1. Introduction

Although most of the previous studies of the interactions between tropical cyclones and extratropical upper-tropospheric troughs have concentrated on the use of diagnostic calculations from analyses, a few have made use of satellite imagery (e.g., Velden 1987; Bosart et al. 2000). Hanley et al. (2001, hereinafter HMK) identified two different types of trough interactions based on the potential vorticity (PV) structure of the upper-tropospheric trough. The first type, defined as superposition, occurs when an upper-tropospheric PV maximum approaches within 400 km of the tropical cyclone center. The second type, defined as distant interaction, occurs when the PV maximum is within 1000–400 km of the center. HMK found that 78% of all superposition cases intensified and 61% of all distant-interaction cases resulted in deepening. Subtle differences in the PV structure of the upper-tropospheric troughs involved in intensifying and weakening distant-interaction cases suggest that PV signatures alone may not be suitable as forecast methods (HMK). The high percentage of intensifying superposition cases observed by HMK suggests that identification of such interactions by numerical guidance or by satellite imagery could improve the forecasting of intensity change during this type of interaction.

The relationship between water vapor and PV has been shown by several investigators (e.g., Appenzeller and Davies 1992; Appenzeller et al. 1996; Mansfield 1997; Demirtas and Thorpe 1999). In several case studies of tropical cyclone–trough interactions, Hanley (1999) noted that there was very strong similarity between the PV structure of the upper-tropospheric trough and signatures observed in satellite water vapor imagery [i.e., Geostationary Operational Environmental Satellite-8 (GOES-8) 6.7-μm imagery]. The size and shape of the PV maxima on selected isentropic surfaces was represented by corresponding dark regions in the water vapor imagery with similar structures. Water vapor image brightness temperature contours in the satellite imagery were noted to be approximately parallel to PV contours. This note illustrates the similarities of the PV and water vapor signatures for the case of Hurricane Bertha (1996), an example of a superposition-type tropical cyclone–trough interaction.

Hurricane Bertha originated on 1 July 1996 from a tropical wave that moved off the coast of West Africa. The storm followed a smooth path around the Atlantic subtropical ridge while a weak midlevel trough persisted in the western North Atlantic. Bertha strengthened into a hurricane on 8 July as it moved across the Leeward and Virgin Islands. At that time, maximum winds of 39 m s⁻¹ were recorded. On 9 July, the track turned to the northwest, and the maximum sustained winds were measured to be 51 m s⁻¹ at 0600 UTC. Bertha began to follow a more north-northwest path on 10–11 July at a forward speed of about 4 m s⁻¹. Maximum winds were gradually decreasing at this time, from 51 m s⁻¹ on 9 July to 36 m s⁻¹ on 11 July. Prior to landfall near Wilmington, North Carolina, at 2000 UTC 12 July, Bertha accelerated to a speed near 8 m s⁻¹. As an upper-level trough approached the hurricane, winds in Bertha abruptly increased to 46 m s⁻¹, which was the estimated 1-min wind speed at landfall. Details of Hurricane Bertha's track and intensity evolution were obtained from Pasch and Avila (1999).

The storm was much stronger at landfall than had been originally forecast. Although warnings were posted prior to landfall, Bertha resulted in approximately $270 million in damage in North Carolina, and 12 deaths were reported that were due to the storm (Pasch and Avila 1999). The failure of official forecasts to describe adequately the intensification of Bertha highlights the challenge of forecasting tropical cyclone intensity. The rapid reintensification of Bertha just prior to landfall indicates the need for more forecasting methods with regard to identifying trough interactions.

2. Data and analysis methods

3. Discussion

AVHRR sea surface temperatures (SST; Fig. 1) indicate that Bertha was moving over regions of SST warmer than 26°C throughout its entire track over water. During the period of intensification near landfall (0600–1800 UTC 12 July; Fig. 2), SSTs along Bertha's track remained relatively constant (Fig. 1). Because intensification occurred over relatively constant water temperatures, SST changes alone are not sufficient to explain the fluctuations in intensity observed in Hurricane Bertha. Prior to the strengthening observed in Fig. 2 (0600–1800 UTC 12 July), Bertha was weakening steadily as the storm moved northward along the east coast of Florida.

The vertical wind shear in the 850–200-hPa layer is calculated by averaging over a 500-km radius about the storm center (Fig. 3). The averaging removes the azimuthal mean component and retains the cross-storm component (Molinari 1993). As the trough and hurricane approach one another (1200 UTC 12 July), the vertical shear increases to more than 15 m s⁻¹ from the northwest. After 1200 UTC 12 July, the shear begins to decrease as it shifts to a southwesterly direction. Although it may appear that weakening shear was the reason for the intensification noted in Fig. 2, the central pressure began to fall 6 h prior to the drop in shear magnitude.

