Surface Observations of Landfalling Hurricane Rainbands

454 MONTHLY WEATHER REVIEW VOLUME 133

G. D. SKWIRA National Weather Service, Lubbock, Texas

J. L. SCHROEDER AND R. E. PETERSON Wind Science and Engineering Research Program, Texas Tech University, Lubbock, Texas

(Manuscript received 29 October 2003, in final form 20 May 2004)

ABSTRACT

This study examines the rainband-scale fluctuations of various meteorological parameters for Hurricanes Bonnie (1998) and Dennis (1999). Hurricane rainbands, identified by Next Generation Weather Surveil- lance Radar-1988 Doppler (NEXRAD WSR-88D) data, are examined shortly after making landfall on the North Carolina coastline. Additionally, Wind Engineering Mobile Instrumented Tower Experiment (WEMITE) data are exploited to provide a unique look into the surface structure of the captured rainbands. The observed meteorological data suggest equivalent potential temperature minima and decreasing hurricane-relative inflow to be commonly associated with intensifying or mature landfalling hurricane rainbands. Available vertical thermodynamic profiles suggest the source of the lower equivalent potential temperature air to range anywhere from 750 to 869 m above the surface, assuming no entrainment. Additionally, no discernable trend in wind speed is found to accompany the rainband’s passage.

1. Introduction rection, higher wind speeds, and temperature decreases. a. Background and objectives More recently, a number of studies (Barnes et al. Hurricane rainbands produce large amounts of rain- 1983; Willoughby et al. 1984; Barnes and Stossmeister fall and often lead to very costly and potentially fatal 1986; Powell 1990a,b; Barnes et al. 1991; and others) flooding. Additionally, since rainbands occupy such a have been completed on various hurricane rainbands large area within tropical cyclones, it is not surprising utilizing aircraft data on hurricanes located well out to that they may play an important role in the evolution sea. Many of these studies found decreasing hurricane- and intensity of the hurricane they are associated with relative inflow and decreasing equivalent potential tem- (Barnes and Stossmeister 1986). Hence, a better under- perature in conjunction with the hurricane rainbands. standing of rainbands will lead to a more complete un- Additionally, some of the studies found hurricane rain- derstanding of hurricanes as a whole. The objective of bands to be favored regions for enhanced wind speeds. this paper is to quantify the surface kinematic and ther- Overall, the studies suggested that the magnitude of the modynamic fields within landfalling hurricane rain- boundary layer modification (interruption in inflow, bands in order to gain a more complete understanding equivalent potential temperature decrease, and wind of rainbands as they impact the landmass they encroach speed increase) was directly related back to the amount upon. of convective activity, with more convectively active rainbands producing greater boundary layer modifica- b. Historical approach tion. Moreover, it has been suggested that this boundary layer modification, if spatially and temporally large Wexler (1947) and Ligda (1955) identified a number enough, could well have a direct impact of the overall of hurricane rainbands that tracked over Florida and intensity of the parent hurricane. determined, from available surface observations, that The purpose of this study is twofold. First, this study they contained regions of heavy rain, veering wind di- expands upon the previous knowledge by examining surface data across various rainbands in an effort to determine whether the more recent results uncovered Corresponding author address: G. D. Skwira, National Weather with aircraft data also apply to the observation made at Service, 2579 S. Loop Suite 100, Lubbock, TX 79423-1400. the surface. Second, this study examines the rainbands E-mail: [email protected] as they make landfall. The hope is to better quantify the

