1Tropical Cyclone Report
|
|  Yüklə 1.72 Mb.
|
| 1Tropical Cyclone Report
Hurricane Dean (AL042007) 13-23 August 2007 James L. Franklin National Hurricane Center 31 January 2008
(for damage statistics in Martinique and Guadeloupe and to add acknowledgments) Revised 7 April 2008 (corrected wind speeds in Table 1 between 16/1200 UTC and 17/0600 UTC) Dean was a classic Cape Verde cyclone that moved through the Caribbean as a major hurricane, passing very close to Jamaica and later making landfall on the east coast of the Yucatan Peninsula as a category 5 hurricane. Dean, the first Atlantic hurricane since Andrew of 1992 to reach land at category 5 strength, was responsible for 32 deaths.
Dean originated from a well-defined tropical wave that crossed the west coast of Africa on 11 August. The wave was associated with a closed surface low even before entering the Atlantic, but strong easterly shear kept the system’s convection displaced from an elongated circulation center for a couple of days. By about 0600 UTC 13 August, however, the circulation became better defined and sufficiently connected to the deep convection to consider the low a tropical depression about 350 n mi west-southwest of Praia in the Cape Verde Islands. The “best track” chart of the tropical cyclone’s path is given in Fig. 1, with the wind and pressure histories shown in Figs. 2 and 3, respectively. The best track positions and intensities are listed in Table 1. Embedded in a strong easterly current, the depression initially moved westward at about 20 kt. The environment was still characterized by easterly shear, and the depression took 30 h to reach storm strength; this occurred at 1200 UTC 14 August, when the system was centered about 1250 n mi east of Barbados. Although the cyclone’s satellite presentation remained ragged, Dean strengthened early the next day as it turned to the west-northwest, still moving briskly. Dean would continue on this heading, to the south of a deep-layer ridge of high pressure, for the next seven days. The easterly shear gradually abated, and by late on 15 August well-defined convective banding had developed around the center and microwave data showed the formation of a partial eyewall. Shortly thereafter, infrared satellite imagery began to show an eye, and Dean became a hurricane early on 16 August about 480 n mi east of Barbados. As upper-level outflow became more pronounced, Dean reached an intensity of 80 kt by 1200 UTC that day, but the eyewall then disappeared and the strengthening trend temporarily halted. Dean entered the Caribbean Sea on 17 August, its center passing between Martinique and St. Lucia around 0930 UTC (Fig. 4). The northern eyewall, accompanied by sustained winds of about 85 kt, passed directly over Martinique. With upper-level outflow increasing in all quadrants, Dean then began to strengthen rapidly in the eastern Caribbean Sea, its winds increasing from 80 to 145 kt (category 1 to category 5 on the Saffir-Simpson Hurricane Scale) in the 24 h ending at 0600 UTC 18 August. At 1200 UTC that day, Dean’s minimum central pressure was 923 mb. During this period of rapid deepening Dean’s forward motion slowed to about 15 kt; its heading, however, remained remarkably constant throughout the Caribbean as high pressure was maintained to the north of the hurricane, while a mid- to upper-level low over south Florida on 18 August retreated westward in tandem with the tropical cyclone. The center of Dean passed about 180 n mi south of Puerto Rico early on 18 August, and continued west-northwestward in the direction of Jamaica. By 1200 UTC that day, microwave imagery showed a concentric eyewall structure (Fig. 5). As the inner eyewall eroded over the next 12 h, Dean’s maximum sustained winds decreased from 145 kt to 120 kt (Category 4). Interestingly, the central pressure fell slightly during this time, dropping below 920 mb for a short period early on 19 August (Fig 3). Dean remained a Category 4 hurricane as its center passed within about 80 n mi of the south coast of Haiti during the morning of 19 August, and within about 20 n mi of the