JUNE 2007 ATALLAH ET AL. 2185

Precipitation Distribution Associated with Landfalling Tropical Cyclones over the Eastern

EYAD ATALLAH

Department of Atmospheric and Oceanic Sciences, McGill University, Montreal, , Canada

LANCE F. BOSART AND ANANTHA R. AIYYER*

Department of Earth and Atmospheric Sciences, University at Albany, State University of New York, Albany, New York

(Manuscript received 28 November 2005, in final form 25 August 2006)

ABSTRACT

Tropical cyclones (TCs) making over the United States are examined by separating those asso- ciated with precipitation predominantly left of their tracks from those with the same to the right of their tracks. Composites of atmospheric variables for these two TC categories are performed and analyzed using potential vorticity (PV) and quasigeostrophic (QG) frameworks. Dynamical signatures are retrieved from these composites to help understand the evolution of precipitation in these storms. Results indicate that a left of track precipitation distribution (e.g., Floyd 1999) is characteristic of TCs undergoing extratropical transition (ET). In these cases, a positively tilted midlatitude trough approaches the TC from the northwest, shifting precipitation to the north-northwest of the TC. Potential vorticity redistribution through diabatic heating leads to enhanced ridging over and downstream of the TC, resulting in an increase in the cyclonic advection of vorticity by the thermal wind over the transitioning TC. A right of track precipitation distri- bution is characteristic of TCs interacting with a downstream ridge (e.g., David 1979). When the down- stream ridge amplifies in response to TC-induced diabatic heating ahead of a weak midlatitude trough, the PV gradient between the TC and the downstream ridge is accentuated, producing a region of enhanced positive PV advection (and cyclonic vorticity advection by the thermal wind) over the TC. The diabatic enhancement of the downstream ridge is instrumental in the redistribution of precipitation about the transitioning TCs in both cases and poses a significant forecast challenge.

1. Introduction 2002. Floyd (1999) proved to be a particularly difficult forecast challenge. While most of the hazardous a. Purpose weather warnings as the storm approached the coast Several recent storms have highlighted the impor- focused on the potential for wind damage, flooding due tance of understanding the effect of the interaction of to heavy precipitation proved to be a graver source of landfalling tropical systems with midlatitude features. destruction. More than 50 cm of precipitation fell over Notable recent events impacting the United States in- sections of , and amounts exceeding 20 clude Floyd and Irene in 1999, and Isidore and Lili in cm occurred over a wide area stretching from the Pied- mont region of North Carolina to southeastern New York. Atallah and Bosart (2003) provided a detailed case study of Floyd and indicated that the precipitation * Current affiliation: Department of Marine, Earth, and Atmo- spheric Sciences, North Carolina State University at Raleigh, Ra- distribution and intensity associated with the storm re- leigh, North Carolina. sulted from its interaction with a potent midlatitude trough and subsequent extratropical transition (ET). This study will attempt to elucidate a general paradigm Corresponding author address: Anantha Aiyyer, Dept. of Ma- rine, Earth, and Atmospheric Sciences, North Carolina State Uni- for understanding the dynamics responsible for modu- versity at Raleigh, Raleigh, NC 27695. lating the distribution, spatial extent, location, and in- E-mail: [email protected] tensity of precipitation associated with tropical cyclones

DOI: 10.1175/MWR3382.1

© 2007 American Meteorological Society

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC

MWR3382 2186 MONTHLY WEATHER REVIEW VOLUME 135 subsequent to landfall, regardless of whether or not the wind and precipitation distribution concentrated near (TC) undergoes a successful ET. the center of the tropical cyclone often evolves into a broad, asymmetric configuration. Klein et al. (2000) b. ET climatologies showed that the inner core of the tropical cyclone tends Past studies suggest that ET is a relatively common to lose its symmetric appearance as strong frontogen- phenomenon in many subtropical ocean basins (e.g., esis develops poleward of the tropical cyclone. Al- Foley and Hanstrum 1994; Klein et al. 2000; Hart and though the expanding cloud field associated with the Evans 2001; Sinclair 2002; Jones et al. 2003; Dare and TC can include large amounts of high clouds due to Davidson 2004). Foley and Hanstrum (1994) found that outflow into the midlatitude westerlies, regions of sig- a key precursor to ET includes a highly meridional nificant precipitation are typically embedded in the trough–ridge pair upstream of the TC and that ET oc- large cloud shield (Jones et al. 2003). curs most frequently when a large-amplitude cold front Several recent papers (e.g., Klein et al. 2002; Ritchie approaches within 1700 km of the tropical cyclone. and Elsberry 2003; McTaggart-Cowan et al. 2003; Sin- Harr and Elsberry (2000) and Harr et al. (2000) noted clair 2004) have attempted to quantify the effect of mid- that, during ET, significant precipitation to the north- latitude troughs on ET. Klein et al. (2002) showed that east of the TC favors diabatic ridging poleward of the reintensification is favored when the upper-level TC TC, increasing the amplitude of the trough–ridge cou- outflow enhances the equatorward jet-entrance region plet, resulting in an enhanced jet streak poleward of of a downstream jet streak. This occurs as the diabatic the TC. heating associated with the TC redistributes potential Hart and Evans (2001) found that, relative to the vorticity (PV) such that the anticyclonic circulation on number of TCs in each basin, ET is more frequent in the equatorward side of the jet streak is strengthened. the North Atlantic as compared with the North Pacific. Similarly, Ritchie and Elsberry (2003) found that the They found that 46% of TCs during 1950–96 underwent proper phasing of the tropical cyclone and the midlati- some form of ET, with the east coast of the United tude trough (i.e., the TC becomes collocated with a States and the Canadian Maritime Provinces most region of differential cyclonic vorticity advection in- likely to experience transitioning storms. Jones et al. creasing with height associated with the midlatitude (2003) showed that ET occurs in nearly every ocean trough) results in substantial enhancement of the up- basin that experiences tropical cyclone activity. An ob- per-level divergence, a feature observed in many case jective definition and quantification of ET has recently studies of ET. Sinclair (2004) showed that cyclonic vor- been proposed in Evans and Hart (2003) and Hart ticity advection (CVA) at jet level is the strongest in- (2003) that uses changes in the low-level thickness field dicator of significant redevelopment during ET. and midtropospheric thermal wind to signal ET onset Dare and Davidson (2004) found that the maximum and completion. Arnott et al. (2004) used an objective intensity of TCs near Australia is generally followed by cluster analysis methodology to quantify the ET pro- recurvature, which suggests the importance of trough cess and showed that distinct frontal and thermal struc- interactions. The importance of the tropical cyclone on tures can be identified as TCs undergo ET. These struc- the eventual strength of ET was addressed by Agustí- tural changes are evident in the Hart (2003) cyclone Panareda et al. (2004) who found that the impact of the phase space analysis and are consistent with the empiri- TC was maximized when the positive PV anomaly was cal formulation of ET as provided by Evans and Hart situated equatorward of the jet in a region of anticy- (2003). clonic equivalent barotropic shear. Hart et al. (2006) examined the factors that deter- mine whether a TC undergoing ET will reintensify as d. ET case studies an . Their results show that the presence of a negatively tilted baroclinic trough up- Palmén (1958) and Anthes (1990) examined the stream of a TC undergoing ET is an important indicator large-scale dynamics associated with the ET of Hurri- of post-ET reintensification, whereas the presence of a cane Hazel (1954) along the east coast of the United positively tilted upstream baroclinic trough precludes States, and concluded that the transport of tropical air significant extratropical reintensification once ET has masses into the midlatitudes resulted in a large increase occurred. in the available potential energy (APE) of the atmo- sphere. Bosart and Carr (1978), DiMego and Bosart c. Storm structures and intensity (1982a,b), and Bosart and Dean (1991) discussed the Palmén (1958), Muramatsu (1985), and Foley and precipitation distribution and dynamics associated with Hanstrum (1994) have found that the nearly symmetric the transition of (1972). DiMego and

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC JUNE 2007 ATALLAH ET AL. 2187