MWR2866 FEBRUARY 2005 SKWIRA ET AL. 455 rainband properties where the greatest human impact is experienced, near the surface over land. c. Rainband definition Rainbands were identified through radar reflectivity data. Areas of relatively high base-level reflectivity (32.5ϩ dBZ), bordered by lower reflectivity on both sides, with a length substantially greater than its width, were chosen. This criterion was chosen to pull out the heavier rain while excluding the lighter stratiform precipitation within the rainbands. In addition to the above stipulations, only rainbands that passed over the Wind Engineering Mobile Instrumented Tower Experiment (WEMITE) towers were chosen for further analysis to ensure the availability of high-resolution surface data. FIG. 1. Plot of the WEMITE 1 and the KMHX radar locations The reflectivity criterion was modeled after the 25-dBZ for Hurricane Bonnie. Also displayed is the best-fit track for Hur- cutoff employed by Barnes et al. (1983). The arbitrary ricane Bonnie [day/time (UTC)]. criterion, for this study, was raised from 25 to 32.5 dBZ to provide a better differentiation between rainbands. riod in which WEMITE data were collected. Data were Using the lower 25-dBZ criterion would have included not collected during the landfall of Tropical Storm most of the precipitation occurring with the hurricanes Dennis, but were gathered during Hurricane Dennis’ near the WEMITE towers; therefore, the lower cutoff initial pass off the North Carolina coastline. would have precluded the differentiation of individual rainbands. b. Radar Level II radar data were obtained from the National 2. Data sources Climatic Data Center (NCDC; available online at http://www.ncdc.noaa.gov/). More specifically, radar a. Wind Engineering Mobile Instrumented data from the Morehead City, North Carolina Tower Experiment (KMHX), Next Generation Weather Surveillance Ra- WEMITE is a project developed by the Texas Tech dar-1988 Doppler (NEXRAD WSR-88D) were ac- University Wind Science and Engineering program quired from 1001 UTC 26 August to 0633 UTC 27 Au- (Schroeder and Smith 2003). WEMITE was initiated in gust 1998 and from 0338 UTC 30 August to 2122 UTC 1998 and at the time consisted of one mobile tower, 30 August 1999 for Hurricanes Bonnie and Dennis, re- approximately 10 m tall, instrumented to collect tem- spectively. KMHX was the nearest functioning perature, pressure, and relative humidity information at NEXRAD with respect to the WEMITE tower location the 1.2-m level, as well as wind speed and direction for both hurricanes. The KMHX radar data were the information at multiple heights (10.7-, 6.1-, and 3.0-m primary source for identifying rainbands as they passed levels). The top anemometer height was lowered to 9.1 over the WEMITE towers. m after the 1998 season. In 1999 WEMITE also gained a second 10-m mobile tower capable of sampling meteorological parameters similar to those of the original tower; the second tower measures wind speed at four heights (2.1, 4.0, 6.1, and 10.1 m). The weather data were acquired at 5 Hz in 1998 and at 10 Hz since then (J. L. Schroeder 2003, personal communication). The goal of WEMITE was to place the tower(s) in the vicinity of landfalling hurricanes with the objective of gathering high-resolution temperature, pressure, humidity, and wind data within the hurricane boundary layer. At the time of this paper, 17 tropical storms and/ or hurricanes have been intercepted. Data gathered from Hurricanes Bonnie (1998) and Dennis (1999) are employed in this paper. Figures 1 and 2 depict the location of the WEMITE tower(s) during the passage of FIG. 2. Plot of the WEMITE 1, WEMITE 2, and the KMHX Hurricanes Bonnie and Dennis, respectively, along with radar locations for Hurricane Dennis. Also displayed is the best- the storm’s best-fit track. Table 1 displays the time pe- fit track for Hurricane Dennis [day/time (UTC)].

Unauthenticated | Downloaded 09/25/21 07:38 AM UTC 456 MONTHLY WEATHER REVIEW VOLUME 133

TABLE 1. Period of data collection for WEMITE platforms during Hurricanes Bonnie and Dennis. Storm Platform Start time End time Bonnie WEMITE 1 0915 UTC 26 Aug 1998 2015 UTC 27 Aug 1998 Dennis WEMITE 1 0448 UTC 30 Aug 1999 0247 UTC 31 Aug 1999 Dennis WEMITE 2 0123 UTC 30 Aug 1999 0123 UTC 31 Aug 1999

c. Dropsondes and radiosondes WEMITE 2 tower. Five rainbands fitting the outlined criteria were selected from Dennis. The selected rain- Dropsonde data obtained from the National Hurri- bands occurred as Dennis made an initial pass off the cane Center (NHC; available online at http:// North Carolina coastline. www.nhc.noaa.gov/) and rawinsonde data obtined from the Forecast Systems Laboratory (FSL; available online 4. WEMITE data manipulation and processing at http://raob.fls.noaa.gov/) were analyzed when present near rainbands that passed over the WEMITE a. Quality control towers. The sondes were used to define the vertical The first step with the WEMITE data was to “clean structure of the atmosphere inside and outside of the it up,” since some of temperature, pressure, and rela- rainband and to quantify the available instability. tive humidity data contained “spikes.” The spikes consisted of large instantaneous changes in the parameters. 3. Study area Although significantly deviant, the spikes generally did not persist for more than2s(5–10 data points). When Hurricane Bonnie (1998) and Hurricane Dennis spikes were found, the values of good data on either (1999) were chosen for this rainband study primarily side of the spike were averaged and inserted in place of because high-resolution surface data were collected for the spikes. For the purpose of this process, a spike was both storms through WEMITE. The high-resolution defined as an instantaneous change (0.1- or 0.2-s surface data within the hurricane environment, coupled change, depending on the sampling rate) in tempera- with nearby radar data, provided a great opportunity to ture greater than 0.25°C, change in relative humidity quantify the low-level meteorological characteristics of greater than 1.5%, or a change in barometric pressure the landfalling rainbands. greater than 0.30 hPa. Hurricane Bonnie made landfall near Wilmington, Next, the wind direction data were adjusted to true North Carolina, as a category 2 hurricane, around 0330 north from the raw tower-relative values by adding the UTC 27 August (Pasch et al. 2001). Bonnie’s best-fit proper adjustment factor determined from a poststorm track, the location of the WEMITE 1 tower, and the site survey (J. R. Howard 2001, personal communica- location of the KMHX radar are shown in Fig. 1. For tion). The wind directions agreed well with the nearby Hurricane Bonnie, WEMITE 1 was located at an azi- Automated Surface Observation System (ASOS) wind muth of 240° and a range of 110 km from the KMHX direction data in all cases (when available). radar. This range corresponded to the lowest elevation b. Standardization angle being approximately 1.7 km above the WEMITE tower. Five rainbands fitting the outlined criteria were After quality control, all the data were averaged and selected from Bonnie. All of the selected rainbands oc- the wind speed data were adjusted to a standard 10-m curred before Bonnie made landfall. height and to an open exposure, as recommended by Hurricane Dennis (1999) made an initial pass off the Powell et al. (1996). The standardization process was North Carolina coastline 30 August 1999 as a category completed to remove the dependence of the wind speed 2 hurricane and then, after meandering out a sea, even- on the upwind terrain, thus allowing any changes to be tually made landfall as a tropical storm near Cape related directly back to hurricane/rainband-scale struc- Lookout National Seashore, North Carolina, at 2100 tures within the storm. UTC 4 September (Lawrence et al. 2001). Dennis’ best- 1) AVERAGING TIME fit track, location of the WEMITE 1 and WEMITE 2 towers, and the location of the KMHX radar are shown The entire dataset was averaged in order to reduce in Fig. 2. WEMITE 1, for Hurricane Dennis, was lo- the dataset size. In order to accomplish this, three av- cated at an azimuth of 76.1° and a range of 49.4 km eraging times were used. All the meteorological param- from the KMHX radar and resulted in the lowest el- eters were averaged into discrete sections of 5-s, 1-min, evation angle being approximately 0.6 km above the and 10-min mean values. WEMITE 1 tower. WEMITE 2, for Hurricane Dennis, 2) HEIGHT was located at an azimuth 103.1° and a range of 21.2 km from the KMHX radar and resulted in the lowest el- All wind speed observations were adjusted to a stan- evation angle being approximately 0.2 km above the dard 10-m height and open exposure. The wind speeds