south coast of Jamaica that evening. Dean’s intensity as it passed Jamaica is estimated to be 125 kt, although reconnaissance data suggest that Dean’s strongest winds remained just offshore. On 20 August Dean moved away from Jamaica over the deep warm waters of the northwestern Caribbean. The convective structure that day was dominated by a single eyewall, and under light shear Dean began to strengthen as it approached the Yucatan Peninsula. As the eyewall contracted, Dean regained Category 5 status near 0000 UTC 21 August, and was still deepening when the center made landfall near the town of Majahual in the Costa Maya tourist region of the Yucatan near 0830 UTC that day. At the time of landfall, Dean is estimated to have had a minimum central pressure of 905 mb and maximum sustained winds of 150 kt, making it the first land-falling Category 5 hurricane in the Atlantic basin since Andrew of 1992. Dean weakened as it moved across the Yucatan Peninsula, emerging into the Bay of Campeche around 1900 UTC. Although Dean maintained hurricane strength throughout its 10-h passage over land, its inner core convective structure was largely disrupted. Aircraft reconnaissance data in the Bay of Campeche showed that the cyclone’s radius of maximum wind had expanded to roughly 55 n mi, and Dean was only able to recapture a small fraction of its former strength. Deep-layer high pressure along the northern coast of the Gulf of Mexico kept Dean on its west-northwestward track until 1200 UTC 22 August, when the cyclone turned to the west. Dean made landfall at 1630 UTC that day near the town of Tecolutla, Mexico, about 90 n mi northeast of Veracruz, as a Category 2 hurricane with winds of 85 kt. Dean weakened rapidly after landfall, becoming a depression by 0000 UTC 23 August, and dissipating over the mountains of central Mexico shortly thereafter.
Observations in Dean (Figs. 2 and 3) include satellite-based Dvorak technique intensity estimates from the Tropical Analysis and Forecast Branch (TAFB) and the Satellite Analysis Branch (SAB), as well as flight-level, Stepped Frequency Microwave Radiometer (SFMR) and dropwindsonde observations from flights of the 53rd Weather Reconnaissance Squadron of the U. S. Air Force Reserve Command (AFRC). Microwave satellite imagery from NOAA polar-orbiting satellites, the NASA Tropical Rainfall Measuring Mission (TRMM), the NASA QuikSCAT, and Defense Meteorological Satellite Program (DMSP) satellites, among others, were also useful in tracking Dean and analyzing its structure. AFRC reconnaissance aircraft conducted 17 missions into Dean over a period of seven days. Unfortunately, communications issues with the aircraft prevented receipt of a substantial quantity of reconnaissance data operationally. In addition, the very strong radial gradients of wind speed across the eyewall of Dean incorrectly triggered SFMR data-blanking algorithms, causing the loss of additional data. A substantial effort was made to reconstruct/reprocess these data to assist in the preparation of this report. Raw SFMR data were reprocessed after the event by of ProSensing, Inc., using their latest retrieval algorithm with improved quality-control thresholds. Only 10-s means from the reprocessed SFMR profiles were considered in the post-storm analysis; this is the same spatial averaging that is applied to AFRC flight-level observations. Complete flight-level data records were obtained for selected missions, as well as raw dropsonde observations from the eyewall of Dean, which were then reprocessed by the author. There were a number of notable reconnaissance observations obtained during Dean, including an SFMR surface wind estimate of 150 kt at 0504 UTC 18 August (Fig. 2). Examination of the SFMR wind profile across the eyewall reveals that this was a broad maximum that is difficult to dismiss as unrepresentative, although the peak flight-level (700 mb) wind observed around this time was 154 kt, which would correspond only to 139 kt at the surface using the standard 90% adjustment factor for this flight level. There were two dropsondes early on 18 August that