1. Introduction

Every year, the North Atlantic Ocean experiences an average of nine tropical storms and hurricanes (Elsberry 1987). Of these nine tropical cyclones (TCs), about 45% (or about four storms per year) transition into extratropical cyclones (Hart and Evans 2001). In their climatology, Hart and Evans (2001) determined that most extratropical transitions occur between 35° and 45°N, with latitudinal variations corresponding to the seasonal progression of the subtropical jet axis; this latitudinal zone encompasses the populated United States East Coast and the Canadian Maritime Provinces, as well as major shipping routes and fishing grounds. Despite the proximity of this extratropical transition (ET) zone to both large population and financial interests, the process of ET remains poorly forecasted.

Hurricane Irene (1999) formed south of Cuba around 1200 UTC 13 October; the storm crossed Cuba, the Florida Keys, and the southern tip of Florida, finally emerging into the Gulf Stream. Irene then paralleled the East Coast along the Gulf Stream until near its time of transition, when it accelerated northeastward away from the coast. The storm presented a forecast problem because of its proximity to the coast and its uncertain track (Avila 1999). Furthermore, Irene intensified 15 hPa in 12 h near its time of transition, as determined by a reconnaissance flight into the storm near 0800 UTC 18 October 1999.

Key characteristics in the extratropical transition of TCs are the loss of symmetry and a marked enhancement in the baroclinicity of the system. Baroclinic temperature gradients can be induced in a TC by interaction with one or more synoptic features, including midlatitude cyclones and upper-level features such as troughs and jet streaks. Interaction with these features, especially in the presence of vertical wind shear, contributes to loss of symmetry in a tropical storm and frontogenesis. The storm experiences a period during which its structure changes from warm-cored and symmetric, typical of TCs, to cold-cored and asymmetric, as characteristic of extratropical cyclones. The associated structural changes include expansion of the area of gale-force wind, surface frontogenesis, a shift in the height and radius of maximum winds, asymmetric development of the wind and rain structures, and transition from a dynamically warm-core structure to dynamically cold-core (e.g., Klein et al. 2000; Thorncroft and Jones 2000). Extratropical transition is a process that can be characterized by distinct objective measures of onset and completion; it is not well defined by a single point in time (Evans and Hart 2003).

The case of Hurricane Irene (1999) is far from the first study of a hurricane interacting with a trough system; Molinari et al. (1995) explored the intensification of Hurricane Elena (1985) due to an interaction with an approaching midlatitude trough. They observed a narrowing and elongating of the approaching trough and concluded that the superposition of the trough and storm lead to an enhanced wind-induced, surface heat exchange (WISHE) feedback (Emanuel 1986, 1997) and ultimately to the observed rapid intensification cycle of Elena. Bosart and Bartlo (1991) considered the subtropical development of Tropical Storm Diana (1984); while this storm initially began as the result of wave breaking of an upper-level trough, the resulting upper potential vorticity (PV) center was able to deepen in the low-stability environment present over the Gulf Stream in September, again allowing for surface flux and convective feedbacks to continue the tropical storm development. In contrast to Hurricane Irene (1999), the common features of both Elena (1985) and Diana (1984) were underlying warm, and relatively uniform, tropical waters and their occurrence early in the season, when the atmospheric support for baroclinic development is confined much farther north (Hart and Evans 2001). By remaining in tropical waters during the entire interaction, each storm was able to maintain a deep convection signature and, ultimately, to intensify through enhanced latent heat release as tropical storms. The influence of convection on the evolution of Irene throughout her life cycle is one of the key factors explored here.

The evolution of Hurricane Irene (1999) through transition is studied here. The methods used in simulating Irene are described in section 2, followed by an overview of the life cycle of Irene in section 3. A description of the features associated with the midlatitude cyclogenesis associated with Irene as she transitioned will be given in section 4. An analysis of the cyclogenesis as well as the intensification of Irene during and after transition is presented in section 5, and concluding remarks are summarized in section 6.