Bosart (1982a,b) found that Agnes reintensified and diagnosis, via its invertibility property, of the role underwent ET as large-scale ascent ahead of the ap- played by the TC and key features of the environment proaching trough overspread the periphery of the rem- (e.g., Agustí-Panareda et al. 2005). nant low-level circulation. However, a transitioned cyclone does not have to be f. Goals intense in order to have significant impact, as flooding can accompany relatively weak remnant cyclones. Hur- The main focus of this paper is to understand the ricane Camille (1969) produced catastrophic flooding in synoptic-scale dynamics modulating the precipitation the Virginia Piedmont as a relatively weak transitioned distribution associated with landfalling TCs. While pre- cyclone (Chien and Smith 1977). Bosart and Lackmann vious studies have generally concentrated on ET and (1995) documented the case of David (1979), which the associated dynamics, this paper will examine all reintensified as a tropical system over land prior to un- landfalling systems exclusively from the point of view of dergoing ET while it interacted with a relatively weak the precipitation distribution, regardless of whether or midlatitude trough. Sinclair (1993) undertook a study not the system successfully undergoes ET. As such, the of the ET of Tropical Cyclone Patsy (1986) and found retrieval of characteristic signatures associated with ET that the greatest ascent in the redevelopment of Patsy is illustrative, but essentially coincidental to the study. also occurred in a region of an upward increase in CVA No assumptions are made or imposed regarding the in the equatorward entrance region of a subtropical jet. initial strength of the TC at landfall or the eventual Evans and Prater-Mayes (2004) also showed that the strength of any transitioned cyclone. Therefore, this ET and reintensification of Irene (1999) was aided by study hopes to provide a more generalized framework an upper-level trough and jet streak. Colle (2003) illus- than previous studies for understanding the dynamics trated the importance of latent heating on frontogen- modulating the precipitation distribution as opposed to esis in the mid- and lower troposphere, and quantified the dynamics modulating the strength or intensity of the impact of the latent heating on the strength and ET. A PV and quasigeostrophic (QG) framework is precipitation distribution associated with Floyd (1999). adopted to help illustrate the essential dynamics. The Abraham et al. (2004) documented observations from paper is organized as follows. Section 2 contains the the ET of (2000) and showed that data and methodology, section 3 presents the results of the entrainment of relatively cool and dry air into the the diagnostics, and section 4 summarizes and discusses western quadrant of the storm effectively diminished the significance of these results. the precipitation in the southwestern quadrant of the cyclone. 2. Data and methodology e. PV thinking a. Gridded fields Studies of tropical systems have also been facilitated All gridded fields for use in the composites are taken by the use of PV (Rossby 1940; Ertel 1942; Hoskins et from the National Centers for Environmental Predic- al. 1985). The “PV thinking” perspective (Morgan and tion–National Center for Atmospheric Research Nielsen-Gammon 1998) has been utilized in numerous (NCEP–NCAR) reanalysis at 0000 and 1200 UTC (Kal- studies of hurricane–trough interactions (e.g., Molinari nay et al. 1996; Kistler et al. 2001). The NCEP–NCAR et al. 1995, 1997; Bosart and Lackmann 1995; Bosart et reanalysis provides a relatively uniform dataset over a al. 2000) and transitioning TCs (e.g., Browning et al. long period, and is useful for the production of com- 1998; Henderson et al. 1998; Thorncroft and Jones posites. The disadvantage of the reanalysis is the rela- 2000; McTaggart-Cowan et al. 2001). tively coarse grid resolution of 2.5° 2.5°. However, Potential vorticity is particularly useful in the study this drawback is mitigated because the focus of this of TC trough interactions. The TC corresponds to a study is synoptic scale in nature. It is understood that positive PV anomaly concentrated in the lower and the representations of the core of any tropical systems midtroposphere at small radii and a broad and shallow in the reanalysis are inadequate. Moreover, no attempt negative PV anomaly at larger radii in the upper tro- is made to diagnose the change of structure in the core posphere. On the other hand, extratropical cyclones are of the systems in this study. Rather, general signatures often associated with maxima of PV in the mid- and and changes in the structure of the mass field will be upper levels of the troposphere at large radii. The util- used to diagnose the sense of the QG forcing for ascent ity of PV lies not only in the visualization of the evo- as a first-order dynamical explanation of the precipita- lution of the interaction between the two, but also in tion distribution around the storms in question.

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC 2188 MONTHLY WEATHER REVIEW VOLUME 135 b. Storms and compositing TABLE 1. A list of TCs included in the study. The dates and times listed represent the initial time that a storm was included in All the storms used in the composites are listed in each of the composites. The numbers in parentheses show the Table 1. (Full details on the life cycle and track of each (kt) of the cyclone at the time listed as storm and the best track data can be found online at taken from the best-track data. http://www.nhc.noaa.gov/pastall.shtml.) The first time Storm LOC ROC period used for each storm in each composite is listed. This time (t 0) is defined as the first time that the Hazel (1954) 1200 UTC 15 Oct (110) storm exhibits a left of track (LOC) or right of track Audrey (1957) 1200 UTC 28 Jun 0000 UTC 27 Jun (ROC) precipitation distribution. The storm is then (45) (90) composited relative to a reference storm track. The Ethel (1960) 0000 UTC 16 Sep evolution in each of the composites is taken as the 24-h (45) period following the time listed in Table 1. To be in- Carla (1961) 1200 UTC 13 Sep 1200 UTC 12 Sep (30) (100) cluded in any of the composites, each storm had to meet Candy (1968) 0000 UTC 24 Jun the following criteria: (60) • Landfall was along either the Gulf or Atlantic coasts Camille (1969) 1200 UTC 18 Aug (65) of the United States. Doria (1971) 1200 UTC 27 Aug • The storm had to display a poleward component in its (50) track. Agnes (1972) 1200 UTC 21 Jun • The storm was far enough inland to permit rain mea- (30) surements in all quadrants of the storm. Eloise (1975) 1200 UTC 23 Sep (110) The storms are composited relative to a specified lati- David (1979) 0000 UTC 5 Sep tude–longitude point based on a reference storm track. (65) All the grids in the LOC composite at time t 0 are Frederic (1979) 0000 UTC 14 Sep (40) moved so that the center of each storm occupies the Alicia (1983) 1200 UTC 19 Aug same latitude–longitude coordinate. This process is re- (25) peated for each of the following time steps in each com- Bob (1985) 0000 UTC 25 Jul posite. The apparent motion of the storm in the com- (65) posites is therefore an artifact of the reference storm Danny (1985) 1200 UTC 18 Aug 1200 UTC15 Aug (25) (80) track. Elena (1985) 1200 UTC 3 Sep Applying the above criteria resulted in 32 storms be- (25) ing chosen from the period of 1950–98. Figure 1 dis- Gloria (1985) 1200 UTC 27 Sep plays the best-track data of a few representative storms (85) and the associated precipitation. Note that if storm po- Bonnie (1986) 1200 UTC 26 Jun (65) sitions were needed after the best-track data was no Chris (1988) 1200 UTC 29 Aug 0000 UTC 28 Aug longer available (i.e., the storm underwent ET), the (20) (30) position of the associated 850-hPa vorticity maximum Bob (1991) 1200 UTC 19 Aug in the reanalysis was used for the purposes of compos- (95) iting. All storms were composited with reference to a Andrew (1992) 0000 UTC 28 Aug 1200 UTC 26 Aug (20) (80) storm track with the same precipitation characteristics Beryl (1994) 1200 UTC 15 Aug and evolution. The dates along the tracks represent the (35) 1200 UTC time and the dots are in 6-h increments. Erin (1995) 1200 UTC 4 Aug Figures 1a,b show the tracks of Audrey (1957) and (20) Carla (1961), respectively, and are cases where the Bertha (1996) 1200 UTC 13 Jul (60) storms can be binned at different times in both the left Danny (1997) 1200 UTC 24 Jul of track and right of track composites. Both storms (30) make landfall with the bulk of the precipitation falling to the right of the storm. After recurvature, precipita- tion shifts to the left of the storm track. For example, in Agnes (1972) and David (1979) are shown as examples Fig. 1b, Carla (1961) would be binned as an ROC storm of storms with a consistent LOC and ROC precipitation from 1200 UTC 11 September to 0000 UTC 13 Septem- distribution, respectively. In general, each storm is used ber 1961and as an LOC storm from 1200 UTC 13 Sep- only when the precipitation distribution about the tember to 1200 UTC 14 September 1961. In Figs. 1c,d, storm is unambiguous. The 32 cases represent about 1/3

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC JUNE 2007 ATALLAH ET AL. 2189

