Danny made landfall as a minimal hurricane on the Alabama coast on 19 July 1997 after drifting over Mobile Bay for over 10 h. Danny’s unusually close proximity to the Doppler radar (WSR-88D) in Mobile provided an unprecedented view of the storm’s complex and dramatic evolution during a prolonged landfall event over a 1-day period.
Base reflectivity and velocity products were combined with aircraft reconnaissance information to detail the formation of concentric eyewalls and complete evolution of an eyewall replacement cycle. This highly symmetric hurricane then underwent a rapid asymmetric transition in Mobile Bay during which a small eyewall mesovortex developed adjacent to intense convection in the western eyewall. Radar-estimated rainfall increased dramatically during the asymmetric phase. Rates exceeded 100 mm h−1 for nine consecutive hours west of the center while precipitation nearly vanished to the east. Changes in the distribution of precipitation corresponded with changes in the low-level wind velocity structure.
A 25-h temporal composite of WSR-88D base velocities displayed axisymmetric intensification and contraction of Danny’s core during the eyewall replacement cycle. Later, the asymmetric phase was dominated by further contraction and intensification on the west side only. In the western eyewall, a persistent boundary layer wind maximum evolved and contracted to a radius of only 10–13 km from the center. Concurrently, eastside boundary layer winds diminished as the maximum winds rose to 1–1.5-km altitude and the radius expanded. Danny reached maximum intensity during this asymmetric phase in Mobile Bay with base velocities >44 m s−1 (85 kt) at 600-m elevation. Eyewall contraction in the meandering storm, combined with climatologically elevated SSTs in shallow Mobile Bay, probably played significant roles in Danny’s continued intensification there.
Discrepancies arose in determining Danny’s structural patterns, intensity, and evolution during landfall. Interpretation varied depending on the type of platform used to observe the storm. The continual sampling of the storm by the nearby WSR-88D provided detail not available from aircraft data alone. Doppler base velocities in the intense westside convection were much stronger than measured at flight level. Yet, the opposite was true in the storm’s drier east side.
Doppler radar showed that the westside base velocity maxima were confined to the boundary layer (600–700 m) and represented a slightly conservative estimate of maximum surface gusts recorded at Dauphin Island during the passage of Danny’s convective eyewall. Thus, winds probably were even stronger at boundary layer levels below the WSR-88D’s lowest scan elevation. The shallowness and persistence of Danny’s boundary layer velocity maxima stressed the need for accurate wind information in the lowest few hundred meters of a tropical cyclone’s eyewall for a better indication of the storm’s true intensity and near-surface wind velocities.
Landfalling tropical cyclones1 (TCs) produce enormous damage from the combined effects of wind, rainfall, and storm surge. Wind and rainfall patterns may vary widely from one storm to another (Foley 1995). Structural changes in a storm, such as those associated with eyewall replacement cycles, topography, or vertical wind shear, may produce intensity fluctuations and a spatial redistribution of strongest winds and heaviest rainfall. A TC’s inner core may exhibit nearly symmetric, or strongly asymmetric, features. Data sources, such as Doppler radar and aircraft, are important for an accurate analysis of the storm (Foley 1995). Because most hurricane damage occurs in the coastal zone, the specific structure and organization of the storm landfall is of great importance. Unfortunately, lack of data often precludes a detailed description and analysis of the surface mesoscale structure at landfall (Powell 1987, 1990;Powell et al. 1996).
Reconnaissance aircraft are a primary data source during TC landfalls in the United States. These reports of storm location and intensity are considered the most reliable data source (Foley 1995) and are cost effective (Gray et al. 1991). Aircraft are nearly always scheduled into landfalling storms; recent policy changes now allow overland flights (pilot’s discretion) at these times. Yet slow-moving storms, meandering near the coast for 1–2 days may present logistical problems for continuous aircraft monitoring, similar to that portrayed by Willoughby (1990). Observational gaps during landfall may result.
Coastal radars can provide frequent information in landfalling storms (Ellsberry 1995), such as accurate storm positioning when the eye is well developed. Airborne radars are mobile and can yield additional data at this and other times. The Weather Surveillance Radar-1988 Doppler (WSR-88D) may provide mesoscale details of wind and precipitation structure in the cyclone core and rainbands at landfall (Foley 1995). The land-based WSR-88D can continuously sample the inner-core features of a storm over long periods of time under certain conditions. First, WSR-88D wind estimates are possible only <100 km from the radar site (Foley 1995). Radar base beam elevations2 at a 100-km distance exceed 1500 m (i.e., the 850-hPa level). Much shorter ranges (<60 km) are required to observe the detailed boundary layer structure. Also, a multihour structural evolution of boundary layer winds and low-level precipitation near the storm center are only observed if the system moves slowly enough to remain in radar range.
Unfortunately, this combination of circumstances is rare. Really close approaches of slow-moving hurricanes to WSR-88Ds are infrequent events. Typically, hurricanes have forward speeds of ∼5 m s−1 along the north Gulf Coast, and even higher for many other U.S. locations (Neumann and Pryslak 1981). So, TCs passing <60 km from a WSR-88D rarely remain in this range for more than a few hours (see Stewart and Lyons 1996), and most are over land and weakening. Hurricane Danny (1997) is an exception.
Danny entered Mobile Bay as a minimal hurricane on 19 July 1997 (Fig. 1) and produced devastating rains over southwest Alabama (Pasch 1997). Remarkably, Danny’s center remained <100 km from Mobile’s WSR-88D for >48 h, and the strongly convective west eyewall remained <60 km from the radar for 12 h. This unique circumstance was a rare opportunity for nearly continuous WSR-88D observation of a landfalling TC’s core during dramatic structural changes.
This paper documents the structural evolution of Hurricane Danny’s inner core (<40 km from the center) before and during landfall on the Alabama coast. The nearby WSR-88D at Mobile (KMOB) observed massive changes in Danny’s low-level wind and precipitation during this time, including 1) an axisymmetric eyewall replacement and contraction, 2) development of a large convective asymmetry accompanied by extreme rainfall rates, 3) further asymmetric contraction adjacent to intense convection, and 4) persistent wind maxima near the top of the boundary layer in the heavily convective eyewall. Also, this study elaborates on Doppler-observed low-level wind maxima that were significantly stronger than maximum winds reported by aircraft at flight level.
