1. Introduction
Nineteen ninety-eight was an active year for tropical cyclones (TCs) in the Atlantic basin. Fourteen tropical storms developed; 10 of these tropical storms became hurricanes. The long-term average numbers of tropical storms and hurricanes per season are 10 and 6, respectively. From 1995 through 1998, 33 hurricanes occurred, the largest 4-yr total ever observed (going back to at least the start of reliable records in the mid-1940s). Three of the 1998 hurricanes strengthened into major hurricanes [maximum winds ≥96 kt (49 m s−1); categories 3, 4, and 5, on the Saffir–Simpson hurricane scale; Simpson (1974)]. It was a hurricane season that will long be remembered for a staggering number of fatalities, over 9000, due to disastrous Hurricane Mitch in Central America. Mitch was the fourth most intense hurricane ever observed in the Atlantic basin, and the strongest ever observed in the month of October. There were many TC landfalls; seven tropical storms or hurricanes struck the United States. Hurricane Georges left a path of devastation across the islands of the northern Caribbean Sea, and caused hundreds of deaths in the Dominican Republic.
After a slightly later than normal start with Alex in late July, followed by a couple more quiet weeks, the season more than made up for lost time. In a remarkable 35-day span from 19 August to 23 September, 10 named TCs formed. On 25 and 26 September, four hurricanes, Georges, Ivan, Jeanne, and Karl, were on the map at the same time in the Atlantic basin, for the first time since 1893.
Table 1 lists the tropical storms and hurricanes of 1998, and Fig. 1 is a map of their tracks. As was the case in 1995 and 1996, most of the TCs originated in the deep Tropics south of latitude 20°N.
Figure 2 shows the sea surface temperature anomalies from the long-term mean for August through October of 1998. Practically all of the 1998 TCs occurred during these months. During this period nearly all of the Atlantic Ocean's surface from the equator to 60°N was warmer than normal. Of particular interest is the tropical region from the Caribbean Sea eastward to near the coast of Africa. Here, sea surface temperatures were as much as 1°C above normal. The warmer than normal waters may have been a contributing factor to the above normal TC activity. The increase in hurricane activity since 1995 is consistent with the multidecadal sea surface temperature fluctuations identified by Landsea et al. (1999). They showed that the North Atlantic oscillates between warm and cold states that last 25–40 yr each. The last cold episode extended from 1971 to 1994. Since 1995, the north Atlantic appears to have switched back to a warm phase and Atlantic hurricanes have also increased to a level of activity similar to that of the late 1920s to late 1960s.
It has been known for some time (e.g., Riehl and Shafer 1944; Gray 1968) that the vertical shear of the horizontal wind is a major controlling factor in TC genesis and intensity change. Figure 3 shows the anomalies of the vertical shear from the long-term mean for August, September, and October of 1998. Superimposed on this chart are positions where TCs developed during these 3 months (initial tropical depression stage), and crosses showing where they reached tropical storm strength. No system developed in an area where the mean shear was above normal. Also, no tropical cyclone strengthened into a tropical storm where the mean shear was stronger than normal except Klaus, which became a tropical storm over subtropical latitudes in an area where the shear was slightly higher than average.
Experience has shown that winds or heights at the 500-mb level usually provide a reasonably good approximation of the TC steering flow. Figure 4a is a map of the mean 500-mb heights for August and September of 1998 with the tracks of the TCs during those months superimposed. Figure 4b shows the anomalies of this height field from the long-term mean. There was a large negative height anomaly centered near the Canadian Maritimes, a large positive anomaly over the eastern Atlantic, and a ridge of slightly higher than normal heights extending westward to near southern Florida. Over the Atlantic, the TC tracks generally followed the periphery of the subtropical ridge, although there was a subtle weakness in the ridge in the vicinity of 40°W where two systems, Ivan and Jeanne, recurved much farther east than the others. Over the Gulf of Mexico, the steering flow appears rather weak and, in fact, TC motion tended to be slow and/or erratic in that area.
2. Tropical storm and hurricane summaries
a. Tropical Storm Alex, 27 July–2 August
A well-organized tropical wave emerged from the west coast of Africa on 26 July and moved westward at 15–20 kt (1 kt = 0.514 m s−1). Early on 27 July, ship reports and satellite scatterometer winds supported the presence of a surface circulation in association with the wave. On this basis, it is estimated that the system attained tropical depression status around 1200 UTC 27 July about 300 n mi (1 n mi = 1.85 km) south-southwest of the Cape Verde Islands.
The depression changed little in organization on 27 July and most of 28 July. It had minimal deep convection near the center, as it moved on a general west-northwest track at 15–20 kt. During this period, satellite imagery characterized the depression as a large and elongated circulation that was still embedded within the intertropical convergence zone. By the evening of 28 July, deep convection increased near the center. Dvorak (1984) satellite intensity estimates indicate that the cyclone strengthened into Tropical Storm Alex by 0000 UTC 29 July.
Alex continued to move on a general west to west-northwest course at 10–15 kt in response to a deep-layer ridge over the tropical eastern Atlantic. During the next several days, Alex's development was hampered by a mid-to upper-level trough, and attendant cyclonic circulation, located to its north and west. By 30 July, satellite imagery indicated that the storm was experiencing southerly vertical wind shear. During the evening of 30 July, satellite imagery showed a burst of deep convection just east of the center. It is estimated that Alex reached a peak intensity of 45 kt from 1800 UTC 30 July to 0600 UTC 31 July, and a minimum central pressure of 1002 mb near 0000 UTC 31 July. Shortly thereafter, increased southerly vertical wind shear induced by the mid- to upper-tropospheric trough to the west of Alex curtailed further strengthening.
Over the next few days the vertical wind shear took its toll. The low-level center of Alex became fully exposed south of the remaining deep convection on 1 August. Alex turned toward the northwest later that day and continued to weaken gradually. It weakened to a depression by midday on 2 August. Later that afternoon, data from an Air Force Reserve Command (AFRC) “Hurricane Hunter” reconnaissance aircraft showed that the system no longer had a closed low-level circulation, and Alex had dissipated.
b. Hurricane Bonnie, 19–30 August
Bonnie was the third hurricane to directly hit the coast of North Carolina during the past three years.
1) Synoptic history
The source of Bonnie was a large and vigorous tropical wave that moved over Dakar, Senegal, on 14 August. The wave appeared in visible satellite imagery as a large cyclonic low- to midlevel circulation void of deep convection. The wave caused 24-h surface pressure changes of −3.5 and −4.0 mb at Dakar and Sal, respectively. There was a well-established 700-mb easterly jet that peaked at 50 kt just before the wave axis crossed Dakar, followed by a well-marked wind shift from the surface to the middle troposphere. The overall circulation left Africa just north of Dakar where the ocean was relatively cool. However, a strong high pressure ridge steered the system on a west-southwest track over increasingly warmer waters and convection began to develop. Initially, there were several centers of rotation within a much larger circulation. It was not until 1200 UTC 19 August that the system began to consolidate into a tropical depression. Although the central area of the tropical depression was poorly organized, the winds north of the circulation were nearly tropical storm strength, as indicated by ship observations and high-resolution low-cloud wind vectors provided in real time by the University of Wisconsin. Based on these winds and satellite intensity estimates, the depression strengthened into Tropical Storm Bonnie by 1200 UTC 20 August. Bonnie moved on a general west to west-northwest track around the circulation of the Azores–Bermuda high toward the northern Leeward Islands.
The first reconnaissance plane into Bonnie arrived late on 20 August and measured a 1004-mb minimum pressure and 61-kt winds at 500-m elevation, northeast of the center. Bonnie skirted the Leeward Islands and most of the associated weather remained to the north over the open Atlantic. During that period, Bonnie's circulation was very asymmetric.
Under favorable upper-level winds, Bonnie gradually strengthened and became a hurricane at 0000 UTC 22 August, when it was located about 200 n mi north of the eastern tip of Hispaniola. At that time, the Hurricane Hunters found a nearly complete eyewall and peak flight-level winds of 76 kt. Bonnie moved on a general west-northwest heading and reached 100-kt maximum winds and 954-mb minimum pressure about 150 n mi east of San Salvador in the Bahamas. Figure 5 shows a visible satellite image of Bonnie near that time.
The ridge to the north of Bonnie temporarily weakened and the steering currents collapsed. The hurricane then drifted northward for a period of 18–24 h. Thereafter, the subtropical ridge reintensified, forcing Bonnie to move northwestward and then northward toward the coast of North Carolina, during which time the hurricane maintained winds near 100 kt. After slight weakening of the hurricane, the eye of Bonnie passed just east of Cape Fear around 2130 UTC 26 August and then made landfall near Wilmington, as a category 2 hurricane, around 0330 UTC 27 August.