Unauthenticated | Downloaded 09/25/21 07:38 AM UTC FEBRUARY 2005 SKWIRA ET AL. 457 were adjusted to a 10-m height using an adjusted form provided by the NHC (http://www.nhc.noaa.gov/). Hur- of the log law (Simiu and Scanlan 1996) shown below: ricanes were assumed to move linearly between each best-fit location for the calculations. For a more thor- U͑10͒ ln͓͑10 Ϫ z ͒րz ͔ ϭ d o ͑ ͒ ough explanation of the standardization procedure and ͑ ͒ ͓͑ Ϫ ͒ր ͔ , 1 U z ln z zd zo the calculation of meteorological variables refer to Skwira (2003). where U(10) is the wind speed at 10 m, U(z) is the wind speed observation, zd is the zero-plane displacement height, zo is the roughness length determined by the 5. Results upwind exposure, and z is the height of the wind speed observation. The WEMITE observations presented consist of those taken by WEMITE 1 for Hurricane Bonnie and 3) EXPOSURE WEMITE 2 for Hurricane Dennis. WEMITE 1 data for Hurricane Dennis were not presented in this paper All observations were adjusted to an open exposure since they were very similar to the results obtained using a combination of the log law and the relationship from the WEMITE 2 tower. The WEMITE 1 wind data between frictional velocity and roughness length (Simiu were adjusted down to 10.0 m from 10.7 m and the and Scanlan 1996). More specifically, first the log law WEMITE 2 wind data were adjusted up to 10.0 m from was solved in terms of the frictional velocity: 6.1 m for the Hurricanes Bonnie and Dennis analysis, U͑10͒k respectively. The 10.1-m wind data were not used from ϭ ͑ ͒ u* ͑ ր ͒ , 2 WEMITE 2 because the data were unreliable. ln z zo where k is von Kármán’s constant (0.4). a. Radar reflectivity Using Eq. (2), the frictional velocity for each obser- Figures 3a–e display vertical cross sections of the vation was calculated. Next, the frictional velocity was KMHX radar reflectivity of the five selected rainbands estimated for open exposure using from Hurricane Bonnie as they are centered over the 0.0706 zos WEMITE 1 tower location. The vertical cross sections u* ϭ u*ͩ ͪ , ͑3͒ s z of the KMHX radar reflectivity of the five selected o rainbands from Hurricane Dennis, as they are centered where u*s is the frictional velocity for an open exposure, between the WEMITE 1 and WEMITE 2 towers, are and zos is the standard “open” roughness length (0.03 m). displayed in Figs. 3f–j. The vertical reflectivity cross The calculated variables were then substituted back sections are taken approximately perpendicular to the into the log law: rainband axis in order to examine the crossband structure. Additionally, Fig. 4 displays the base-level radar us* z U͑10, open͒ ϭ ͩ ͪ lnͩ ͪ, ͑4͒ reflectivity for each rainband, including the location of k zo the vertical cross section for Fig. 3. In order to provide a more complete understanding where U(10, open) is the 10-m open exposure wind, u*s is the value from Eq. (3), z is the wind height (10 m), of each rainband, Table 2 displays some of the properties of each rainband including the rainband stage, rain- and zo is the new standardized roughness length (0.03 m). In order to complete the above calculations the band type, approximate distance from the hurricane roughness lengths were determined from written de- center, position within the hurricane (quadrant), and scriptions of site location, pictures of the deployment approximate width and length. The rainband stages are site taken before and following the storm (J. R. Howard classified as intensifying, mature, or dissipating, and are 2002, personal communication), and satellite images defined by the base-level reflectivity trend within an of the nearby area (TerraServer home page, http:// hour of the rainband passing over the WEMITE tow- terraserver.homeadvisor.msn.com/default.asp). Rough- er(s). A rainband increasing in overall area and maxi- ness lengths calculated from the turbulence intensity mum reflectivity is classified as intensifying, one rela- method (Beljaars 1987) using the WEMITE data dur- tively steady in area and maximum reflectivity is clas- ing Hurricanes Bonnie and Dennis were also consulted sified as mature, and one decreasing in areal coverage (Conder 1999; Smith et al. 2001). and maximum reflectivity is classified as dissipating. The rainband types are classified as outer, secondary, principal, connecting, or eyewall. The secondary, prin- c. Calculations of various meteorological variables cipal, and connecting rainbands are identified using the Some additional meteorological variables, including definitions outlined in Willoughby et al. (1984). Rain- the dewpoint, mixing ratio, potential temperature, bands outside of the principal band are assigned an equivalent potential temperature, and the storm- outer rainband classification, while rainbands immedi- relative inflow were calculated. The storm-relative in- ately adjacent to the eye are classified as eyewall rain- flow was calculated with respect to the best-fit locations bands.