Discussion of Danny’s features is limited to those readily identifiable in archive level II WSR-88D velocity and reflectivity data, U.S. Air Force Reserve (USAFR) C-130 aircraft reconnaissance vortex reports, and surface observations from Dauphin Island’s (DPIA1) Coastal-Marine Automated Network (C-MAN)3 site. The low-level aspects of the storm in or close to the boundary layer are emphasized. Section 2 details data and analysis techniques. The storm history prior to landfall is summarized in section 3. Section 4 details the structural morphology of Danny’s inner core precipitation and wind fields at landfall. Then, section 5 compares maximum winds observed by WSR-88D, aircraft, and DPIA1 in Danny’s eyewall. Finally, section 6 contains a summary of Danny’s structural evolution and a discussion of future research.
2. Data and analysis
The WSR-88D system and product characteristics are detailed in OFCM (1992). The National Weather Service (NWS) WSR-88D at KMOB operates in a volume-scan mode with 6-min intervals between scans, and with a wavelength of 10–11 cm. The beam pulse duration is 1.66 μs, providing a pulse length of approximately 500 m. Pulse repetition frequencies of 1014 and 1181 s−1 are used in Doppler velocity calculations. The horizontal and vertical beamwidth is 0.95° (0.83 km at 50-km range).
The WSR-88D archive level II digital data are manipulated using the WSR-88D Algorithm Testing and Display System (WATADS) (NSSL 1997). The KMOB NWS office is located in hills west of the city and has a station elevation of 64 m above mean sea level (MSL) (NCDC 1990). The WSR-88D antenna is adjacent to the office on a 24-m tower (J. Garmon, KMOB NWS, 1999, personal communication). Thus, an 88-m antenna height is added to WSR-88D beam elevations from WATADS to obtain elevations above MSL. The archive level II data are complete for 25 h between 2300 UTC 18 July and 0000 UTC 20 July 1997, except for a 1-h, 20-min gap between 0909 and 1029 UTC 19 July.
Conventional surface wind observations from DPIA1 are used to augment and verify the radar and USAFR reconnaissance information. The DPIA1 wind sensor is located 17.4 m above MSL (NDBC 1998) and provides, among other things, consecutive 10-min wind averages and the hourly maximum 5-s wind gust (Meindl and Hamilton 1992).
Two hodographs (Fig. 2) were generated near Danny from 1) a Slidell, Louisiana, rawinsonde sounding before Danny reached TC status, and 2) a storm-centered location where the hurricane’s circulation was factored out. For the latter, the Hurricane Research Division4 (HRD) averaged environmental winds from airborne Doppler and from coastal WSR-88Ds in a 90-km circle around Danny over multiple layers for 1915–2006 UTC 18 July. Both hodographs depict vertical wind shear and possible steering currents in Danny’s immediate environment.
The KMOB WSR-88D provided near-surface reflectivity and Doppler radial winds in Danny’s core. To facilitate evaluation of the storm’s structural evolution, radial snapshots of inbound and outbound base velocities were generated that represent instantaneous, nearly horizontal wind structure on each side of the hurricane. These snapshots were produced along radials extending outward 1) from the zero isodop5 through the storm’s center, and 2) through the inbound or outbound maximum base velocity (MBV). This radial, normal to the radar beam at the MBV location, extended >65 km outward (Fig. 3). Digital base velocity information and its distance from the storm’s central zero isodop were extracted outward at 3 m s−1 intervals. The point where the radial through the MBV intersects the central zero isodop was assumed to approximate the storm’s wind center. The Doppler winds presented only an incomplete two-dimensional picture of inner-core velocities and failed to provide meaningful winds where the flow was more perpendicular to the radar beam. These MBVs were manually extracted using WATADS for each volume scan between 2300 UTC 18 July and 2400 UTC 19 July. They were subjected to manual quality control and were required to maintain reasonable continuity in direction and speed with surrounding patterns in both space and time. Observations were deleted when suspected velocity aliasing produced unlikely or absurd wind patterns, a condition most common in pixels adjacent to range folding. But the limited extent of aliasing along a radial had a negligible impact. Fewer than 0.5% of MBVs were deleted over the 25-h archive period. Thus, a history of Danny’s radial base velocity patterns during landfall was produced.
Eye positions were identified both by WSR-88D and reconnaissance vortex information. Core feature interpretations in TCs from different WSR-88D products can differ. Stewart and Lyons (1996), after Wood and Brown (1992), discussed different center location interpretations in WSR-88D reflectivity and velocity products, including differences between the apparent and actual wind centers in WSR-88D velocity fields due to parallax errors and elliptical eyes. Given Danny’s unusually small core diameter (Da), combined with a 40–100-km distance (Ra) between the eye and radar site, a ratio of Ra/Da ≥ 1 generally was satisfied and any correction for parallax was small. A distortion of the eye caused by asymmetrical convection (discussed in section 4) could promote some center location discrepancies. Thus, the WSR-88D-derived radius of MBV (RMBV) may not exactly match Danny’s actual radius of maximum wind (RMW) derived from other sources. However, WSR-88D RMBVs appeared to be a good indicator of the true RMW and seemed to capture significant aspects of Danny’s evolution.
Hourly isohyetal analyses were completed for 1100 UTC 19 July–0300 UTC 20 July and encompassed the life cycle of a large eyewall convective system (EWCS). An EWCS was defined as analogous to a mesoscale convective system, but one that is embedded in highly rotational background flow of a TC’s eyewall (H. E. Willoughby 1999, personal communication). One-hour precipitation (OHP) products from the WSR-88D were used to compute the areal coverage of selected OHP rates.
