940 hPa. Two other GPS sondes semble the soundings found in the mesolow area of the were jettisoned near the circulation center and they dis- tropical squall line. She found that with light rain rates (;1– play similar structures to the GPS sonde in Fig. 6. All 5mmh21) and a preexisting mesoscale downdraft origi- three of the soundings look very similar to the soundings nating at the base of the anvil of 20.5 m s21, the model discussed by Zipser (1977) and Leary (1980). Bister and soundings closely resembled the thermodynamic structures Emanuel (1997) suggest that after an MCS and a mid- found in these mesoscale unsaturated downdrafts. tropospheric MCV has formed, the troposphere becomes Near the LLCC of Humberto under the trailing edge nearly saturated on the mesoscale in the vortex core. of the anvil, from about 600 to about 850 hPa, the Observations in Humberto demonstrate that the core is

Unauthenticated | Downloaded 09/27/21 12:53 PM UTC APRIL 2012 D O L L I N G A N D B A R N E S 1183

FIG. 6. Skew T–logp diagram for a GPS sonde jettisoned at 1854 UTC 22 Sep 2001, from the DC-8 near the circulation center. The red horizontal lines are pressure, the blue lines slanted from bottom left to top right indicate temperature, the dark blue lines slanted in opposite direction are dry adiabats, the curved green lines are moist adiabats, the blue dashed lines are water vapor, the thick red line is the sounding temperature, and the thick blue dashed line is the sounding dewpoint (Dolling and Barnes 2012). not saturated. The high dewpoint depressions indicate the circulation center. The WP-3D track is from west- a deep dry layer in the lower troposphere. southwest to the east-northeast and the red arrow (Fig. 3) Figure 7 displays the vertical velocity observed by depicts the path of the aircraft as it passed by the LLCC. the WP-3D as it travels under the stratiform area near A downdraft of signiﬁcant magnitude is present during

21 FIG. 7. Vertical velocity (m s ) as a function of time (UTC) from the WP-3D ﬂying at 4–5 km. The aircraft is near the circulation center from 1852 to 1855 UTC.

Unauthenticated | Downloaded 09/27/21 12:53 PM UTC 1184 MONTHLY WEATHER REVIEW VOLUME 140

FIG. 8. East–west vertical cross section of the temperature anomaly, on an axis through the surface circulation center. The color bar indicates the temperature anomaly (K). The storm center is located at 66.78W. The composite time is 2 h, 48 min, from 1843 FIG. 9. Plan view of temperature at 400 hPa (8C) from the GPS to 2131 UTC. sondes. The temperature is contoured (solid black) every degree. The black dot marks low-level circulation center. the passage near the circulation center. The magnitude of the downdraft is 2–6 m s21 for about 3 min corresponding to a distance of 25 km. The area of mesoscale descent Subsidence collocated with the trailing edge of the may be larger as the WP-3D ﬂew adjacent to the LLCC. stratiform precipitation forms a low-level warm core and

Equivalent potential temperature ue on the same transect decreases the surface pressure. In the upper troposphere increases by 2 K as the WP-3D traverses the downdraft. thermodynamic conditions are similar to the surround-