The hurricane slowed down and weakened over eastern North Carolina. It dropped to tropical storm status based on surface observations and Weather Surveillance Radar-1988 Doppler (WSR-88D) winds. Bonnie turned northeastward over water ahead of a middle-level trough and rapidly regained hurricane strength as indicated by aircraft reconnaissance data. Thereafter, the hurricane moved on a general northeast to east track and became extratropical near 1800 UTC 30 August, about 240 n mi south-southeast of Newfoundland.
2) Meteorological statistics
The Hurricane Bonnie event was characterized by a high density of observations. During Bonnie, the National Oceanic and Atmospheric Administration (NOAA) Gulfstream jet and P-3 aircraft deployed a very large number of Global Positioning System (GPS) dropsondes (Hock and Franklin 1999) over a large portion of the Atlantic as a part of a major synoptic flow experiment. These observations were primarily used to initialize the numerical models.
The maximum winds measured were 116 kt at the 700-mb level at 0113 UTC 25 August and then again at 1659 UTC 26 August. These measurements were taken during AFRC and NOAA reconnaissance missions, respectively. Table 2 displays selected surface observations during Bonnie, primarily over the area where the hurricane made landfall. There were several important observations from amateur observers relayed to the Tropical Prediction Center/National Hurricane Center (TPC/NHC) and to the local National Weather Service forecast offices. These include reports of peak wind gusts of 104 kt near North Carolina State Port at 0138 UTC and 100 kt at Wrightsville Beach at 1951 UTC 27 August. Rainfall totals of about 200–280 mm were recorded in portions of eastern North Carolina. Storm tides of 1.5–2.4 m above normal were reported mainly in eastern beaches of Brunswick County, North Carolina, while a storm surge of 1.8 m was reported on the Albemarle Sound in Pasquotank and Camdern Counties. A tornado was reported in the town of Edenton, North Carolina, in Chowan County.
3) Casualty and damage statistics
Three people died as a consequence of Bonnie. A 12-year-old girl was killed when a large tree fell on her home in Currituck County, North Carolina. Another person was caught in rip currents and drowned in Rehoboth Beach, Delaware. A third person died on Cape Cod when choppy seas overturned a rowboat. This last death may have been only indirectly related to Bonnie.
There are numerous reports of many trees down, roof and structural damage, and widespread power outages primarily in eastern North Carolina and Virginia where a federal disaster was declared for several counties. The area hardest hit appears to have been Hampton Roads, Virginia, where the damage probably reached well into the hundreds of millions of dollars. The Property Claims Services Division of the Insurance Services Offices reports that Bonnie caused an estimated $360 million in insured property damage to the United States. This estimate includes $240 million in North Carolina, $95 million in Virginia, and $25 million in South Carolina. A conservative ratio between total damage and insured property damage, based on past landfalling hurricanes, is two to one. Therefore, the total U.S. damage estimate is $720 million.
4) Warnings
A hurricane warning was issued from Murrells Inlet, South Carolina, to the North Carolina–Virginia border, including the Pamlico and Albermarle Sounds, at 0900 UTC on 25 August, about 39 h prior to the landfall of Bonnie on the coast of North Carolina.
c. Tropical Storm Charley, 21–24 August
1) Synoptic history
The origin of Tropical Storm Charley is unclear. It may have evolved from a large swirl of clouds that left the coast of Africa on 9 August, mainly to the north of Dakar, Senegal. Charley's immediate precursor consisted of a small area of deep convection first noted a few hundred miles to the northeast of the Leeward Islands on 15 August. Intermittent convective activity continued while the system moved just north of west for the following few days. On 19 August, animation of satellite pictures showed a cyclonic rotation of the clouds over the southeastern Gulf of Mexico.
The first formal position estimate from satellite analysts came on the evening of 19 August. Dvorak T numbers for gauging intensity were first assigned the next day (1.5) over the central Gulf [see Dvorak (1984) for description of T numbers]. By the morning of 20 August, surface winds had begun to increase. NOAA's central Gulf buoy 42001 measured sustained winds as high as 31 kt and gusts to 45 kt at 1700 UTC. These stronger winds were fleeting, however, and an investigation of the system late that day by AFRC reconnaissance aircraft did not indicate a closed low-level circulation center.
A center “fix” was made aboard reconnaissance aircraft around 1300 UTC the next day, indicating that the system became a tropical depression around 0600 UTC on 21 August. At that time, the depression was centered about 275 n mi off of the south Texas coast. The TC moved toward the west-northwest to northwest at about 10 kt during its 3-day lifetime.
Although the center was not well formed initially, the amount of deep convection steadily increased, particularly over the northern semicircle. This portion of the cyclone swept over the oil platforms of the northern Gulf; data from these platforms (e.g., Table 3) suggest that the depression reached tropical storm status by 1800 UTC on 21 August. Winds of hurricane force were noted at a flight level of 300 m in intense convection to the northeast of the center early on 22 August. Charley was likely then at its peak strength, near 60 kt. Subsequent wind speeds measured by reconnaissance aircraft were considerably lower, and it is estimated from these and other observations that surface winds were closer to 40 kt when Charley's center made landfall near Port Aransas about 1000 UTC on 22 August.
The surface circulation weakened further after landfall and likely dissipated early on 24 August along the Rio Grande near Del Rio, Texas. Although the winds diminished inland, and a closed surface circulation could no longer be identified, a slow-moving circulation aloft persisted in the Del Rio vicinity and generated flooding rains that were most devastating in that area on 23 and 24 August. By late on 25 August, most of the remnant cloud system had deteriorated and precipitation diminished.
2) Meteorological statistics
Charley's primary legacy will be the rainfall and associated flooding near Del Rio. On 23 August, 427 mm of rain fell in Del Rio. This easily surpassed the previous daily record of 223 mm set on 13 June 1935. A nearby site recorded 447 mm for the 24-h period ending in the morning hours of 24 August. Along the coast, maximum rainfall totals were near 125 mm except for an unofficial report of 230 mm near the mouth of the San Bernard River in Brazoria County. River flooding along the Rio Grande occurred well downstream from Del Rio, in the Laredo area. Storm tides of 0.6–1.1 m above normal astronomical levels were reported along the coast.
The Automated Surface Observing System (ASOS) sites at Rockport (RKP) and Galveston (GLS), Texas, were the only two surface reporting stations on land to measure sustained tropical storm force winds. They recorded 2-min winds of 36 kt and 34 kt, respectively. A gust to 55 kt was reported from the Port O'Connor Coast Guard Station. A minimum pressure of 1000 mb is estimated at landfall from the observation of 1000.7 mb at RKP an hour later.
3) Casualty and damage statistics
Charley's death toll stands at 13 in Texas, with 6 people missing. All were apparently flood victims located well inland. The total consists of four people including two toddlers who were in a pickup truck that was swept away by rising water in Real County on 23 August. Seven other people from the truck were rescued. Nine deaths due to drowning occurred in Del Rio (Val Verde County) along the San Felipe Creek during the late night of 23 August. Emergency operations personnel in Mexico reported that seven people died in Ciudad Acuña, Mexico, across the border from Del Rio. Media reports indicate that three of these victims drowned while trying to cross a flooded gully.
Total U.S. losses due to the inland flood are estimated at $50 million. Property losses were reported in several counties and consisted of damages to residences, businesses, roads, bridges, and agriculture. About 1500 houses, 200 mobile homes, and 300 apartments were damaged or destroyed in Val Verde County, where about $40 million in losses occurred. Minor beach erosion was reported along portions of the Texas coast.
4) Warnings
The NHC issued a tropical storm warning from Brownsville to High Island, Texas, on its first advisory (as a tropical depression) at 1500 UTC on 21 August, about 19 h prior to landfall in that area. The warning was extended eastward to Cameron, Louisiana, 6 h later.
d. Hurricane Danielle, 24 August–3 September
Danielle had a long track across the Atlantic. Although it did not significantly impact land as a TC, it battered portions of the United Kingdom as an extratropical system.
1) Synoptic history
A tropical wave, accompanied by disorganized cloudiness and showers, moved off the west coast of Africa on 21 August. Within 24 h, deep convection became somewhat more consolidated in clusters near an ill-defined center of cyclonic cloud rotation. Initial Dvorak technique classifications were assigned at 1100 UTC 22 August. Thereafter, the organization of the disturbance continued to improve gradually as cloudiness and showers became concentrated in a circular area. By 0600 UTC 24 August, the Dvorak T number was analyzed at 2.0 and it is estimated that Tropical Depression 4 formed around this time, centered a little less than 600 n mi west-southwest of the Cape Verde Islands. Strengthening continued, as satellite images showed convection becoming more tightly wrapped around the center, and the TC is estimated to have become Tropical Storm Danielle by 1800 UTC 24 August. Upper-tropospheric outflow was well defined over the area, and Danielle intensified further. The first visible satellite pictures on 25 August revealed a “pinhole” eye, indicating that the system had become a hurricane. Danielle was a quite compact system, with tropical storm force winds covering an area estimated to be only a little more than 100 n mi in diameter. Based on satellite data, this rapidly strengthening hurricane reached a peak intensity of near 90 kt around 0600 UTC 26 August, while centered about 900 n mi east of the Leeward Islands.
