Structural and Intensity Changes of Hurricane Bret (1999). Part I: Environmental Influences

4320 MONTHLY WEATHER REVIEW VOLUME 136

ALEXANDER LOWAG* AND MICHAEL L. BLACK Hurricane Research Division, NOAA/AOML, Miami, Florida

MATTHEW D. EASTIN Department of Geography and Earth Sciences, University of North Carolina, Charlotte, North Carolina

(Manuscript received 25 October 2007, in final form 2 April 2008)

ABSTRACT

Hurricane Bret underwent a rapid intensification (RI) and subsequent weakening between 1200 UTC 21 August and 1200 UTC 22 August 1999 before it made landfall on the Texas coast 12 h later. Its minimum sea level pressure fell 35 hPa from 979 to 944 hPa within 24 h. During this period, aircraft of the National Oceanic and Atmospheric Administration (NOAA) flew several research missions that sampled the environment and inner core of the storm. These datasets are combined with gridded data from the National Centers for Environmental Prediction (NCEP) Global Model and the NCEP–National Center for Atmo- spheric Research (NCAR) reanalyses to document Bret’s atmospheric and oceanic environment as well as their relation to the observed structural and intensity changes. Bret’s RI was linked to movement over a warm ocean eddy and high sea surface temperatures (SSTs) in the Gulf of Mexico coupled with a concurrent decrease in vertical wind shear. SSTs at the beginning of the storm’s RI were approximately 29°C and steadily increased to 30°C as it moved to the north. The vertical wind shear relaxed to less than 10 kt during this time. Mean values of oceanic heat content (OHC) beneath the storm were about 20% higher at the beginning of the RI period than 6 h prior. The subsequent weakening was linked to the cooling of near- coastal shelf waters (to between 25° and 26°C) by prestorm mixing combined with an increase in vertical wind shear. The available observations suggest no intrusion of dry air into the circulation core contributed to the intensity evolution. Sensitivity studies with the Statistical Hurricane Intensity Prediction Scheme (SHIPS) model were conducted to quantitatively describe the influence of environmental conditions on the intensity forecast. Four different cases with modified vertical wind shear and/or SSTs were studied. Dif- ferences between the four cases were relatively small because of the model design, but the greatest intensity changes resulted for much cooler prescribed SSTs. The results of this study underscore the importance of OHC and vertical wind shear as significant factors during RIs; however, internal dynamical processes appear to play a more critical role when a favorable environment is present.

1. Introduction tions have shown little improvement. While the Na- tional Hurricane Center (NHC) uses a combination of Forecasting fluctuations in intensity of tropical cy- dynamical and statistical computer models, the average clones (TCs) remains one of the major challenges for error of its Atlantic basin intensity forecasts between meteorologists. Even though forecast models have con- 1997 and 2006 has not significantly improved (Fig. 1). sistently become more sophisticated, intensity predic- Inaccurate intensity forecasts are partially a result of an incomplete understanding of the various physical * Current affiliation: Rosenstiel School of Marine and Atmo- mechanisms that control the cyclone’s intensity (Eman- spheric Science, University of Miami, Miami, Florida. uel 1999). Such controls can be broadly grouped into three categories: environmental, oceanic, and internal. Environmental mechanisms include vertical wind shear Corresponding author address: Alexander Lowag, Rosenstiel School of Marine and Atmospheric Science, University of Miami, (DeMaria 1996; Frank and Ritchie 2001; Zehr 2003), 4600 Rickenbacker Causeway, Miami, FL 33149. the interaction of the storm with synoptic upper-level E-mail: [email protected] features (Molinari and Vollaro 1989; DeMaria et al.

DOI: 10.1175/2008MWR2438.1

MWR2438 NOVEMBER 2008 LOWAG ET AL. 4321

of a TC are associated with periods of rapid intensification (RI) and weakening. The definition of RI is non- uniform. Holliday and Thompson (1979) related it to a decrease in minimum sea level pressure (MSLP) greater than 42 hPa over a 24-h period, whereas Brand (1973) associated it with an increase in wind speed of 50 kt dayϪ1. A RI in the Statistical Hurricane Intensity Prediction Scheme (SHIPS) model by DeMaria and Kaplan (1994a) is linked to an increase in wind speed of at least 30 kt within 24 h (Kaplan and DeMaria 2003), whereas rapid weakening is related to a decrease in wind speed of more than 20 kt dayϪ1 (Brand 1973). Even though only a small percentage of TCs undergo RI (DeMaria and Kaplan 1994a), the consequences of such events can be significant if RI occurs in close proximity to coastal regions, many of which have experienced near-exponential population growth in recent de- cades (Pielke and Pielke 1997). For example, Hurricane Andrew (1992) gained strength over the Bahamas as its FIG. 1. Official NHC intensity forecast errors (kt) between 1997 1-min maximum sustained wind speed increased from and 2006 for the Atlantic basin. The solid and dashed lines show 110 to 150 kt within 18 h and made landfall on the the errors for a 12- and 36-h forecast, respectively. The dotted line South Florida coast as a category 5 hurricane. The represents errors for a 72-h forecast. (Data source: NHC.) storm killed 15 people directly and caused $25 billion in property damage (Mayfield et al. 1994; Rappaport 1993; Hanley et al. 2001), and the moisture content of 1994). More recently during the active 2004 and 2005 midlevel air (Fink and Vincent 2003; Dunion and Atlantic Hurricane seasons, a total of six hurricanes Velden 2004). Oceanic contributions to intensity varia- (Charley, Dennis, Emily, Katrina, Rita, and Wilma) ex- tions arise through air–sea interactions (Chan et al. perienced RI events within 48 h of making landfall. 2001) and thus knowledge of the thermal structure of Although several environmental factors present prior the upper ocean is essential for their accurate predic- to or during RI events have been identified (Holliday tion (Mainelli et al. 2008). Heat fluxes from the ocean and Thompson 1979; Kaplan and DeMaria 2003), the to the atmosphere influence the internal dynamics of a physical mechanisms and interactions that cause sud- TC. Hence, it is suggested that more precise locations den intensity changes in TCs are not well understood. of oceanic warm and cold pools in numerical models One notable RI event that has received considerable should be beneficial to TC forecasts (Walker et al. study is Hurricane Opal. In October 1995, Opal inten- 2005), since heat fluxes strongly depend on the sea sur- sified rapidly over the Gulf of Mexico where its MSLP face temperature (SST; Cione and Uhlhorn 2003). A decreased by 47 hPa within 16 h. During the first stage study by Lin et al. (2005) showed that incorporation of of intensification between 1800 UTC 3 October and additional eddy information from satellite-derived sea 0600 UTC 4 October 1995, the storm was influenced by surface height anomalies (SHAs) resulted in an im- a synoptic trough that enhanced Opal’s outflow and proved intensity forecast for Supertyphoon Maemi increased the divergence in the upper troposphere. (2003) in the Coupled Hurricane Intensity Prediction Bosart et al. (2000) showed that a consequence of this Scheme (CHIPS) model. Finally, internal processes enhancement process was convective growth in the eye- that contribute to intensity changes include eyewall re- wall region and the onset of intensification. Low verti- placement cycles (Willoughby et al. 1982; Willoughby cal wind shear also contributed to Opal’s initial 1990), vortex Rossby waves (Montgomery and Kallen- strengthening (Shay et al. 2000). The second stage of bach 1997), mesovortices in the eye–eyewall region intensification occurred when Opal moved over a warm (Schubert et al. 1999; Kossin and Schubert 2001; Kossin core eddy shed by the Loop Current. During this pas- and Eastin 2001; Eastin et al. 2005) and convective sage the surface winds increased from 35 to over 60 bursts (Heymsfield et al. 2001; Kelley et al. 2004). In- msϪ1. The effect of the warm core prevailed despite tensity changes experienced by any given TC are likely the cooling of the sea surface due to upwelling and a combination of these mechanisms working in concert. vertical mixing in the upper ocean (Shay et al. 2000). Some of the larger errors in predicting the intensity The weakening of Opal was related to an eyewall re-