This warmer ue is likely from the convective arc that has ing environment with no evidence of the expected pos- been advected downwind in the anvil and over the LLCC. itive temperature perturbation that has been observed Two soundings found near the LLCC are compared in mature hurricanes (e.g., Hawkins and Imbembo 1976). with soundings jettisoned at the periphery of the analysis There is only a small positive temperature perturbation area to determine the impact of the warm dry layer that of about 18C in the upper troposphere over the circula- was apparent from 500 to 1000 hPa on the surface pres- tion center (Fig. 9). The National Aeronautics and Space sure. Calculating the hydrostatic pressure deficit from this Administration (NASA) DC-8, which was flying at warm dry layer reveals that this layer accounts for about 10 km, also observed a 18C temperature perturbation a 9–11-hPa drop in pressure, depending on which pe- confirming the GPS sonde data. One might expect some ripheral sounding is used. This accounts for over 90% warming to the north and the possibility of a tilted warm of the total pressure drop observed at the surface in TS core because of the moderate vertical shear of the hori- Humberto. zontal wind; however, this is not present. In these early Using all the GPS sondes jettisoned on 22 September, stages of development, the maximum temperature per- a three-dimensional matrix is constructed and then a turbation is not in the upper atmosphere as would be vertical cross section of the temperature perturbation is expected from the WISHE theory. made as described in the data and methodology section. In a related study using the vortex data messages, This vertical crosssection is an east–west slice directly Vigh (2010) investigates eye formation. He observes that through the LLCC. The temperature perturbation cross many storms display evidence of subsidence warming and section (Fig. 8) exhibits a maximum of 68–78Clocatedat drying in the lower troposphere up to 3 days before the about 2 km. This low-level positive temperature pertur- aircraft eye (visible on radar) is observed. The pre-eye bation is found beneath the anvil with rain rates of ap- warming events become more frequent 30–24 h before proximately 1–4 mm h21. This suggests that in Tropical eye formation and are present up to the time that the eye Storm Humberto, conditions occur that are similar to has formed. those discussed by Zipser (1977) and Leary (1980), ex- The schematic in Fig. 10 highlights the difference in cept that the downdraft observed just below the anvil is warm-core formation for early stage TC development for an order of magnitude larger. Humberto and a system if it were to form from a WISHE

Unauthenticated | Downloaded 09/27/21 12:53 PM UTC APRIL 2012 D O L L I N G A N D B A R N E S 1185

FIG. 10. (a) Schematic of skew T–logp based on WISHE. The black line is a moist adiabat and represents environmental temperature structure, while the red line represents an undilute eyewall updraft. The red hatched area displays warming. (b) A schematic of the temperature anomaly associated with (a) displays a maximum temperature anomaly in the upper troposphere. (c) As in (a), but for the observations of Humberto. Below 500 hPa, the red line follows a dry adiabat until encountering the inversion and the BL. (d) As in (b), but that the temperature anomaly is associated with the temperature structure of (c) and displays maximum warming in the lower troposphere as observed in Humberto. The values in (b) and (d) are not actual observed values, but simply representative of T structure. perspective. The top two panels reveal how a warm core responsible for the formation of the nascent warm core would appear if WISHE was in action. As Bister and located in the lower levels of the troposphere (Fig. 8). Emanuel (1997) suggest, a nearly saturated core with The area of mesoscale descent is also capping the area increased fluxes from the sea surface would cause con- beneath it, which allows for a buildup of ue via fluxes from vection in the core of the system. Having nearly saturated the sea surface. As discussed by Dolling and Barnes the core of the storm and assuming an undilute updraft, (2012), this reservoir of ue is about 750 m deep and is as air would rise along a slightly warmer moist adiabat than warm as 365 K (Fig. 11). The reservoir was created over the environment (Fig. 10a). Since moist adiabats spread a number of hours as air swirled into the LLCC and ac- with height, the maximum warming would be present quired heat and moisture from the sea. The high ue air is in the upper troposphere (Fig. 10b). In the case of not located in the high wind area of the storm as is ex- Humberto, a mesoscale downdraft on the trailing edge pected from WISHE. Humberto is yet to be driven by the of the anvil causes dry adiabatic descent under the anvil sea to air fluxes that are so important in mature TCs. (Fig. 10c). This warms the lower troposphere, decreases The cap (Fig. 12) has values of CIN as strong as the surface pressure, and causes a maximum tempera- 2250 J kg21 over the LLCC. CAPE near the LLCC is as ture perturbation at low altitudes (Fig. 10d). high as 2500 J kg21 and values of ;2000 J kg21 are also found in a small area collocated with the deep convec- c. Trapping of air below warm core and u buildup e tive band to the northeast of the circulation center. The The area of mesoscale descent (Figs. 6 and 7) located large CAPE collocated with the LLCC could contribute on the trailing edge of the anvil observed from the WP- to the acceleration of the updraft and spinup via the 3D and GPS sondes is performing two functions. It is stretching term of the absolute vorticity equation. This is