After Danielle reached its first peak in strength, southeasterly vertical shear appeared to disrupt its organization. By the time the first reconnaissance aircraft reached the hurricane around 0000 UTC 27 August, it was not as well organized in satellite imagery as it had been. This first aircraft mission found a maximum wind of 90 kt at the 850-mb flight level, but a remarkably high central pressure of 993 mb. Such values of wind and pressure show how much deviation from the typical wind versus pressure relationship can occur in compact hurricanes. In contrast to this, the much larger Hurricane Bonnie, which was in progress over the western Atlantic around the same time with a comparable maximum wind speed, had a minimum central pressure that was 25 mb lower than Danielle's.
Moderate vertical shear continued to preclude much strengthening of Danielle. However, aircraft data indicated that a second 90-kt intensity peak occurred around 1200 UTC on 27 August. For the next few days, some weakening took place even though the eyewall structure was generally maintained and atmospheric conditions seemed to be favorable for intensification. By 30 August, Danielle was barely a hurricane. Movement over waters cooled by the earlier passage of Hurricane Bonnie may have contributed to the weakening.
From the time the cyclone formed, and for about 6 more days, the motion was toward the west-northwest, with the forward speed gradually slowing from 18–20 kt over the eastern Atlantic to 9 or 10 kt on 30 August. By that time, Danielle was nearing the western periphery of the subtropical anticyclone that had steered it across much of the Atlantic. As it continued to decelerate, the hurricane turned toward the northwest and north, reaching its westernmost longitude, about 74°W, early on 31 August. The hurricane restrengthened, and reached a third 90-kt intensity peak around 1200 UTC that day.
In response to increasing southwesterly midtropospheric steering flow, ahead of a trough near the U.S. east coast, Danielle completed its recurvature and began to move northeastward on 1 September. It also regained 90-kt maximum winds for the fourth, and final, time. The center of the accelerating hurricane passed slightly less than 200 n mi northwest of Bermuda early on 2 September. Danielle began to lose its tropical characteristics on 3 September, as its center passed about 200 n mi south of Cape Race, Newfoundland. Danielle became an extratropical cyclone around 0000 UTC 4 September, although it was still a strong storm with hurricane force winds. The storm moved eastward to east-northeastward across the North Atlantic for the next couple of days, with only slow weakening. It turned northeastward about 300 n mi west of the British Isles on 6 September, its forward speed slowed to 6–10 kt. The cyclone became indistinct when it merged with another extratropical low about 200 n mi north of Ireland on 8 September.
2) Meteorological Statistics
The highest wind reported in Danielle was 97 kt, at 700 mb from the Hurricane Hunter reconnaissance aircraft at 1141 UTC 27 August. Additional Hurricane Hunter wind observations of 95 kt at 850 mb and 92 kt at 700 mb were taken at 1235 UTC 31 August and 1656 UTC 1 September, respectively. These data, along with satellite-based intensity estimates on 26 August, are the bases for the four intensity peaks of 90 kt in the best track. The minimum pressure estimate, 960 mb, was derived from a lowest pressure observation of 962.6 mb from the Canadian Laurentian Fan buoy, 44141, at 0900 UTC 3 September. It is also noteworthy that this buoy measured a significant wave height of 15.9 m with a maximum wave height of 26.8 m at the time of its lowest pressure. Sustained winds of 34 kt with gusts to 47 kt were observed at Bermuda at 1100 UTC 2 September.
3) Casualty and damage statistics
No reports of casualties due to Danielle have been received at the National Hurricane Center. The western part of Great Britain was lashed by Danielle as an extratropical storm on 6 September. Several people were rescued from treacherous sea conditions in the area. On the coast of Cornwall, beach areas had to be evacuated after waves became so high that they were breaking over some houses. A police all-terrain vehicle on the Isles of Scilly was swept into the sea by a rogue wave as it was being driven down a concrete pier in one of the island's main towns.
4) Warnings
A tropical storm warning was issued for Bermuda at 1500 UTC 1 September since the southern portion of Danielle's circulation was likely to affect that island. Sustained winds of tropical storm force occurred at Bermuda about 20 h after the issuance of this warning as the center passed well to the northwest and north. No other warnings or watches were necessary for this TC.
e. Hurricane Earl, 31 August–3 September
Earl made landfall in the Florida panhandle as a category 1 hurricane. Significant storm surge flooding resulted in the “Big Bend” area of Florida.
1) Synoptic history
Hurricane Earl formed from a strong tropical wave that emerged from the west coast of Africa on 17 August. Persistent convection accompanied the wave as it moved westward across the tropical Atlantic. A weak low-level cyclonic circulation was suggested in animation of satellite imagery, as well as in limited aircraft reconnaissance and island reports as the system passed through the Lesser Antilles on 23 August. Tropical cyclone development appears to have been inhibited by unfavorable winds aloft while the system moved through the Caribbean. These unfavorable conditions were a result of the upper-level outflow from large and powerful Hurricane Bonnie located over the southwestern North Atlantic. Nevertheless, the tropical wave continued to be easily tracked in satellite imagery as it moved into the Gulf of Mexico where cloudiness and thunderstorms increased. It is estimated that the system became a tropical depression over the southwest Gulf of Mexico midway between Merida and Tampico, Mexico, at 1200 UTC 31 August.
Based on aircraft reconnaissance, the tropical depression became Tropical Storm Earl about 500 n mi south-southwest of New Orleans, Louisiana, near 1800 UTC 31 August. The center remained difficult to locate by satellite, and, in fact, multiple centers were reported by aircraft reconnaissance for the next couple of days. Occasionally, a new center would appear to form, which made tracking extremely difficult. Although the best track shown in Fig. 1 indicates a general motion toward the north and then northeast near 10 kt while Earl was over the Gulf of Mexico, a certain amount of “smoothing” was necessary to account for multiple centers and any possible center reformations.
Based on aircraft reconnaissance data, Earl is estimated to have reached hurricane status at 1200 UTC 2 September while centered about 125 n mi south-southeast of New Orleans. The system never exhibited a classical hurricane appearance. Instead, satellite imagery showed the deepest convection confined primarily to the eastern semicircle and aircraft reconnaissance data indicated a very asymmetric wind field, with the strongest winds located well east and southeast of the center.
After briefly reaching category 2 status, Earl made landfall near Panama City, Florida, as a category 1 hurricane near 0600 UTC 3 September. Because the strongest winds remained well to the east and southeast of the center, the highest storm surge occurred in the Big Bend area of Florida, well away from the center. The TC weakened to below hurricane strength soon after making landfall, and became extratropical at 1800 UTC 3 September as it moved northeastward through Georgia. The deepest convection became well removed from the center by this time and the strongest winds were located over the Atlantic waters off the U.S. southeast coast. The extratropical cyclone moved off the mid-Atlantic coast near 1800 UTC 4 September, crossed over Newfoundland on 6 September, and was tracked across the North Atlantic until it was absorbed by a larger extratropical cyclone (formerly Hurricane Danielle) on 8 September.
2) Meteorological statistics
The operational aircraft reconnaissance flights into Earl were provided by the AFRC. The minimum central pressure reported by aircraft was 985 mb at 0045 UTC on 3 September. This minimum pressure was measured by dropsonde and was the lowest pressure reported during Earl's existence. The maximum winds of 104 kt from a flight level of 850 mb (near 1.5 km) were measured at 1638 UTC 2 September. These peak winds were in a limited area about 80 n mi east of the center. The Hurricane Hunters never reported an eyewall. Reconnaissance data and land-based radar presentations suggest the hurricane weakened before it moved onshore.
Satellite analyses underestimated the intensity of Earl, probably because the system never exhibited a classical TC pattern. For example, the maximum winds estimated from the Tropical Analysis and Forecast Branch (TAFB), the tropical branch of the Air Force Weather Agency (AFWA), and the Satellite Analysis Branch (SAB) were 55, 55, and 45 kt, respectively.