Increasing values of vertical shear on 11 July (Fig. 3) result in a downshear tilt in the hurricane, as evidenced by the presence of an exposed eye in Figs. 4a,b, which depict Bertha at 1700 and 2000 UTC 11 July, respectively. This disruption of the vertical alignment of the hurricane results in rapid weakening during this period (Fig. 2). Velden (1997) has proposed that the weakening observed in Hurricane Bertha may be related to “backdoor” vertical wind shear. This wind shear scenario arises when a short-wave trough and associated jet downstream of the hurricane begin to amplify through baroclinic processes. The resulting jet entrance region is located just to the north of the main hurricane outflow channel and results in air accelerating from the hurricane into the jet entrance region. High-density multispectral satellite winds for 1800 UTC 11 July and 0000 UTC 12 July (Figs. 5a,b) show an increase in winds over the northeast quadrant of Hurricane Bertha and increased outflow to the northeast, into the jet entrance region. The satellite signatures observed in Figs. 4 and 5 exhibit features of the backdoor shear process described by Velden (1997) during the period of rapid weakening observed on 11 July.

Despite further increases in the magnitude of the vertical shear, visible imagery at 1200 UTC 12 July (Fig. 4d) indicates redevelopment of the hurricane with the eye no longer exposed, a much better organization of the upper-level circulation, and vigorous convective rainbands. In a study of the relationship between vertical shear and the resulting intensity response of a tropical storm, Gallina and Velden (2000) suggest that there may be an inverse relationship between shear and intensity changes, with a time lag of 12–36 h, depending on the strength of the storm. When the 6-h intensity changes for Hurricane Bertha are plotted versus the vertical shear (not shown), there appears to be such a relationship in this case, with an approximately 24-h lag in response of the hurricane to the changes in the strength of the vertical shear. This relationship does not appear to be valid during the period in which the hurricane rapidly reintensifies on 12 July. During this time period, the shear and 6-h intensity change are both increasing, suggesting that some process is acting to oppose the negative impact of the increasing strength of the vertical shear. HMK and others have suggested that the positive effects of eddy momentum flux convergence (or potential vorticity advection) on the tropical cyclone may help to offset the negative impacts of increasing vertical shear.

The time evolution of EFC at 200 hPa for Hurricane Bertha is shown in Fig. 6. DeMaria et al. (1993) define a moderate trough interaction to be one in which the EFC at 200 hPa is between 10 and 20 m s⁻¹ day⁻¹ and a strong interaction to be one with values of EFC greater than 20 m s⁻¹ day⁻¹. The values of EFC plotted in Fig. 6 indicate the occurrence of a strong interaction beginning at 0600 UTC 12 July, with values exceeding 40 m s⁻¹ day⁻¹ by 1200 UTC 12 July. The timing of the strong trough interaction coincides with the period of increasing vertical shear from 11 to 12 July. A maximum of EFC propagates inward with time on the 345-K surface (not shown) and reaches the center of the hurricane at the time intensification begins. Molinari and Vollaro (1989) showed a similar relationship between the vertical shear and intensification during Hurricane Elena (1985), which underwent a strong trough interaction and began to intensify under conditions of increasing vertical shear during the interaction.

Satellite water vapor imagery (GOES-8, 6.7 μm) of Hurricane Bertha is shown in Fig. 7, and the evolution of PV on the 345-K isentropic surface at the same times is shown in Fig. 8. The 345-K surface was chosen because it contained the largest eddy fluxes at the time of superposition (following Molinari et al. 1995). Images are shown every 6 h beginning 1200 UTC 11 July, which is 18 h prior to the onset of intensification as noted in Fig. 2. The hurricane is clearly visible off the east coast of the United States, represented by a region of high moisture (Fig. 7). The region of dry air (dark area) to the northwest of the hurricane, extending from Kansas northeast across to New England, is associated with an upper-level short-wave trough (Fig. 7a). The dark area corresponds to the region of high PV visible in Fig. 8a, and the edge of the dry air lies approximately along the 2-PVU contour (1 PVU = 1 × 10⁻⁶ m² K s⁻¹ kg⁻¹). The band of moisture between the dry air (high PV) and the hurricane is in the region of high PV gradient (Fig. 8a). Outflow from Bertha at the 345-K level is not well defined at this time, and the storm is rapidly weakening.