Unauthenticated | Downloaded 09/27/21 12:53 PM UTC 1186 MONTHLY WEATHER REVIEW VOLUME 140

21 21 FIG. 11. Plan view at 20-m height of equivalent potential tem- FIG. 12. Plan view of CAPE (J kg ) and CIN (J kg )asa perature. Streamlines superposed as blue lines with arrows. Lati- function of latitude and longitude. CAPE is shown by the values tude and longitude delineated every 0.58. The color bar at right on the color bar. CIN is represented by the blue contours. The corresponds to equivalent potential temperature (K). The white circulation center is marked by a blue ‘‘1’’ (Dolling and Barnes dot indicates the surface circulation center. 2012).

the dominant term in the spinup process given that the (350 K) at 1942 UTC. The 350-K ue is advected cy- environment of a developing TC has low vertical shear clonically around the nascent eye and mixing out from of the horizontal wind (Gray 1968; McBride and Zehr the strong convection on the north side of the nascent 1981) and density variations (solenoidal term) are neg- eyewall. ligible. The convective cells that feed upon this air would The arc of convection at 2109 UTC (Fig. 15a) wrap- have a higher concentration of vorticity and may be akin ping from southeast to north of the LLCC in Fig. 3 has to the vortical hot towers discussed by Hendricks et al. reformed closer to the circulation center. At this time (2004), Reasor et al. (2005), and Montgomery et al. (2006). the WP-3D is far enough away from the LLCC that at- tenuation is a problem when viewing the convection d. Evolution of reflectivity structures near the storm center. The plan view of the reflectivity at and building an eyewall 2123 UTC (Fig. 15b) reveals that the convection near the Figure 13a is a plan view of the reflectivity structure circulation center has changed from the plan view in Fig. at 1942 UTC. The WP-3D is flying from north to south 13a. Instead of two distinct areas of convection, the arc at an altitude of 4 km. Comparing Figs. 13a and 3 50 km north and the new cells nearer the center, there is (1900 UTC), there is a marked change in the reflectivity one broad arc of convection. In observing a loop of the structure. In contrast to Fig. 3, the structures in Fig. 13a radar images, at the time that warm ue BL air breaks out exhibit reflectivity as high as 46 dBZ adjacent to the from under the nascent warm core and initiates con- north side of the capping inversion within about 25 km vection just north of the circulation center, the arc of of the LLCC. The arc of deep convection 50 km to the convection 50 km to the north breaks apart and new north of the circulation center looks less organized at convective cells start to form adjacent to the low-level this time. As the WP-3D travels north and passes just warm core with the trailing southeast portion of the west of the new convection near the LLCC at 2002 UTC, convective arc remaining as a major rainband. an RHI scan (Fig. 13b) shows that the anvil to the west is As the WP-3D first entered Humberto the convective still present as vigorous new convection reaches heights arc was at a distance more than double the radius (20– of 15 km. 25 km) of the low-level warm perturbation. As the WP- At 1941 UTC the WP-3D is flying directly through the 3D left Humberto, the new unified arc is adjacent to the new convection (Fig. 13a) on the north side of the low-level warm core and has a structure more like LLCC. Equivalent potential temperatures as warm as a mature TC with a warm core, a nascent eyewall, and 356 K are recorded by the WP-3D (Fig. 14). Such warm a major rainband. Equivalent potential temperatures ue can only have originated in the reservoir found in the will increase in the eyewall as convection transports nascent eye and under the dry adiabatic layer. As the higher ue up the eyewall column. A ring of higher ue WP-3D flies south it encounters another area of warm ue around the eyewall would eventually create a TC with

Unauthenticated | Downloaded 09/27/21 12:53 PM UTC APRIL 2012 D O L L I N G A N D B A R N E S 1187

FIG. 14. Equivalent potential temperature as measured from the NOAA WP-3D ﬂying at approximately 600 hPa from 1938 to 1947 UTC 22 Sep 2001. The y axis is the equivalent potential temperature (K) and the x axis is time.