3. Storm history prior to 19 July 1997
Hurricane Danny originated from a weak cold-core upper-tropospheric trough that migrated into the Gulf of Mexico from the eastern United States (Pasch 1997). A tropical depression began forming south of Louisiana between 14 and 16 July 1997. Early on 15 July, northerly vertical wind shear of 23.7 × 10−4 s−1 from 0 to 10 km (see Fig. 2) slowed intensification. A weak upper-level trough north of Tropical Storm Danny steered the system slowly northeastward on 17 July. By 0000 UTC 18 July, Slidell rawinsonde data (not shown) revealed a relaxation of the wind shear as the TC strengthened to a minimal hurricane and crossed the mouth of the Mississippi River.
Danny, a relatively small storm, developed an axisymmetric inner-core cloud pattern on 18 July 1997 (Fig. 4) within weak 0–10-km vertical wind shear of 14.1 × 10−4 s−1 (see Fig. 2). Strengthening continued and Danny’s central pressure fell >6 hPa during the day (Fig. 5); still the well-organized storm failed to intensify beyond a minimal category 1 hurricane.
Weak westerly steering currents of 0.5–3.0 m s−1 between 5- and 9-km altitude guided the storm on a slow eastward course on 18 July. The current study period began as Danny approached point B in Fig. 1 and moved within 100 km of the WSR-88D at KMOB.
4. Structure at landfall
Danny sequentially underwent a symmetric, then asymmetric, contraction of its inner core over a 19-h period before landfall. The classical symmetric contraction culminated with an intensifying single-eyewall storm 12 h later, followed by an asymmetric, highly convective contraction and further intensification over Mobile Bay. Data from aircraft and the KMOB WSR-88D chronicled the changes in precipitation and wind structure during this eyewall replacement and contraction.
a. Concentric eyewalls in a minimal hurricane
Eyewall replacement cycles, discussed by Willoughby (1990), begin with the formation of an outer convective ring (eyewall) around a preexisting inner eyewall. This outer eyewall, resulting from a coalescence of outer bands, eventually contracts and forces dissipation of the inner eyewall. During a classical axisymmetric contraction, a major hurricane initially may lose intensity as the inner eyewall collapses, pressure rises, RMW expands, and a flat wind profile develops (Willoughby et al. 1982, 1984; Willoughby 1990; Marks and Dodge 1997). However, renewed strengthening often occurs shortly thereafter as the outer eyewall continues its contraction over a 12–36-h period, provided landfall does not occur first. [Willoughby (1979), Willoughby et al. (1982), and Shapiro and Willoughby (1982) discuss the physics of the eyewall cycle.] Unlike Danny, most minimal hurricanes with concentric eyewalls exhibit little intensity fluctuation because the outer eyewall becomes as strong as the inner by the time the replacement is complete. Thus, Danny’s weakening during eyewall replacement represented an unusual situation.
Figure 6 shows the evolution of Danny’s precipitation pattern during landfall. Concentric eyewalls (Fig. 6a) first developed in Danny near point B at 2300 UTC 18 July, indicating the storm was at or near maximum intensity (Jordan and Schatzle 1961; Holliday 1977; Willoughby et al. 1982). Danny obtained a 984-hPa pressure minimum (the first of two) near point B shortly after the end of a Tropical Cyclone Windfields at Landfall experiment conducted by HRD (Dodge et al. 1999). Danny’s outer eyewall formed as the storm approached the coast, consistent with Hawkins (1983) and Willoughby (1990). A USAFR reconnaissance aircraft penetrated Danny at 2325 UTC 18 July, but did not report concentric eyewalls until subsequent penetrations at 0455 and 0612 UTC 19 July (see Table 1, rows B, C, and D). Concentric eyewalls in hurricanes are sometimes not identified by aircraft, even though postanalysis of radar loops indicates their presence (Willoughby et al. 1982); this was the case in Danny’s situation.
Concentric eyewalls also were found in WSR-88D velocities. A temporal profile of manually derived radials of inbound and outbound base velocities was created from 2300 UTC 18 July to 2400 UTC 19 July (Fig. 7). Both of Danny’s eyewalls displayed separate wind maxima between 2300 UTC 18 July and 0200 UTC 19 July, consistent with Willoughby et al. (1982).
b. Eyewall replacement morphology
Convectively driven contracting maxima of the swirling wind constitute the primary mechanism for intensification of hurricanes (Willoughby 1990). This type of intensification occurs in axially symmetric storms and may be manifest by eyewall replacement and contraction cycles (Willoughby et al. 1982). Danny experienced such a cycle during the 13-h period between 2300 UTC 18 July and 1200 UTC 19 July as it approached and entered Mobile Bay. Dramatic changes in precipitation and low-level wind structure were observed in WSR-88D images during this time.
Early on 19 July near point C, the storm turned slowly northward. The central pressure started rising and maximum MBVs began oscillating between western and eastern eyewalls with very little correlation to observed fluctuations in broad sea level pressure (SLP) gradients (−∇p) between the eye and the surrounding nonstorm environment. Westside MBVs decreased most noticeably in both eyewalls after 0200 UTC 19 July, while eastside velocities increased above 31 m s−1 during this time, even though Danny’s central pressure was rising. By 0400 UTC 19 July, westside velocities once again increased and were >36 m s−1 in the outer eyewall 1 h later. Central pressure falls were not observed until after 0600 UTC 19 July; thus, wind increases in Danny’s outer eyewall preceded pressure falls in the eye itself by 2–4 h. By 0700 UTC 19 July, MBVs once again increased to 36 m s−1 on Danny’s east side, while the west side weakened. This oscillation continued past 0830 UTC 19 July as winds once again decreased (increased) on the east (west) side.
Outer eyewall MBV increases several hours before central pressure falls and the oscillation of maximum winds from one side of the storm to the other deserves further discussion. Changes in −∇p may be local to the inner edge of the eyewall, and so not observed in coarser gradient measurements between the eye and environment. The local −∇p in the eyewall may tighten or relax to a greater degree than reflected by SLP changes in the eye’s center or outside the storm. Willoughby (1990) shows that the acceleration of the swirling wind is concentrated just inside the RMW and that wind profiles inside the eye become more U shaped during intensification. Thus, conceivably, pressures can fall between the eyewall and the center while staying constant or even rising at the center itself. Also, low-level winds converging into the eyewall’s diabatically induced updrafts can become supergradient, and become stronger than expected from −∇p alone.