reported instantaneous surface winds of 144 and 149 kt, but in both cases the profile shapes strongly suggest that these particular estimates were unrepresentative of a sustained wind. The best track intensity for 0600 UTC 18 August is set at 145 kt, and represents a compromise between the SFMR and flight-level estimates. It is worth noting that Dean’s minimum sea-level pressure at this time was only 929 mb, and that the pressure fell another 10 mb or so over the next 18-24 h. During the time of this pressure fall, however, important structural changes were occurring: Dean’s inner eyewall eroded, resulting in a re-distribution of the pressure gradient. There were 30-40 kt reductions in both peak SFMR and dropsonde winds observed during this period, and the peak flight-level winds concurrently fell by over 20 kt. Accordingly, the best track shows a decrease in winds from 145 kt to 120 kt (Fig. 2). This well-monitored hurricane illustrates some of the pitfalls of using central pressure as a proxy or measure of tropical cyclone intensity. Dean is estimated to have made landfall near Majahual as a Category 5 hurricane on 21 August with maximum sustained winds of 150 kt and a minimum pressure of 905 mb, the latter estimate being based on a dropsonde report of 906 mb at 0814 UTC, accompanied by a surface wind speed of 15 kt. At 905 mb, Dean ties Camille (1969) and Mitch (1998) for the seventh-lowest central pressure on record in the Atlantic basin, and the third-lowest landfall pressure, behind only Gilbert and the 1935 Labor Day hurricane. The maximum winds at landfall are harder to estimate, as there is a considerable spread among the various observations taken around this time. The most extreme of these is a dropsonde surface report of 177 kt at 0728 UTC. This was associated with a very thin layer of strong winds immediately at and just above the surface, which because of friction is an impossible steady-state configuration; consequently this observation is rejected as unrepresentative of a sustained wind. Several hours prior to landfall, around 0000 UTC, there was good agreement between the SFMR and surface-adjusted flight-level winds on an intensity of 145 kt. Subsequently, the pressure fell from 914 mb to 905 mb, while the flight-level winds increased slightly: at 0653 UTC and 0819 UTC, flight level winds of 165 and 164 kt were recorded, respectively, corresponding to surface winds of 149 and 148 kt. Interestingly, the SFMR did not encounter surface winds higher than 136 kt around the time of landfall, further complicating the analysis. Given the eyewall/convective structure, however, greater weight is placed on the flight-level observations, resulting in a landfall intensity estimate of 150 kt. Ship and buoy reports of winds of tropical storm force associated with Dean are given in Table 2, and selected surface observations from land stations are given in Table 3. In Martinique, a sustained (1-min mean) wind of 81 kt was observed at Vauclin, a site along the southeast coast of the island but outside of Dean’s eyewall. A gust to 101 kt was reported at Sainte-Anne; this site did experience the eyewall but has a more sheltered exposure. The highest storm-total rainfall in Martinique was 13.07 inches at Fort de France-Colson. There are few authoritative observations from Dean’s passage over Jamaica, as many instruments did not survive the storm. At Norman Manley International Airport in Kingston (MKJP), the weather station tower was blown over around 1800 UTC 19 August, causing a loss of data. Numerous rain gauges were blown or washed away. There was a ham radio report of an 89 kt sustained wind in Munro, St. Elizabeth, before the instrument was disabled. The highest rainfall report was 13.50 inches at Ingleside, Manchester. No official wind observations of significance were received in association with either of Dean’s landfalls in Mexico, which occurred in relatively-sparsely populated areas. The maximum rainfall report was 15.39 inches at Requetemu, San Luis Potosi in central mainland Mexico. In Sabancuy, Campeche, 10.87 inches of rain was recorded.