2. Methods of analysis

Hurricane Irene (1999) was simulated using the nonhydrostatic fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU– NCAR) Mesoscale Model (MM5), version 2 (Grell et al. 1994). Gridded fields from the Navy Operational Global Atmospheric Prediction System (NOGAPS) (Hogan and Rosmond 1991) were used for the initial and boundary conditions. The MM5 was run with grid spacing of 45 km × 45 km and 29 vertical sigma levels, an improvement on the 1° × 1° grid spacing of the available NOGAPS operational analyses, but still too coarse to resolve hurricane eye processes.

In the simulations presented here, the MM5 was initialized at 1200 UTC 16 October 1999. The initialization time was selected because, compared to simulations using other onset dates, (i) it was at least 24 h before the first signs of extratropical transition, allowing adequate time for spinup of the tropical phase of the cyclone before the transition period began; (ii) the vortex was completely over the ocean, minimizing land surface effects; and (iii) the resulting storm track best captured the landfall and recurvature locations in the National Hurricane Center (NHC) best-track analysis.

The MM5 simulation used here is a continuous data assimilation run (Stauffer and Seaman 1994). The MM5 “first guess” analysis was directly interpolated from the 1° × 1° NOGAPS analyses. The subsequent data assimilation cycle incorporated satellite-derived winds (Velden et al. 1997, 1998) into the model at 3-h intervals beginning at 0000 UTC 18 October and continuing until the end of the simulation (Prater 2002). The veracity of these results was assessed by Prater and Evans (2002) in a separate study. In particular, sensitivity testing of the impact of the choice of convective parameterization scheme on the storm simulation revealed that the Kain– Fritsch scheme (Kain and Fritsch 1990, 1993) more accurately represented Irene than the Betts–Miller scheme. Evaluation of the impact of the convective parameterization scheme was based on evolution of the dynamical structure, precipitation signature, and track throughout the storm life cycle. These results provide a measure of confidence in the MM5 assimilation incorporating the Kain–Fritsch convection scheme, and thus that assimilation is employed here.

Finally, as a comparison to the model data for analysis, the NHC best-track data were used here to provide Dvorak-estimated and observed wind and intensity data, as well as track position (Avila 1999). Advance Microwave Sounding Unit (AMSU) data from around 0000 UTC 16 October were used to evaluate the broad structural characteristics of the initial tropical vortex.

3. The life cycle of Irene

The NHC best track of Irene from 1200 UTC 16 October to 1200 UTC 19 October and the MM5 simulated track are plotted in Fig. 1. Before ET commenced, Irene was analyzed by the NHC at an intensity of about 985 hPa (Fig. 2) with maximum winds at 65 to 70 kt. Between 1200 UTC 17 October and 0600 UTC 18 October, rapid intensification was analyzed, with the pressure falling to 964 hPa and winds rising to 95 kt; in addition, the storm began accelerating with a more eastward component of motion, toward the northeast and then the east-northeast, with translation speed reaching 30 kt by 0900 UTC 18 October. The intensification and acceleration were coincident with the commencement of extratropical transition around 1800 UTC 17 October. The NHC discussion from 0900 UTC 18 October reported a “Big surprise!” when a reconnaissance aircraft found winds of 129 kt at 902 hPa, producing a “conservative estimate” of 95-kt surface winds and central pressure of 958 hPa at 0756 UTC 18 October, when Irene was centered just off the Outer Banks of North Carolina (Avila 1999). This peak in intensity represents the peak of the rapid intensification phase as Irene began to interact with an approaching midlatitude upper-level trough and jet streak. At this time, the model storm was over SST of 26°C; the storm experienced monotonically decreasing SSTs (to 12°C by 1200 UTC 19 October) for the remainder of the simulation. The storm evolution prior to the time of transition onset will be examined in more detail below.

After reaching its peak intensity, Irene weakened slightly over 10 h, to maximum winds of 80 kt and minimum central pressure of 968 hPa by 1800 UTC 18 October; according to Dvorak satellite estimates, the storm remained at this intensity for the remainder of the best-track records. In addition, Irene continued to accelerate to the northeast under the influence of an approaching upper-level trough, with translation speed reaching 55 kt toward the ENE. Thus, within a period of about 30 h, Irene weakened slightly from a weak category 2 hurricane to a strong category 1 and then transitioned to an extratropical cyclone, decreasing in intensity by only 10 hPa. The NHC declared Irene extratropical at 0300 UTC 19 October (Avila 1999).