FIG. 1. Storm track (black lines) and precipitation (shaded starting at 10 mm every 40 mm) for Hurricanes (a) Audrey (1957), (b) Carla (1961), (c) Agnes (1962), and (d) David (1979). The dots show the storm positions and the numbers mark the 1200 UTC position for each date. of all storms making landfall in the eastern United It should be noted that he UPD underestimates the States and approximately 1/2 of all storms that exhibit rainfall maximum by about 50% on approximately 13% some northward component in their tracks over land. of the storm days (not shown). However, the spatial or The precipitation distribution was taken from the qualitative distribution of the precipitation for most NCEP Unified Precipitation Dataset (UPD), which storms is well replicated. Figures 2a,b show subjective provides gridded daily precipitation amounts (1200– hand analysis of the storm totals for Bertha (1996; Fig. 1200 UTC) on a 0.25° 0.25° latitude–longitude grid 2c) and Fran (1996; Fig. 2d) as taken from the National covering the region from 20°–60°Nto140°–60°W. The Weather Service forecast offices in Sterling, Virginia, data were analyzed using a Cressman (1959) objective and Raleigh, North Carolina, respectively. It is clear analysis scheme. (Computational details can be found when comparing Figs. 2a,b with Figs. 2c,d that the UPD online at http://www.cpc.ncep.noaa.gov/research_ qualitatively represents the distribution of the precipi- papers/ncep_cpc_atlas/7/toc.html.) tation well, albeit the maxima are somewhat reduced.

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC 2190 MONTHLY WEATHER REVIEW VOLUME 135

FIG. 2. (a) Subjective analysis of rainfall associated with (1996) from the Climate Prediction and National Hurricane Centers; (b) subjective analysis of rainfall associated with (1996) from the Forecast Office in Sterling, VA; storm total precipitation (mm) from the UPD for (c) Bertha (1996) and (d) Fran (1996).

3. Results the storm track was still on the west side of the Appa- lachian Mountain chain. This suggests that while orog- The tracks of the storms used to produce the LOC raphy and cold-air damming may be significant in many and ROC precipitation composites are shown in Fig. 3. cases, it is not a necessary condition to achieve the LOC The precipitation associated with some of the storms precipitation distribution. The LOC composites are switches sides during their evolution, and consequently created in storm relative coordinates with respect to the some tracks appear in both Figs. 3a,b. The reader is track of (1999) as shown in Fig. 4. referred back to Table 1 for the starting date for each Clearly, not all storms in the composite moved at the storm for the respective composites. Of the 14 storms same speed. Consequently, the 12-h position of each used in the LOC composite, six made landfall along the storm was relocated to the 12-h position of Floyd. It east coast of the United States while eight of the storms should be noted that Floyd was chosen as a reference made landfall along the Gulf Coast. A total of six of the track since it is roughly representative of the average 14 storms had a LOC precipitation distribution while LOC track vector.

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC JUNE 2007 ATALLAH ET AL. 2191

FIG. 3. The tracks of the tropical cyclones (based on the National Hurricane Center best-track data) for (a) storms with precipitation to the left of track and (b) storms with precipitation to the right of track.

The tracks of the storms used to produce the ROC a. Synoptic overview composites are shown in Fig. 3b. The storms in the The 1000–500-hPa thickness, the 1000-hPa heights, ROC bin were composited with respect to Audrey at and 850–200-hPa wind shear for both composites are the time of landfall (1957; Fig. 1a) as it is roughly rep- shown in Fig. 5. At time t 0 (Fig. 5a), the LOC com- resentative of the average ROC track vector. As in the posite TC is still offshore with a minimum central LOC composites, the first time in the composite is the height value of about 45 m at 1000 hPa. The cyclone first time the storm exhibits a clear ROC precipitation circulation center is located near the thickness ridge distribution over land. Of the 16 storms used in the (576 dam), indicative of its warm-core structure. A ROC composite, 11 made landfall along the Gulf Coast, while only five storms made landfall along the East large synoptic-scale thermal trough is located in the Coast. Interestingly, all of the storms that made landfall Midwest, with the trough axis lying approximately along the East Coast made landfall near the – along the Mississippi River. Warm-air advection is al- border, and then proceeded to track ready evident along the Virginia and North Carolina along or just east of the spine of the Appalachians. Piedmont, with cold-air advection becoming prevalent While the location of landfall may be incidental, along and to the west of the Appalachians. A weak storms making landfall farther north are inherently trough axis in the 1000-hPa height field lies north of the more likely be associated with cold-air damming and/or TC, presumably a surface reflection of the midlatitude a significant midlatitude trough, and consequently ex- thermal trough. hibit a LOC precipitation distribution (as will be dis- Twelve hours later (t 12), the cyclone has moved cussed later). The track along the spine of the Appala- onshore and the cyclonic curvature of the thermal chians, on the other hand, does seem intuitive for the trough has increased concomitantly with an increase in ROC composite, as the wind field around the tropical the anticyclonic curvature in the thickness field imme- cyclone center would produce upslope (downslope) diately along the East Coast (Fig. 5b). This is indicative conditions in the eastern (western) quadrant of the of an increase in the vorticity gradient across the storm storm. The western extent and strong curvature of the and represents an increase in the differential cyclonic tracks over Gulf Coast indicates that these systems of- vorticity advection with height and consequently the ten encounter a strong east–west ridge axis with west- quasigeostrophic forcing for ascent. The S-shaped con- erlies off to the west and north. Also, 13 of the 16 figuration now evident in the thickness field, is a pat- storms exhibited an ROC precipitation distribution tern indicative of midlatitude cyclogenesis. The implied while still south of 38°N and at least eight of the storms maximum in warm-air advection is now located near in the composite later evolved into LOC storms. populated areas of the Northeast, stretching from

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC 2192 MONTHLY WEATHER REVIEW VOLUME 135

FIG. 4. Schematic of the track and storm total precipitation (in.; 1 in. 2.54 cm) associated with Hurricane Floyd as reproduced from the Weekly Weather and Crop Bulletin (Vol. 86, No. 38). southeastern Pennsylvania to southern Massachusetts. hPa resides over southeast Texas, west of a relatively The decrease in minimum 1000-hPa height of 30 m may strong high pressure system with a maximum central be indicative of storm strengthening and/or an artifact height of about 180 m located northeast of of the expanding scale of the transitioning storm. (Fig. 5d). The relative weakness of the ROC cyclone is Twenty-four hours after first exhibiting a LOC pre- more an artifact of the relatively small scale of the com- cipitation distribution (t 24), the cyclone center is posite cyclones than an actual indication of intensity as located over Long Island, New York (Fig. 5c). A local indicated in Table 1. Although strong geostrophic maximum in the thickness field has propagated pole- southerlies are evident in the eastern quadrant of the ward and is situated almost due east of Virginia, ap- composite TC, the TC remains isolated from any sig- proximately 300–400 km southeast of the cyclone cen- nificant baroclinic zone, precluding any notable tem- ter. Meanwhile, the cyclone continues to move toward perature advections. The only appreciable region of colder thickness values (576 dam). The axis of the ther- warm-air advection is located near the main band of mal trough has become negative as indicated by the westerlies to the northeast of the cyclone, on the north- solid line (an indication of the increasing magnitude of west periphery of the Bermuda high. differential cyclonic vorticity advection) with the The cyclone weakens (true for almost every member trough axis stretching from southeastern Michigan to of the composite) as it moves north toward an east– coastal Georgia. The minimum central height of the west-oriented baroclinic zone with a central minimum storm continues to decrease to 15 m at 1000 hPa. The height now of 90 m at 1000 hPa (t 12; Fig. 5e). The S-shaped configuration of the thickness field has be- strong high pressure system east of the cyclone center come more pronounced as the cyclone entrains colder remains the dominant feature on the map. As the ridge air into the southwestern quadrant of the storm. The axis retreats poleward of the cyclone, the geostrophic implied warm-air advection increases in magnitude as winds turn southerly on the west side of the anticy- the gradients in both the thickness field and height field clone, and the warm-air advection becomes more pro- intensify. It is now essentially impossible to distinguish nounced to the north-northeast of the TC center. As the structure of this system from that of an extratropical the TC continues to track to the northeast (t 24), it cyclone, suggesting that ET is occurring or has taken comes into closer contact with the baroclinic zone on place. the east side of a weak trough in the 1000–500-hPa At t 0 in the ROC composite, a relatively weak thickness field that has moved out of the Rockies and cyclone with a minimum central height of 75 m at 1000 into the western Plains (Fig. 5f). The orientation and

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC JUNE 2007 ATALLAH ET AL. 2193

FIG. 5. The 1000–500-hPa thickness (dashed black lines, contoured every 60 m), 1000-hPa geopotential height (solid black lines, contoured every 30 m), and the 850–200-hPa wind shear (shaded, contoured every 5 m s1, starting at 20 m s1) for (a) LOC at t 0, (b) LOC at t 12, (c) LOC at t 24, (d) ROC at t 0, (e) ROC at t 12, and (f) ROC at t 24. Thick solid lines denote thermal trough axes. The thick arrows in (a) and (d) depict the composite storm tracks.