Unauthenticated | Downloaded 09/25/21 07:38 AM UTC 458 MONTHLY WEATHER REVIEW VOLUME 133

FIG. 3. Vertical cross section of reflectivity for (a) Bonnie rainband 1 at 1225 UTC 26 Aug 1998; (b) Bonnie rainband 2 at 1559 UTC 26 Aug 1998; (c) Bonnie rainband 3 at 1819 UTC 26 Aug 1998; (d) Bonnie rainband 4 at 2206 UTC 26 Aug 1998; (e) Bonnie rainband 5 at 2355 UTC 26 Aug 1998; (f) Dennis rainband 1 at 0750 UTC 30 Aug 1999; (g) Dennis rainband 2 at 1154 UTC 30 Aug 1999; (h) Dennis rainband 3 at 1214 UTC 30 Aug 1999; (i) Dennis rainband 4 at 1418 UTC 30 Aug 1999; and (j) Dennis rainband 5 at 2022 UTC 30 Aug 1999. The horizontal and vertical dimensions are in kilometers, and the reflectivity scale (dBZ) is at the bottom of the figure. The inside (closest to the hurricane center) and outside of each rainband are noted by an I and O, respectively.

Unauthenticated | Downloaded 09/25/21 07:38 AM UTC FEBRUARY 2005 SKWIRA ET AL. 459

FIG. 4. Base-level radar reflectivity for (a) Bonnie rainband 1; (b) Bonnie rainband 2; (c) Bonnie rainband 3; (d) Bonnie rainband 4; (e) Bonnie rainband 5; (f) Dennis rainband 1; (g) Dennis rainband 2 (leftmost) and Dennis rainband 3 (rightmost); (h) Dennis rainband 4; and (i) Dennis rainband 5. The reflectivity scale (dBZ) is at the bottom of the figure. The inside (closest to the hurricane center) and outside of each rainband are noted by an I and O, respectively. The dark black lines correspond to the location of the cross sections shown in Fig. 3. Time of images match those given in Fig. 3, except for Dennis rainband 3 [(g)], which is overlaid on the Dennis rainband 2 image, with the cross-section location approximated.

Unauthenticated | Downloaded 09/25/21 07:38 AM UTC 460 MONTHLY WEATHER REVIEW VOLUME 133

TABLE 2. Properties of Hurricanes Bonnie and Dennis rainbands. The rainband quadrant location is abbreviated FQ, LFQ, and LRQ for the forward quadrant, left-forward quadrant, and left-rear quadrant, respectively. Distance from Rainband Stage Rainband type eye (km)/quadrant Width (km) Length (km) Bonnie band 1 Dissipating Outer 165/FQ 30 Ͼ100 Bonnie band 2 Mature Connecting 115/FQ 30 Ͼ100 Bonnie band 3 Mature Eyewall 75/FQ 30 Ͼ100 Bonnie band 4 Intensifying New eyewall 50/FQ 20 Ͼ100 Bonnie band 5 Intensifying Inside eye 20–25/FQ 10 Ͻ100 Dennis band 1 Mature Principle from outer 220/LFQ 10 Ͼ100 Dennis band 2 Mature Secondary 150/LFQ 5 Ͻ100 Dennis band 3 Mature Secondary 170/LFQ 10 Ͻ100 Dennis band 4 Mature to dissipating Connecting/edge of principal 150/LRQ 50 Ͼ100 Dennis band 5 Intensifying to mature Outer 220/LRQ 15 Ͼ100