The WSR-88Ds at Slidell, Louisiana; Mobile, Alabama; Eglin Air Force Base, Florida; and Tallahassee, Florida, were helpful in locating the center and areas of strongest winds aloft as the cyclone moved near shore. As is often the case in landfalling hurricanes, there were no reports from land stations of sustained hurricane force winds in Earl. Table 4 lists selected U.S. surface observations. The NOAA Coastal Marine Automated Network (C-MAN) station at Cape San Blas (near Apalachicola, Florida) reported 10-min sustained winds of 48 kt between 0400 and 0500 UTC and gusts to 61 kt at 0436 UTC 3 September. The strongest winds at the time of landfall likely remained over water near the Big Bend area of Florida. Several wind reports from north Florida were relayed to the NHC through amateur radio volunteers. The highest measured wind gust was 79 kt at an elevation of 10 m from a Davis wind instrument located in the middle of St. George Island at 29.40°N, 84.53°W at 0102 UTC 3 September. The storm surge was estimated to be near 2.4 m in Franklin, Wakulla, Jefferson, and Taylor Counties and approximately 2 m in Dixie County. These values tapered off to less than 1 m in Lee County. Rainfall totals of 75–150 mm were common near the path of Earl, although much higher amounts were recorded in a few areas. A storm total of 416 mm near Panama City, Florida, was the highest reported. Several tornadoes were reported in central and north Florida, Georgia, and South Carolina.
There was an extensive sampling of the Gulf of Mexico with GPS dropsondes from the NOAA jet around 0000 UTC 2 September. These data showed that a midtropospheric trough over the east-central United States extended into the central Gulf of Mexico. This atmospheric feature provided the steering current that moved Earl northeastward into the Florida panhandle.
3) Casualty and Damage Statistics
Hurricane Earl was directly responsible for three deaths. Two occurred as a result of a boat that capsized off Panama City. One death occurred as a result of a tornado near St. Helena, South Carolina.
The Property Claims Services Division of the Insurance Services Offices estimated that Earl caused insured property damage of $15 million in Florida, $1 million in Georgia, and $2 million in South Carolina. These estimates do not include storm surge damage. In addition, the National Flood Insurance Program reported $21.5 million of insured (storm surge related) losses in Florida. Using a two to one ratio of total to insured property losses gives a total U.S. damage estimate of $79 million for Earl.
4) Warnings
Since the NHC forecasts are based, in part, on the computer guidance, which in the case of Earl generally had a westward bias, hurricane warnings were not extended eastward over the landfall location until 1300 UTC 2 September. These warnings were not issued with as much lead time (17 h prior to the arrival of the center) as the NHC desires. Fortunately, appropriate preparations appear to have been completed anyway.
f. Tropical Storm Frances, 8–13 September
Frances brought very heavy rainfall to portions of eastern Texas and southern Louisiana.
1) Synoptic history
Frances formed within a broad area of low pressure that first showed signs of organization of its associated convective cloudiness on 4 September. The convection was widespread over the western Caribbean and southern Gulf of Mexico, but there was no well-defined low-level circulation center. This situation persisted for several days as the system moved slowly west-northwestward. During this time a tropical wave moved into the area, and the system consolidated. By 8 September, the system developed a 1000-mb central surface pressure and considerable organized deep convection over a large area of the western Gulf of Mexico. It is estimated that a tropical depression formed by 1800 UTC on this date about 140 n mi east of Brownsville, Texas. At this stage in its development, with its large size, loosely organized convection, and lack of a distinct center, the TC was similar to the so-called monsoon depression of the western North Pacific basin.
The tropical depression drifted southward for about a day. By 1800 UTC on 10 September, wind observations from a data buoy, reconnaissance aircraft, and several oil rigs indicated that the depression had strengthened to a 35-kt tropical storm. Frances began moving north to northwestward at 10–15 kt. The center moved inland across the Texas coast just north of Corpus Christi at 0600 UTC on 11 September. By this time, Frances had strengthened to 55 kt under a large anticyclone aloft, in weak vertical shear, over SSTs near 30°C.
After moving inland, the center moved in a small cyclonic loop for 12 h between Corpus Christi and Victoria, and then moved northward across eastern Texas as a weakening tropical depression. The best track ends at 1800 UTC on 13 September, when the center was near the Texas–Oklahoma border north of Dallas, but the remnant low pressure and rainfall were tracked northward to Iowa during the next 24 h.
2) Meteorological statistics
Table 5 lists selected surface observations, including selected rainfall totals. Tropical storm force wind speeds were observed at several data buoys and oil rigs in the western Gulf of Mexico. The C-MAN station at Sabine, Texas, reported a maximum 2-min wind speed of 44 kt; this is the highest observed sustained surface wind speed. Sustained winds of tropical storm force were observed over land at Galveston and Victoria, Texas, and at Jefferson County Airport in Louisiana. Frances was a large storm; 34-kt winds extended approximately 300 n mi north and east of the center. Storm surge flooding of up to 1.8–2.4 m occurred along the middle and upper Texas coast and up to 1.5 m along the Louisiana coast. This flooding persisted for about 48 h. Freshwater flooding from rainfall was the most significant weather effect. Frances dropped copious rainfall over east Texas and southern Louisiana. The highest total reported in Texas was 411.5 mm in Brazoria County, and the highest total from Louisiana was over 289 mm. Undoubtedly, even higher amounts are likely to have accumulated in these areas. Over a dozen tornadoes were reported over southwestern and south-central Louisiana on 11–12 September. These caused one death (see next section), and extensive damage to schools in Acadia and Evangeline Parishes, and to homes in Lafayette Parish.
3) Casualty and damage statistics
The only known fatality directly attributable to Frances was in Lafourche Parish, Louisiana, where a man was killed when his trailer home was destroyed by a tornado spawned by the tropical storm. Six others were injured by this tornado. An indirect death occurred in the New Orleans area where a woman died in an automobile accident.
Moderate beach erosion occurred along much of the upper Texas and western Louisiana coastlines. Three Texas counties and four Louisiana parishes were declared federal disaster areas, primarily due to the rainfall flooding. These include Brazoria, Galveston, and Harris Counties and the parishes of Cameron, Jefferson, Lafourche, and Terrebonne. The Insurance Services Offices reported that a total of $110 million in insured property damage has been claimed in Texas, Louisiana, and Mississippi. The Houston Chronicle reported that $256 million in damage was inflicted in Galveston County. The total damage estimate for Frances is $500 million.
4) Warnings
Tropical storm warnings were issued along the Gulf of Mexico coast from Tampico, Mexico, northward and eastward including all of Texas and Louisiana. The warnings for the central Texas coast were issued at 2100 UTC on 9 September, some 33 h before landfall and almost 24 h prior to the time tropical storm force winds reached the coast.
g. Hurricane Georges, 15 September–1 October
Georges was the second-strongest and second-deadliest hurricane in the Atlantic basin during the 1998 season. Its 17-day journey resulted in eight landfalls, from the northeastern Caribbean to the coast of Mississippi, and 602 fatalities—mainly in the Dominican Republic and Haiti.
1) Synoptic history
Georges originated from a tropical wave that crossed the west coast of Africa late on 13 September. Rawinsonde data from Dakar, Senegal, showed a 35–45-kt easterly jet between 550 and 650 mb. On 14 September, visible satellite imagery depicted a large, well-defined cloud system in association with the wave; and meteorologists at the TAFB, SAB, and AFWA began satellite-based Dvorak intensity classifications. By early on 15 September, ship reports indicated the presence of a closed surface circulation, and a tropical depression formed at 1200 UTC about 300 n mi south-southwest of the Cape Verde Islands in the far eastern Atlantic. During the next 24 h the depression became better organized as banding features developed and deep convection formed over the center. The system became a tropical storm at 1200 UTC on 16 September centered about 620 n mi west-southwest of the Cape Verde Islands. Georges moved on a persistent west-northwest course for the next 10 days, a classic Cape Verde–type track, in response to a mid- to upper-level tropospheric ridge that strengthened with height.
After it became a tropical storm, Georges continued to gradually strengthen over the next several days. Around 1800 UTC on 17 September, a banding-type eye feature appeared, indicating that Georges had reached hurricane intensity. By 19 September, an upper-level anticyclone was well established over Georges and satellite pictures suggested that the hurricane was beginning to strengthen rapidly, as indicated by colder cloud tops, increased symmetry of the deep convection, and the warming and contracting of the well-defined eye (Fig. 6).
By early afternoon on 19 September, the first AFRC reconnaissance aircraft reached the hurricane and measured maximum flight-level winds of 146 kt and a minimum central pressure of 938 mb, confirming the intensification trend noted in satellite imagery. Georges's surface winds increased to near 125 kt at 1800 UTC on 19 September, making it a category 4 hurricane. Several GPS dropsondes were deployed in the eyewall of the hurricane as it neared peak strength. Near-surface (below 60 m) wind estimates from these drops indicated maximum winds of 134–150 kt. On this basis, Georges is estimated to have reached a peak intensity of 135 kt at 0600 UTC on 20 September while located about 285 n mi east of Guadeloupe in the Lesser Antilles.