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC 2194 MONTHLY WEATHER REVIEW VOLUME 135 strength of the thickness field indicate the storm is be- PVU). The upper-level PV anomaly takes on a negative ginning to experience increased west-southwesterly tilt as it wraps cyclonically around the surface cyclone. shear. The negative tilt indicates an increasingly potent trough owing to an increase in the differential positive PV ad- b. A PV perspective vection. Also at this time, a northeastward extension of the low-level PV anomaly becomes evident in associa- 1) HORIZONTAL MAPS tion with a developing warm front (Fig. 6c). A PV perspective shows that the composite TC ex- In contrast, the ROC composite at time t 0 shows hibits a maximum of approximately 0.75 potential vor- only a weak midlatitude trough located over Montana, ticity units (1 PVU 106 Kkg1 m2 s1) in the 850– approximately 2000 km to the north-northwest of the 700-hPa layer (lower troposphere; Fig. 6a). The wind composite TC (0.6-PVU anomaly) over eastern Texas field in this layer shows that the circulation center is not (Fig. 6d). A low-level southerly jet is located east of the quite closed, as evidenced by a trough in the wind field TC (not shown), while winds in the upper troposphere extending north of the low center (not shown). The are from the south-southwest across the TC center in storm is already experiencing some wind shear as indi- response to the weak thermal trough to the west (see cated by the 20 m s1 southwesterly wind barb in the Fig. 5d). The lower-tropospheric PV anomaly dissipates 300–200-hPa layer (upper troposphere) over the cy- slightly as the composite storm moves to the northeast clone center. Meanwhile, a strong upper-level trough ahead of the midlatitude trough approaching the west- with a maximum of 5 PVU approaches the storm from ern Great Lakes (Fig. 6e). The low-level jet provides the northwest. This trough exhibits a positive tilt and is moisture into and ahead of the low-level PV anomaly located approximately 1500 km to the northwest of the lying across the upper Midwest. This implied moisture cyclone center. A band of positive PV advection is oc- transport can lead to enhanced precipitation ahead of curring on the east side of the trough, stretching from the midlatitude trough to the north, similar to the sche- northern Ohio to the Labrador coast of Canada, sug- matic illustration shown in Bosart and Carr (1978, their gestive of synoptic-scale forcing for ascent in this re- Fig. 3). As the storm moves toward the main band of gion. westerlies, the upper-level flow strengthens across the Twelve hours later (t 12), as the region of positive center of the TC from 5 to 10 m s1 as the flow veers to potential vorticity advection overspreads the composite the southwest. TC, the area covered by the 0.75 PVU contour in the By t 24 h, the low-level PV anomaly is strengthen- 850–700-hPa layer increases considerably, while the dis- ing slightly as a region of weak positive PV advection tance between the cyclone and upper-level trough de- spreads equatorward (Fig. 6f). Brisk low-level south- creases (Fig. 6b). The positive PV advection now over- easterlies of 10 m s1 (not shown) are now feeding into spreading the cyclone indicates that there is synoptic- a confluent zone beneath an equatorward jet-entrance scale forcing for ascent occurring over the cyclone region as indicated by the upper-level wind field. The resulting in an expansion in the scale of the low-level strength of the upper-level westerlies continues to in- PV anomaly. Shear has increased slightly as indicated crease across the TC as the storm moves poleward to- by the upper-tropospheric 25 m s1 southwesterly ward the main baroclinic zone. winds over the cyclone center. There is a change in the 2) CROSS SECTIONS structure of the upper-level PV anomaly, which has ex- perienced considerable erosion over southern Ontario In Fig. 7a, the TC is represented by the relatively and eastern New York, presumably in association with isolated PV maximum at 700 hPa (0.7 PVU) at time the diabatic redistribution of PV by latent heat release t 0 in the LOC composite. A column of exceptionally (precipitation). Note that values of PV in this region low-PV air lies to the right of the TC in the cross section have decreased, in spite of the positive PV advection in and the TC itself is located underneath a steeply sloped this region evident in Fig. 6a. The front edge of this PV dynamic tropopause (DT; PV wall as discussed by anomaly now exhibits more anticyclonic curvature, re- Hoskins and Berrisford 1988) as measured by the 1.5- sulting in a trough with a more neutral tilt. In response PVU contour. At this time, the storm is still warm core to this rearrangement of the upper-level PV, the winds as indicated by the downward-bulging isentropes ahead of the trough turn more southerly, allowing the stretching from 200 hPa down to near the surface. The storm to track closer to the coast. midlatitude trough is located about 1500 km to the By t 24, the upper-level trough and the cyclone northwest of the storm as defined by the combination center are superposed (Fig. 6c). The lower-level PV of the lowest depression of the DT and the associated anomaly continues to grow and strengthen (0.9 upward bulge (cold-core structure) in the isentropes

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC JUNE 2007 ATALLAH ET AL. 2195

FIG. 6. The 850–700-hPa PV (shaded in cool colors every 0.15 PVU starting at 0.45 PVU where 1 PVU 10 6 Kkg 1 m2 s 1) and 300–200-hPa PV (shaded in warm colors every 1 PVU starting at 2 PVU) and wind (barbs, knots convention) for (a) LOC at t 0, (b) LOC att 12, (c) LOC at t 24, (d) ROC at t 0, (e) ROC at t 12, and (f) ROC at t 24. Note that the thick black lines represent the position of the cross sections in Fig. 8. Thick dashed contours indicate areas of positive PV advection in the 300–200-hPa layer starting at 1 105 PVU s1, and contoured every 2 105 PVU s1. The double arrow lines indicate the approximate trough axis in the 300–200-hPa layer as defined by the wind field.

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC

Fig 6 live 4/C 2196 MONTHLY WEATHER REVIEW VOLUME 135

FIG. 7. Cross sections of PVU (shaded, starting at 0.4 PVU as given by the color bar), isentropes (thin solid lines, contoured every 3 K), and relative humidity (dashed lines, contoured every 10%) for (a) LOC at t 0, (b) LOC at t 12, (c) LOC at t 24, (d) ROC at t 0, (e) ROC at t 12, and (f) ROC at t 24. Note that the thick black line denotes the 1.5-PVU surface, taken to be the DT. Thick solid lines indicate the position of the warm core associated with the TC.