The salient observations in Humberto are as follows: (i) the warm core forms away from deep convection under the anvil in the stratiform area of precipitation, (ii) the warm core forms in the lower troposphere con- trary to what would be expected if WISHE were acting, and (iii) the MCV is offset from the LLCC and advects the exhaust from the anvil over the LLCC where a me-

FIG. 13. (a) Plan view reflectivity image at 1942 UTC 22 Sep soscale area of subsidence is observed to warm the lower 2001. The image is 120 km 3 120 km. The white dot shows the low- troposphere. A schematic of the structures that are ob- level circulation center. The location of the WP-3D is marked by served in Humberto appears in Fig. 16. As the WP-3D the white plus symbol in the center of the image. The areas of low first enters Humberto on 22 September, when it is a reflectivity on the perimeter of the image are due to sea contami- 1000-hPa tropical storm, the reflectivity pattern consists nation, not rain. (b) RHI scan at 2002 UTC from west to east. Line in (a) shows the position of the scan. The high reflectivities adjacent of an arc of convection approximately 50 km north of to the sea are due to surface contamination effects, not rain. the low-level circulation center. GPS sonde observations reveal that an MCV is present with its center about 50 km north-northeast of the LLCC. RHI scans clearly a warm core in the upper troposphere, essentially like show that an anvil created from the convective arc is WISHE theory. advected over the LLCC by the midlevel winds associated with the MCV. Under the trailing edge of the anvil and over the LLCC, subsidence is evident from aircraft 4. Summary and discussion observations and multiple GPS sondes. The soundings in Plan views and vertical cross sections of kinematic the vicinity of the LLCC have lapse rates below 5 km and thermodynamic structures constructed from GPS that are approximately dry adiabatic with dewpoint sondes acquired over 3 h on 22 September 2001 are depressions of over 12 K. The maximum temperature utilized to examine Tropical Storm Humberto. High- perturbation over the LLCC is about 7 K at 2 km. The altitude GPS sondes from NASA’s ER-2 and the DC-8 warm dry layer beneath 5 km accounts for 90% of the along with GPS sondes from the NOAA WP-3D allows surface pressure reduction. High-altitude GPS sondes for an investigation of Humberto through the depth of and the NASA DC-8 record only a 1-K temperature the troposphere. A thorough investigation of every GPS perturbation in the upper troposphere. At this point in sonde is made to correct the key variables and to omit formation the warm dry layer below 5 km is caused by spurious data. The GPS sonde data are combined with subsiding air and not warmer ue air moving up the eye- WP-3D flight level data and radar in an effort to docu- wall column. Rain rates over the circulation center are a ment the important aspects of the formation of Tropical few millimeters per hour. The thermodynamic and re- Storm Humberto. flectivity structures we observe in Humberto are similar

Unauthenticated | Downloaded 09/27/21 12:53 PM UTC 1188 MONTHLY WEATHER REVIEW VOLUME 140

FIG. 16. Schematic of Tropical Storm Humberto. The convective cell height and intensity depicted by colors with the white depicting tops of 16 km and the red up to 10 km. The warm core is located in subsidence region below the base of the trailing edge of the anvil. The low-level ﬂow (blue arrows) and LLCC are offset from the center location of the MCV located at 500 hPa.