Unauthenticated | Downloaded 09/28/21 10:02 PM UTC 4322 MONTHLY WEATHER REVIEW VOLUME 136 placement cycle that was completed shortly after the hurricane reached peak intensity (Lawrence et al. 1998). Bosart et al. (2000) also found that increased vertical wind shear and a shallower mixed layer depth also contributed to Opal’s weakening. The number of detailed observational analyses of the environment and inner core of rapidly intensifying hurricanes is quite limited. This paper, which is Part I of a two-paper series, aims to reduce such deficiency by documenting the available observations from the atmospheric and oceanic environment of Hurricane Bret (1999) during a period of RI and weakening just prior to landfall. Inner-core observations directly relevant to the observed environmental changes are also discussed. The paper is organized as follows. Section 2 describes the history of Hurricane Bret, followed by a detailed overview of the available data in section 3. Section 4 discusses the evolution of external forcing and the storm’s response. Sensitivity studies with the SHIPS FIG.2.NHC’s best track for Hurricane Bret between 1800 UTC 18 Aug and 0000 UTC 25 Aug 1999. The solid line with filled dots model are provided in section 5, and a summary of our represents the MSLP (hPa), and the solid line with squares shows findings is in section 6. In Part II, a detailed analysis of the maximum sustained wind speed (kt). Each dot and square the inner-core observations during the RI event and the indicates individual values for pressure and wind speed, respec- subsequent weakening prior to landfall will be pre- tively. The two solid vertical lines indicate the period of Bret’s sented. rapid intensification. The best-track positions of Bret’s center are given in Fig. 7.

2. Storm history Hurricane Bret originated from a tropical wave that over Texas and Mexico, the system dissipated on 25 moved westward from Africa over the tropical Atlantic August. Ocean (Lawrence et al. 2001). The wave did not de- As for many other storms that rapidly deepened, the velop until the evening of 18 August 1999 when the intensity forecasts for Bret did not capture the RI event system became a tropical depression situated over the well (Fig. 3). Both NHC and SHIPS forecasts are com- Bay of Campeche. Twenty-four hours later, Bret was pared to the NHC best track that consists of 6-hourly upgraded to a tropical storm after an upper-level estimates of the storm center location, 1-min sustained trough causing high wind shear over the western Gulf surface wind speeds, and MSLP. Note that Bret’s fore- of Mexico moved away from the cyclone. The storm cast intensity was underestimated in both cases. How- then headed to the north and intensified progressively. ever, the difference between the NHC forecasts and the Figure 2 shows the central pressure and wind speed of best track is larger than for SHIPS, especially during Bret according to NHC’s best track. The period of RI the period of RI. On 0300 UTC 19 August forecasters occurred between 1200 UTC 21 August and 1200 UTC at NHC predicted a maximum sustained wind speed for 22 August. In the 6-h period beginning at 1800 UTC 21 0000 UTC 22 August of 50 kt, while the SHIPS forecast August, the MSLP fell 21 hPa from 975 to 954 hPa, and from 0000 UTC 19 August was 78 kt for the same time. over a 24-h period it dropped 35 hPa from 979 to 944 The actual intensity was 120 kt. The higher intensities hPa. The deepening made Bret a category 4 hurricane in SHIPS are due to the fact that land effects were not with a sustained wind speed of 125 kt. A weak ridge included in the 1999 version of the model, while NHC over the northwestern Gulf of Mexico and a cyclonic accounted for Bret’s proximity to the western Gulf circulation over the Rio Grande River valley in the coast (L. A. Avila 2008, personal communication). In- midtroposphere caused the hurricane to track north- terestingly, Bret’s weakening is also not captured in the westward on 22 August. Despite forecasts that called forecasts. A SHIPS forecast from 0600 UTC 20 August for constant intensity, Bret weakened to a category 3 predicted maximum winds of 131 kt for 0000 UTC 23 storm with a maximum sustained wind speed of 100 kt August, while NHC at 0300 UTC 22 August predicted and a MSLP of 951 hPa at landfall along the south a constant wind speed of 120 kt for the 12-h period Texas coast around 0000 UTC 23 August. After moving starting at 1200 UTC on the same day. The best track,

Unauthenticated | Downloaded 09/28/21 10:02 PM UTC NOVEMBER 2008 LOWAG ET AL. 4323

(NCEP–NCAR) reanalyses and Aviation [(AVN; now Global Forecast System (GFS)] global model analyses were used to describe the large-scale atmospheric conditions surrounding Hurricane Bret. The AVN analyses have a spatial resolution of 1° in the horizontal and are obtained every 6 h for 28 vertical levels. NCEP–NCAR reanalyses are also based on the AVN model. The data are available for the same pressure levels as the AVN data, but have a coarser horizontal and temporal resolution of 2.5° and 12 h, respectively. The SSTs and temperature characteristics of the oceanic upper layer are measured by airborne expendable bathythermographs (AXBTs). These instruments profile the water temperature to a depth of approximately 350 m. After the instrument has splashed onto the sea surface, it releases a temperature probe that descends at a constant rate of 1.5 m sϪ1. The water temperature is inferred from frequency fluctuations of the signal sent to the airplane (Boyd 1987). Consideration of SST alone is not sufficient to understand the ocean’s role on the intensification process of a tropical cyclone (Palmen 1948). Vertical mixing processes induced by minimal tropical storm force winds result in a cooling of the water at the sea surface (Kraus and Turner 1967). Con- sequently, knowledge of the conditions in the upper ocean is crucial to determine the oceanic influence on TC intensity variations. Besides SST, the oceanic heat content (OHC) is used to describe the oceanic contri- bution to Bret’s intensity changes. It is defined as the energy stored in the ocean that can be extracted by the storm through evaporation and conduction and is pro- portional to the temperature difference between the sea surface and 26°C isotherm, as well as the depth of the latter (Leipper and Volgenau 1972). The methodology used for estimation of the OHC is based on Shay et al. (2000). The ocean’s vertical structure is consid- ered as a two-layer fluid in which the depth of the 20°C isotherm separates the two regimes. Estimation of the average upper-layer thickness and reduced gravity from climatological temperature and salinity fields in asso- ciation with SHA data from altimeters, including the FIG. 3. Forecasts of maximum sustained wind speed (kt) from Ocean Topography Experiment (TOPEX)/Poseidon (top) NHC and (bottom) SHIPS. The symbols indicate different forecasts between 0000 UTC 19 Aug and 0000 UTC 23 Aug 1999. (T/P), allows for the determination of the upper-layer The solid line with filled dots represents the best track. The first thickness. The study by Shay et al. (2000) also found value for each SHIPS forecast corresponds to the best track. Each that the depth of the 26°C isotherm is approximately NHC forecast starts 9 h after it was issued. half of the upper-layer thickness. Weekly averaged SSTs needed for the OHC estimation are obtained however, indicates maximum winds of only 100 kt on from the Advanced Very High Resolution Radiometer 0000 UTC 23 August. (AVHRR). A detailed description of the methodology is presented in Huber (2000). 3. Data Even though the T/P data have an accuracy of 2 cm Both the National Centers for Environmental Pre- on the mesoscale (Cheney et al. 1994), it should be diction–National Center for Atmospheric Research noted that OHC values derived from this platform may