The MM5-simulated intensity of Irene is less than the NHC best-track data because of the model resolution; gridded data does not capture the strongest point value but rather averages it with all the winds in a grid box, thus decreasing the peak intensity of the storm (Fig. 2). The interpretations of the pattern of intensification between the NHC and the MM5 agree well until later in the storm life cycle, where they diverge; while the NHC (based on satellite estimates) maintains an unchanging intensity posttransition, the MM5 intensifies the storm somewhat at the end of ET and during the extratropical phase. Because the NHC was only estimating the storm intensity based on satellite estimates—and because in the extratropical phase of its life cycle, the storm is big enough to be well resolved by the MM5—the model interpretation of intensification tendency during and after transition is taken to be reasonable.

Cyclone phase diagnostics (Hart 2003) using the MM5 assimilation as input are presented in Fig. 3. Based on these analyses, Irene commenced extratropical transition near 1800 UTC 17 October, the first point that the storm becomes asymmetric (Evans and Hart 2003). Symmetry is defined in the phase diagnostics by the difference in 900–600-hPa thickness across the storm relative to storm motion; a thickness difference of 10 m was determined empirically to define the onset of ET (Evans and Hart 2003). Extratropical transition completion is characterized by the first point in time after the storm has become both asymmetric and cold-cored; the sign of the thermal wind in the 900–600-hPa layer defines the core structure, where a negative slope indicates a warm-cored system and a positive slope indicates a cold core (Evans and Hart 2003). This phase diagnosis places completion of ET at 0000 UTC 19 October, 3 h prior to the NHC declaration and within the temporal resolution of the NHC best track (Fig. 3).

Irene was near the climatological mean of transitioning North Atlantic TCs with respect to its duration (30 h), location (45°N), and date (mid-October) of extratropical transition; the climatological average for ET is 36 h, taking place from about 34°–44°N, with 50% of October tropical cyclones experiencing ET (Hart and Evans 2001; Evans and Hart 2003; Hart 2003). Since only a small percent of storms that form in the Gulf of Mexico undergo ET and reintensify, Irene's early lifetime was unusual for a transitioning Atlantic storm (Hart and Evans 2001). Irene was also one of the 50% of transitioning TCs that undergo posttransition reintensification (Hart and Evans 2001).

4. Cyclogenetic features

Hart and Evans (2001) demonstrated that the months in which ET was most frequent corresponded to the times when the region favorable for midlatitude development was closest geographically to the region supporting tropical development. Thus, to determine mechanisms supporting ET, we explore factors favorable for midlatitude cyclogenesis and development.

During ET, Irene interacted with an upper-level trough and jet streak, providing a favorable environment for cyclogenesis and intensification of the developing storm. In addition, a warm-core seclusion provided by the northward advection of warm, moist air in the core and ahead of Irene also contributed to the baroclinicity of the region, further enhancing cyclogenesis. The contributions of these factors to the rapid intensification of Irene during ET and also to the intensification of the storm after it became extratropical are explored here. A more detailed analysis can be found in Prater (2002).

5. Analysis of cyclogenesis and intensification

The upper-level jet streak and trough, along with the warm-core anomaly advected northward by Irene, provided support for cyclogenesis in the vicinity of the jet streak as Irene approached. These features can also be linked to upward motion, and their interactions can be described by the quasigeostrophic omega equation as well as by isentropic potential vorticity analysis.

a. Quasigeostrophic dynamics

In this section, we use the quasigeostrophic (QG) omega equation to aid in interpreting the evolution of Irene through the transition process. While QG theory is not necessarily valid in the core of a tropical cyclone, it is a reasonable approximation for synoptic motions, such as the interactions of the broad cyclone vortex (as Irene undergoes extratropical transition) and the upper-level trough discussed above. In quasigeostrophic theory, the presence of upward motion depends on both differential cyclonic vorticity advection and positive temperature advection:

View Expanded

where R_d is the gas constant for dry air, s̃ is the local horizontal average of static stability, ζ_g represents the geostrophic vorticity, ∇_p is the spatial gradient w.r.t. a constant pressure surface, V_g represents the horizontal (with respect to constant pressure surfaces) geostrophic winds, and Z is the geopotential height (Carlson 1991). The two forcing terms on the right correspond to the vertical gradient of horizontal vorticity advection and the Laplacian of thickness advection (thermal advection), respectively. Given realistic boundary conditions for ω at the surface and upper troposphere, positive values for the forcing terms—that is, CVA increasing with height or warm-air advection (WAA)—generally correspond to regions of vertical ascent.