Similar to TS Chantal (Heymsﬁeld et al. 2006) and TS Isabel (Stossmeister and Barnes 1992), Humberto had a LLCC that was displaced tens of kilometers from convection. Both Chantal and Humberto had their highest temperature perturbation in the lower troposphere. The analysis of Isabel also displayed a warm dry layer in the lower troposphere. Another commonality between de-

veloping tropical storms is the high ue reservoir collo- FIG. 15. (a) Plan view reflectivity image at 2109 UTC 22 Sep cated with the LLCC in TS Chantal, Humberto, and 2001.The image is 240 km 3 240 km. The white dot shows the low- Danny (Molinari et al. 2004). level circulation center and the white ‘‘plus’’ sign in the center of the image is the WP-3D location. (b) As in (a), but at 2123 UTC. Although Humberto forms at a relatively high lati- The areas of low reflectivities 40–50 km from the WP-3D are due to tude, the environmental conditions on 22 September are sea contamination, not rain. The shear vector and magnitude are ideal for formation and intensification of a tropical sys- indicated by the arrow. tem. A primary theme of many past observational studies has been the transformation of a low-level cold core to to the model results by Leary (1980). An existing me- a warm core (Riehl 1948; Hubert 1955; Yanai 1961). In soscale downdraft with light rain rates allows the air to Humberto, the observations reveal a warm dry layer in continue its descent without becoming saturated and the lower troposphere. The warm dry layer lowers the causes a deep layer of warm dry air to form. The pres- surface pressure and caps the BL allowing for an in- sure reduction of 11 hPa at the sea surface is due to this crease in ue.Itisthiswarmue air that feeds the nascent warm dry layer. If a mesoscale area with such a pressure eyewall convection. Concurrently with the outbreak of lowering could be maintained for ;12 h, then we may be eyewall convection, the convective arc 50 km to the viewing a critical step in tropical cyclone formation. north reforms adjacent to the low-level warm core. In Underneath the warm perturbation the boundary this paradigm, the formation of the nascent warm core layer acquires more heat and moisture, but the warm is the first step in a sequence of events that transform layer above the BL traps this air, allowing ue to reach the Humberto from a tropical system that resembles an warmest values seen anywhere in the storm. When this MCS to a system that resembles a TC. More studies air eventually escapes it has a CAPE of 2500 J kg21 that on systems in the pre-WISHE phase of development feeds the deep vigorous cumulonimbi that later become are needed to verify whether this is a common path the nascent eyewall. This high CAPE would tend to con- for formation. Humberto does, however, continue to centrate vorticity via the stretching term. develop and the minimum central pressure drops by

Unauthenticated | Downloaded 09/27/21 12:53 PM UTC APRIL 2012 D O L L I N G A N D B A R N E S 1189