Unauthenticated | Downloaded 09/28/21 10:02 PM UTC 4324 MONTHLY WEATHER REVIEW VOLUME 136 have distinct discrepancies to the actual values. In the GFS forecasts instead of the analyses. It is assumed that presence of a warm core eddy, the estimated SHA val- errors for dependent runs are approximately 20% ues from the T/P altimeter depend on the location of lower than for real-time runs except for cases where the the eddy relative to the ground track of the platform forecast track went over land, whereas the best track (Shay et al. 2000). The larger the distance between the did not (M. DeMaria 2008, personal communication). track and the center of the eddy, the higher the discrep- Also, since it is the purpose to examine the SHIPS ancy between measured and real SHA values. Due to a sensitivity to shear and SST, further complications by 10-day sampling period of the T/P altimeter, actual introducing track errors in operational runs were SHA values can be underestimated by approximately avoided. 10%. An increased availability of in situ measurements is needed to determine the error associated with the estimation of the OHC and the depths of the isotherms 4. Discussion (Huber 2000). Observations from global positioning system (GPS) a. Atmosphere dropwindsondes released in the eyewall and National To examine the external influences on Bret’s inten- Oceanic and Atmospheric Administration’s (NOAA) sity variations, its large-scale atmospheric environment WP-3D lower fuselage (LF) radar provided informa- was analyzed. Frank and Ritchie (2001) have shown the tion about the inner-core structure of the storm. Radar influence of vertical wind shear on cyclone intensity. reflectivity from the LF radar (see Jorgensen 1984a for The definition of vertical shear and methods for its cal- radar details) were mapped into a storm-relative hori- culation vary. Shear calculations in Palmer and Barnes zontal plane following Marks (1985). GPS dropwind- (2002) and Zehr (2003) were based on the difference sondes measure temperature, humidity, pressure, and between two pressure levels, usually 850 and 200 hPa or winds between the flight level and the sea surface at similar heights. Gallina (2002) used layer-averaged half-second intervals (Hock and Franklin 1999). Each winds between 925 and 700 hPa and between 300 and sonde moves approximately with the horizontal wind as 150 hPa. There are also variations in the horizontal it falls and is also influenced by updrafts and down- domain relevant to shear calculation. Zehr (2003) con- drafts during its descent. The horizontal wind speed sidered the area between 200 and 800 km away from and direction are determined from the GPS positions. the storm center, while Gallina (2002) removed the vor- Vertical motions are estimated from the difference be- tex by discarding winds closer than 400 km to the storm tween the actual fall rate of the sonde, which is deter- center at upper levels and 800 km in lower levels. De- mined hydrostatically from the thermodynamic data, Maria and Huber (1998) and Paterson et al. (2005) av- and its theoretical fall velocity. The vertical resolution eraged over a storm-centered area and consequently for all measured variables is ϳ5 m, and the accuracy of did not remove the vortex in order to cancel out the Ϫ the winds ranges between 0.5 and 2 m s 1. A total of 42 symmetric part of the primary circulation. GPS dropwindsondes deployed by the NOAA aircraft In this study, wind shear was obtained by calculating in the eyewall region of the hurricane were used. wind vector differences between two pressure levels at Our environmental analysis throughout the RI and each grid point. The model grid point closest to the subsequent weakening periods reveals significant best-track position was defined as the storm center. changes in the vertical wind shear and underlying SST/ Consequently, shear was not calculated over the same OHC. To test the sensitivity of intensity forecasts dur- area for the AVN and NCEP–NCAR analyses, which ing this period, we employ SHIPS and alter the initial may have contributed to the somewhat different shear environmental conditions. The SHIPS model is based values. Values for the 200–850 hPa vertical wind shear on statistical relationships between climatological, per- for the AVN and SHIPS analyses and NCEP–NCAR sistence and synoptic predictors (DeMaria and Kaplan reanalyses for an annulus between 200 and 800 km 1994a; DeMaria et al. 2005). The SHIPS version used in away from Bret’s center are provided in Table 1. Even this study is slightly different from the operational one. though the shear magnitudes for the three analyses dif- The runs were dependent, which is a “best case” sce- fer from each other, they show a similar trend. Shear nario. For the dependent runs the predictors are calcu- remained low within a 12-h period starting at 1200 UTC lated along the observed storm track, and analyses are 21 August 1999, but increased significantly afterward. used to calculate the atmospheric predictors. In real AVN shear at 0000 UTC 22 August indicated an in- time, the predictors are determined along the forecast crease from 7.8 to 14.0 kt within 6 h, while SHIPS val- track, and the atmospheric fields are obtained from ues increased from 8.8 to 12.0 kt. Shear based on

Unauthenticated | Downloaded 09/28/21 10:02 PM UTC NOVEMBER 2008 LOWAG ET AL. 4325

TABLE 1. Values for vertical wind shear between 200- and 850-hPa pressure levels within a domain 200 and 800 km away from the storm center. Values are provided for three different models: AVN, SHIPS, and NCEP. Shear values for AVN and SHIPS are available every 6 h between 0600 UTC 21 Aug and 0000 UTC 23 Aug 1999, whereas NCEP values are presented every 12 h. The magnitude is given in knots and the direction in degrees. Note that a direction of 90° indicates pure westerly shear.

AVN SHIPS NCEP Date Magnitude Direction Magnitude Direction Magnitude Direction 0600 UTC 21 Aug 1999 11.1 92 6.1 207 —— 1200 UTC 21 Aug 1999 9.2 108 5.6 106 4.5 140 1800 UTC 21 Aug 1999 7.8 89 8.8 87 —— 0000 UTC 22 Aug 1999 14 73 12 89 13.8 89 0600 UTC 22 Aug 1999 15.9 92 11.6 102 —— 1200 UTC 22 Aug 1999 18.2 107 13.8 118 16.7 137 1800 UTC 22 Aug 1999 16.3 99 11 118 —— 0000 UTC 23 Aug 1999 18.4 98 13 126 14.6 139

NCEP–NCAR reanalyses was more than 3 times higher influenced by upper-level wind shear. Within the 24-h than 12 h before. Within the following 24 h, the shear period starting at 0000 UTC 22 August, the shear mag- remained nearly constant or slightly increased (AVN). nitude between 200 and 850 hPa ranged from 14.0 to Because of the coarse spatial and temporal resolution, 18.4 kt. For the same time period shear values for the the NCEP shear analysis seemed less representative of 500–850-hPa levels (not shown) were between 1.7 and the atmospheric conditions than the AVN analysis and 6.2 kt indicating that strong upper-level winds were re- was not included in the further discussion. As wind sponsible for the high shear. AVN shear values for dif- shear in Bret’s vicinity increased at 0000 UTC 22 Au- ferent domains (Table 2) show lower shear for smaller gust, the storm was still rapidly intensifying for 6 more radii. Conditions for intensification were especially fa- hours. Similar results were found in a model study by vorable for an area less than 400 km away from the Persing et al. (2002) who showed that wind shear im- circulation center at 1200 and 1800 UTC 21 August. posed on Hurricane Opal (1995) did not weaken the This low shear was in best agreement with the RI of storm immediately. Also, Frank and Ritchie (2001) Bret and makes physical sense since Bret was a small found that it may take up to 36 h until the cyclone storm. At its maximum intensity, hurricane-force winds responds to increased wind shear. Hurricane Jimena only extended to 40 miles from Bret’s center. (1991) did not weaken for 2 days even though it was The AVN horizontal wind fields at the 200-hPa level exposed to shear ranging between 25 and 39 kt (Black at 1200 UTC 21 and 22 August (Fig. 4) reveal a large et al. 2002). One possible mechanism for the time lag anticyclone over Mexico. The high pressure system be- between the onset of wind shear and the vortex re- came stronger on 22 August and increased the wind sponse to it was suggested by Reasor et al. (2004) who shear in the vicinity of Bret (Table 1). At 0600 UTC 21 argued that shear-induced vortex Rossby waves reduce August the analysis (not shown) indicated a divergent the tilt of the vortex and lead to a resiliency of the flow pattern over and in proximity to Bret’s center, storm to vertical shearing. which lasted until landfall. Upper-level winds may have AVN analyses revealed that Bret was predominantly contributed to the intensification of the storm as