All 10 of the hurricane rainbands (Figs. 3a–j) possess study that was not included in these cases was a dissi- some common characteristics. First, all of the rainbands pating outer rainband that did not exhibit any regen- contain little high reflectivity (32.5ϩ dBZ, denoted by eration or redevelopment anywhere. the interface between the 29- and 34-dBZ shades) Although similar in vertical development and reflec- above the 5–6-km level, with a majority of the midlevel tivity regeneration, the rainbands exhibit great range in high reflectivity due primarily to the “bright band” lo- width varying between 5 and 50 km, with a mean of 21 cated near a freezing level of 5 km for both storms. This km. The smallest of the rainbands are the two second- result suggests that although all of these rainbands are ary bands in Dennis (Figs. 3g, 3h, or 4g). Conversely, associated with modest hurricanes, the landfalling hur- the largest rainband, at 50 km in width, occurs near the ricane rainbands traversing over the WEMITE towers interface of the principal and connecting band in Den- consist primarily of stratiform precipitation. Even the nis (Figs. 3i or 4h), and although quite wide, it is rela- more convective rainbands with greater low-level re- tively shallow. The shallowness and homogeneity of the flectivity gradients including Bonnie band 4, Bonnie connecting rainband is consistent with Willoughby et al. band 5, Dennis band 1, and Dennis band 3 (Figs. 3d, 3e, (1984), who found connecting rainbands commonly 3f, and 3h, respectively) possess little high reflectivity contain much less cellular convection than the principal above the freezing level, with the two Dennis rainbands rainbands. showing signs of a bright band adjacent to the strongest Additionally, the rainbands occurred anywhere from reflectivity cores. 20 to 230 km from the circulation center, with Bonnie’s Second, although not apparent from Figs. 3 or 4, the rainbands being generally closer, providing for a wide reflectivity animation (not shown) of the rainbands de- range in rainband types (outer rainbands to eyewall picts that the reflectivity tended to regenerate/ rainbands). The one obvious limitation in the diversity redevelop on the inside portion of the rainband in 6 of of the rainbands is that only two of the rainbands oc- the 10 cases, consistent with Barnes et al. (1983). All curred in the rear half of the storm. The relative lack of cases that had redevelopment on the inside of the rain- rainbands from the rear section of the storm was in part band were larger rainbands (Ͼ100 km in length) and due to the storm’s motion relative to the tower loca- were either mature or intensifying rainbands. The re- tions and partially because the rear portion of the flectivity redevelopment on the inside of the larger storms over land tended to form a large area of pre- rainbands suggests that these larger bands were orga- cipitation that, although had banded features, did not nized in a manner that promoted updrafts and reflec- fall into a rainband classification. tivity redevelopment to their inside. Examination of the b. Kinematics velocity data (both base data and cross sections) for convergent signals on the inside of the rainbands A 3-h window of storm-relative inflow and standard- proved difficult for a number of reason: First, there was ized 1-min wind speed (as explained in section 4b) is a significant range-folding problem for three of the shown in Figs. 5a–j. The vertical lines in Figs. 5a–h cases; second, four of the six rainbands were oriented represent the approximate time the rainband is cen- roughly parallel to the radar beam over the WEMITE tered over the WEMITE tower. Figures 5i and 5j have towers, resulting in the detection of the tangential flow no vertical line since the two rainbands were stationary, but not the radial flow. Nonetheless, five of the six and thus the location relative to the WEMITE tower rainbands did possess a confluent signal collocated with did not change significantly. With the exception of the their redevelopment along at least a portion of the rain- two stationary rainbands, the remaining rainbands gen- band. The above-stated hypothesis worked well, in this erally moved off to the west or northwest as they passed limited study, for the outer, principal, connecting, and over the WEMITE towers. The rainband motion, eyewall rainbands. The one large rainband from this coupled with their position in the front or left-front

Unauthenticated | Downloaded 09/25/21 07:38 AM UTC FEBRUARY 2005 SKWIRA ET AL. 461

FIG. 5. Standardized WEMITE observations of the 1-min wind speed and storm-relative inflow during the passage of (a) Bonnie rainband 1; (b) Bonnie rainband 2; (c) Bonnie rainband 3; (d) Bonnie rainband 4; (e) Bonnie rainband 5; (f) Dennis rainband 1; (g) Dennis rainband 2; (h) Dennis rainband 3; (i) Dennis rainband 4; and (j) Dennis rainband 5. The vertical lines represent approximately when the rainband is centered over the WEMITE tower, while graphs with no vertical lines denote rainbands that did not translate over the tower but remained relatively stationary. The wind speed (m sϪ1) is denoted by a solid line and storm-relative inflow (degree) by the dotted line. Time is in UTC. Hurricane Bonnie data [(a)–(e)] occur on 26 Aug 1998, while Hurricane Dennis data [(f)–(j)] occur on 30 Aug 1999.