The trough is moving rapidly, and, by 0000 UTC 12 July (Fig. 8c), the trough and Bertha begin to approach one another with the high PV of the trough located approximately 800 km from the storm center. Outflow from the storm is better defined to the east of its center. The increase in PV gradient between the trough and the hurricane (Figs. 8c,d) is reflected in the organization of moisture in water vapor imagery (Figs. 7c,d). As the region of high PV associated with the trough interacts with the low-PV air of the hurricane outflow anticyclone, the trough PV sharpens and becomes elongated (Fig. 8d). Between 0000 and 1200 UTC 12 July (Figs. 8c,e), the speed of the trough has reduced significantly. By 1200 UTC 12 July, the main part of the trough has moved east of the hurricane center while the base of the trough has been retarded in the vicinity of the hurricane (Fig. 8e), and the region of high PV is within approximately 400 km west of the storm center. The wrapping of high PV into the center of Bertha at 1200 UTC 12 July (Fig. 8e) can also be seen in the water vapor imagery in Fig. 7e as a dark, dry region with a similar structure. Strong outflow is now evident to the northeast of the storm center, and rapid intensification begins at 0006 UTC 12 July (Fig. 8d). As the hurricane intensifies, the convective bands become better organized (Fig. 7e), and a well-defined outflow develops to the south of the hurricane (Fig. 8e). The deformation of the trough PV in this case resembles that of a “trough fracture,” observed in Hurricane Diana (1984) by Dean and Bosart (1996) and in Hurricane Elena (1985) by Molinari et al. (1995). This has also been referred to as “wave breaking” by Thorncroft et al. (1993).

The vertical structure of PV (Fig. 9) is shown in a series of northwest–southeast cross sections through the center of Bertha for the same times as in Fig. 8. The approach of the upper-level trough and the scale reduction of the PV maximum as it interacts with the outflow layer anticyclone are clearly visible in Fig. 9. The initial PV maximum approaching the hurricane becomes narrower and shallower in depth by 1200 UTC 12 July (Fig. 9e). As the vertical shear increases in magnitude (Fig. 3) with the approach of the upper-level trough, the tilt of the PV maxima of the hurricane becomes more pronounced (Figs. 9a–e). Intensification begins at 0600 UTC 12 July and continues until landfall at 2000 UTC 12 July. During that period, the upper-level PV maximum superposes with the PV maximum of the hurricane (Fig. 9e). This superposition is also noted in the case of Hurricane Elena [see Fig. 2 of Molinari et al. (1995)]. In Bertha, the troughs are moving rapidly, and the destruction of the initial upper PV maximum occurs between observation times. The magnitude of the vertical shear drops after 1200 UTC 12 July, and the PV maximum of the hurricane becomes less tilted in the vertical while the storm intensifies (Fig. 9f).

The trough interaction described in Figs. 8 and 9 resembles the progression of events in the case of Hurricane Elena (1985), described by Velden (1987) and Molinari et al. (1995). In both cases, an upper-level trough undergoes synoptic-scale wave breaking (trough fracture) as it moves west to east to the north of the storm center. In the case of Bertha, a small-scale PV maximum (Fig. 8e) remains to the west of the hurricane center after the trough experiences wave breaking, while the main part of the synoptic-scale trough (Fig. 8f) continues eastward well to the north of the hurricane center.

The exact mechanism by which a hurricane intensifies after interacting with an upper-tropospheric trough is unclear. In the case of Hurricane Bertha, weakening appears to have been caused by increasing vertical wind shear associated with an amplifying short-wave trough/jet downstream of the center, resulting in exposure of the low-level eye. Inward propagating maxima of EFC during this period of strong shear are suggestive of a strong trough interaction. Enhanced upper-level eddy momentum flux convergence is related to low-level spinup of the tropical cyclone and enhanced vertical motion near the core (Molinari and Vollaro 1990). The favorable effects of PV advection may have overcome the negative impacts of increasing vertical shear, leading to intensification while the shear was still increasing. As an alternative, one can consider the near superposition of the low-level and upper-level PV anomalies seen in Fig. 9. As argued by Molinari et al. (1995) for the case of Hurricane Elena (1985), the observed reintensification may represent “constructive interference” of the lower- and upper-level PV anomalies (Hoskins 1990). As the PV anomalies come within a Rossby radius, the resulting combined pressure and wind anomalies are greater than those associated with each separate anomaly. The stronger wind and pressure anomalies are thought to then initiate the wind-induced surface heat exchange mechanism (Emanuel 1986; Rotunno and Emanuel 1987; Emanuel et al. 1994), leading to intensification of the hurricane (Molinari et al. 1995). Because the trough never crosses over the hurricane center, intensification is not reversed.

4. Summary

The evolution of a hurricane–trough interaction from the perspective of satellite imagery was described for the case of Hurricane Bertha. The similarity of signatures observed in water vapor imagery to those seen in gridded analyses of PV suggests that some types of trough interactions may be identified by the signatures described in the PV fields shown in HMK. This identification would only be possible in the case of superposition types of trough interactions, for which the majority of cases are found to intensify (HMK). The differences in PV signatures in distant-interaction-type trough interactions were found to be subtle and may be difficult to identify in satellite imagery. The ability to identify superposition types of trough interactions may prove to be a useful tool in forecasting intensity change during these specific interactions; however, more cases need to be studied before a reliable method can be established.