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC JUNE 2007 ATALLAH ET AL. 2197 from the surface to about 300 hPa. The juxtaposition of structure in the lower troposphere. However, it should the warm-core TC and the cold-core midlatitude trough be noted that ET often leads to an increase in storm results in a region of enhanced baroclinicity throughout scale resulting in better representation of the storms in the depth of the troposphere. Cool-dry air is situated to the LOC composite. The lack of a warm-core structure the northwest of the cyclone as indicated by relative in the lower troposphere in conjunction with the ab- humidity values at or below 20%. The ingestion of this sence of any cold-core troughs in proximity of the TC, air into the storm will help to diminish precipitation in precludes the presence of a any significant baroclinic the southeastern quadrant of the storm, and is consis- zone near the surface. However, a weak midlatitude PV tent with the findings of Abraham et al. (2004). anomaly in the upper troposphere is located to the east Over the next 12 h, two key developments take place. of the tropical system. The difference between the First, the lower-tropospheric PV anomaly strengthens warm core of the tropical system and the cold-core and becomes displaced to the left of the warm core, structure in the midlatitude PV anomaly results in an while the midlatitude PV anomaly is displaced to the enhanced baroclinic zone in the upper troposphere, im- right of the cold core (Fig. 7b). This results in the partial plying enhanced values of wind shear in the mid- and superposition of the two features and an increase in the upper troposphere. Above and just to the east of the QG forcing for ascent as the PV becomes embedded in TC lies a region of relatively low PV air, resulting in a the potential temperature () gradient. At the same less pronounced version of the PV wall shown in the time, the slope of the DT over the lower-tropospheric LOC composite. High values of relative humidity are PV anomaly increases. This suggests the presence of confined to near and just east of the positive low-level strong upper-tropospheric frontogenesis. By the end of PV anomaly. the 24-h period, a nearly coherent and tropospheric Twelve hours later (t 12), a weak midlatitude PV deep PV tube is apparent (Fig. 7c). The presence of a anomaly appears to interact weakly with the TC as in- local maximum of PV within the envelope of the PV dicated by the connection of the 0.6-PVU surface with tube in the lower troposphere suggests that at least the stratospheric PV reservoir (Fig. 7e). The thermal some tropical characteristics remain, even though the structure of the TC appears unchanged, while the cold system exhibits a strong up-shear tilt. The second key anomaly at 300 hPa just to the west has dissipated. A feature is the strengthening of the temperature gradient more potent midlatitude PV anomaly (cold trough) is as indicated by a 500-km decrease in distance between entering the edge of the cross section along the western the remnant warm and cold cores. The steeply sloped U.S. coastline as indicated by the penetration of the DT isentropes in this region represent areas of strong po- to the 250-hPa level. By the end of the 24-h period, the tential isentropic ascent (descent) given an easterly midtropospheric PV column has taken on a bit of a (westerly) component in the wind field and is consistent downshear tilt as the composite storm has moved north with a redistribution of precipitation to the northwest into a relatively zonally oriented baroclinic zone (Fig. quadrant of the storm. These results are in agreement 7f). In spite of the previous interaction, the tropical PV with the findings of Atallah and Bosart (2003, their Fig. anomaly remains isolated from the stratospheric PV 7) and Colle (2003), although it should again be noted reservoir and ET does not take place. that Floyd is not included in this composite. Further- 3) DYNAMIC TROPOPAUSE ANALYSIS more, in their ensemble based investigations of the ET of Hurricane Floyd (1999), Roebber et al. (2004) A DT analysis of LOC TCs shows that initially (t showed that despite the underprediction of precipita- 0), the core of the TC is largely embedded beneath a tion in their relatively coarse-resolution simulations, ridge (west of the ridge axis) near the eastern periphery most of the ensemble members were able to identify of the upstream trough–baroclinic zone and the associ- the potential for the heaviest precipitation within the ated jet streak (Fig. 8a). Over time, the pattern evolves northwest quadrant of the storm. into a more amplified trough–ridge couplet as the In the ROC composite, the TC at t 0 is represented 335-K isentrope moves southward over the Midwest, by a fairly upright column of PV that has a maximum in and the 345-K isentrope moves northward along the the midtroposphere (Fig. 7d). The thermal structure of Atlantic Coast (Fig. 8b). Positive advection of PV is the cyclone is weakly cold core in the lower tropo- evident ahead of the midlatitude trough near and over sphere, with a strong warm-core structure in the mid- the southern Appalachians, with westerly winds blow- and upper troposphere. The lack of a warm-core struc- ing across the isentropes near the base of the trough. ture in the lower troposphere may be an artifact of the Diabatically enhanced ridging is taking place over a Reanalysis resolution as many case studies of storms in large portion of the northeastern United States as in- the composites (not shown) fail to exhibit a warm-core dicated by the shaded regions depicting the difference

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC 2198 MONTHLY WEATHER REVIEW VOLUME 135

FIG. 8. Potential temperature (dashed black lines, contoured every 10 K) and wind (barbs, knots convention) on the DT along with the 850-hPa relative vorticity (contoured every 2 105 s1, starting at 2 105 s1), and diabatic residual ridging (shaded in K with intervals given by the color bars) for (a) LOC at t 0, (b) LOC at t 12, (c) LOC at t 24, (d) ROC at t 0, (e) ROC at t 12, and (f) ROC at t 24.

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC JUNE 2007 ATALLAH ET AL. 2199 between local and advective changes in [i.e., (/t) the geostrophic wind, and is the vertical motion with for each 12-h period. This calculation serves as a respect to pressure. While this equation does neglect [١ · v rough estimate of the strength of the diabatic effects deformation terms, the quasi-circular vorticity struc- under the assumption that nonconservation of on the tures of the composites TCs suggests that these terms DT is primarily attributable to diabatic processes. At are not important here. Furthermore, an investigation this time, the TC is approaching the gradient on the of the Q vector formulation of the QG omega equation DT. By t 24, the core of the TC has finally become (not shown) yields essentially identical results. This embedded in the gradient (Fig. 8c). Furthermore, the does not imply that deformation and frontogenesis do magnitude of the gradient has increased as the 335-K not play a key role in individual cases. Indeed, Colle isentrope has moved to the east, while the 355-K isen- (2003) showed that these terms are key players in the trope remains situated near the coast. This differential precipitation distribution of Hurricane Floyd (1999). propagation of the isentropes acts to reorient the tilt of However, in the present context the above formulation the trough from positive to negative. of the omega equation represents a good first-order The ROC composite at time t 0 differs in that there approximation of the synoptic-scale vertical motion as is a zonally oriented gradient on the DT with a weak is illustrated in Fig. 9. trough in the western United States (Fig. 8d). The TC is In the LOC composite at time t 0, a strong dipole located just northwest of the center of the outflow an- of cyclonic and anticyclonic vorticity advection by the ticyclone. The main equatorward jet-entrance region thermal wind is already evident oriented from north- lies poleward of the system in Nebraska and South east to southwest across the TC center. The region of Dakota. As the cyclone moves northeastward, a ridge ascent is mostly collocated with the region of CVA by develops to the east of the cyclone with the 355-K sur- the thermal wind, and stretches from the eastern quad- face moving northeastward from extreme northeastern rants to the northwestern quadrant of the storm (Fig. Texas to southern Illinois–Indiana (Fig. 8e). This dia- 9a). The QG forcing for ascent is strongest along the batically enhanced ridging results in a more pro- eastern seaboard to the north of the TC as evidenced by nounced trough structure and a tighter gradient up- the minimum in of 3 103 hPa s1 in this region. stream of the TC, resulting in increased wind speeds Note that while the air is rising in the eastern quadrant over Missouri, placing the storm in proximity of a jet- of the TC, it has a recent history of subsidence as indi- entrance region. The associated secondary circulation cated by the maximum of over the Tennessee River in this region enhances ascent (see Fig. 9) northeast of Valley. Furthermore, this air is relatively dry to begin the storm, which, in combination with low-level mois- with (recall Fig. 8a), resulting in a minimum of precipi- ture transport on the eastern side of the TC (Fig. 7e), tation in the eastern quadrant of the TC. This is analo- helps explain the ROC precipitation distribution. In gous to the “dry slot” seen in extratropical cyclones Fig. 8f (t 24), the TC has moved out of the ridge and (e.g., Carr and Millard 1985). So while the air is being into the edge of southwesterly flow associated with the lifted in the eastern quadrant of the storm, parcel satu- gradient over Missouri. In the ROC composite, the ration is only realized in the ascent region north and ridge propagates away from the TC circulation in northwest of the TC. marked contrast to the LOC composite. This precludes The QG forcing for ascent expands to the west over a shortening of the wavelength of the trough–ridge cou- the mid-Atlantic and Northeast by t 12 (Fig. 9b). This plet resulting in relatively weak QG forcing for ascent. is consistent with an expansion of the precipitation shield to the west. It should be noted that for cases of c. A quasigeostrophic perspective ET occurring along the eastern seaboard, the land– In this section, we adopt a QG perspective to help ocean interface can serve to enhance any synoptic-scale diagnose the synoptic-scale-forcing mechanisms that baroclinic zone in the region, further enhancing/focusing modulate the precipitation distribution associated with precipitation in the Piedmont of the mid-Atlantic. Also the composite TCs. This task is accomplished through at this time, the region of descent (anticyclonic vorticity an investigation of the advection of geostrophic relative advection by the thermal wind) has expanded to the vorticity (if we ignore the small contribution of the Co- east in the southern quadrant of the TC. By the end of riolis force) by the thermal wind (Sutcliffe 1947; Sutc- the 24-h period, the CVA–anticyclonic vorticity advec- liffe and Forsdyke 1950; Trenberth 1978) as given by tion (AVA) couplet has become completely aligned 2 2 2 2 the equation [ p (f o / )( / p)] (fo / )2[( vg / along the motion vector of the cyclone, indicating a ١ ١ • p) p g], where p is the horizontal gradient opera- completion of ET (Fig. 9c). tor, fo is the Coriolis parameter at a reference latitude, In comparison, the ROC composite initially has com- T[ ln( )/ p] is the static stability parameter, g is paratively little indication of QG forcing for TC ascent

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC 2200 MONTHLY WEATHER REVIEW VOLUME 135

FIG. 9. The 700–400-hPa advection of relative geostrophic vorticity by the thermal wind (thin solid lines denote cyclonic vorticity advection while thin dashed denote anticyclonic vorticity advection, contoured starting at 2 1010 m1 s2), 700–400-hPa layer average omega (shaded dark to light for ascent and light to dark for descent in bs1), and 850-hPa geostrophic relative vorticity (thick solid lines, contoured starting at 2 105 s1) for (a) LOC at t 0, (b) LOC at t 12, (c) LOC at t 24, (d) ROC at t 0, (e) ROC at t 12, and (f) ROC at t 24.