Shortly after 0600 UTC 20 September, the hurricane began to weaken. The eye became indiscernible in satellite pictures, and later that afternoon aerial reconnaissance could no longer find a closed eyewall. Examination of water vapor satellite imagery and satellite-derived wind analyses from the Cooperative Institute for Meteorological Satellite Studies at the University of Wisconsin suggest that one factor possibly responsible for the weakening could have been upper-level northerly vertical wind shear induced by an upper-level anticyclone located over the eastern Caribbean. By the evening of 20 September, the central pressure had risen 26 mb. Georges made the first two of its eight landfalls in the Lesser Antilles (on Antigua, and then St. Kitts and Nevis) early on 21 September with maximum sustained surface winds near 100 kt.
By midmorning of 21 September, an upper-level low over Cuba moved westward away from Georges, and discouraged a northwestward turn of the hurricane away from Puerto Rico. Later in the afternoon, the shear appeared to diminish and the outflow aloft improved, but Georges never fully reintensified due in part to the circulation's interaction with land. Georges made landfall in southeast Puerto Rico with sustained surface winds near 100 kt on the evening of 21 September. The hurricane moved inland over Puerto Rico, weakened slightly, and then moved into the Mona Passage early on 22 September. Georges began to reintensity over the Mona Passage and made landfall later that morning, about 75 n mi east of Santo Domingo in the Dominican Republic, with estimated sustained surface winds of 105 kt. Figure 7 is a radar image of Georges at landfall in the Dominican Republic, and Fig. 8 is a satellite image of the hurricane around the same time.
During the next 21 h, Georges weakened as it moved slowly across the mountains of the Dominican Republic and Haiti, where it produced heavy rain, deadly flash floods, and mud slides. The system emerged into the Windward Passage on the morning of 23 September with 65-kt maximum winds. Georges changed little before making landfall about 25 n mi east of Guantanamo Bay in eastern Cuba later that afternoon. The system remained a hurricane as it moved slowly west-northwestward across the northern coast of Cuba, and it crossed the northern coast to the sea by late afternoon on 24 September. Satellite imagery showed that Georges retained fairly impressive upper-level outflow during its crossing of both Hispaniola and Cuba.
Once back over water, the hurricane began to reintensify. Early on 25 September, a band of deep convection developed east of the center, which expanded throughout the morning. Georges made landfall during midmorning of 25 September in Key West, Florida, with a minimum central pressure of 981 mb and maximum winds of 90 kt. After moving away from Key West, Georges turned more to the northwest, then north-northwest, and gradually slowed down on 26 and 27 September. This occurred in response to the midtropospheric anticyclone north of the hurricane shifting eastward into the southeastern United States. The hurricane made landfall near Biloxi, Mississippi, on the morning of 28 September with estimated maximum sustained 1-min winds of 90 kt and a minimum central pressure of 964 mb. After landfall, the system meandered over southern Mississippi and was downgraded to a tropical storm on the afternoon of 28 September.
Moving in a cyclonic loop over southern Mississippi, Georges became quasi-stationary for the next 6–12 h. The tropical storm began moving in a generally northeast to east direction early on 29 September and was downgraded to a tropical depression by midmorning while located about 30 n mi north-northeast of Mobile, Alabama. Georges continued to move eastward at 5–10 kt on 29 and 30 September. By the early morning of 1 October, the system dissipated near the northeast Florida–southeast Georgia coast, although a very weak remnant low did emerge over the western Atlantic during the day. This remnant circulation merged with a frontal zone by late on 1 October.
2) Meteorological statistics
(i) Wind and pressure data
The bulk of the aerial reconnaissance flights into Georges were done by the AFRC Hurricane Hunters. The Hurricane Hunters flew 17 missions, and made 81 center fixes while NOAA aircraft performed 6 missions contributing 24 center fixes. The highest wind speed reported was 152 kt (at 700 mb) at 0112 UTC 20 September by the NOAA aircraft. The lowest central pressure reported was 937 mb at 0613 UTC 20 September by the Hurricane Hunters with a corresponding maximum flight-level wind of 144 kt. During this period, subjective Dvorak intensity estimates from the TAFB, SAB, and AFWA were T6.5 (127 kt/935 mb) and objective-based Dvorak estimates ranged between T6.5 and T7.0 (140 kt/921 mb).
Georges's track brought it within range of WSR-88Ds at San Juan, Puerto Rico; Key West, Florida; New Orleans, Louisiana; and Mobile, Alabama. At 2213 UTC 21 September, the WSR-88D in San Juan measured winds near 100 kt 30–60 m above the antenna as the center was making landfall in Puerto Rico. A dual-doppler radar (Doppler-on-Wheels, DOW) was operating during Georges's landfall in Biloxi, Mississippi (J. Wurman 1998, personal communication). Around 0855 UTC 28 September, the radar showed a maximum 2–5-s wind gust of 107 kt.
Several land-based locations in the Caribbean recorded sustained hurricane force winds during Georges passage including Hamilton Airport and Virgin Islands Territorial Emergency Management Agency (VITEMA) at Herman Hill in St. Croix, Cyril E. King Airport in St. Thomas, and all the official reporting sites in Puerto Rico. The highest sustained wind and gust reported at an official site were 78 and 93 kt, respectively, at Roosevelt Roads Naval Station at 2302 UTC 21 September. These, as well as other selected surface observations for Georges, are listed in Table 6a. The highest unofficial wind report received in the Caribbean was a wind gust of 153 kt (at an elevation of about 213 m) from the island of Saba of the Netherlands Antilles at 1044 UTC 21 September. The corresponding minimum pressure recorded at the site was 971.9 mb. As is often the case in the Caribbean, many unofficial weather reports are relayed to the NHC via amateur radio operators. These observations are invaluable in helping to determine conditions in locations with no official weather reporting equipment. One of the most important observations reported was in Fajardo, Puerto Rico, where the Civil Defense office measured a sustained wind of 96 kt with gusts to 113 kt at 2130 UTC 21 September. Operationally, this report was the basis of making Georges a category 3 hurricane at landfall in Puerto Rico. The AFRC Hurricane Hunter reconnaissance reported a maximum flight-level wind of 117 kt and a minimum central pressure of 962 mb near the time of landfall in the southeastern Dominican Republic. Surface reports received from the Instituto de Meteorologica in Cuba indicate that the maximum 1-min surface wind observed was 71 kt at Punta Lucrecia, Holguin, while the highest gust of 80 kt was measured at Sagua La Grande, Villa Clara. The minimum central pressure recorded over Cuba was 988 mb in Cayo Coco. All of these reports occurred as Georges moved out of Cuba and into the Florida Straits where it began to restrengthen.
The maximum 2-min average wind recorded at Key West, Florida, was 48 kt at 1353 UTC 25 September with the peak gust of 76 kt; the minimum central pressure reported was 982.5 mb. It should be noted that higher winds and a lower pressure value likely occurred after an equipment/power failure at this site around 1500 UTC. The highest gust recorded in the Florida Keys was 96 kt at the Monroe County Emergency Operations Center in Marathon. The Sombrero Key C-MAN buoy recorded a maximum sustained wind of 82 kt with a peak gust to 92 kt at 1500 UTC 25 September. Moreover, this buoy recorded hurricane force winds for a 3-h period (1300–1600 UTC). This, along with other National Data Buoy Center (NDBC) observations, can be found in Table 6b.
Georges made its final landfall near Biloxi, Mississippi, around 1130 UTC on 28 September with maximum sustained surface winds of 90 kt and a minimum central pressure of 964 mb. The AFRC aircraft reported a 960-mb pressure at 0503 UTC. The lowest pressure measured by a land station was 964.9 mb at 1055 UTC 28 September at Keesler Air Force Base (KBIX) in Biloxi; Harrison County Civil Defense in Gulfport, Mississippi, recorded 967.2 mb at 1015 UTC. The NOAA ship Oregon II measured a minimum central pressure of 970 mb at 0830 UTC 28 September while in port in Pascagoula, Mississippi. On 28 September, KBIX reported sustained hurricane force winds (65 kt) at 0855 UTC. At 0755 UTC, KBIX reported wind gusts of 109 kt, and 149 kt at 0855 UTC 28 September. The latter value is questionable because 1) near-simultaneous DOW maximum wind measurements were 107 kt; 2) the anemometer at KBIX is a hot-wire anemometer, known to have serious errors in heavy rain, for example, the erroneous 205-kt wind gust in Typhoon Paka (Hagemeyer 1998); and 3) AFRC dropsonde data from the same time period measured a peak wind of 101 kt at 920 mb. A Texas Instrument WR25 anemometer, operated by Mississippi Power and Light 1 mi north of the beach in Biloxi, measured a wind gust of 102 kt. Reconnaissance data from the AFRC aircraft suggest that the boundary layer and inner core of Georges never fully recovered from its passage across Hispaniola and Cuba. Despite an apparently healthy cloud and outflow pattern, and a gradual drop in minimum central pressure of 13 mb (from 975 to 962 mb) over a 36-h period, from early on 26 September to the evening of 27 September the eye never became reestablished to its former stature. Most of the aircraft reports from 26 to 28 September indicated a partially formed eyewall, open to the west or southwest. Also, eyewall GPS dropsonde data near landfall in Mississippi suggest that the winds at the surface were 20%–30% below those at flight level (3 km). This contrasts with eyewall observations taken when Georges was near peak intensity just east of the Leeward Islands where surface winds were generally equal to or greater than those at the 3-km level.