20 hPa in the next 24 h, close to the threshold for rapid REFERENCES intensification. Aberson, S. D., M. L. Black, R. A. Black, R. W. Burpee, J. J. Cione, Future experiments in developing storms should use C. W. Landsea, and F. D. Marks, 2006: Thirty years of tropical both GPS sondes and Doppler radar with the two WP- cyclone research with the NOAA P-3 aircraft. Bull. Amer. 3Ds and upper-level aircraft. More aircraft will lower Meteor. Soc., 87, 1039–1055. the sampling time and allow sampling of a larger area. Barnes, G. M., 2008: Atypical thermodynamic profiles in hurricanes. Mon. Wea. Rev., 136, 631–643. GPS sondes jettisoned to capture the mesobeta scale Beven, J. L., S. R. Stewart, M. B. Lawrence, L. A. Avila, J. L. near the circulation with a larger spacing farther away Franklin, and R. J. Pasch, 2003: Annual summary—Atlantic from the center is necessary. Doppler analysis together season of 2001. Mon. Wea. Rev., 131, 1454–1484. with GPS sondes analysis might reveal the role of both Bister, M., and K. A. Emanuel, 1997: The genesis of Hurricane VHTs and the thermodynamics of the inner core. It will Guillermo: TEXMEX analyses and a modeling study. Mon. also be important to jettison high-altitude GPS sondes Wea. Rev., 125, 2662–2682. Bogner, P. B., G. M. Barnes, and J. L. Franklin, 2000: Conditional to map the MCV/LLCC displacement. instability and shear for six hurricanes over the Atlantic Ocean. Wea. Forecasting, 15, 192–207. Bosart, L. F., and F. S. Sanders, 1981: The Johnstown flood of July 5. Speculation 1977: A long-lived convective system. J. Atmos. Sci., 38, 1616– The prevalence of anvils in nearly all developing TCs 1642. Brandes, E. A., 1990: Evolution and structure of the 6–7 May 1985 suggests that subsidence in a light rain region resulting mesoscale convective system and associated vortex. Mon. in a warm core should occur often. What are the critical Wea. Rev., 118, 109–127. conditions that make some anvils more potent than Byers, H. R., 1944: General Meteorology. McGraw-Hill, 645 pp. others in forming a low-level warm core? The VWS di- Chen, S. C., and W. M. Frank, 1993: A numerical study of the genesis rection and magnitude likely affect the spatial arrange- of extratropical convective mesovortices. Part I: Evolution and ment of the MCV relative to the LLCC. An example dynamics. J. Atmos. Sci., 50, 2401–2426. Davis, C. A., and L. F. Bosart, 2001: Numerical simulations of the may be TS Chantal. The lower-tropospheric warming genesis of Hurricane Diana (1984). Part I: Control simulation. was observed 80 km southwest of the deep convection; Mon. Wea. Rev., 129, 1859–1881. this region was possibly a remnant of warming that oc- Dolling, K., 2010: The evolution of Hurricane Humberto (2001). curred when Chantal was in a lower VWS environment. Ph.D. dissertation, Dept. of Meteorology, University of Ha- Increasing VWS may have separated the anvil from waii at Manoa, Honolulu, HI, 180 pp. ——, and G. M. Barnes, 2012: The creation of a high equivalent the LLCC. The rain rate under the anvil cannot be too potential temperature reservoir in Tropical Storm Humberto heavy, otherwise, the temperature will approach a moist (2001) and its possible role in storm deepening. Mon Wea Rev., adiabatic profile with little warming. The anvil and 140, 492–505. stratiform rain need to be maintained for many hours. Emanuel, K. A., 1986: An air–sea interaction theory for tropical To achieve this, deep convection, which supplies the cyclones. Part I: Steady-state maintenance. J. Atmos. Sci., 43, anvil with much of its condensate, must remain in 585–604. Enagonio, J., and M. T. Montgomery, 2001: Tropical cyclogenesis proximity to the stratiform rain. It is important that the via convectively forced vortex Rossby waves in a shallow subsidence under the anvil remains in an area with high water primitive equation model. J. Atmos. Sci., 58, 685–705. vorticity and survives long enough that a reservoir of Frank, W. M., and S. Chen, 1991: Simulations of vortex formation in convective weather systems. Preprints, 19th Conf. on Hur- warm ue can form. The reservoir and the associated convective burst may be critical in the transition of the ricanes and Tropical Meteorology, Miami, FL, Amer. Meteor. Soc., 241–244. MCS/TS to a mature TC with its wind maximum adja- Fritsch, F. N., and R. E. Carlson, 1980: Monotone piecewise cubic cent to the warm core. interpolation. SIAM J. Numer. Anal., 17, 238–246. Gamache, J. F., and R. A. Houze Jr., 1982: Mesoscale air motions Acknowledgments. The National Science Foundation associated with a tropical squall line. Mon. Wea. Rev., 110, Grants ATM0735867 and AGS-1042680 supported this 118–135. Gray, W. M., 1968: Global view of the origin of tropical distur- research. The data collection efforts of NOAA/AOML/ bances and storms. Mon. Wea. Rev., 96, 669–700. HRD, NOAA/Aircraft Operations Center, and NASA ——, 1979: Hurricanes: Their formation, structure and likely role were essential to our work. A special thanks to everyone in the tropical circulation. Meteorology over the Tropical who worked on the CAMEX-4 and COVES field pro- Oceans, D. B. Shaw, Ed., Royal Meteorological Society, gram. Without your work this study would not have been 155–218. possible. The deep soundings dropped by the NASA DC- Harr, P. A., M. S. Kalafsky, and R. L. Elsberry, 1996: Environ- mental conditions prior to formation of a midget tropical cy- 8 and ER-2 were especially helpful in mapping the upper clone during TCM-93. Mon. Wea. Rev., 124, 1693–1710. atmosphere. We would also like to thank Jonathan Vigh Hawkins, H. F., and S. M. Imbembo, 1976: The structure of a small, and an anonymous reviewer for their helpful comments. intense hurricane: Inez, 1966. Mon. Wea. Rev., 104, 418–442.