Unauthenticated | Downloaded 09/25/21 07:38 AM UTC 462 MONTHLY WEATHER REVIEW VOLUME 133 quadrant, placed all the data gathered earlier than the The remaining rainbands (Figs. 5b, 5c, and 5g) pos- vertical line on the outside portion of the rainband axis sess fluctuating degrees of storm-relative inflow with a with the data following the vertical line to the inside slight decreasing trend within the 3-h window. The portion of the rainband (closer to the vortex center). slight decreasing trend does not appear associated with The storm-relative inflow is calculated so that a positive the rainbands passage in Figs. 5b and 5g, but instead is value represents the angle that the wind is blowing into likely just part of the larger-scale hurricane wind field. the storm center (in comparison to a tangential wind The onset of the decreasing trend with the Bonnie eye- with zero inflow). A negative storm-relative inflow is wall rainband (Fig. 5c), however, is coincident with the outflow. rainband passage. One interesting feature is that the standardized wind As a whole, the storm-relative results suggest that the speed, both in the rainband windows and within the more convective rainbands tend to impede the inflow, entire WEMITE record (not shown), contains no hur- supportive of previous research (Wexler 1947; Ligda ricane-force (sustained 1-min 33 m sϪ1) values for Hur- 1955; Barnes et al. 1983; Powell 1990a; etc.), although ricanes Bonnie or Dennis. It should be noted that the not to a point of total interruption. Moreover, the re- standardization procedure resulted in slightly higher sults also suggest that a dissipating rainband may be a wind speeds than the observed values, primarily due to favored region for increased storm-relative inflow. the exposure adjustment from rougher land to open c. Thermodynamics terrain, and is not responsible for the relatively low values of wind speed found throughout the records. A 3-h window of temperature, dewpoint, and equiva- The lack of hurricane-force wind is likely due to the fact lent potential temperature encompassing each rainband that both hurricanes were weakening, and, as seen in passage is shown in Figs. 6a–j. The graphs display section 5a, the portion of the storm passing over the equivalent potential temperature decreases associated WEMITE towers was not very convectively active. with 9 of the 10 rainbands, with the remaining rainband Regardless, the wind records do not show any dis- (Fig. 6b) possessing an interruption in its increasing cernable trend associated with the hurricane rainbands. trend. Four of the rainbands that contain equivalent Half of the rainbands (Figs. 5a, 5c, 5e, 5i, and 5j) show potential temperature decreases (Figs. 6a, 6c–e) con- relatively steady wind speeds (Ϯ5msϪ1) coupled with tain relative minima, with all minima occurring in the the rainband passage. The remaining rainbands contain outside portion of the rainband. The remaining rain- slight increasing trends (Figs. 5b and 5f–h) and even a bands that possess equivalent potential temperature wind speed decrease (Fig. 5d) associated with the rain- decreases (Figs. 6f–j) do not contain a corresponding band passage. There are minor (ϩ5msϪ1) wind speed equivalent potential temperature recovery. It should relative maxima near the passage of a few rainbands be noted that Figs. 6g and 6h do share the same drop over the tower, but nothing significant. The lack of since they are located very close together (both spa- wind speed fluctuation within the rainbands may again tially and temporally), and it is difficult to distinguish be related back to the relative lack of strong convection whether the decrease should be attributed to one or and general weakening of the hurricanes. both of the rainbands. Additionally, the final two rain- The storm-relative inflow data, however, do display band equivalent potential temperature decreases (Figs. some interesting tendencies. First, the only dissipating 6i,j) are due primarily to drier (and cooler for Fig. 6j) rainband (Fig. 5a) boasts increasing inflow throughout continental air working into the system from the west, its passage. The four most convective rainbands (Figs. and not rainband-scale processes. Regardless, the rea- 5d–f, 5h) exhibit decreasing storm-relative inflow fol- son for the differences between the two observed lowing the center of the rainband passage over the modes of equivalent potential temperature variation tower, although the actual magnitude and duration var- with the rainband passage (decrease followed by in- ies. The one area of spikes in Fig. 5d (near 2205 UTC) crease versus decrease followed by steady or slower is due to sensor error. The wind direction record at 10.1 decrease) is not known, although three of the four rain- m changes dramatically for 2 min and then returns to bands that do possess an increase following the rain- normal. The drastic change is likely sensor error since band passage are located near the eye and thus have the anemometers at the other two levels did not show higher equivalent potential temperature air nearby to any such fluctuations. The reason for the brief sensor help in the recovery. The remaining rainbands are not error is unknown, but the record beyond 2210 UTC located near the eye and hence do not have a significant appears correct and agrees with the other sensors. higher equivalent potential temperature source to The final two stationary rainbands in Hurricane Den- readily draw upon. nis (Figs. 5i,j) reveal a steadily decreasing inflow The common presence of equivalent potential tem- throughout. The steady decrease is primarily a result of perature decreases with rainband passage is further ex- the storm translating off to the east-northeast, and not plored with available dropsonde and rawinsonde data any significant changes in the wind direction. Thus, the (not shown). A dropsonde released inside of Bonnie decreasing storm-relative radial inflow is more a prod- band 1 found the 344.2-K equivalent potential tempera- uct of the analysis than the observations. ture minimum (Fig. 6a) to correspond to air from the

Unauthenticated | Downloaded 09/25/21 07:38 AM UTC FEBRUARY 2005 SKWIRA ET AL. 463

FIG. 6. Same as Fig. 5, except for the temperature (thin solid black line), dewpoint (thin gray line) in degrees Celsius, and equivalent potential temperature (thick black line with dark triangles) in kelvins.