TABLE 2. As in Table 1, but only for AVN model for three different domains: 200–400, 0–400, and 0–800 km.

200–400 km 0–400 km 0–800 km Date Magnitude Direction Magnitude Direction Magnitude Direction 0600 UTC 21 Aug 1999 12 61 11.8 59 11.1 90 1200 UTC 21 Aug 1999 9.5 96 7.8 96 8.5 107 1800 UTC 21 Aug 1999 4.3 39 3.1 40 7.4 90 0000 UTC 22 Aug 1999 11.1 32 11.3 37 14 71 0600 UTC 22 Aug 1999 13.8 65 14.4 64 15.7 91 1200 UTC 22 Aug 1999 14.6 88 11.1 88 17.3 108 1800 UTC 22 Aug 1999 9.5 43 11.1 41 15.7 96 0600 UTC 23 Aug 1999 12 48 10.1 43 17.1 97

Unauthenticated | Downloaded 09/28/21 10:02 PM UTC 4326 MONTHLY WEATHER REVIEW VOLUME 136

Ϫ FIG. 4. AVN analysis of horizontal wind speed (m s 1) at 200 hPa at (a) 1200 UTC 21 Aug and (b) 1200 UTC 22 Aug 1999 between 15° and 40°N and 70° and 120°W. The hurricane symbol shows the position of Bret according to the best track. Circulation centers of upper- level high and low pressure systems are indicated with H and L, respectively. The black dashed line shows the trough axis associated with the upper-level low. was already discussed by Bosart et al. (2000) who sug- cycle the fluxes never exceeded the threshold of 10 gested that increased divergence related to a trough in msϪ1 dayϪ1 as defined by DeMaria et al. (1993) for the upper troposphere enhanced the convection in the hurricane–trough interaction. It is assumed that the up- inner core of Hurricane Opal (1995). The analysis for per-level low and the trough related to it did not have 1200 UTC 21 August (Fig. 4a) also indicated a small an influence on the hurricane’s intensity. In addition, upper-level low over the U.S.–Mexican border, which AVN analyses, water vapor satellite imagery, and drop- retrograded to the northwest within the next 24 h. Op- windsonde data showed that there was no intrusion of erational SHIPS forecasts were used to determine the dry air into the storm. momentum fluxes in Bret’s environment at the 200-hPa During the RI and weakening of Bret, subtle changes level within a 100–600-km annulus away from the cir- in the storm structure were noticeable (Fig. 5). Radar culation center. It was found that during the storm’s life reflectivity in the inner-core region showed a similar

Unauthenticated | Downloaded 09/28/21 10:02 PM UTC NOVEMBER 2008 LOWAG ET AL. 4327

pattern between 2250 UTC 21 August and 2118 UTC 22 August. Note that Bret was rapidly deepening during the NOAA flight on 21 August, whereas it was already weakening during the mission 24 h later. The eyewall was characterized by a wavenumber 1 asymmetry, and spiral rainbands outside of the eyewall were present on both days. The radar reflectivity on 22 Au- gust was lower than on 21 August, and its structure changed significantly late on 22 August (Fig. 5c) as the storm appears to have become more asymmetric. While the area of high reflectivity in the eyewall (Ͼ40 dBZ) encompassed the western semicircle at 2118 UTC 22 August (Fig. 5b), it was significantly reduced 2.5 h later and only extended to the northwestern quadrant. Also note that the reflectivity in the rainband region was considerably lower indicating reduced convection and weakening of the storm. The weakening was related to changes in vertical motions, which were apparent in the profiles of dropwindsonde vertical velocity observations in Fig. 6. Profiles were calculated using 50-m intervals in the vertical. Mean motions in a vertical bin were computed from all available dropwindsondes. The maximum and minimum values were the most extreme data points observed by any of the available sondes. The profiles were obtained from 22 sondes for 21 August and 20 sondes for 22 August. Mean up- and downdrafts were stronger on 21 than on 22 August. Updrafts on 21 August remained nearly constant at about 2 m sϪ1 up to 1000 m and gradually increased aloft to approximately 6 m sϪ1 at 3800 m. On 22 August they showed roughly the same magnitude of 2 m sϪ1 throughout the whole atmospheric layer below the flight level. Averaged downdrafts revealed a constant profile on both days, but were stronger on 21 August as are maximum values for up- and downdrafts. Also note that the mean profile of all drafts shows stronger vertical motions above 2500 m on 21 August, and was similar below on both days. The higher vertical velocities on the first day were a consequence of stronger vertical overturning due to enhanced convection during Bret’s intensification. Down- drafts are typically located radially inside and outside from strong updrafts (Jorgensen 1984b; Black and Hal- lett 1999). The weaker convection on 22 August is in agreement with the lower radar reflectivity in the inner core (Fig. 5). The mean profiles on 21 August are similar to the findings of Black et al. (1996) who derived vertical motions in hurricanes from Doppler radar data. Values for mean up- and downdrafts were approxi- FIG. 5. Composite of LF radar reflectivity (dBZ) observed in mately 1 m sϪ1 weaker in Bret’s case. Also note that the Hurricane Bret at (a) 2250 UTC 21 Aug, (b) 2118 UTC 22 Aug, and (c) 2340 UTC 22 Aug 1999. The black cross in the eye indi- negative values for the average of all drafts in Bret’s cates the position of the NOAA aircraft at the time of the scan. case were not observed in the aforementioned work. The domain of the scan is 240 km ϫ 240 km. The release of latent heat through condensation in

Unauthenticated | Downloaded 09/28/21 10:02 PM UTC 4328 MONTHLY WEATHER REVIEW VOLUME 136