(ii) Storm surge data
The storm surge plus breaking wave effects was about 3 m near Fajardo, Puerto Rico. On the U.S. mainland, high water mark surveys were conducted in the Florida Keys and along the northern Gulf of Mexico shoreline from Mississippi to Apalachicola, Florida. The goal of the survey was to measure both still water marks in buildings and debris lines. Generally, still water marks in buildings represent the storm surge while debris lines represent the storm surge and breaking wave effects combined. In almost all cases the measured debris line height outside a structure is higher than the measured still water mark inside the structure. In the Florida Keys the range of the still water marks was 0.9–2.3 m while the debris line heights were 1.1–3.3 m. Most of these measurements were taken on the Atlantic-facing shoreline. Along the Gulf coast the range of the still water marks was 2.1–3.7 m in Mississippi, 1.5–3.3 m in Alabama, and 1.1–1.4 m in the Florida panhandle. Similarly, the debris line heights ranged from 1.7 to 3.8 m in Mississippi, 1.5 to 3.8 m in Alabama, and 1.4 to 5.6 m in the Florida panhandle. The high debris line heights in Florida reflect the very deep water that lies just offshore that allows large waves to break very close to the shoreline and “run up” the beach.
(iii) Rainfall data
Georges was a prodigious rain producer. In the U.S. Virgin Islands, rainfall totals were generally between 76 and 203 mm. In Puerto Rico, the maximum official 2-day United States Geologic Survey (USGS) rain gauge measurement was 625 mm in Lago El Guineo near Villalba, while the maximum Cooperative Observer (COOP) 2-day total reported was 720 mm in Jayuya. No surface-based rainfall estimates are available from the Dominican Republic or Haiti. Satellite-derived estimates suggest that as much as 1000 mm of rain may have fallen over portions of the Dominican Republic and Haiti during the 24-h period ending at 1200 UTC on 23 September. Over Cuba, the Instituto de Meteorologica reported a maximum storm total of 620 mm in Limonar.
Rainfall in the Florida Keys was considerably less than in Cuba or Hispaniola. Key West recorded 213 mm. In contrast, storm totals along the Gulf coast were higher, because of the hurricane's marked deceleration. The maximum rainfall total from an official observation site was 616 mm at Eglin AFB in the Florida panhandle while the highest storm total was 753 mm from a COOP in Bay Minette, Alabama. Rainfall totals generally ranged from 250 to 500 mm over most of southern Mississippi and Alabama, and the Florida panhandle. As a result of the heavy rains, widespread river flooding occurred in southern Mississippi from 30 September through 2 October flooding homes and forcing evacuations. The Tchoutacabouffa River at D'lberbville, Mississippi, set a record crest of 5.8 m at 0200 UTC 30 September.
(iv) Tornadoes
Most of the reported tornado activity associated with Georges occurred in Florida and Alabama. Twenty-eight tornadoes are estimated to have touched down, mostly in northwest Florida. No deaths were directly attributable to these tornadoes. Two tornadoes were also reported in Puerto Rico.
3) Casualty and damage statistics
The 602 direct deaths attributed to Georges make it the 19th-deadliest TC in the Atlantic basin during the twentieth century. Most of the deaths occurred in the Dominican Republic and Haiti, due mainly to flash flooding and subsequent mud slides in high terrain regions. The lone direct death in the United States, a freshwater drowning, occurred in Mobile, Alabama.
Insured property damage estimates supplied by the Insurance Services Office yielded a total of $2.955 billion in the United States including Puerto Rico and the U.S. Virgin Islands. These estimates exclude storm surge damage. Based on the insured losses, the total estimated damage from Georges is $5.9 billion, $2.31 billion of which occurred in the continental United States. In Puerto Rico, there was extensive damage to homes throughout the island. More than 72 000 homes were damaged, of which about 28 000 were completely destroyed. During the hurricane, over 26 000 people were in shelters. In the Dominican Republic some 185 000 persons were left homeless by Georges, and 100 000 remained in shelters for several weeks due to a lack of electricity and water service. In Haiti, government officials stated that over 165 000 had been left homeless by the hurricane. Agriculture in Puerto Rico was hit hard by Georges: 95% of the plantain and banana crop was destroyed, along with 75% of the coffee crop. Despite Georges's weakened state when it moved across Cuba, it had a substantial impact. A total of about 60 000 homes were reported damaged, of which nearly 3500 were completely destroyed. As in Puerto Rico, agriculture was hard hit with major losses at banana plantations in eastern Cuba.
The damage to dwellings in the United States was not so extensive as in the Caribbean. In the Florida Keys, 1536 homes were damaged, of which 173 were completely destroyed. Many of these were mobile homes. Some roof and structural damage was also reported along the coast of Mississippi. In the first 60 days or so after Georges's final landfall in Mississippi, the American Red Cross spent $104 million on relief services in the United States Virgin Islands, Puerto Rico, Alabama, Louisiana, Mississippi, the Florida Keys, and the Florida panhandle. This makes Georges the most expensive disaster aid effort in the organization's 117-year history.
4) Warnings
Since Georges was well forecast, the lead times on the hurricane warnings were more than sufficient for the completion of typical protective actions. Nearly 897 000 residents evacuated in south and west-central Florida, including about 100 000 people in Dade County. About 35 000 persons left the Florida Keys in response to the mandatory evacuation order issued by the Monroe County Emergency Management Center. It should be noted however that about 40% of the Keys' residents did not evacuate, in spite of the threat of a direct hit by a major hurricane.
h. Tropical Storm Hermine, 17–20 September
Hermine was barely of tropical storm strength when it made landfall on the coast of Louisiana.
1) Synoptic history
Hermine developed from a tropical wave that crossed Dakar, Senegal, on 5 September and moved westward across the Atlantic. The wave had no significant deep convection until it reached the Windward Islands. There, cloudiness and showers increased and there was a large 24-h pressure change on 12 September. The wave continued westward close to the coast of South America and then moved northwestward through the northwest Caribbean Sea, to the Yucatan Channel, where a low pressure system developed. The system began to interact with an upper-level low in the Gulf of Mexico and another tropical wave that reached the area. During this period, a large and well-defined monsoon-type flow prevailed over Central America, the northwestern Caribbean Sea, and the Gulf of Mexico. It was not until 1200 UTC 17 September that the system acquired enough organization to be classified as a tropical depression.
The depression made a cyclonic loop over the central Gulf of Mexico as it interacted with the upper-level low located in the Bay of Campeche. The depression gradually became organized, despite the unfavorable upper-level wind shear that prevailed in the area, and reached tropical storm status at 1200 UTC 19 September. Hermine moved on a general northward track and made landfall as a weakening 35-kt tropical storm near Cocodrie, Louisiana, at 0500 UTC 20 September.
2) Meteorological statistics
Hermine's landfall resulted in rainfall of generally less than 25 mm, and there were no reports of tropical storm force winds.
3) Casualty and damage statistics
There were no reports of casualties due to Hermine, and damage was minor.
4) Warnings
Due to the unusually large uncertainties in the forecast track of Hermine while the cyclone was looping in the central Gulf of Mexico, it was necessary to issue tropical storm watches and warnings for a large portion of the Gulf coast, from the upper Texas coast to the Florida panhandle.
i. Hurricane Ivan, 19–27 September
Hurricane Ivan was the first in a trio of hurricanes to form and persist over the eastern Atlantic Ocean during the latter part of September. The wave from which Ivan developed was easily identified by a concentration of thunderstorms over western Africa near the Greenwich Meridian on 14 September. Although deep convection diminished when the wave neared the coast, cloudiness and sounding data showed the system's passage into the Atlantic on 16 September.
Strong convection associated with the wave redeveloped to the south of the Cape Verde Islands on 17 September. Position estimates from satellite analyses began that day. The cloudiness near the apex of the wave became more concentrated on 18 September. Late that day, Dvorak T-numbers first reached 2.0 from the TAFB and 1.5 from the SAB. Slow development followed and it is estimated that the system became a tropical depression near 0000 UTC on 19 September, about 175 n mi to the southwest of the Cape Verde Islands.