Unauthenticated | Downloaded 09/25/21 07:38 AM UTC 464 MONTHLY WEATHER REVIEW VOLUME 133

910-hPa (800 m) level. Another dropsonde, released not fall into a rainband classification. A second limita- inside of Bonnie band 3, found the 348.1-K minimum tion is that five of the seven rainbands that have 32.5- (Fig. 6c) to correspond to air from the 900-hPa (750 m) dBZ radar reflectivity vertically to the freezing level level. Additionally, a 0600 UTC sounding launched possess bright bands, suggesting that this dataset is pre- from KMHX immediately outside of Dennis band 1 dominately stratiform. Hence, the results may not hold found the 340-K minimum (Fig. 6f) to correspond to air true for more convectively active hurricane rainbands. from the 910-hPa (869 m) level. Assuming no dilution, Regardless, the rainband dataset still reveals a num- the above available vertical thermodynamic profiles ber of common features. Table 3 summarizes the rain- suggest the source of the lower equivalent potential band results. First, all the rainbands possessed a de- temperature air to range anywhere from 750 to 869 m crease (or interruption in the rise) of the equivalent above the surface. In reality, there is some entrainment potential temperature, although not all decreases were and the source region of the lower equivalent potential directly attributed to the rainbands. The few cases that temperature air may be well above the values deter- vertical temperature information in close proximity to mined assuming no entrainment. the rainband exist suggested the source of the lower equivalent potential temperature air to range anywhere 6. Summary, conclusions, and recommendations from 750 to 869 m above the surface, assuming no entrainment. a. Summary Moreover, while none of the cases experienced a total interruption to the inflow, many of the rainbands did Ten rainbands, five from Hurricane Bonnie (1998) and five from Hurricane Dennis (1999), were selected exhibit decreasing hurricane-relative inflow, with the for detailed analysis. This study focused on the radar more convectively active rainbands tending to impede Ͼ reflectivity and rainband-scale fluctuations of various the inflow the most. Additionally, the larger ( 100 km dynamic and thermodynamic meteorological param- in length) mature and intensifying rainbands were or- eters as the rainbands passed over the WEMITE tow- ganized in a manner that promoted updrafts and reflec- ers. In order to complete the analysis, meteorological tivity redevelopment to their inside. The reflectivity re- data were obtained from WEMITE tower data, drop- development to the inside may appear counterintuitive, sondes, rawinsondes, and NEXRAD WSR-88D data. since one might expect the greatest surface conver- Once ascertained, all surface-level wind data were ob- gence on the outside of the rainband, where the lower jectively adjusted in an effort to standardize them to a equivalent potential temperature air associated with ϭ the rainband would impede the inflow. Moreover, the 10-m height, open exposure (zo 0.03 m), and a 1-min averaging time. Additionally, storm-relative inflow was observations of surface inflow indicate decreases to the calculated. inside of the band. Although both above arguments suggest convergence, and thus new development to the outside of the rainband, they do not take into account b. Conclusions and speculation anything above the surface. Previous studies (Barnes et This collection of landfalling rainbands, although al. 1983; Barnes and Stossmeister 1986; Barnes et al. relatively small, does represent a wide array of rain- 1991; Powell 1990a,b) have suggested weak near- band types, stages, hurricane-relative locations, and surface convergence generally on the outside portion of sizes (see Table 2). One obvious limitation is that only the rainband. However, the same studies have also two of the rainbands occur in the rear half of the storm, shown the axis of convergence sloping to the inside of partially because of the storm’s motion relative to the the rainband above the surface, with the maximum low- tower locations and partially because the rear portion level convergence and vertical wind speed on the inside of the storms over land tend to form a large area of of the rainband, resulting in redevelopment there. The precipitation that, although had banded features, did larger rainbands examined in the study support the

TABLE 3. Summary of Hurricanes Bonnie and Dennis rainband characteristics. Rainband Intensity (motion) Equivalent potential temperature Hurricane-relative inflow Bonnie band 1 40 dBZ (NW) 3-K min Increased to inside Bonnie band 2 40–45 dBZ (NW) Steady in band with adjacent rises Decreased to inside Bonnie band 3 40 dBZ (NW) 2-K min Decreased to inside Bonnie band 4 45 dBZ (NW) 1-K min outside, 4-K rise inside Decreased to inside Bonnie band 5 40ϩ dBZ (rapidly W) 2-K min outside Increased to center, decreased to inside Dennis band 1 45 dBZ (NNW) 10-K decrease Decreased to inside Dennis band 2 40–45 dBZ (rapidly W) 4-K decrease Increased to inside, then decreased Dennis band 3 45 dBZ (rapidly W) 4-K decrease Max at center Dennis band 4 40 dBZ (stationary) Gradual 4-K decrease Decreased to inside Dennis band 5 34–40 dBZ (stationary) Steady 6-K decrease Decreased to inside