Ϫ FIG. 6. Average vertical velocity (m s 1) as a function of height (m) measured by GPS dropwindsondes deployed on NOAA flight missions on (a) 21 Aug and (b) 22 Aug 1999. The thick black solid line shows the averaged profile for all winds. The medium black solid line represents the averaged updrafts, while the thin black solid line indicates maximum updrafts. The medium and thin gray solid lines show averaged and maximum downdrafts, respectively. the eyewall region leads to TC intensification. There- the SSTs measured by AXBTs deployed during the pe- fore it seems plausible that because of the reduced con- riod of RI and subsequent weakening. At the beginning vection in Bret’s eyewall region on 22 August, the of the deepening at 1200 UTC 21 August, the hurricane transport of moist air from the boundary layer to higher moved over relatively warm waters of about 29°C. With altitudes (as described in Black et al. 1996) was dimin- wind shear values of less than 10 kt, the storm encoun- ished and could not maintain the strength of the cy- tered very favorable conditions for RI (Kaplan and De- clone. The size of Bret’s eye on late 21 August (Fig. 5a) was about 30 km in diameter and showed a slight contraction to approximately 28 km 24 h later (Fig. 5c). Even though eye contraction is indicative of storm strengthening (Willoughby et al. 1982; Willoughby 1990), Bret was weakening because of the less vigorous convection. The composites presented herein were only used to document gross features and evolution of Bret’s secondary circulation during the deepening and weakening period. Finally, the southwesterly shear caused a wavenumber 1 asymmetry in the radar reflectivity images of the inner core (Fig. 5). It has been shown that the highest reflectivity is generally observed to the left of the shear vector (Marks and Gamache 1992; Franklin et al. 1993; Reasor et al. 2000; Black et al. 2002; Eastin et al. 2005). The highest reflectivity in the eyewall of Bret was in the northwest quadrant on both 21 and 22 August, consistent with the findings of these studies. b. Ocean FIG. 7. Selected SST observations (°C) for AXBT locations on 21 Aug (squares) and 22 Aug 1999 (inverted triangles). The Along with atmospheric interactions, Hurricane dashed line shows the best track. Best-track values for MSLP in Bret’s intensity changes were associated with the oce- hPa and maximum sustained winds (kt) are given every 6 h be- anic conditions in the Gulf of Mexico. Figure 7 shows ginning at 1200 UTC 21 Aug 1999.

Unauthenticated | Downloaded 09/28/21 10:02 PM UTC NOVEMBER 2008 LOWAG ET AL. 4329

Ϫ2 TABLE 3. Values for the OHC (kJ cm ), and the depth of the 26°C isotherm (h26; m) for model grid points closest to the best-track positions for Hurricane Bret every 6 h between 0000 UTC 21 Aug and 0000 UTC 23 Aug 1999. The model has a spatial resolution of 0.5°. Hence, discrepancies between model grid points and best-track positions are limited to a few tenths of a degree.

OHC h26 Date 21 Aug 22 Aug 23 Aug 21 Aug 22 Aug 23 Aug 0000 UTC 21 Aug 1999 66.3 66.3 66.1 81.2 81.2 81 0600 UTC 21 Aug 1999 74.1 74.4 74.5 89.4 89.7 89.8 1200 UTC 21 Aug 1999 90.4 87.5 86.9 109.6 106.1 105.3 1800 UTC 21 Aug 1999 69.1 59.2 57.3 84.7 72.6 70.2 0000 UTC 22 Aug 1999 54.8 45.7 44.5 67.7 56.5 54.9 0600 UTC 22 Aug 1999 58.5 58.4 58.1 77.4 77.3 77 1200 UTC 22 Aug 1999 70.6 70.4 70.4 100 99.8 99.8 1800 UTC 22 Aug 1999 26 26 26 41.8 41.8 42 0000 UTC 23 Aug 1999 13.2 13.2 13.2 22 22 22

Maria 2003). Afterward, the SSTs were gradually in- ables remained nearly constant for the best-track posi- creasing, and Bret encountered temperatures of ap- tions at 0000 and 0600 UTC 21 August, and the 18-h proximately 30°C at 0600 UTC 22 August, 6 h before period starting at 0600 UTC 22 August. The oceanic the storm reached its lowest MSLP. The high SSTs were conditions changed markedly during the RI and Bret’s related to two warm eddies that Bret passed to its right subsequent weakening phase. The OHC at Bret’s posi- (Huber 2000). With air temperatures measured by GPS tion at 1200 UTC 21 August, when the pressure drop dropwindsondes of approximately 25°C close to the sea began, was 90.4 kJ cmϪ2, and about 20% higher than surface in the eyewall region, strong sensible and latent where the storm center was 6 h before. The passage heat fluxes probably contributed to the strengthening over this first warm pool contributed to Bret’s intensi- of the cyclone. As it approached the Texas coast, the fication. Similar results were found for Hurricane Opal SSTs were significantly reduced. AXBT measurements (1995) by Hong et al. (2000) who showed in a numerical on 22 August revealed relatively cool shelf waters be- experiment that the hurricane’s MSLP while passing tween 25° and 26°C while the air temperature at the over a warm-core ring (WCR) in the Gulf of Mexico air–sea interface in the eyewall slightly dropped to dropped more than twice as much as compared to when 24°C. Cione and Uhlhorn (2003) found that small varia- the WCR was removed. OHC values for the same po- tions in SST in the inner core of a TC can tremendously sition are 3.2% lower on 22 August, and 3.9% on 23 modify air–sea fluxes in this region. For example, under August. The lower heat contents were a consequence of high-wind conditions a 1-K reduction of inner-core SST reduced h26, since weekly averaged Reynolds SST used may alter sensible and latent heat fluxes by as much as in this analysis method remained constant over the 40%. Reduced heat fluxes and consequently Bret’s 3-day period. Note that due to the deep thermocline the weakening to a category 3 storm were attributed to the upper ocean was not mixed and SSTs consequently re- lower SSTs in the coastal region. mained high as indicated by Lin et al. (2005) who sug- Cooler SSTs through hurricane-induced upwelling of gest that warm eddies insulate TCs from cold water by colder water may influence the intensity of subsequent preventing entrainment at the bottom of the oceanic storms when passing over the cold wake (Brand 1971). mixed layer (OML). Walker et al. (2005) also showed that cold SSTs in- Six hours later at 1800 UTC 21 August, OHC values duced by upwelling in the northern Gulf of Mexico at Bret’s storm center were lower, and relative changes caused a negative feedback to Hurricane Ivan (2004) more pronounced with Ϫ14.3% and Ϫ17.1% on 22 Au- resulting in a weakening of the storm. Consequently, gust and 23 August, respectively. Lower OHCs were a the authors suggest that AXBT measurements of SST result of stress-induced upwelling of cool water and are crucial in order to improve the intensity forecast for heat fluxes from the ocean to the atmosphere (Kraus landfalling TCs. and Turner 1967) where both stress and heat fluxes Since SST observations are necessary but not suffi- increase with wind speed. Shear-induced entrainment cient for TC intensity studies, the conditions of the up- of cooler water at the base of the OML (Pollard et al. per ocean were also examined. Values for the OHC and 1973) also contributed to the decrease of the 26°C iso- the depth of the 26°C isotherm (h26) along Bret’s best therm depth. The hurricane passed the second warm track between 0000 UTC 21 August and 0000 UTC 23 core on 1200 UTC 22 August. Even though the OHC in August are presented in Table 3. Values for both vari- the inner core was 20% higher than 6 h before, the