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC JUNE 2007 ATALLAH ET AL. 2201

(Fig. 9d). A minimum of is situated over and just to A key factor in determining the precipitation distri- east of the TC as a result of weak westerly shear over bution around landfalling TCs is the effect that storm- the vortex and is consistent with the weak CVA by the induced diabatic ridging over and downstream of the thermal wind in this region. Over the next 12 h, a de- storm has on the rearrangement of the mass field. Dia- crease in the separation of the midlatitude and tropical batic ridging increases the amplitude of the flow, mak- features results in an increase of the strength of the ing any midlatitude troughs more dynamically active thickness gradient and the associated thermal wind as a upstream via the mechanism of increased positive PV thickness ridge develops from Louisiana to the eastern advection (or cyclonic vorticity advection by the ther- Great Lakes (Fig. 9e), resulting in an expansion and mal wind) over the surface cyclone center. The extent elongation of the region of ascent. to which this “activation” occurs depends on a number Eventually, the thermal ridge associated with the of factors, including the depth and strength of the cold remnant TC weakens as the storm core approaches the core associated with the trough. main baroclinic zone situated over the southern Plains In the case of the LOC precipitation distribution pat- and the Missouri Valley (see Fig. 8f). However, diabatic tern, the relevant dynamics are consistent with storms downstream ridge building well poleward of the TC undergoing the process of ET and are summarized results in increased forcing for ascent ahead of the mid- schematically in Fig. 10a. The interaction generally be- latitude trough (Fig. 9f). Also, the proximity of the TC gins when a strong midlatitude trough approaches the to the equatorward entrance region of the jet streak to TC from the northwest at a distance of less than 1500 the northeast (see Fig. 5f) results in a broad region of km. The length scale of this interaction agrees well with ascent to the east and north of the storm. previous work done on ET such as Foley and Hanstrum (1994) who identified 1700 km as the climatological dis- 4. Discussion and conclusions tance for the beginning of “cyclone capture” or ET in Forecasts of precipitation distribution associated Australia. with landfalling tropical systems continue to be a chal- The strength of the cold-core structure of the trough lenge despite recent advances in numerical modeling is important in that the juxtaposition between the cold and techniques. Several recent storms (e.g., Floyd, core of the midlatitude trough and the warm core of the Dennis, and Irene 1999) have illustrated some of the TC produces a pronounced tropospheric deep baro- shortcomings of the conventional wisdom of relating clinic zone. The steepness of the isentropes in these precipitation totals to storm motion. The threat of situations allows for isentropic ascent between the mid- flooding was anticipated in the case of Dennis due to its latitude trough and the TC in the presence of the cir- relatively slow movement, but the heaviest amounts fell culation of the TC. The favored orientation of the mid- in relatively isolated areas and were on the order of latitude trough in these trough–TC interactions is a 10–15 cm. In contrast, the warnings issued as Floyd positive tilt (thick black lines in Fig. 10a) as this allows approached the coast emphasized the threat of wind for the baroclinic zone to stretch from west to north of damage in light of the expected increase in its forward the TC. Furthermore, a southwesterly jet on the front speed. However, flooding associated with Floyd proved side of these troughs can interact with the convective to be the greatest source of destruction as widespread outflow associated with the approaching TC, resulting amounts greater than 20 cm fell on parts of the mid- in an enhanced jet streak and more pronounced sec- Atlantic with isolated amounts exceeding 50 cm falling ondary circulations in the jet-entrance and exit regions. in eastern North Carolina. Thus, Hurricane Floyd As the TC begins to overrun the strong baroclinic zone, served as the impetus for developing a paradigm by a strong and deep region of warm-air advection devel- which the distribution and intensity of precipitation as- ops, resulting in an expansion of the precipitation shield sociated with landfalling TCs may be understood (see to the north-northwest of the TC (shaded region in Fig. also Atallah and Bosart 2003). Toward this end, in this 10a). The latent heat release from this expanding area study, several landfalling TCs were composited strictly of precipitation then serves as a positive feedback based on the distribution of precipitation relative to the mechanism into the dynamics of the storm through the TC center and track in order to identify synoptic-scale redistribution of PV. This occurs as the northeastern dynamic signatures that modulate the precipitation dis- edge of the midlatitude PV anomaly (negative PV ten- tribution. Potential vorticity thinking was used to help dency in Fig. 10a) erodes in response to latent heat elucidate the interactions between midlatitude and release in the midtroposphere. Meanwhile, the south- tropical systems, while QG dynamics were utilized to ern end of the midlatitude PV anomaly tends to cy- account for the displacement of the greatest vertical clonically wrap into the TC (positive PV tendency in motions from the center of the TC. Fig. 10a) as the TC ingests cold air into its southwestern

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC 2202 MONTHLY WEATHER REVIEW VOLUME 135

tion center approaches a zonally oriented baroclinic zone to the north, the storm tilts downshear (see Fig. 7d) causing the precipitation to become displaced to the east of the cyclone. The basis of this reasoning was discussed in detail by Jones (1995, 2004) and DeMaria (1996) and will be addressed only briefly here. Under adiabatic flow conditions, vertically aligned PV anoma- lies of equal magnitude on the upper and lower bound- aries of a two-layer model (a representation of an equivalent barotropic vortex) are tilted downshear. The PV anomalies then begin to rotate cyclonically about each other as the upper and lower anomalies advect each other, helping minimize the effect of the shear. Also, the tilt of the vortex creates regions of localized vertical wind shear (note that this is independent of the ambient wind shear tilting the vortex) on either side of the vortex tube, leading to a disruption in the gradient thermal-wind balance (GTWB). Secondary circulations arise in order to restore GTWB with ascent (descent) downshear (upshear) of the vortex tube. The differen- tial ascent across the vortex tube in turn acts against the sense of the wind shear. Schematically, the evolution of the ROC composite is summarized in Fig. 10b. As the composite TC moves northward, a ridge develops downstream of it in re- sponse to the diabatic heating and associated outflow from the TC. The enhanced downstream ridging ahead FIG. 10. Schematics for landfalling tropical cyclones for (a) the of the TC amplifies the flow pattern and increases the LOC composite and (b) the ROC composite. The curved black thermal gradient between the TC and a weak midlati- lines represent streamlines of the upper-tropospheric (i.e., 250 hPa) flow. Arrows represent motion and deep tropospheric shear tude trough approaching the system from the north- with the relative magnitudes given by the length of the arrow. The west. As the TC moves across this gradient toward curved green line represents the trajectory of a parcel starting cooler thicknesses, the shear over the cyclone center near the surface in the warm sector and ending in the mid- to increases and backs from westerly to southwesterly. upper troposphere in the cool sector. The gray shaded area rep- The QG forcing for ascent, by the advection of vorticity resents regions of precipitation and pluses and minuses represent the local PV tendency resulting from a combination of advection by the thermal wind, increases and shifts to the north- and the diabatic redistribution of PV. east quadrant of the storm. Eventually, the cyclone be- comes thoroughly embedded in the thickness gradient in the southwesterly flow between a relatively broad quadrant. The end result is the reorientation of the upstream trough and downstream ridge, indicating that trough from a positive to a negative tilt as the wave- a weak transition may be taking place. This signature is length between the trough and the ridge collapses. This consistent with about 1/3 of the storms in the ROC leads to increased values of differential CVA and a composite that experience a shift to the left in the pre- more dynamically active trough. Essentially this pro- cipitation distribution. cess can be though of as a type of self-development Since ridging associated with the outflow of the TC is aided by the diabatic redistribution of PV and is not in such a key player in eventual precipitation distribution general confined to tropical systems as shown by Dick- and intensity, the question arises as to how well do inson et al. (1997) in their study of the 12–14 March numerical models represent the bulk upscale effects of 1993 Superstorm. deep . Figure 11 displays the 48- and 72-h Storms that exhibit an ROC precipitation distribu- forecast error in the NCEP Global Forecast System tion tend to be the ones that weakly interact with mid- (GFS) [formerly Aviation (AVN)] model in the mass latitude troughs in comparison to the LOC storms. field during the ETs of Floyd (1999) and Fabian (2003), ROC storms are in general characterized by small cir- respectively. Figures 11a,c show the 200-hPa height culation centers at the time of landfall. As the circula- field and wind for 0000 UTC 17 September 1999