The TC was influenced by mid- to upper-level cyclonic flow centered to its northwest, over the central Atlantic. At first, that pattern consisted of a trough elongated southward from 30°N. During Ivan's development, however, water vapor imagery showed the trough become a closed circulation that partially enveloped the TC. A second trough later affected Ivan as well. These features influenced the details of Ivan's track, which was generally northwestward at 10–15 kt from 20 to 25 September. The troughs also produced vertical shear that slowed Ivan's development. Ivan became a tropical storm late on 20 September, and during the following 48 h wind speeds increased to around 55 kt, as estimated from the appearance of what could have been an eye on satellite pictures for about an hour.
Ivan's hurricane stage occurred rather far to the north. At 1400 UTC on 23 September, the eye reappeared and was more prominent than earlier, indicating that Ivan was becoming a hurricane as it neared 30°N. After again disappearing, the eye became its most distinct with a diameter of about 20 n mi. Ivan is estimated to have reached its peak intensity of about 80 kt on the morning of 26 September, when it was centered about 300 n mi to the west of the Azores Islands. At that time, the influence of the upper troughs on Ivan was decreasing and the track of the hurricane was increasingly controlled by the westerlies just to the north. Late on 26 September, the eye disappeared. The inner convective structure deteriorated rapidly as Ivan passed eastward over cooler waters north of the Azores Islands on 27 September. Ivan then became a weakening extratropical storm, and then gale, which moved northeastward.
j. Hurricane Jeanne, 21 September–1 October
Jeanne formed from a tropical wave that was slow to emerge from western Africa. The associated disturbed weather lingered near the African coast from 19 through 20 September, and gradually became better organized. An initial Dvorak classification was made by the TAFB at 1800 UTC 19 September, locating a center about 120 n mi offshore of the coast of Guinea. Only a slight increase in organization and little motion was noted during the following 24 h. By 0600 UTC 21 September, deep convection had increased and it is estimated that the system had become a tropical depression, while centered about 140 n mi southwest of the coast of Guinea–Bissau. According to NHC records beginning in 1886, only Tropical Storm Christine of 1973 formed farther east in the Atlantic basin than Jeanne. However, it should be noted that prior to the satellite surveillance era (which began in the 1960s) some storms may have gone undetected in this region.
The cyclone moved generally west-northwestward, gradually strengthening into a tropical storm later on 21 September. Jeanne was situated in an environment of slight east to southeasterly shear, which is typical for systems in the eastern tropical Atlantic. Early on 22 September, Jeanne began to intensify at a faster pace, and by 1800 UTC that day it became a hurricane about 120 n mi southwest of the Cape Verde Islands. This was the closest point of approach to those islands. For the next couple of days, Jeanne continued moving toward the west-northwest, strengthening to its estimated peak intensity of 90 kt while located about 580 n mi west of the westernmost Cape Verde Islands. The forward speed slowed and the hurricane turned toward the northwest, and then north, during 25–27 September. On 25–26 September Jeanne weakened, mainly due to increased southwesterly vertical shear. These events were likely caused by an amplifying mid- to upper-tropospheric trough located about 10° to the west. This trough assured that Jeanne would remain in the eastern Atlantic for its life cycle.
Under the continued influence of the trough, Jeanne accelerated toward the north-northeast on 28 September. The hurricane reintensified somewhat, to near 80 kt, while located about 550 n mi west-southwest of the Azores. As the system turned toward the northeast and east-northeast on 29 September, its forward speed slowed and it weakened to a tropical storm. Jeanne continued toward the east-northeast while gradually weakening. Around 0000 UTC 1 October, the cyclone reached the Azores but had degenerated to a depression with few tropical characteristics. The extratropical low moved eastward from the Azores, generating an area of gale force winds before reaching the coast of Portugal just north of Lisbon around 0000 UTC 4 October. Jeanne's extratropical remnants became unidentifiable over Spain later that day.
Jeanne's peak intensity of 90 kt on 24 September is based on satellite estimates. A French drifting buoy, identifier 41599, reported winds of 55 kt at 60°, 48 kt at 60°, and 75 kt at 110° near 23.3°N, 40.6°W at 1000, 1100, and 1900 UTC, respectively, on 26 September. Although this buoy's data could be considered questionable, the 75-kt wind was used for the best track intensity, since it was reported very near the center of the hurricane where, simultaneously, a burst of deep convection occurred. The island of Horta in the Azores reported wind gusts to 35 kt around 1800 UTC 30 September.
k. Hurricane Karl, 23–28 September
Karl developed from a small nontropical surface low pressure area that appeared off the coast of the Carolinas on 21 September. Deep convection became better organized as the low moved eastward, and a tropical depression formed from the disturbance near 1200 UTC 23 September about 50 n mi west-northwest of Bermuda. Convective banding increased and the system became Tropical Storm Karl that evening. Embedded in the flow on the southwest side of a broad deep-layer trough over the northwest Atlantic, the TC began moving east-southeastward about this time.
Satellite imagery showed the gradual development of a more symmetrical cloud pattern with the center embedded among the coldest convective tops. Karl became a hurricane near 1200 UTC 25 September about 550 n mi east-southeast of Bermuda. At this time, Hurricane Georges was over the Straits of Florida, Hurricane Ivan was over the North Atlantic about 500 n mi west-southwest of the Azores, and Hurricane Jeanne was over the tropical Atlantic about midway between Africa and the Lesser Antilles. According to records at the NHC, the last time four hurricanes were present in the Atlantic at the same time was on 22 August 1893. Records also note that on 11 September 1961, three hurricanes and possibly a fourth existed.
Karl moved northeastward around the southeastern side of the large deep-layer trough. A well-defined eye developed, and it is estimated that Karl first reached a maximum intensity of 90 kt at 0000 UTC 27 September while centered about 875 n mi east-northeast of Bermuda. The eye remained distinct for at least 6 h, after which time the hurricane started to weaken primarily due to increasing upper-level shear. The hurricane accelerated toward the northeast and weakened to a tropical storm by 0000 UTC 28 September over 23°C water 175 n mi west-northwest of the westernmost Azores. Karl continued moving over increasingly cooler waters and became extratropical later on 28 September as the circulation center became well removed from any deep convection. The extratropical cyclone was tracked to south of Ireland by late on 29 September.
l. Hurricane Lisa, 5–9 October
Lisa originated from a tropical wave that moved westward from Africa into the eastern tropical Atlantic Ocean on 29 September. The associated cloudiness was fairly well organized and centered at about 10°N. By the next day, it was an almost indistinguishable part of the intertropical convergence zone (ITCZ) which was active across the entire tropical Atlantic. By 3 October, the system became better defined as its convection increased and the ITCZ cloudiness dissipated to its east and west. On 4 October, midway between Africa and the Lesser Antilles, there were signs of a low-level circulation and it is estimated that a tropical depression formed at 0000 UTC on 5 October.
The depression strengthened into Tropical Storm Lisa on 5 October, although it was in an environment of strong vertical shear, as evidenced by the low-level center's appearance to the west of the deep convection. This shear was caused by an upper-level low located northwest of the storm. The presence of this low also weakened the ridge to the north, causing the storm to turn northward. During the next 2 days, a strong baroclinic trough in the westerlies evolved into a deep low in the central North Atlantic. This resulted in an acceleration toward the northeast. The forward speed exceeded 50 kt by the afternoon of 9 October. The vertical shear relaxed over the storm and it gradually strengthened. Lisa turned northward on 9 October, steered by the deep low to its west and a 1032-mb high to its east. This strong east–west pressure gradient also resulted in increasing the surface winds well to the east of the center. Lisa briefly strengthened to a 65-kt hurricane on 9 October before merging with an extratropical frontal system in the far North Atlantic. On 10 October, it was no longer possible to identify a well-defined circulation in satellite imagery.
A NOAA drifting buoy (16.6°N, 46.9°W) in the central tropical Atlantic observed 35-kt winds at 0850 UTC on 5 October and 36-kt winds at 2138 UTC. These observations were essential in determining that Lisa had become a tropical storm, as satellite-based intensity estimates were well below storm strength at these times. The estimate that Lisa acquired 65-kt hurricane force winds on 9 October was based on satellite intensity estimates and on a report of 61 kt from the ship ZCBD9 at 46.9°N, 33.3°W (approximately 240 n mi east of the center) at 1800 UTC. By that time the system was rapidly becoming extratropical, and it is not certain that the strongest winds were near the center.
m. Hurricane Mitch, 22 October–5 November
Mitch was responsible for over 9000 deaths, predominately from rain-induced flooding in Honduras and Nicaragua, making it one of the deadliest Atlantic TCs in history. The 905-mb minimum central pressure and estimated maximum sustained wind speed of 155 kt over the western Caribbean make Mitch the strongest October hurricane observed since records began in 1886. Mitch moved across the Yucatan Peninsula and southern Florida as a tropical storm.