CVA at 500 hPa is weak during the tropical phase of Irene but increases ahead of the surface cyclone as the storm undergoes ET (Fig. 7, contours). Temperature advection is weak in the tropical phase of Irene, before the midlatitude surface cyclone reaches the naturally baroclinic environment of the coastal zone; despite a thermal anomaly associated with the TC, the tropical environment is associated with weak pressure gradients, thus limiting advection (Fig. 7a). Once the midlatitude surface cyclone reaches the coastal thermal gradient (seen in Fig. 8a), the temperature advection by the quasigeostrophic motions increases (Figs. 7b–d). Cold-air advection behind the midlatitude surface feature is particularly strong, but WAA ahead of both Irene and the upper-level trough increases throughout extratropical transition, leading to cyclogenesis ahead of the center of the surface storm caused by both features through quasigeostrophic mechanisms described above. The advection of the warm-core anomaly (discussed in section 4c) with the motion of Irene also introduces a strengthening thermal gradient with the system. The WAA and CVA associated with Irene as it undergoes ET are collocated, especially near the end of ET (Fig. 7c) and when the storm has become extratropical (Fig. 7d), indicating the potential for developing and strengthening the cyclone.

Prater and Evans (2002) explored the evolution of the midlatitude development region both with and without Hurricane Irene (1999). They found that without Irene, cyclogenesis may still have occurred as the weak upper-level trough reached the coastal thermal gradient and jet streak. In this case, the trough moved into a zone of low-level convergence and stronger thermal gradient (the coastal front) that contributed to upward vertical motion just ahead of the trough, resulting in surface pressure falls and the development of a surface cyclone. The presence of Irene, however, contributed stronger CVA and somewhat stronger WAA ahead of the midlatitude surface feature. Thus, the extratropically transitioning TC strengthened the cyclogenetic processes ahead of the upper-level trough, generating a stronger surface cyclone more rapidly than if the TC had not been present (Prater and Evans 2002). Similarly, Thorncroft and Jones (2000) found that while an existing trough likely would have developed a large, deep, surface cyclone, the presence of extratropically transitioned Hurricane Iris (1995) contributed to explosive cyclonic development.

b. Isentropic potential vorticity analysis

Another approach to interpreting the dynamics is through isentropic PV analysis. This analysis is useful in examining TCs because the PV anomaly of a TC has large amplitude in the near-zero PV tropical environment. Because the mean height of an isentrope increases from equator to pole, PV analysis on an isentropic surface can capture both the midlatitude trough and the TC well (Hoskins et al. 1985). Further, the influence of the diabatic heating can be explored through the temporal evolution of this forcing term in the PV equation:

View Expanded

where PV = ρ⁻¹ω_a · ∇θ is the isentropic potential vorticity (IPV), ρ is density, ω_a is the absolute vorticity vector, θ̇ is the warming rate, θ is potential temperature, and K = ∇ × ν∇²v.

The IPV field depicts the two systems approaching to within about 500 km on the 320-K isentrope (lower to midtroposphere in the region surrounding the storm) as they interact. Hanley et al. (2001) categorize a “favorable distant interaction” as an interaction between two upper-level PV maxima at a distance of 400–1000 km; intensification results from upper-level divergence, with the center of the tropical cyclone near the right entrance region of a jet streak. The evolution of Irene through ET relied on the storm's movement into the westerlies and interaction with the midlatitude trough (Prater and Evans 2002). This interaction is explored from the time of pressure change discontinuity at 0000 UTC 18 October (Fig. 2), first through the evolution of the 320-K IPV (Fig. 10) and then through the convective heating tendency (Fig. 11). Immediately prior to transition (0000 UTC 18 October), an upper-level trough is visible approaching the Great Lakes area, while the PV center associated with Irene remains isolated (Fig. 10a). As ET progresses, the PV center elongates toward the north and east; meanwhile, the approaching trough lifts northeastward (Fig. 10b; 1200 UTC 18 October). Near the completion of ET (0000 UTC 19 October), the PV signature of the trough begins to interact with that of Irene, as the spatial scale of PV above 0.5 PVU increases (Fig. 10c). Finally, after transition has been completed, the interaction of the original upper-level trough and the transitioned TC continues as the centers of PV approach (Fig. 10d).