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC

Fig 10 live 4/C JUNE 2007 ATALLAH ET AL. 2203

FIG. 11. (a) 200-hPa geopotential heights (solid lines, contoured every 12 dam) and winds (one barb 5ms 1) for 0000 UTC 17 Sep 1999. The star represents the position of Hurricane Floyd. (b) The GFS 48-h forecast of the 200-hPa geopotential heights and wind verifying at 0000 UTC 17 Sep 1999. (c) The difference between (a) and (b) in the 200-hPa geopotential height (shaded every 30 dam; dark to light for negative errors and dark to light for positive errors) and wind. (d) Same as in (a) but for 0000 UTC 8 Sep 2003. The star represents the position of Hurricane Fabian. (e) The GFS 72-h forecast of the 200-hPa geopotential heights and wind verifying at 0000 UTC 8 Sep 1999. (f) The difference between (d) and (e) in the 200-hPa geopotential height (shaded every 30 dam; dark to light for negative errors and dark to light for positive errors) and wind.

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC 2204 MONTHLY WEATHER REVIEW VOLUME 135

(Floyd) and 0000 UTC 8 September 2003 (Fabian), re- the TC displaces the forcing for ascent to the right of spectively, where the position of the storm is indicated the track of the TC. by the star. The 48- and 72-h GFS 200-hPa forecasts for Hart et al. (2006) showed that the presence of nega- Floyd and Fabian verifying at the above times are give tively tilted trough upstream of the TC undergoing ET in Figs. 11b,d. Note that in Fig. 11b, the forecast height helped to compact the positive PV advection over and and wind field display significantly less anticyclonic cur- downshear of the surface cyclone and was a key factor vature over the northeastern United States as com- in the subsequent reintensification of the resulting ex- pared to verification. The most significant error is as- tratropical cyclone. Although not shown in Hart et al. sociated with the diabatically enhanced ridging over the (2006), it is likely that precipitation would have shifted eastern Great Lakes as values of the 200-hPa geopo- to the left of the track of the reintensifying extratropical tential heights in this region are underforecast by over cyclone in response to vigorous ascent associated with 12 dam (Fig. 11c). The anticyclonic gyre associated with cyclonic vorticity advection by the thermal wind over this error in the height field had associated errors in the and downshear of the surface cyclone as warm air as- wind exceeding 15 m s1. The error in the ridge struc- sociated with the remnant TC is forced upward as it ture not only leads to a failure of the model’s ability to wraps cyclonically around and over the low-level baro- capture the dynamic nature of the ET of Floyd (e.g., clinic zone. The dynamical processes identified in this Atallah and Bosart 2003), but also results in errors paper as important to the shift of precipitation to the downstream. However, it should be noted that Roebber left of the track of the center of the landfalling and et al. (2004) showed that the ensemble simulation of transitioning TCs are also likely playing a significant Floyd in the fifth-generation Pennsylvania State Uni- role in the extratropical cyclones undergoing reintensi- versity–National Center for Atmospheric Research fication subsequent to ET as identified by Hart et al. (PSU–NCAR) Mesoscale Model (MM5) exhibited (2006). greater sensitivity to the initial conditions (ICs) than These results underscore the need for further im- the convective parameterization scheme. The success- provement in the ability of operational forecast models ful simulation came from the Eta Model ICs while the in accurately representing the bulk upscale effects of AVN ICs resulted in a poor simulation. However, it deep convection. As noted by Roebber et al. (2004), should be noted that both sets of ICs produced poor improvement in forecasts of precipitation distribution forecasts in an operational setting at time scales greater and intensity can be achieved through a combination of than 36 h. This suggests that model physics and resolu- ensemble based prediction and guidance from targeted tion still play a vital role in ET simulation on a case- high-resolution numerical simulations (e.g., Colle by-case basis. A similar error structure is evident in the 2003). This approach has the potential to combine com- forecast of Fabian with a 12-dam error located just putational efficiency and predictive skill provided a dy- downstream of the cyclone (Fig. 11f). In the case of namical framework exists that can help forecasters syn- Fabian, the ridge appears to be dislocated to the west thesize the two streams of information. It is hoped that relative to the forecast (Figs. 11e,f). This error is the results presented in this work, in conjunction with present in spite of the fact that the forecast position of related past studies, will further contribute toward the Fabian is relatively accurate (not shown). development of such a paradigm. While Floyd and Fabian are only two cases, the in- ability of the models to replicate diabatic ridging in Acknowledgments. This work is supported by NSF association with convection is systematic in nature as Grant ATM-0304254. Valuable guidance and insight indicated by Dickinson et al. (1997). It appears clear was provided by Dr. Frank Marks of the Hurricane that diabatic ridging plays an essential role in determin- Research Division. Comments and suggestions made ing the precipitation distribution associated with land- by two anonymous reviewers and Dr. David Schultz falling TCs. In the case of ET, precipitation shifts to the were very beneficial in improving the content of this left of track as differential CVA from an approaching paper. midlatitude trough overspreads the TC. The down- stream diabatic ridging acts to enhance the magnitude REFERENCES of this differential CVA, leading a dynamically active Abraham, J., J. W. Strapp, C. Fogarty, and M. Wolde, 2004: Ex- system in the QG sense. In the ROC case, diabatic tratropical transition of Hurricane Michael: An aircraft in- downstream ridging results in the “activation” of a vestigation. Bull. Amer. Meteor. Soc., 85, 1323–1339. Agustí-Panareda, A., C. D. Thorncroft, G. C. Craig, and S. L. weak upstream trough through an enhanced thermal Gray, 2004: The extratropical transition of gradient and subsequent shear (Bosart and Lackmann (1999): A potential-vorticity perspective. Quart. J. Roy. Me- 1995). The resultant southwesterly thermal wind across teor. Soc., 130, 1047–1075.

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC JUNE 2007 ATALLAH ET AL. 2205