After 0900 UTC 19 July, larger-scale gradients increased as SLP rose 3.5 hPa in 6 h west of the center and coincided with a decrease of 1 hPa in the eye at the beginning of Danny’s asymmetric transition and steady strengthening of westside winds. A simultaneous 2-hPa SLP increase to the east, however, was accompanied by a 5 m s−1 decrease in MBVs there (opposite of expectations).
Between 0455 and 0612 UTC 19 July, aircraft reported inner and outer eye diameters of 25 and 56 km. During this time, inner and outer wind maxima merged as >30 m s−1 base velocities fused across the earlier moat between the eyewalls, and the overall wind field intensified.
Reconnaissance reports (Table 1) show that the inner eye expanded from 19 to 28 km between 2325 UTC 18 July and 0612 UTC 19 July (Fig. 8). The expansion of the inner eye as the outer succeeds is not unusual. A similar eye expansion was not reflected in WSR-88D base velocities of the across-eye diameter of inner outbound-to-inbound 26 m s−1 isodops (26I) (Fig. 8). This isodop diameter remained close to 24 km between 2300 UTC 18 July and 0400 UTC 19 July, then contracted to 20 km by 0600 UTC 19 July. The east- and westside portions of the isodop diameter contracted in concert with the strengthening of respective outer-eyewall MBVs. The eastside 26I (see Fig. 8) contracted after 0100 UTC 19 July, followed by the westside 26I nearly 3 h later, coincident with the base velocity increases. Thus, Danny’s inner winds experienced some strengthening and contraction between 0455 and 0612 UTC 19 July even while the aircraft-reported inner-eye diameter continued to expand. The storm’s central pressure during this time remained constant at 987 hPa, after rising from its minimum 6 h earlier.
By 0726 UTC 19 July, aircraft reported an elliptical eye 19 km by 36 km wide. Only traces of the inner eye remained in reflectivity images (not shown) as it merged with the contracting outer eyewall. The western (eastern) RMBV contracted from 30 to 11 km (28 to 22 km) between 0600 and 1130 UTC 19 July (Fig. 7). During this time, the eastside RMBV remained larger than its western counterpart; however, the inner 26I radii displayed a pronounced symmetric contraction on both sides of the storm over the same period.
Danny developed a spectacular axisymmetric, single eyewall by 1039 UTC 19 July, evident in the precipitation and velocity patterns in Figs. 6b, 7, and 8. Heavy convection (≥50 dBZ reflectivity) was limited only to a small region in the southwest quadrant of the storm near DPIA1, while the rest of the storm’s precipitation was primarily stratiform. Contraction of the wind and precipitation pattern continued until a second 984-hPa pressure minimum occurred shortly before 1200 UTC 19 July. Aircraft reported a small circular eye only 11 km in diameter and both the 26I radii were symmetric at ≈7 km either side of the center (≈15 km diameter). The eye was near point G and was almost completely enclosed in Mobile Bay, a warm tidal estuary 48 km by 37 km.
c. Asymmetric displacement and contraction of Danny’s center
A second episode in Danny’s prolonged landfall began after 1200 UTC 19 July 1997. A dramatic asymmetry developed in both base reflectivity and velocity as the storm changed from a stratiform precipitation pattern to a heavily convective one. The eye contraction continued on the west side only and coincided with a dramatic increase in convection there. Reflectivities rapidly diminished on the stratiform east side. Westside MBVs consistently exceeded 38 m s−1 and eventually reached >40 m s−1. This was opposite to the pattern of bands normally favoring a TC’s east side (Willoughby et al. 1984).
Surface streamline analyses (not shown) using local observing stations showed strong (very weak) confluence of northerly (southerly) winds into the western (eastern) eyewall. The strongest convection persisted along the meridional extent of Mobile Bay. Also, an elevated low-level jet (see section 5a) occupied the rain-free east side of Danny at this time and provided a conveyor belt of Gulf air to the intense westside convection. The environment contained only weak vertical wind shear and probably was not a factor in Danny’s asymmetry. But the origin of this asymmetry requires additional study and is beyond the scope of this paper.
Precipitation rates and their spatial distributions experienced significant changes. Areal coverages of various WSR-88D OHP rates6 appear in Fig. 9 immediately preceding the development of westside convective asymmetry. Precipitation rates and coverages were low as the storm entered the bay. The area bounded by the 25–50 mm h−1 (1–2 in. h−1) isohyet was only 100–200 km2 around 1100–1200 UTC 19 July and coincided with the storm’s greatest axial symmetry. Only a 25 km2 area experienced OHP rates over 50 mm h−1 at this time.
As the storm became asymmetric and the intense EWCS developed in the west eyewall, these rates and coverages steadily increased. By 1400 UTC, the 25–50 mm h−1 isohyet encompassed 580 km2 and maximum rates exceeded 100 mm h−1 (4 in. h−1) over a small area in the southwest eyewall. The area bounded by the 25–50 mm h−1 isohyet continued increasing and ultimately reached 1400 km2 by 2100–2200 UTC 19 July, shortly after the eye made landfall. Precipitation rates in the west eyewall were >100 mm h−1 from 1300 to 2200 UTC and expanded to a maximum coverage of 130 km2 shortly after Fig. 6c. Composite reflectivities (not shown) at times exceeded 65 dBZ here. Danny’s western convective asymmetry (Fig. 6d) persisted for at least another 24 h and produced disastrous flooding as the storm drifted slowly inland over southwest Alabama.