Unauthenticated | Downloaded 09/27/21 12:53 PM UTC 1190 MONTHLY WEATHER REVIEW VOLUME 140

Hendricks, E. A., M. T. Montgomery, and C. A. Davis, 2004: The ——, and J. Enagonio, 1998: Tropical cyclogenesis via convectively role of ‘‘vortical’’ hot towers in the formation of Tropical Cy- forced vortex Rossby waves in a three-dimensional quasi- clone Diana (1984). J. Atmos. Sci., 61, 1209–1232. geostrophic model. J. Atmos. Sci., 55, 3176–3207. Heymsfield, G. M., E. Ritchie, J. Simpson, J. Molinari, and L. Tian, ——, and J. L. Franklin, 1998: An assessment of the balance ap- 2006: Structure of highly sheared tropical storm Chantal dur- proximation in hurricanes. J. Atmos. Sci., 55, 2193–2200. ing CAMEX-4. J. Atmos. Sci., 63, 268–287. ——, M. E. Nicholls, T. A. Cram, and A. B. Saunders, 2006: A Hock, T. F., and J. L. Franklin, 1999: The NCAR GPS dropwind- vortical hot tower route to tropical cyclogenesis. J. Atmos. Sci., sonde. Bull. Amer. Meteor. Soc., 80, 407–420. 63, 355–386. Houze, R. A., Jr., 1977: Structure and dynamics of a tropical squall- Ooyama, K. V., 1964: A dynamical model for the study of tropical line system. Mon. Wea. Rev., 105, 1540–1567. cyclone development. Geofis. Int., 4, 187–198. Hubert, L. F., 1955: A case study of hurricane formation. J. Meteor., Reasor, P. D., M. T. Montgomery, and L. F. Bosart, 2005: Meso- 12, 486–492. scale observations of the genesis of Hurricane Dolly (1996). Jorgensen, D. P., 1984: Mesoscale and convective-scale charac- J. Atmos. Sci., 62, 3151–3171. teristics of mature hurricanes. Part I: General observations by Riehl, H., 1948: On the formation of typhoons. J. Meteor., 5, 247– research aircraft. J. Atmos. Sci., 41, 1268–1285. 264. ——, and P. T. Willis, 1982: A Z–R relationship for hurricanes. Ritchie, E. A., and G. J. Holland, 1997: Scale interactions during the J. Appl. Meteor., 21, 356–366. formation of Typhoon Irving. Mon. Wea. Rev., 125, 1377–1396. Kakar, R., M. Goodman, R. Hood, and A. Guillory, 2006: Over- ——, and R. L. Elsberry, 2001: Simulations of the transformation view of the convection and moisture experiment (CAMEX). stage of the extratropical transition of tropical cyclones. Mon. J. Atmos. Sci., 63, 5–18. Wea. Rev., 129, 1462–1480. Kleinschmidt, E., Jr., 1951: Grundlagen einer theorie des tropic- Rodgers, F. R., and J. M. Fritsch, 2001: Surface cyclogenesis from schen zyklonen (The underlying principles of a theory of tropi- convectively driven amplification of midlevel mesoscale con- cal cyclones). Arch. Meteor. Geophys. Bioklimatol., 4A, 53–72. vective vortices. Mon. Wea. Rev., 129, 605–637. Leary, C. A., 1980: Temperature and humidity profiles in meso- Rotunno, R., and K. A. Emanuel, 1987: An air–sea interaction scale unsaturated downdrafts. J. Atmos. Sci., 37, 1005–1012. theory for tropical cyclones. Part II: Evolutionary study using ——, and E. N. Rappaport, 1987: The life cycle and internal a non-hydrostatic axisymmetric numerical model. J. Atmos. structure of a mesoscale convective complex. Mon. Wea. Rev., Sci., 44, 542–561. 