Unauthenticated | Downloaded 09/28/21 10:02 PM UTC 4330 MONTHLY WEATHER REVIEW VOLUME 136 storm did not intensify. One plausible explanation is Note that the modified SSTs were 0.2°C cooler on av- that the vertical wind shear reached its peak at the same erage within the first 24 h than the weekly averaged time (Table 1 and Table 2) and outweighed the positive Reynolds SSTs along the best track used in CTRL. For feedback of the ocean. the last 2 times, the modified SST were 2.4° As Bret approached the Texas coast, it encountered and 3.3°C, respectively, lower. Finally, the AVN28TEMP waters with OHCs of about 26 kJ cmϪ2 due to the run was a combination and used shear values from relatively shallow shelf and cooler SSTs. Three-day SST AVN28 and SSTs from TEMP in order to simulta- composites from AVHRR data showed warm SSTs neously study the effects of shear and SST on the in- (ϳ28°–29°C) along the Texas coast on 22 August. How- tensity forecasts. ever, the composite for 24 August indicated SSTs of For all forecasts between 1800 UTC 18 August and less than 27°C for the same area, which is in agreement 1800 UTC 22 August the average deviation from the with AXBT measurements. It is suggested that storm- CTRL was calculated for the other three cases. The induced mixing ahead of the storm was responsible for control run showed the highest intensity forecast. The the lower SSTs. Moreover, the storm’s reduced trans- predicted maximum sustained winds for TEMP were lational speed of only 2.5 m sϪ1 near landfall enhanced 0.91 kt lower than for CTRL. The largest differences the mixing of the upper-ocean layer (Huber 2000) and were obtained for AVN28 (1.47 kt) and AVN28TEMP led to a negative feedback on the hurricane. Conse- (2.43 kt). The relatively small differences are due to the quently, the convection in Bret’s inner-core region was linear regression method that SHIPS is based on and is reduced due to a decrease in surface heat fluxes. not indicative of the physical interactions for a particu- Emanuel et al. (2004) showed in a modeling study lar storm. Each predictor only explains a small percent- that Hurricane Bret underwent a RI shortly before age of the total variation in the forecasts (DeMaria and landfall due to the shoaling effect where the rising sea- Kaplan 1994a). The intensity forecasts for selected floor meets the base of the OML and thus prevents cases at 0600 UTC 21 August and 0000 UTC 22 August further ocean cooling. Flight-level observations from are shown in Fig. 8. Both dates were chosen to illustrate Air Force reconnaissance aircraft indicated a short pe- the sensitivity of the SHIPS model during the period of riod of reintensification for Bret. However, the authors RI and subsequent weakening. suggest that this period was related to internal dynam- The forecasts from 0600 UTC 21 August for the ex- ics of the hurricane, as will be shown in Part II of this amined cases differed only by 1 kt within the first 18 h study, since the waters in the very shallow shelf along of the forecast period. After that, the differences be- the Texas coast have already been well mixed prior to came a bit more distinct. Note that the cyclone was still Bret’s approach and thus prevented the shoaling effect. intensifying at 0000 UTC 22 August, even though the shear in the AVN model (AVN28) nearly doubled from 6 h before. The shear in the control run increased by 5. Sensitivity studies with the SHIPS model approximately one-third at the same time. TEMP As shown in Fig. 3, SHIPS intensity forecasts for showed lower wind speeds since the modified SST val- Hurricane Bret were inaccurate. The quality of model ues were slightly cooler than in CTRL. Bret also had a forecasts partially relies on the accuracy of the initial lower intensity in AVN28 than in CTRL. This was a conditions. Sensitivity studies with the SHIPS model consequence of an averaged 5.7-kt-higher shear in the were conducted for which the initial conditions for wind AVN model. The lowest wind speeds were predicted in shear and SST were modified. Four different cases were AVN28TEMP, which was a result of the higher shear studied. First, a control run (CTRL) using vertical wind and lower SST values compared to the original initial shear from NCEP–NCAR reanalyses as the back- conditions. ground field was performed. Second, vertical shear val- For the forecasts from 0000 UTC 22 August, the in- ues were derived from AVN analyses within a domain tensity for TEMP was lower than in CTRL, but higher between 200 and 800 km away from the storm center than in AVN28 because SHIPS is less sensitive to SST (AVN28, see also Table 1). Third, the vertical shear changes when the storm is closer to its maximum po- from the CTRL run was combined with modified SSTs tential intensity (DeMaria and Kaplan 1994b). Also, to produce the TEMP run. In particular, the SST for the influence of shear in the SHIPS model is increasing 1200 UTC 21 August was set to 29.0°C and then in- with storm intensity. Consequently, the influence of creased to 29.5°C for the 18-h period beginning at 1800 cooler SSTs in TEMP on Bret’s weakening was smaller UTC 21 August. Cooler SST of 26.5°C were imposed than the impact of strong shear in the AVN model in on 1800 UTC 22 August, and 25.5°C 6 h later, that AVN28. Due to warmer SST and lowest shear values, represented the cool shelf waters along the Texas coast. CTRL predicted the highest intensity for Bret. It

Unauthenticated | Downloaded 09/28/21 10:02 PM UTC NOVEMBER 2008 LOWAG ET AL. 4331

shows that SHIPS and, therefore intensity forecasts, are very sensitive to SST and OHC. Since SST is directly related to OHC beneath an intense hurricane, it implies that OHC is a very important parameter. This is consistent with the results of observational studies of Bosart et al. (2000), Shay et al. (2000), and Cione and Uhlhorn (2003) and the theoretical work by Schade and Emanuel (1999). The general lack of sensitivity further underscores the importance of the internal processes during an RI event, which are completely ignored in SHIPS. It also indicates that initial conditions play an important role in model predictions of cyclone intensities. Rhome et al. (2006) have done a similar study with the SHIPS model in which they correlated vertical shear estimates to changes in intensity. They found for the 2005 hurricane season that the 850–500 hPa shear over an annulus with inner and outer radius of 200 km and 800 km, respectively, was most highly correlated with intensity changes. Finally, it should be noted that SHIPS is based on a regression technique that uses climatological, persis- tence, and synoptic predictors. Since each predictor only explains a small fraction of the intensity variation (DeMaria and Kaplan 1994a), extreme events such as RI tend to be underpredicted.

6. Summary Hurricane Bret underwent an RI and subsequent weakening between 1200 UTC 21 August and 1200 UTC 22 August 1999 in the Gulf of Mexico before it made landfall on the Texas coast 12 h later. Its minimum sea level pressure fell 35 hPa from 979 to 944 hPa within 24 h. Both the National Hurricane Center and SHIPS did not forecast the observed intensity changes very well for this time period, underestimating both the amount of strengthening and weakening. AVN analyses of Bret’s large-scale environment on 21 August revealed favorable atmospheric conditions at the beginning of the RI with vertical wind shear relax-

FIG. 8. SHIPS forecasts for maximum sustained wind speed (kt) ing below 10 kt. The simultaneous passage over a warm for the time between (top) 0600 UTC 21 Aug and 0000 UTC 23 eddy with increased OHC and SSTs above 29°C sup- Aug 1999 and (bottom) 0000 UTC 22 Aug and 0000 UTC 23 Aug ported the strengthening of the storm. Increasing shear 1999 for four different cases: CTRL (case 1), AVN28 (case 3), due to an anticyclone in the upper troposphere in as- TEMP (case 6), and AVN28TEMP (case 7). Forecasts are given sociation with reduced OHC values and cooler SSTs of every 6 h. All forecasts are compared to the best track (solid line). ϳ26°C along the Texas coast contributed to Bret’s weakening on 22 August. should also be pointed out that the modeled intensities Shear values based on the AVN and NCEP model were higher than the observed ones. Nevertheless, the output were calculated for different radii and compared discrepancies between the forecasts and the best track to the SHIPS wind shear. Differences in the magnitude during the weakening stage were significantly smaller between the two models were shown, but both AVN than during the RI. and NCEP shear had a similar trend that was also rep- Even though the differences between the three cases resented in SHIPS. AVN shear values were lower for and the control run were small, this sensitivity study smaller radii and indicated the most favorable condi-