The strong convection probably influenced the behavior of Danny’s eye over Mobile Bay. In a well-organized TC, the dynamic (wind) center nearly always lies within the eye depicted by radar reflectivity (Foley 1995). Willoughby et al. (1984) define a storm’s dynamic center as a region of calm winds encircled by closed streamlines. Sometimes the dynamic center migrates across the eye toward intense asymmetric convection in the eyewall. Also, mesovortices can form adjacent to persistent heavy rain in a sector of the eyewall (Gamache et al. 1997). In Danny’s case, the dynamic center became very diffuse after 1200 UTC as the intense westside EWCS developed. An eyewall mesovortex (EWMV) developed in western sections of the eye after 1400 UTC 19 July, adjacent to the EWCS. This small EWMV is clearly visible in Fig. 6c and continued to contract. The EWMV’s track between 1504 and 2013 UTC, as estimated from WSR-88D data, is represented in Fig. 10 by a series of storm positions labeled 1–4. This EWMV may have originated 1) from a further contraction of the 11-km-wide eye reported by reconnaissance aircraft near point G (see G1 and G2 in Table 1), or 2) as a separate entity within the eye. It is also unclear as to whether the EWMV represented a single vortex, or a cyclonic sequence of edge-of-the-eye vortices that amplified (weakened) upon approaching (leaving) the westside EWCS.
As the EWMV developed, the eye to the east near point H (see Fig. 10) became more diffuse. Aircraft observations at 1410 UTC indicated that although the pressure remained at 984 hPa, the eye had expanded by 7 km in ≈1 h. A corresponding expansion of the 26I diameter also was evident in Danny’s core. On the east side of the eye, a broad and diffuse expansion of the RMBV continued for at least another 10 h, whereas the convective west side’s RMBV quickly resumed a multihour contraction after 1500 UTC.
The eye remained recognizable in the base velocity pattern, but became difficult to recognize in reflectivity images as eastside precipitation eroded. The EWMV was most apparent in the reflectivity images and contracted to a diameter of only 4–6 km (2–3 n mi) before landfall on the bay’s eastern shore.
The EWMV was not so distinct in the base velocity profiles; however, the WSR-88D mesocyclone algorithm repeatedly indicated small mesocyclone and 3D correlated shear symbols in the vicinity of the EWMV’s reflectivity center. But the algorithm often lost the EWMV signature possibly because of its superposition upon the general rotation around Danny’s diffuse eye. The algorithm required the convective-scale rotation to be symmetric, and Danny’s was not. Some volume scans indicated overlapping large (eye) and small (EWMV) mesocyclone symbols. Operationally, forecasters should closely investigate small (possibly sporadic) Doppler-indicated mesocyclones near the eyewall for the possible existence of EWMVs offset from the eye center.
Danny strengthened during the ongoing contraction of the EWMV over Mobile Bay. H. E. Willoughby (1999, personal communication) suggests that once Danny’s strong eyewall winds moved over shallow and warm Mobile Bay, increased surface heat flux (Emanuel 1995, 1997) probably sustained the EWCS. Air converged into the EWCS from both the concave and convex sides in response to convective heating. Surface observations adjacent to Mobile Bay, combined with WSR-88D base velocities, clearly showed pronounced confluence of winds here. Willoughby speculates that vorticity stretching in the EWCS updraft and PV generation below the level of maximum condensational heating may have forced the vortex to reform or propagate westward. A weak eastward steering current over the central Gulf Coast, south of a weak upper-tropospheric trough (at 1200 UTC 19 Jul), opposed the westward vortex propagation, resulting in a nearly stationary storm. Thus, the westside EWCS remained over Mobile Bay for several more hours.
Prior to Danny’s arrival, sea surface temperatures (SSTs) in the northern Gulf (0000 UTC 18 Jul) were 29°–30°C at 1-m depth (buoys 42040 and 42007), and 30°C at the mouth of Mobile Bay [DPIA1 at 0.5 m below mean low water (MLW)].7 Yet, DPIA1’s SSTs in summer are more representative of Gulf SSTs and are generally cooler than SSTs in the bay’s interior.8 For instance, interior bay SSTs were not available during Danny, but a recent comparison on 6 August 1999 provides an example of warmer waters in the bay. Here, the 24-h average SST was 33.2°C near Cedar Point9 (8 km northwest of DPIA1), compared to 31.5°C at Gulf buoy 42040 (119 km south of DPIA1).10 Bay SSTs reached 35°C that day, compared to 32.5°C in the Gulf. Cooler Gulf water failed to invade the bay during Danny’s landfall because the eye crossed east of the bay’s mouth and north winds produced a 0.6-m tide deficit (Pasch 1997) at the northern end.
The bay averages 3 m in depth, except for a 12-m-deep ship channel extending the bay’s length, so subsurface vertical mixing under Danny’s eyewall probably failed to cool the water. However, surface heat transfer and cold rain (∼1000 mm) helped reduce DPIA1 SSTs to 24°C by 0100 UTC 20 July, a 6°C decline during Danny’s passage. Possibly as the bay cooled, westward vortex regeneration succumbed to eastward steering and Danny moved slowly onshore.
The westside MBVs exceeded 39 m s−1 almost continuously for 9 h beginning around 1130 UTC near point G in Mobile Bay. Danny reached peak intensity between 1804 and 1824 UTC, when numerous MBVs exceeded 44 m s−1 and the westside 26I contracted to a minimum of 6 km (3 n mi).11 No corroborating aircraft observations were available from 1500 to 2000 UTC 19 July; however, Danny’s central pressure probably was <984 hPa prior to final landfall.
The EWMV crossed the coast near Mullet Point, Alabama, on the bay’s eastern shore (see point 4), between 1900 and 2000 UTC 19 July (Pasch 1997), and a rapid decrease in outbound base velocities and expansion of the RMBV occurred west of the center. Once over land, the EWMV lost identity. East of the EWMV landfall point, the elongated eye of Danny moved inland closer to Weeks Bay (see point I).