115, 1503–1527. Simpson, J., E. Ritchie, G. J. Holland, J. Halverson, and S. Stewart, Malkus, J. S., and H. Riehl, 1960: On the dynamics and energy 1997: Mesoscale interactions in tropical cyclone genesis. Mon. transformation in steady-state hurricanes. Tellus, 12, 1–20. Wea. Rev., 125, 2643–2661. Marks, F. D., Jr., 1985: Evolution of the structure of precipitation in Stossmeister, G. J., and G. M. Barnes, 1992: The development of Hurricane Allen. Mon. Wea. Rev., 113, 909–930. a second circulation center within Tropical Storm Isabel Martin, C., 2007: ASPEN user manual. NCAR, 61 pp. [Available (1985). Mon. Wea. Rev., 120, 685–697. online at http://www.eol.ucar.edu/isf/facilities/software/aspen/ Velasco, I., and J. M. Fritsch, 1987: Mesoscale convective com- Aspen%20Manual.pdf.] plexes in the Americas. J. Geophys. Res., 92, 9591–9613. McBride, J. L. and R. Zehr, 1981: Observational analysis of Vigh, J. L., 2010: Formation of the hurricane eye. Ph.D. disserta- tropical cyclone formation. Part II: Comparison of non- tion, Dept. of Atmospheric Science, Colorado State Univer- developing versus developing systems. J. Atmos. Sci., 38, sity, Fort Collins, CO, 538 pp. 1132–1151. Wang, Z., M. T. Montgomery, and T. J. Dunkerton, 2010a: Genesis Menard, R. D. and J. M. Fritsch, 1989: A mesoscale convective of pre–Hurricane Felix (2007). Part I: The role of the easterly complex-generated inertially stable warm core vortex. Mon. wave critical layer. J. Atmos. Sci., 67, 1711–1729. Wea. Rev., 117, 1237–1261. ——, ——, and ——, 2010b: Genesis of pre–Hurricane Felix Molinari, J., D. Vollaro, and K. L. Corbosiero, 2004: Tropical (2007). Part II: Warm core formation, precipitation evolution, cyclone formation in a sheared environment: A case study. and predictability. J. Atmos. Sci., 67, 1730–1744. J. Atmos. Sci., 61, 2493–2509. Willoughby, H. E., and M. B. Chelmow, 1982: Objective de- Montgomery, M. T., and B. F. Farrell, 1993: Tropical cyclone for- termination of hurricane tracks from aircraft observations. mation. J. Atmos. Sci., 50, 285–310. Mon. Wea. Rev., 110, 1298–1305. ——, and R. J. Kallenbach, 1997: A theory for vortex Rossby waves Yanai, M., 1961: A detailed analysis of typhoon formation. J. Me- and its application to spiral bands and intensity changes in teor. Soc. Japan, 39, 187–214. hurricanes. Quart. J. Roy. Meteor. Soc., 123, 435–465. Zipser, E. J., 1977: Mesoscale and convective-scale downdrafts as ——, and C. Lu, 1997: Free waves on barotropic vortices. Part I: distinct components of squall-line structure. Mon. Wea. Rev., Eigenmode structure. J. Atmos. Sci., 54, 1868–1885. 105, 1568–1589.

Unauthenticated | Downloaded 09/27/21 12:53 PM UTC