Based largely on reports provided by the meteorological services of the affected countries, the number of direct deaths associated with Dean is estimated to be 32, with 14 occurring in Haiti, 12 in Mexico, 3 in Jamaica, 2 in Dominica, and 1 in St. Lucia. Remarkably, the deaths in Mexico occurred in the states of Hidalgo, Puebla, Veracruz, and San Luis Potosi, in association with Dean’s second (weaker) landfall. Dean is estimated to have been a Category 2 hurricane when its northern eyewall passed over Martinique. Flooding was reported throughout the island. Approximately 1300 homes were destroyed and another 7500 experienced severe damage. Media and government reports indicate the complete destruction of the island’s banana crop, and a 70% loss of the sugar cane crop. Damage has been estimated at 400 million Euros. Although there were no direct deaths, there were at least three indirect deaths, including one from a heart attack and one suicide. Over 200 post-storm injuries were reported during the cleanup after the event. In Guadeloupe, damage has been estimated at 100 million Euros, with about 75% of the banana plantations destroyed. The fishing industry also suffered significant losses, and resorts along the south coast of the island reported damage. Strong winds and heavy seas caused extensive damage along the north and west coasts of St. Lucia, including damage to bridges, roofs, and utility poles. Damage there was estimated at $18 million. In St. Vincent, some homes lost their roofs, and roof damage was also reported in Dominica. There was a complete loss of banana crop in Dominica. In Barbados, storm surge flooding described as serious was reported along the south coast. The center of Dean passed about 90 n mi south of the Dominican Republic. Rainfall associated with outer bands there was relatively light, although very heavy surf was reported along the south coast that media reports indicate destroyed several homes. Landslides reportedly destroyed several hundred homes in Haiti, and were responsible for most of the fatalities there. In Jamaica, the most severe impacts were reported in the southeastern parishes of Clarendon, St. Catherine, and Kingston/St. Andrew, where it is estimated that roughly two-thirds of the homes were completely destroyed or would require major repairs. Flooding in Jamaica was not generally significant, although some occurred in eastern parishes where grounds had become saturated by rains the previous week. Agriculture, particularly the banana crop, was severely impacted. Dean made its initial landfall in Mexico in a relatively uninhabited area, sparing the popular tourist destinations of Cancun and Cozumel, and consequently the damage was relatively light. Majahual was the only town to experience the full force of the hurricane, where hundreds of buildings were destroyed and steel girders were crumpled. Puerto Costa Maya, the nearby cruise port, was severely damaged and was expected to be closed for many months. Farther to the south, the government of Belize reported about $100 million in damage in that country. At its second landfall near Tecalutla, Mexico, extensive roof damage was reported, along with downed trees and power lines.
The genesis of Dean was well anticipated in Tropical Weather Outlook (TWO) products. The wave from which Dean formed was introduced into the TWO as a tropical cyclone candidate as it emerged from the African continent into the Atlantic. The National Weather Service Global Forecast System (GFS) model was consistently calling for development of this wave well before it entered the Atlantic basin. A verification of official and guidance model track forecasts is given in Table 4. Average official track errors for Dean were 21, 39, 58, 82, 141, 198, and 274 n mi for the 12, 24, 36, 48, 72, 96, and 120 h forecasts, respectively. The number of forecasts ranged from 37 at 12 h to 19 at 120 h. These errors are lower than the average long-term official track errors at all time periods (Table 4), and significantly lower than the long-term averages out to 48 h. As can be seen in Fig. 6, the official forecasts had relatively low cross-track errors; although it is difficult to determine from the diagram, the official forecasts primarily had a slow bias, as did many of the guidance models. Among the models, the GFS performed exceptionally well, while the GFDL did unusually poorly, taking Dean too far to the north on many occasions. Several of the guidance models outperformed the official forecast at the longer intervals. A verification of official and guidance model intensity forecasts is given in Table 5. Average official intensity errors were 10, 12, 13, 14, 21, 33, and 32 kt for the 12, 24, 36, 48, 72, 96, and 120 h forecasts, respectively. For comparison, the average long-term official intensity errors are 6, 10, 12, 14, 18, 20, and 22 kt, respectively. As is often the case with intense hurricanes, official intensity errors for Dean were above the long-term means, and particularly so at the longer ranges. However, the official forecasts were generally superior to the objective guidance. The official forecasts had a modest low bias. Watches and warnings associated with Dean are given in Table 6. Acknowledgements. Radar imagery, as well as information on conditions and impacts in Martinique and Guadeloupe was provided by of Meteo-France. The meteorological services of St. Lucia, Jamaica, and Mexico also provided some of the quantitative information included in this report. 1Table 1. Best track for Hurricane Dean, 13-23 August 2007.
1Table 2. Selected ship and fixed buoy reports with winds of at least 34 kt for Hurricane Dean, 13-23 August 2007.
|