The most severe tree damage inflicted by Danny occurred near the EWMV’s landfall point between Fairhope and Weeks Bay (NCDC 1997). Strong near-surface winds, undetected by the WSR-88D, likely occurred on the north and northwest sides of the EWMV and very likely accounted for much of this tree damage. A site survey by the author found that trees between Fairhope and Weeks Bay fell primarily from the north-northeast through east-northeast, indicating a wind direction nearly perpendicular to the WSR-88D beam radials; thus, no strong base velocities were observed there at landfall. All Doppler-observed hurricane-force MBVs near the EWMV were westside outbound winds located well offshore over Mobile Bay.
The eastside wind pattern during landfall was considerably weaker and less defined. Base velocities to 31 m s−1 remained in a broad 20-km-wide zone over the eastern sector of the storm during the entire EWMV episode. Diffuse eastside base velocity patterns resulted from lack of radar scatterers in that sector of the storm (see Fig. 6d), which precluded RMBV estimates after 1748 UTC 19 July. But the eastside 26I close to the center was observable for several more hours.
5. Comparison of WSR-88D base velocities to maximum aircraft and surface winds
How intense was Danny over Mobile Bay and the Alabama coast? Because of Danny’s close proximity to the KMOB WSR-88D, base velocities were sampled at very low levels within Danny’s inner core. At 0.5° scan elevations, the base velocities were several hundred meters below aircraft flight-level winds. Figure 11 shows a comparison of wind sampling altitudes from aircraft and the KMOB WSR-88D within Danny’s RMW. Also, some winds at 17.4-m elevation above MSL (NDBC 1998) were available from the DPIA1 C-MAN in Danny’s eyewall.
a. Maximum wind comparisons
Reconnaissance flights by USAFR aircraft reported the maximum wind encountered at flight level, and the bearing of maximum wind from Danny’s center. Although stronger winds possibly resided in sections of Danny that were not observed by aircraft or radar, the aircraft’s maximum 10-s (1-km spatial average) flight-level winds provided a quick comparison against the corresponding WSR-88D MBVs and RMBVs from 2300 UTC 18 Jul to 2400 UTC 19 July. Maximum aircraft winds were posted over the Doppler base velocity profile in Fig. 7 (and provided in Table 1), but were not exactly collocated with each other. Instead, the implication was only that the aircraft and WSR-88D winds were observed in the same half semicircle12 of the storm and at the same radius from the center; therefore, actual azimuth angles between WSR-88D and aircraft winds differed.
Difficulties arose when comparing flight-level winds to WSR-88D base velocities, partly due to temporal gaps between aircraft observations (as in Danny between 1410 and 2003 UTC 19 Jul). Large differences can occur between flight-level and surface winds, especially when flight level is above the top of the boundary layer (Powell 1987). The maximum horizontal wind speed in a TC is usually at between 1000 and 1500 m; however, airborne Doppler radar data from Powell and Black (1984) shows maximum winds near rainbands can be lower (300–500 m). Likewise, WSR-88D maximum velocities within TC Ed were at the 400–500-m level (Stewart and Lyons 1996).
Thus, aircraft flying at 1500 m (850 hPa) may be above the strongest winds and may underrepresent a storm’s intensity closer to the ground. A nearby Doppler radar (if available) may provide better estimates of maximum winds if base velocities are sampled below aircraft flight level, but WSR-88D base velocities below 1500 m are only possible <100 km from the radar site. Within 50 km, 0.5° beam elevations are <650 m. There, however, is no guarantee that WSR-88D velocities, even at these low levels, correspond to the maximum tangential winds, since stronger winds could be at lower levels, or may not be oriented directly along beam radials. Nonetheless, WSR-88D base velocities in Danny over Mobile Bay were stronger than winds found by aircraft and consistently lay at altitudes below flight level.
When a TC makes landfall, reconnaissance aircraft generally fly above the boundary layer because of overland flight restrictions (Powell 1990; Willoughby 1990). The 850- and 700-hPa levels represent the usual overland flight levels. During Danny’s landfall, aircraft flew at the 850-hPa level (roughly 1300–1500 m MSL), while the WSR-88D 0.5° beam elevations in the MBVs on either side of Danny’s eye ranged from 600 to 800 m. In the present study, little attempt was made to examine radar observations at aircraft flight level, or use equivalent sampling intervals for aircraft and WSR-88D. The comparison between the two platforms was made strictly to examine differences between aircraft winds and WSR-88D base velocities. Collocation was another limitation because of sharp wind changes over small time–space increments (Powell et al. 1996).
Figure 7 shows that in spite of the comparison problems just discussed, radar and aircraft winds agreed reasonably well in some locations of the storm. Maximum flight level and WSR-88D winds compared fairly well to each when the storm was symmetric and less convective13 (prior to 1100 UTC). Less agreement occurred after 1200 UTC when the western convective asymmetry developed and base velocities were weaker (stronger) to the east (west) than the 850-hPa flight-level winds. These differences deserve further discussion, starting with the west side.
Doppler MBVs were consistently >39 m s−1 (75 kt), in the convective western semicircle, while aircraft there found maximum winds of only 31–35 m s−1 (60–67 kt). After the aircraft departed the storm at 1410 UTC, MBVs increased and consistently exceeded 41 m s−1 (80 kt) from 1500 to 1900 UTC with MBVs >44 m s−1 (85 kt) on several occasions. This period corresponded with the development of the EWMV.
A vertical cross section of radial velocities (Fig. 12a) across the EWMV (Fig. 12b) shows winds >41 m s−1 were confined to the western eyewall’s lowest levels (i.e., only at the 0.5° scan elevation) and were significantly weaker at higher elevations. The westside RMW and convection tilted outward from the center with height, similar to Powell et al. (1996), Marks (1985), and Jorgensen (1984a,b). Much less tilt of the RMW was observed in the nonconvective east side. Powell (1990) indicates that most research flights into storms collect data at flight levels above 1500 m. Recently, Global Positioning System dropsonde data collected from hurricane eyewalls have detected the existence of low-level wind maxima as low as 200 m above MSL (Dodge et al. 1997, 1999), similar to these observed by radar in Danny’s western eyewall.