Unauthenticated | Downloaded 09/28/21 10:02 PM UTC 4332 MONTHLY WEATHER REVIEW VOLUME 136 tions for storm intensification. This is in agreement with Bosart, L. F., C. S. Velden, W. E. Bracken, J. Molinari, and P. G. observations, since Bret was a relatively small storm. Black, 2000: Environmental influences on the rapid intensi- The shear influence was noticeable in the inner-core fication of Hurricane Opal (1995) over the Gulf of Mexico. Mon. Wea. Rev., 128, 322–352. structure of the hurricane. Reflectivity data from lower Boyd, J. D., 1987: Improved depth and temperature conversion fuselage radar showed pronounced wavenumber asym- equations for Sippican AXBTs. J. Atmos. Oceanic Technol., metries left of the shear vector on both days. 4, 545–551. Significant structural changes were observed within 2 Brand, S., 1971: The effects on a tropical cyclone of cooler surface h on late 22 August, as the storm appeared to be more waters due to upwelling and mixing produced by a prior tropical cyclone. J. Appl. Meteor., 10, 865–874. asymmetric. The area of high reflectivity in the rain- ——, 1973: Rapid intensification and low-latitude weakening of band region became considerably smaller and indicated tropical cyclones of the western North Pacific Ocean. J. Appl. weaker convection in the inner core. Profiles of vertical Meteor., 12, 94–103. motions in the eyewall region obtained from GPS drop- Chan, J. C. L., Y. Duan, and L. K. Shay, 2001: Tropical cyclone windsonde measurements revealed average updrafts of intensity change from a simple ocean–atmosphere coupled Ϫ1 Ϫ1 model. J. Atmos. Sci., 58, 154–172. up to 6 m s , while they reached only 2 m s on the Cheney, R. L., L. Miller, R. Agreen, N. Doyle, and J. Lillibridge, next day. Averaged downdrafts reveal a constant pro- 1994: TOPEX/POSEIDON: The 2 cm solution. J. Geophys. file on both days, but are stronger on 21 August as are Res., 99, 24 555–24 563. maximum values for up- and downdrafts. Cione, J. J., and E. W. Uhlhorn, 2003: Sea surface temperature Sensitivity studies with the SHIPS model were con- variability in hurricanes: Implications with respect to intensity change. Mon. Wea. Rev., 131, 1783–1796. ducted. Four different cases were studied for which DeMaria, M., 1996: The effect of vertical shear on tropical cyclone modified shear and SST values were used for model intensity change. J. Atmos. Sci., 53, 2076–2087. initialization. The results indicate that SHIPS and ——, and J. Kaplan, 1994a: A Statistical Hurricane Intensity Pre- therefore intensity forecasts are very sensitive to SST diction Scheme (SHIPS) for the Atlantic Basin. Wea. Fore- and OHC and stress the importance of the OHC in casting, 9, 209–220. forecasting tropical cyclone intensities consistent with ——, and ——, 1994b: Sea surface temperature and the maximum intensity of Atlantic tropical cyclones. J. Climate, 7, 1324– the results of observational studies and theoretical 1334. work. The general lack of sensitivity underscores the ——, and M. M. Huber, 1998: The effect of vertical shear on importance of internal processes, which are not incor- tropical cyclone intensity change: An historical perspective. porated into SHIPS, during an RI event. Preprints, Symp. on Tropical Cyclone Intensity Change, While the first part of the paper solely described the Phoenix, AZ, Amer. Meteor. Soc., 22–29. ——, J. J. Baik, and J. Kaplan, 1993: Upper-level eddy angular environmental influences and the temporal changes momentum fluxes and tropical cyclone intensity change. J. thereof on Bret’s intensity, Part II will discuss the dra- Atmos. Sci., 50, 1133–1147. matic evolution of the inner core in detail. ——, M. Mainelli, L. K. Shay, J. A. Knaff, and J. Kaplan, 2005: Further improvements to the Statistical Hurricane Intensity Acknowledgments. The authors thank Dr. Mark De- Prediction Scheme (SHIPS). Wea. Forecasting, 20, 531–543. Maria and John Kaplan for rerunning the SHIPS model Dunion, J. P., and C. S. Velden, 2004: The impact of the Saharan air layer on Atlantic tropical cyclone activity. Bull. Amer. and their assistance with data interpretation. Dr. Lynn Meteor. Soc., 85, 353–365. K. Shay and Jodi Brewster of the University of Miami Eastin, M. D., W. M. Gray, and P. G. Black, 2005: Buoyancy of are kindly acknowledged for providing data about the convective vertical motions in the inner core of intense hur- oceanic conditions and helpful comments. Discussions ricanes. Part II: Case studies. Mon. Wea. Rev., 133, 209–227. with Dr. Sim Aberson of the Hurricane Research Di- Emanuel, K. A., 1999: Thermodynamic control of hurricane intensity. Nature, 401, 665–669. vision (HRD) significantly improved the quality of the ——, C. DesAutels, C. Holloway, and R. Korty, 2004: Environ- paper, and HRD’s Ms. Sonia Otero is thanked for her mental control of tropical cyclone intensity. J. Atmos. Sci., 61, editorial help on an earlier draft of the paper. 843–858. Fink, A. H., and D. G. Vincent, 2003: Tropical cyclone environ- REFERENCES ments over the northeastern Pacific, including mid-level dry intrusions. Meteor. Atmos. Phys., 84, 293–315. Black, M. L., R. W. Burpee, and F. D. Marks, 1996: Vertical mo- Frank, W. M., and E. A. Ritchie, 2001: Effects of vertical wind tion characteristics of tropical cyclones determined with air- shear on the intensity and structure of numerically simulated borne Doppler radial velocities. J. Atmos. Sci., 53, 1887–1909. hurricanes. Mon. Wea. Rev., 129, 2249–2269. ——, J. F. Gamache, F. D. Marks, C. E. Samsury, and H. E. Wil- Franklin, J. L., S. J. Lord, S. E. Feuer, and F. D. Marks, 1993: The loughby, 2002: Eastern Pacific Hurricanes Jimena of 1991 and kinematic structure of Hurricane Gloria (1985) determined Olivia of 1994: The effect of vertical shear on structure and from nested analyses of dropwindsonde and Doppler radar intensity. Mon. Wea. Rev., 130, 2291–2312. data. Mon. Wea. Rev., 121, 2433–2451. Black, R. A., and J. Hallett, 1999: Electrification of the hurricane. Gallina, G. M., 2002: Environmental vertical wind shear and J. Atmos. Sci., 56, 2004–2028. tropical cyclone intensity change utilizing enhanced satellite

Unauthenticated | Downloaded 09/28/21 10:02 PM UTC NOVEMBER 2008 LOWAG ET AL. 4333