The convective warming rate, or the change in temperature with time due to convective processes alone, directly impacts the evolution of the PV through the total warming rate θ̇, as seen in Eq. (2). Prior to 0600 UTC 17 October, the convective warming maximum is in the lower levels (Fig. 11), but prior to Irene commencing ET, the convective warming maximum shifts to the mid- to upper levels, while downdraft cooling dominates near the surface (Fig. 11). This shift in position of convective warming maximum is maintained through transition and coincides with the strengthening of the PV anomaly at upper levels as the storm undergoes ET, consistent with the convective warming contribution to the local PV increases around Irene [Eq. (2)].

Geostationary Operational Environmental Satellite-8 (GOES-8) infrared imagery (Fig. 12) provides support for the model-derived convective warming rate profiles contoured in Fig. 11. Prior to and immediately after the onset of ET, the storm core retains a symmetric convective signature (Figs. 12a and 12b), with stratiform clouds farther to the north along the approaching cold front; as Irene completes ET, the clouds near the surface center are predominantly stratiform (Figs. 12c and 12d), implying both a higher elevation for the heating and an unsaturated subcloud layer, consistent with the heating profile evolution mapped in Fig. 11.

The height of the convective warming maximum has been linked to intensity in previous studies (Anthes and Keyser 1979; Gyakum 1983; Prater and Evans 2002); in these studies, convective heating in the low to midlevels contributed to the destabilization, surface convergence, and subsequent intensification of a cyclone. Both Anthes and Keyser (1979) and Gyakum (1983) also noted the sensitivity of the growth rate of a cyclone to the level at which the maximum occurred.

6. Concluding summary

Irene experienced extratropical transition for a 30-h period beginning around 1800 UTC 17 October and ending near 0000 UTC 19 October. The end of extratropical transition, as deduced from the objective indicators of ET (Fig. 3), coincided with the NHC declaration of Irene as an extratropical storm at 0300 UTC 19 October, within the temporal resolution of the NHC forecast updates. Following transition, Irene exhibited some strengthening in the MM5 simulation of the storm, placing Irene among the majority of storms that intensify following ET but in the atypical category of a tropical cyclone that had beginnings in the Caribbean Sea and then experienced posttransition intensification.

The extratropical transition of Irene was enhanced by dynamical support for midlatitude cyclogenesis near the right entrance region of an upper-level jet streak associated with an approaching upper-level trough. Thus, as ET commenced Irene was located in a region of enhanced ascent (Fig. 5), a vorticity maximum (Fig. 4), and a potential vorticity anomaly (Fig. 10). The development of a thermal gradient aided by an upper-level θ_e maximum (Fig. 9) and warm-air advection associated with the TC itself. The warm and moist air advected with the transitioning storm (Fig. 7) decreased the static stability of the environment into which Irene moved, enhancing the development and maintenance of convection over low SSTs. Both the warm-anomaly advection and the convective warming contributed to the baroclinic development of Irene during and after ET.

Further research on the extratropical transition of tropical cyclones in the North Atlantic is still necessary. Forecasting the onset and completion of ET remains difficult (Jones et al. 2003); in addition, forecasting the structural changes associated with a transitioning TC, including wind field changes, translation speed, and asymmetry, is still a challenge. The further development of objective criteria for ET, such as the diagnostic ET life cycle of Klein et al. (2000) and the cyclone phase space of Evans and Hart (2003), is necessary to aid forecasters in the recognition of ET onset and subsequent expectations of behavior in a transitioning TC as well as the extratropical remnants of a TC that has undergone posttransition intensification.