The eastside velocity maximum was characteristic of an elevated low-level jet with >31 m s−1 (60 kt) winds at 1300 m MSL above winds <28 m s−1 (55 kt) near (and possibly below) 600-m elevation. Eastside winds from aircraft, sampled at the 1300–1500-m elevation, probably represented winds in this elevated jet core that were not observed at lower levels in the eastside base velocities. The elevated jet wrapped cyclonically around the east and north sides of the storm (not shown) and fed into the large westside EWCS over Mobile Bay.
b. Boundary layer wind representativeness
Hypothetically, if aircraft flew 1) along the velocity cross section’s path in Fig. 12b, and 2) at the typical 850-hPa level in Fig. 12a, the aircraft would observe stronger winds in the elevated jet core near 1500 m, but would miss the weaker winds in the boundary layer near 600 m. Powell (1982) found that 10-m-level gusts (mean winds) at coastal stations may be estimated as 80% (56%) of the flight level mean winds at 500–1500 m, and that maximum surface gusts occurred in heavy convection. In Hurricane Andrew at landfall, Powell et al. (1996) found that surface gusts were 64%–103% of the nearest flight-level 10-s wind maximum. The low end of this range was more appropriate for Danny’s east side, so 64% of the flight-level wind yielded surface gusts of only 21–23 m s−1 over open flat terrain. Indeed, a visual inspection by the author to the east of the landfall point found very little tree damage. Unfortunately, no corroborating wind observations were available.
The velocity profile was quite different on the western side. The same aircraft at 850 hPa would find maximum winds close to the 35 m s−1 WSR-88D velocities depicted at 1500 m in Fig. 12a. Thus, >41 m s−1 base velocities near 600-m altitude would not be sampled by aircraft. Actually, the western eyewall’s MBV at 600 m was 44 m s−1 at only a 10-km radius from the EWMV, and was 125%–130% of the maximum outbound velocity at flight level.
How representative of surface winds were these flight-level and base velocity wind maxima? To the author’s knowledge, DPIA1 was the only archive-capable surface-based anemometer to experience the strong west eyewall of Danny along the Alabama coast. Pasch (1997) mentioned:
The Dauphin Island C-MAN on the end of the island measured 10-min average winds of 65 kt (33 m s−1) at 1145 UTC and a gust to 88 kt (45 m s−1) 21 min earlier. Interestingly, the Mobile WSR-88D radar showed that around these times, the strongest eyewall convection was occurring in this vicinity over the southwest quadrant of the hurricane. At 1139 UTC, aircraft reported maximum winds of 64 kt (33 m s−1) at the 850 mb flight level in the southwest quadrant. Thus, surface and flight-level winds were about the same in this highly convective region of the hurricane.
Observed winds at DPIA1 were compared to surface wind estimates derived from aircraft and WSR-88D data. First, the 33 m s−1 aircraft wind mentioned by Pasch closely matched the 35 m s−1 WSR-88D velocity at 1500 m over DPIA1. Here, the upper 103% value of Powell et al.’s gust range14 produced surface gust estimates of only 34 m s−1 from the reported 33 m s−1 aircraft wind. Thus, the estimated surface gusts were 11 m s−1 (33%) too low when compared to the maximum 5 s−1 gust of 45 m s−1 at DPIA1 (Meindl and Hamilton 1992). Second, the comparison between WSR-88D MBVs and the DPIA1 gust was closer. The MBVs in the southwest eyewall from 1100 to 1200 UTC 19 July were 38–41 m s−1 at 600 m, only 4 m s−1 (10%) short of the DPIA1 gust.
A much better aircraft estimate of DPIA1 surface winds occurred when the maximum flight-level wind (a 1-km spatial average over 10 s) was used to estimate sustained surface winds (in this case, a 10-min average) instead of 5-s surface gusts. Powell et al. (their appendix A) stated that the 103% of flight level value also represents the upper end of the sustained (1 min) surface wind range (57%–103%) in Andrew: sometimes the maximum 1-min surface wind in Andrew was just as strong as the maximum 10-s wind observed nearby at flight level. Danny’s surface wind gusts were much stronger over Mobile Bay and Dauphin Island than those measured at flight level. Indeed, the NHC increased Danny’s maximum sustained wind estimate from 33 to 36 m s−1 (gusting to 44 m s−1) in their 1500 UTC 19 July advisory after these reports were received from DPIA1. Thus, the 39–44 m s−1 MBVs in Danny, observed for several hours at 600 m, may have represented the maximum marine-exposure surface wind gusts in the western eyewall.
Danny and Andrew were not the only landfalling storms with stronger near-surface wind gusts than those indicated at flight level. During Hurricane Alicia’s 1983 landfall in Texas, surface gusts measured by the Coast Guard ship Buttonwood exceeded flight-level (1500 m) peak gusts during each of the four offshore aircraft passes (Powell 1987). Danny’s situation was unique among these three landfalls because a nearby WSR-88D recorded the low-level wind maxima.
The present discussion indicates that, while in the bay, Danny’s heavily convective (nonconvective) westside (eastside) eyewall contained near-surface wind gusts significantly greater than (probably much less than) 100% of maximum flight-level winds. Damage inspections by the author and NWS personnel over Dauphin Island, and between Weeks Bay and Fairhope (NCDC 1997), definitely confirm stronger surface winds in Danny’s western semicircle; surface winds and associated damage appear to be much less to the east.
6. Summary and discussion
The slow motion of Hurricane Danny, combined with its proximity to the KMOB WSR-88D, provided an exceptional opportunity for close, low-level sampling of the storm’s inner core at landfall over a 1-day period. Similar sampling continuity would be difficult or impossible by aircraft, or in a faster-moving storm. Many interesting properties were observed in Danny during this time.
Danny experienced concentric eyewalls and a symmetric eyewall replacement/contraction offshore, followed by an asymmetric and heavily convective contraction over Mobile Bay.
Danny weakened during its replacement cycle, a behavior unusual for a minimal hurricane experiencing eyewall succession. This weakening is more characteristic of major storms.