wind information. M.S. thesis, Dept. of Atmospheric and inner core of Hurricane Norbert (1984). Part I: Kinematic Oceanic Sciences, University of Wisconsin—Madison, Madi- structure. J. Atmos. Sci., 49, 919–942. son, WI, 133 pp. Mayfield, M., L. A. Avila, and E. N. Rappaport, 1994: Atlantic Hanley, D., J. Molinari, and D. Keyser, 2001: A composite study hurricane season of 1992. Mon. Wea. Rev., 122, 517–538. of the interactions between tropical cyclones and upper- Molinari, J., and D. Vollaro, 1989: External influences on Hurri- tropospheric troughs. Mon. Wea. Rev., 129, 2570–2584. cane intensity. Part I: Outflow layer eddy angular momentum Heymsfield, G. M., J. B. Halverson, J. Simpson, L. Tian, and T. P. fluxes. J. Atmos. Sci., 46, 1093–1105. Bui, 2001: ER-2 Doppler radar investigations of the eyewall Montgomery, M. T., and J. R. Kallenbach, 1997: A theory for vor- of Hurricane Bonnie during the Convection and Moisture tex Rossby waves in a three-dimensional quasigeostrophic Experiment-3. J. Appl. Meteor., 40, 1310–1330. model. Quart. J. Roy. Meteor. Soc., 123, 435–465. Hock, T. F., and J. L. Franklin, 1999: The NCAR GPS dropwind- Palmen, E., 1948: On the formation and structure of tropical cy- sonde. Bull. Amer. Meteor. Soc., 80, 407–420. clones. Geophysika, 3, 26–38. Holliday, C. R., and A. H. Thompson, 1979: Climatological char- Palmer, C. K., and G. M. Barnes, 2002: The effects of vertical acteristics of rapidly intensifying typhoons. Mon. Wea. Rev., wind shear as diagnosed by NCEP/NCAR reanalysis data on 107, 1022–1034. Northeast Pacific hurricane intensity. Preprints, 25th Conf. Hong, X., S. W. Chang, S. Raman, L. K. Shay, and R. Hodur, on Hurricanes and Tropical Meteorology, San Diego, CA, 2000: The interaction between Hurricane Opal (1995) and a Amer. Meteor. Soc., 122–123. warm core ring over the Gulf of Mexico. Mon. Wea. Rev., Paterson, L. A., B. Hanstrum, N. Davidson, and H. Weber, 2005: 128, 1347–1365. Influence of environmental vertical shear on the intensity of Huber, M. M., 2000: On the role of the upper ocean in tropical hurricane-strength tropical cyclones in the Australian region. cyclone intensity change. M.S. thesis, Division of Meteorol- Mon. Wea. Rev., 133, 3644–3660. ogy and Physical Oceanography, University of Miami, Mi- Persing, J., M. T. Montgomery, and R. E. Tuleya, 2002: Environ- ami, FL, 73 pp. mental interactions in the GFDL hurricane model for Hur- Jorgensen, D. F., 1984a: Mesoscale and convective-scale charac- ricane Opal. Mon. Wea. Rev., 130, 298–317. teristics of mature hurricanes. Part I: General observations by Pielke, R. A., Jr., and R. A. Pielke Sr., 1997: Hurricanes: Their research aircraft. J. Atmos. Sci., 41, 1268–1286. Nature and Impact on Society. John Wiley and Sons, 279 pp. ——, 1984b: Mesoscale and convective-scale characteristics of ma- Pollard, R. T., P. B. Rhines, and R. O. R. Y. Thompson, 1973: The ture hurricanes. Part II: Inner core structure of Hurricane deepening of the wind-mixed layer. Geophys. Fluid Dyn., 3, Allen (1980). J. Atmos. Sci., 41, 1287–1311. 381–404. Kaplan, J., and M. DeMaria, 2003: Large-scale characteristics of Rappaport, E. N., 1994: Hurricane Andrew. Weather, 49, 51–61. rapidly intensifying tropical cyclones in the North Atlantic. Reasor, P. D., M. T. Montgomery, F. D. Marks Jr., and J. F. Wea. Forecasting, 18, 1093–1108. Gamache, 2000: Low-wavenumber structure and evolution of Kelley, O. A., J. Stout, and J. B. Halverson, 2004: Tall precipita- the hurricane inner core observed by airborne dual-Doppler tion cells in tropical cyclone eyewalls are associated with radar. Mon. Wea. Rev., 128, 1653–1680. tropical cyclone intensification. Geophys. Res. Lett., 31, ——, ——, and L. D. Grasso, 2004: A new look at the problem of L24112, doi:10.1029/2004GL021616. tropical cyclones in vertical shear flow: Vortex resiliency. J. Kossin, J. P., and M. D. Eastin, 2001: Two distinct regimes in the Atmos. Sci., 61, 3–22. kinematic and thermodynamic structure of the hurricane eye Rhome, J. R., C. A. Sisko, and R. D. Knabb, 2006: On the calcu- and eyewall. J. Atmos. Sci., 58, 1079–1090. lation of vertical shear: An operational perspective. Pre- ——, and W. H. Schubert, 2001: Mesovortices, polygonal flow prints, 27th Conf. on Hurricanes and Tropical Meteorology, patterns, and rapid pressure falls in hurricane-like vortices. J. Monterey, CA, Amer. Meteor. Soc., 14A.4. [Available online Atmos. Sci., 58, 2196–2209. at http://ams.confex.com/ams/pdfpapers/108724.pdf.] Kraus, E. B., and J. S. Turner, 1967: A one-dimensional model of Schade, L. R., and K. A. Emanuel, 1999: The ocean’s effect on the the seasonal thermocline. Part II: The general theory and its intensity of tropical cyclones: Results from a simple coupled consequences. Tellus, 19, 98–106. atmosphere–ocean model. J. Atmos. Sci., 56, 642–651. Lawrence, M. B., B. M. Mayfield, L. A. Avila, R. J. Pasch, and Schubert, W. H., M. T. Montgomery, R. K. Taft, T. A. Guinn, E. N. Rappaport, 1998: Atlantic hurricane season of 1995. S. R. Fulton, J. P. Kossin, and J. P. Edwards, 1999: Polygonal Mon. Wea. Rev., 126, 1124–1151. eyewalls, asymmetric eye contraction, and potential vorticity ——, L. A. Avila, J. K. Beven, J. L. Franklin, J. L. Guiney, and mixing in hurricanes. J. Atmos. Sci., 56, 1197–1223. R. J. Pasch, 2001: Atlantic hurricane season of 1999. Mon. Shay, L. K., G. J. Goni, and P. G. Black, 2000: Effects of a warm Wea. Rev., 129, 3057–3084. oceanic feature on Hurricane Opal. Mon. Wea. Rev., 128, Leipper, D. F., and D. Volgenau, 1972: Hurricane heat potential 1366–1383. of the Gulf of Mexico. J. Phys. Oceanogr., 2, 218–224. Walker, N. D., R. R. Leben, and S. Balasubramanian, 2005: Hur- Lin, I. I., C. C. Wu, K. A. Emauel, I. H. Lee, C. R. Wu, and I. F. ricane-forced upwelling and chlorophyll a enhancement Pun, 2005: The interaction of supertyphoon Maemi (2003) within cold-core cyclones in the Gulf of Mexico. Geophys. with a warm ocean eddy. Mon. Wea. Rev., 133, 2635–2649. Res. Lett., 32, L18610, doi:10.1029/2005GL023716. Mainelli, M., M. DeMaria, G. Goni, and L. K. Shay, 2008: Appli- Willoughby, H. E., 1990: Temporary changes in the primary cir- cation of oceanic heat content estimation to operational fore- culation of tropical cyclones. J. Atmos. Sci., 47, 242–264. casting of recent Atlantic category 5 hurricanes. Wea. Fore- ——, J. A. Clos, and M. G. Shoreiba, 1982: Concentric eyewalls, casting, 23, 3–16. secondary wind maxima, and the evolution of the hurricane Marks, F. D., Jr., 1985: Evolution of the structure of precipitation vortex. J. Atmos. Sci., 39, 395–411. in Hurricane Allen (1980). Mon. Wea. Rev., 113, 909–930. Zehr, R. M., 2003: Environmental vertical wind shear with Hur- ——, and J. F. Gamache, 1992: Dual-aircraft investigation of the ricane Bertha (1996). Wea. Forecasting, 18, 345–356.

Unauthenticated | Downloaded 09/28/21 10:02 PM UTC