Unauthenticated | Downloaded 09/25/21 07:38 AM UTC FEBRUARY 2005 SKWIRA ET AL. 465 above model, with five of the six larger rainbands hav- insights and suggestions. This work has been supported ing a confluent signal, from radar velocity data, collo- in part by National Science Foundation Grant ATM- cated with their redevelopment along at least a portion 0134188 and Department of Commerce National Insti- of the rainband. Interestingly, no common wind speed tute of Standards and Technology/Texas Tech Univer- trend was found with the analyzed rainbands. The lack sity Cooperative Agreement Award 70NANB8H0059. of wind speed fluctuation within the rainbands may be related back to the relative lack of strong convection REFERENCES and the general weakening of the hurricanes within Barnes, G. M., and G. J. Stossmeister, 1986: The structure and which the rainbands exist. decay of a rainband in Hurricane Irene (1981). Mon. Wea. The above results promote the argument that rain- Rev., 114, 2590–2601. bands are favored locations for hurricane boundary ——, D. P. Jorgensen, F. D. Marks Jr., and E. J. Zipser, 1983: layer dynamic and thermodynamic modification, in- Mesoscale and convective structure of a hurricane rainband. cluding the transport of air to the surface from aloft. J. Atmos. Sci., 40, 2125–2137. ——, J. F. Gamache, M. A. LeMone, and G. J. Stossmeister, 1991: The boundary layer modification, if present over a A convective cell in a hurricane rainband. Mon. Wea. Rev., large enough region, also suggests the potential for 119, 776–794. rainbands to affect the evolution and intensity of the Beljaars, A. C. M., 1987: The influence of sampling and filtering parent hurricane. It is also speculated that the hurri- on measured wind gusts. J. Atmos. Oceanic Technol., 4, 613– 626. cane boundary layer modifications (impedance of in- Conder, M. R., 1999: Estimation of roughness length through gust flow and decreased equivalent potential temperature) factor analysis. M.S. thesis, Dept. of Atmospheric Science, may indeed be even greater than detected in this study Texas Tech University, 61 pp. with more vigorous convective rainbands. Lawrence, M. B., L. A. Avila, J. L. Beven, J. L. Franklin, J. L. Guiney, and R. J. Pasch, 2001: Atlantic hurricane season of 1999. Mon. Wea. Rev., 129, 3057–3084. c. Recommendations Ligda, M. G., 1955: Hurricane squall lines. Bull. Amer. Meteor. More studies/analyses should be completed on rain- Soc., 36, 340–342. Pasch, R. J., L. A. Avila, and J. L. Guiney, 2001: Atlantic hurri- bands in order to expand this very limited dataset. The cane season of 1998. Mon. Wea. Rev., 129, 3085–3123. expansion of the dataset would allow for universal char- Powell, M. D., 1990a: Boundary layer structure and dynamics in acteristics, if they exist, to be better defined and would outer rainbands. Part I: Mesoscale rainfall and kinematic also enhance the possibility of creating rainband sub- structure. Mon. Wea. Rev., 118, 891–917. ——, 1990b: Boundary layer structure and dynamics in outer hur- sets that might share more common traits. ricane rainbands. Part II: Downdraft modification and mixed Additionally, it would be beneficial to perform dual- layer recovery. Mon. Wea. Rev., 118, 918–938. Doppler analysis of hurricane rainbands to obtain the ——, S. H. Houston, and R. A. Reinhold, 1996: Hurricane An- middle- and upper-level wind field associated with the drew’s landfall in South Florida. Part I: Standardizing mea- landfalling rainbands. Moreover, coupling dual- surements for documentation of surface wind fields. Wea. Forecasting, 11, 304–328. Doppler analysis with high-resolution meteorological Schroeder, J. L., and D. A. Smith, 2003: Hurricane Bonnie wind data gathered by towers (especially in a horizontal ar- flow characteristics as determined from WEMITE. J. Wind ray) from the surface would provide an even more com- Eng. Ind. Aerodyn., 91, 767–789. plete picture of the three-dimensional rainband. Simiu, E., and R. H. Scanlan, 1996: Wind Effects on Structures. John Wiley & Sons, 688 pp. Furthermore, the expansion of this dataset to include Skwira, G. D., 2003: Surface observations of landfalling hurricane more intense landfalling hurricanes and hurricane rain- rainbands: Case studies of Hurricane Bonnie (1998) and Hur- bands (if and when possible) would greatly help in im- ricane Dennis (1999). Ph.D. dissertation, Texas Tech Univer- proving the understanding of the properties of these sity, 267 pp. most intense and dangerous storms. Smith, D. A., J. L. Schroeder, and J. R. Howard, 2001: Final report of activities: 1998–1999 Wind Engineering Mobile Instru- mented Tower Experiment (WEMITE). Tech. Memo. to the Acknowledgments. Special thanks is extended to all Insurance Friends of the National Hurricane Center, Texas the members of the WEMITE team who have given Tech University, Lubbock, TX, 78 pp. and continue to give generously of their time and effort Willoughby, H. E., F. D. Marks Jr., and R. J. Feinberg, 1984: Stationary and moving convective bands in hurricanes. J. At- to make the collection of high-resolution data in a hur- mos. Sci., 41, 3189–3211. ricane environment possible. Additionally, the authors Wexler, H., 1947: Structure of hurricanes as determined by radar. thank the two anonymous reviewers for their helpful Ann. N.Y. Acad. Sci., 48, 821–844.

Unauthenticated | Downloaded 09/25/21 07:38 AM UTC