Early in the symmetric replacement cycle, the central pressure rose and eyewall radial velocities initially weakened. However, as the pressure continued to rise, the outer eyewall MBVs strengthened and began oscillating between east and west sides of the storm with very little correlation to environmental SLP changes. The central pressure fell only after the strengthening outer eyewall winds contracted and merged with the inner eyewall. Thus, central pressure reduction in the eye lagged 2–4 h behind a general increase in central core wind intensity.
Eyewall contraction and a sustained 6-h pressure fall culminated in an axisymmetric single-eyewall storm composed mainly of stratiform precipitation.
Danny became asymmetric and extremely convective over shallow, warm Mobile Bay. An intense EWCS persisted over the bay for 10–11 h west of the eye. As this western asymmetry developed, 25–50 mm h−1 (1–2 in. h−1) precipitation rates expanded from 200 to 1400 km2. Extreme rainfall rates of >100 mm h−1 (4 in. h−1) persisted for nine consecutive hours, reached maximum coverage just before landfall, and occurred for up to 4 h after landfall. Catastrophic flooding of bayous and rivers adjacent to Mobile Bay occurred.
An EWMV formed adjacent to the westside EWCS and coincided with continued asymmetric contraction and intensification there. Westside RMBVs contracted to 11 km (6 n mi) prior to landfall as WSR-88D base velocities at 600 m peaked at 45 m s−1 (87.4 kt); this period probably coincided with Danny’s greatest intensity and lowest central pressure.15 The greatest tree damage occurred near the EWMV’s landfall point on Mobile Bay’s eastern shore.
The eastside RMBV expanded and boundary layer winds weakened before landfall as precipitation there decreased and strongest winds evolved into a broad elevated low-level jet.
The WSR-88D showed that Danny’s strongest winds persisted for several hours in the boundary layer (600–700 m) within the western EWCS, and were observed several hundred meters below much weaker flight level (1300–1500 m) winds there. These WSR-88D MBVs indicate that Danny was significantly stronger in the bay than aircraft flight-level winds suggested.
Winds at flight level (1300–1500 m) in the nonconvective east side of the storm were stronger than WSR-88D winds in the boundary layer (600–700 m).
A 45 m s−1 (88 kt) gust at DPIA1 in Danny’s southwest eyewall was only 10% stronger than 600-m Doppler MBV values there, but was 33% stronger than the maximum wind measured by aircraft at 1300–1500 m. Thus, the WSR-88D MBVs at 600-m elevation probably represented slightly conservative estimates of Danny’s maximum near-surface gusts in the storm’s EWCS.
Results from this study show the importance of sampling hurricane eyewall boundary layers for an accurate assessment of surface wind speeds and overall storm intensity. Danny’s heavily convective (nonconvective) westside (eastside) eyewall contained near-surface wind gusts significantly greater than (probably much less than) 100% of maximum flight-level winds. More investigation of eyewall boundary layer wind maxima at levels closer to the surface is warranted.
Danny demonstrated that, given favorable environmental conditions, a slow-moving small storm may maintain, or even increase, intensity for several hours over a shallow confined tidal estuary, as long as convective portions of the eyewall remain over water with SSTs possibly higher than those in the open ocean.
The origin of Danny’s asymmetry, EWMV development and evolution, and the influence of the westside EWCS on the intensification and motion of Danny while in Mobile Bay need further investigation. Environmental influences, such as coastline proximity and possible low-level wind surges, on concentric eyewall initiation warrants further study. The influences of vertical shear, boundary layer discontinuities, and coastal topography on Danny’s structural transition, vortex maintenance, and resulting rainfall patterns pose interesting problems for further research. Further examination of Danny’s extreme rainfall is in progress.
I wish to express my deep appreciation to Jeff Medlin and the National Weather Service Office in Mobile, Alabama, for assisting in data acquisition and display capabilities. I am also indebted to Jeff for creating Fig. 9. Dr. Hugh Willoughby provided many helpful suggestions for improving the initial manuscript and provided invaluable insight regarding certain aspects of Danny’s structural evolution. Dr. Mark Powell provided some helpful comments regarding flight-level to surface wind comparisons in hurricanes. Finally, I thank two anonymous reviewers and Dr. Aaron Williams for their constructive comments.
Corresponding author address: Dr. Keith G. Blackwell, Department of Geology, Geography, and Meteorology, LSCB 136, University of South Alabama, Mobile, AL 36688.
Tropical cyclone usage here refers to systems of tropical depression through hurricane strength.
This base elevation represents the lowest elevation within the volume scan and denotes a 0.5° beam angle.
DPIA1 is the only surface observation platform with archive capability to experience Danny’s landfalling eyewall.
Thanks to P. Dodge and F. Marks, NOAA/AOML/HRD.
Isodops represent contours of constant Doppler velocity; the zero isodop bisects the wind center in Danny’s eye.
WSR-88D precipitation rates were derived from the tropical Z–R rainfall algorithm.
Sensor information supplied by P. Newkirk, National Data Buoy Center, Stennis Space Center, Mississippi.
Dr. W. Schroeder at the Dauphin Island Sea Lab concurs with this statement based on his own studies.
SSTs taken every 15 min at MLW, courtesy C. Yanny, U.S. Army Corps of Engineers, Mobile, Alabama.
SSTs at DPIA1 were unavailable in 1999 due to a damaged sensor.
Some of this wind increase may be attributable to lower radar 0.5° beam elevations while the storm is in the bay compared to when the storm is in the Gulf.
Semicircles are bounded by the radial extending from the WSR-88D location and through the storm center.
The 36 m s−1 aircraft reconnaissance wind at 2325 UTC was observed at the 700-hPa flight level, while all successive winds were at the 850-hPa level (i.e., closer to the WSR-88D 0.5° scan elevation).
The upper end of the 64%–103% gust range is used because of strong convection over DPIA1 and stronger Doppler base velocities at 600 m as compared to velocities at flight level.
Independent confirmation of central pressure by reconnaissance aircraft was not available at that time.