——, S. L. Gray, G. C. Craig, and C. Thorncroft, 2005: The ex- cycle evolution of extratropical transition for Atlantic tropi- tratropical transition of Tropical Cyclone Lili (1996) and its cal cyclones. Mon. Wea. Rev., 131, 909–925. crucial contribution to a moderate extratropical develop- ——, and B. E. Prater-Mayes, 2004: Factors affecting the post- ment. Mon. Wea. Rev., 133, 1562–1573. transition intensification of Hurricane Irene (1999). Mon. Anthes, R. A., 1990: Recent applications of the Penn State/NCAR Wea. Rev., 132, 1355–1368. Mesoscale Model to synoptic, mesoscale, and climate studies. Foley, G. R., and B. N. Hanstrum, 1994: The capture of tropical Bull. Amer. Meteor. Soc., 71, 1610–1629. cyclones by cold fronts off the west coast of Australia. Wea. Arnott, J. M., J. L. Evans, and F. Chiaromonte, 2004: Character- Forecasting, 9, 577–592. ization of extratropical transition using cluster analysis. Mon. Harr, P. A., and R. L. Elsberry, 2000: Extratropical transition of Wea. Rev., 132, 2916–2937. tropical cyclones over the western North Pacific. Part I: Evo- Atallah, E. H., and L. F. Bosart, 2003: Extratropical transition and lution of structural characteristics during the transition pro- precipitation distribution: A case study of Floyd (1999). Mon. cess. Mon. Wea. Rev., 128, 2613–2633. Wea. Rev., 131, 1063–1081. ——, ——, and T. F. Hogan, 2000: Extratropical transition of Bosart, L. F., and F. H. Carr, 1978: A case study of excessive tropical cyclones over the western North Pacific. Part II: The rainfall centered around Wellsville, New York, 20–21 June impact of midlatitude circulation characteristics. Mon. Wea. 1972. Mon. Wea. Rev., 106, 348–362. Rev., 128, 2634–2653. ——, and D. B. Dean, 1991: The Agnes rainstorm of June 1972: Hart, R. E., 2003: A cyclone phase space derived from thermal Surface feature evolution culminating in inland storm rede- wind and thermal asymmetry. Mon. Wea. Rev., 131, 585–616. velopment. Wea. Forecasting, 6, 515–537. ——, and J. L. Evans, 2001: A climatology of the extratropical ——, and G. M. Lackmann, 1995: Postlandfall tropical cyclone transition of Atlantic tropical cyclones. J. Climate, 14, 546– reintensification in a weakly baroclinic environment: A case 564. study of (September 1979). Mon. Wea. Rev., ——, ——, and C. Evans, 2006: Synoptic composites of the extra- 123, 3268–3291. tropical transition life cycle of North Atlantic tropical cy- ——, C. S. Velden, W. E. Bracken, J. Molinari, and P. G. Black, clones: Factors determining posttransition evolution. Mon. 2000: Environmental influences on the Wea. Rev., 134, 553–578. of (1995) over the Gulf of Mexico. Mon. Henderson, J. M., G. M. Lackmann, and J. R. Gyakum, 1998: An Wea. Rev., 128, 322–352. analysis of Hurricane Opal’s forecast track errors using quasi- Browning, K. A., G. Vaughan, and P. Panagi, 1998: Analysis of an geostrophic potential vorticity inversion. Mon. Wea. Rev., ex-tropical cyclone after reintensifying as a warm-core extra- 127, 292–307. tropical cyclone. Quart. J. Roy. Meteor. Soc., 124, 2329–2356. Hoskins, B. J., and P. Berrisford, 1988: A potential vorticity view Carr, F. H., and J. P. Millard, 1985: A composite study of comma of the storm of 15–16 October 1987. Weather, 43, 122–129. clouds and their association with severe weather over the ——, M. E. McIntyre, and A. W. Robertson, 1985: On the use and Great Plains. Mon. Wea. Rev., 113, 370–387. significance of isentropic potential vorticity maps. Quart. J. Chien, H. H., and P. J. Smith, 1977: Synoptic and kinetic energy Roy. Meteor. Soc., 111, 877–946. analysis of (1969) during transit across the Jones, S. C., 1995: The evolution of vortices in vertical shear. I: southeastern United States. Mon. Wea. Rev., 105, 67–77. Initially barotropic vortices. Quart. J. Roy. Meteor. Soc., 121, Colle, B. A., 2003: Numerical simulation of the extratropical tran- 821–851. sition of Floyd (1999): Structural evolution and responsible ——, 2004: On the ability of dry tropical-cyclone-like vortices to mechanisms for the heavy rainfall over the northeast United withstand vertical shear. J. Atmos. Sci., 61, 114–119. States. Mon. Wea. Rev., 131, 2905–2926. ——, and Coauthors, 2003: The extratropical transition of tropical Cressman, G. P., 1959: An operational objective analysis system. cyclones: Forecast challenges, current understanding, and fu- Mon. Wea. Rev., 87, 367–374. ture directions. Wea. Forecasting, 18, 1052–1092. Dare, R. A., and N. E. Davidson, 2004: Characteristics of tropical Kalnay, E. M., and Coauthors, 1996: The NCEP/NCAR 40-Year cyclones in the Australian region. Mon. Wea. Rev., 132, 3049– Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–471. 3065. Kistler, R., and Coauthors, 2001: The NCEP–NCAR 50-Year Re- DeMaria, M., 1996: The effect of vertical shear on tropical cyclone analysis: Monthly means CD–ROM and documentation. intensity change. J. Atmos. Sci., 53, 2076–2088. Bull. Amer. Meteor. Soc., 82, 247–267. Dickinson, M. J., L. F. Bosart, W. E. Bracken, G. J. Hakim, D. M. Klein, P. M., P. A. Harr, and R. L. Elsberry, 2000: Extratropical Schultz, M. A. Bedrick, and K. R. Tyle, 1997: The March transition of western North Pacific tropical cyclones: An 1993 superstorm cyclogenesis: Incipient phase synoptic- and overview and conceptual model of the transformation stage. convective-scale flow interaction and model performance. Wea. Forecasting, 15, 373–396. Mon. Wea. Rev., 125, 3041–3072. ——, ——, and ——, 2002: Extratropical transition of western DiMego, G. J., and L. F. Bosart, 1982a: The transformation of North Pacific tropical cyclones: Midlatitude and tropical cy- tropical storm Agnes into an extratropical cyclone. Part I: clone contributions to reintensification. Mon. Wea. Rev., 130, The observed fields and vertical motion computations. Mon. 2240–2259. Wea. Rev., 110, 385–411. McTaggart-Cowan, R., J. R. Gyakum, and M. K. Yau, 2001: Sen- ——, and ——, 1982b: The transformation of tropical storm sitivity testing of extratropical transitions using potential vor- Agnes into an extratropical cyclone. Part II: Moisture, vor- ticity inversions to modify initial conditions: Hurricane Earl ticity, and kinetic energy budgets. Mon. Wea. Rev., 110, 412– case study. Mon. Wea. Rev., 129, 1617–1636. 433. ——, ——, and ——, 2003: The influence of the downstream state Ertel, H., 1942: Ein neuer hydrodynamischer wirbelsatz. Meteor. on extratropical transition: Hurricane Earl (1998) case study. Z., 59, 271–281. Mon. Wea. Rev., 131, 1910–1929. Evans, J. L., and R. E. Hart, 2003: Objective indicators of the life Molinari, J., S. Skubis, and D. Vollaro, 1995: External influences

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC 2206 MONTHLY WEATHER REVIEW VOLUME 135

on hurricane intensity. Part III: Potential vorticity structure. Rossby, C. G., 1940: Planetary flow patterns in the atmosphere. J. Atmos. Sci., 52, 3593–3606. Quart. J. Roy. Meteor. Soc., 66, 68–87. ——, D. Knight, M. J. Dickinson, D. Vollaro, and S. Skubis, 1997: Sinclair, M. R., 1993: A diagnostic study of the extratropical pre- Potential vorticity, easterly waves, and eastern Pacific tropi- cipitation resulting from Tropical Cyclone Bola. Mon. Wea. cal cyclogenesis. Mon. Wea. Rev., 125, 2699–2708. Rev., 121, 2690–2707. Morgan, M. C., and J. W. Nielsen-Gammon, 1998: Using tropo- ——, 2002: Extratropical transition of southwest Pacific tropical pause maps to diagnose midlatitude weather systems. Mon. cyclones. Part I: Climatology and mean structure changes. Wea. Rev., 126, 2555–2579. Mon. Wea. Rev., 130, 590–609. Muramatsu, T., 1985: A study on the changes of the three- ——, 2004: Extratropical transition of southwest Pacific tropical dimensional structure and movement speed of the typhoon cyclones. Part II: Midlatitude circulation characteristics. through its life time. Tech. Rep. 14, Meteorological Research Mon. Wea. Rev., 132, 2145–2168. Institute, Japan, 117 pp. Sutcliffe, R. C., 1947: A contribution to the problem of develop- Palmén, E., 1958: Vertical circulation and release of kinetic en- ment. Quart. J. Roy. Meteor. Soc., 73, 370–383. ergy during the development of into an ex- tratropical cyclone. Tellus, 10, 1–23. ——, and A. G. Forsdyke, 1950: The theory and use of upper air Ritchie, E. A., and R. L. Elsberry, 2003: Simulations of the extra- thickness patterns in forecasting. Quart. J. Roy. Meteor. Soc., tropical transition of tropical cyclones: Contributions by the 76, 189–217. midlatitude upper-level trough to re-intensification. Mon. Thorncroft, C., and S. C. Jones, 2000: The extratropical transitions Wea. Rev., 131, 2112–2128. of Hurricanes Felix and Iris in 1995. Mon. Wea. Rev., 128, Roebber, P. J., D. M. Schultz, B. A. Colle, and D. J. Stensrud, 947–972. 2004: Toward improved prediction: High-resolution and en- Trenberth, K. E., 1978: On the interpretation of the diagnostic semble modeling systems in operations. Wea. Forecasting, 19, quasi-geostrophic omega equation. Mon. Wea. Rev., 106, 936–949. 131–137.

Unauthenticated | Downloaded 09/27/21 02:06 PM UTC