1) Synoptic history
Mitch developed from a tropical wave that moved across the southern portion of west Africa on 8–9 October. Rawinsonde data from Abidjan, Ivory Coast (980 n mi southeast of Dakar), suggest that the wave passed through the region around 8 October. The system crossed the west coast of Africa, south of 15°N, on 10 October. Over the next 7 days, west-southwesterly upper-level winds prevented significant development as the wave progressed across the tropical Atlantic.
After the system moved through the eastern Caribbean Sea on 18 and 19 October, satellite pictures on 20 October showed an organizing cloud pattern over the south-central Caribbean. Shower and thunderstorm activity continued to become better organized over the southwest Caribbean Sea early on 21 October. That afternoon, an AFRC reconnaissance aircraft investigated the disturbance, and found winds of 39 kt at the 300-m flight level and a central pressure of 1001 mb. Thus, the system became a tropical depression at 0000 UTC 22 October, about 360 n mi south of Kingston, Jamaica. The depression strengthened to a tropical storm later that day, as it moved in a cyclonic loop about 225 n mi east-southeast of San Andres Island. By 23 October, intensification was disrupted by westerly vertical wind shear associated with an upper-level low north-northwest of the TC. Later on 23 October, the upper low weakened, the shear diminished, and Mitch began to strengthen while moving slowly northward.
Mitch became a hurricane at 0600 UTC 24 October centered about 255 n mi south-southwest of Kingston. Later that day, as it turned toward the west, Mitch began a period of rapid intensification. During the 24-h period beginning on the afternoon of 24 October, its central pressure dropped 52 mb, to 924 mb. A symmetric, well-established upper-tropospheric outflow pattern was evident in satellite imagery, and the hurricane continued to strengthen. On the afternoon of 26 October, the central pressure reached a minimum of 905 mb, while the cyclone was centered about 50 n mi southeast of Swan Island. This pressure is the fourth lowest ever measured in an Atlantic hurricane, tied with Hurricane Camille in 1969. This is also the lowest pressure ever observed in an October hurricane in the Atlantic basin. Prior to Mitch, the strongest measured hurricane in the northwest Caribbean was Hurricane Hattie in 1961, with a central pressure of 924 mb. At its peak on 26 October, Mitch's maximum sustained 1-min surface winds were estimated to be 155 kt, category 5 intensity. Figure 9 is a satellite image of Mitch near peak strength.
After passing over Swan Island on the 27 October, Mitch began to gradually weaken while moving slowly westward. It then turned southwestward and southward toward the Bay Islands off the coast of Honduras. By around 0000 UTC on 28 October, even though the central pressure had risen to 933 mb, and maximum flight-level (700 mb) winds were near 130 kt, GPS dropsondes indicated that surface winds were about 140 kt, that is still category 5 intensity. Even though Mitch passed very near the island of Guanaja as a category 5 hurricane, it is unlikely that winds of that strength were experienced on the island. Mitch slowly weakened as its circulation interacted with the landmass of Honduras. From midday on 27 October to early on 29 October, the central pressure rose 59 mb. The hurricane meandered near the north coast of Honduras from late on 27 October through 28 October before landfall about 70 n mi east of La Ceiba during the morning of 29 October, with estimated surface winds of 70 kt and a minimum central pressure of 987 mb.
After landfall, Mitch moved slowly southward, then southwestward and westward, over Honduras, weakening to a tropical storm by 0600 UTC 30 October, and to a tropical depression by 1800 UTC 31 October. The overall motion was slow, less than 4 kt, for a week. This resulted in a tremendous amount of rainfall, as high as 900 mm or more, primarily over Honduras and Nicaragua (Table 7). A large east–west mountain range, with peaks approaching 3 km, covers this part of Central America. Thus, upslope rainfall enhancement likely contributed to the large totals. The heavy rainfall produced flash floods and mud slides that killed thousands of people. Some heavy rains also occurred in other portions of Central America. Although Mitch's surface circulation center dissipated near the Guatemala–Mexico border on 1 November, the remnant circulation aloft continued to produce locally heavy rainfall over portions of Central America and eastern Mexico for the next couple of days.
By the afternoon of 2 November, TAFB and SAB meteorologists began to follow a cloudsystem center, the remnants of Mitch, in satellite imagery over the Bay of Campeche. Shower and thunderstorm activity began to increase later on 2 November. On 3 November, a low-level circulation became evident in the eastern Bay of Campeche. Later that day, an AFRC aircraft found 45-kt winds at 500 m and a minimum central pressure of 997 mb, indicating the redevelopment of Tropical Storm Mitch about 130 n mi southwest of Merida, Mexico. Mitch moved northeastward and weakened to a depression early on 4 October after it made landfall over the northwestern Yucatan Peninsula. The center reemerged over the south-central Gulf of Mexico by midmorning on 4 October, and Mitch regained tropical storm strength. The storm began to accelerate northeastward as it became involved with a frontal zone moving through the eastern Gulf of Mexico. Mitch made landfall on the morning of 5 November in southwest Florida near Naples, with estimated maximum sustained winds of 55 kt. Mitch continued to move rapidly northeastward and by midafternoon of 5 October, moved offshore of southeastern Florida and became extratropical. The extratropical cyclone accelerated northeastward across the North Atlantic Ocean from 6 October through 9 October.
2) Meteorological statistics
The AFRC Hurricane Hunters flew 19 missions, and made 41 center fixes while NOAA aircraft performed 2 missions and 9 center fixes. The highest 700-mb flight-level wind report was 168 kt at 1900 UTC 26 October by the AFRC. This wind speed was observed 14 n mi northeast of the center near the time of a 905-mb GPS dropsonde-measured pressure. A dropsonde in the northeast eyewall showed winds to near 160 kt at 900 mb, but lower speeds below that altitude. There was a 14-m wind speed of 130 kt measured by a dropsonde at 2244 UTC 27 October. The highest satellite-based intensity estimate, obtained by both objective and subjective methods, was 155 kt on 26 and 27 October. Table 7a lists rainfall observations from Honduras, with a maximum of 911.6 mm from Choluteca. Even higher values may have gone unobserved. Table 7b lists selected surface observations from Florida, where the highest observed sustained wind speed was 52 kt, at an elevation of 44 m, from the Fowey Rock C-MAN station just offshore of Miami. Mitch spawned five tornadoes in south Florida: two in the Florida Keys, one each in Broward, Palm Beach, and Collier Counties. The most significant of these (F2 intensity) occurred in the upper Florida Keys, Islamorada to North Key Largo.
3) Casualty and damage statistics
The estimated death toll from Mitch is 9086. The U.S. Agency for International Development reported the following death totals: Honduras, 5677; Nicaragua, 2863; Guatemala, 258; El Salvador, 239; Mexico, 9; and 7 in Costa Rica. The death toll also includes 31 fatalities when the schooner Fantome sank near Guanaja (Carrier 2001). In addition, another 9191 persons were listed as missing. The exact death toll will probably never be known. However, Mitch was one of the deadliest Atlantic TCs in history, ranking below only the 1780 “Great Hurricane” in the Lesser Antilles (Rappaport and Fernandez-Partagas 1995), and comparable to the Galveston hurricane of 1900, and Hurricane Fifi of 1974 (which also struck Honduras). Mitch also claimed two lives in Monroe County, Florida. Both deaths were drowning-related incidents resulting from a fishing boat capsizing. Sixty-five persons were injured by the tornadoes in Florida.
In Honduras, there was an estimated 50% loss to crops. At least 70 000 houses were damaged, and more than 92 bridges were damaged or destroyed. There was severe damage to the infrastructure of Honduras; entire communities were isolated from outside assistance. To a lesser extent, damage was similar in Nicaragua, where a large mud slide inundated 10 communities situated at the base of La Casitas Volcano. Guatemala and El Salvador also suffered from flash floods that destroyed thousands of homes, and extensively damaged bridges and roads. The Florida tornadoes damaged or destroyed 645 homes. Insured property damage supplied by the Florida Insurance Council puts the insured damage estimate for Florida at $20 million. These estimates exclude storm surge damage. The total estimated U.S. damage from Mitch is $40 million.
4) Warnings/forecast critique
Hurricane warnings were issued for Jamaica, Honduras, Guatemala, Belize, and the Caribbean coast of the Yucatan Peninsula, Mexico. A tropical storm warning was issued for the Cayman Islands, the Gulf of Mexico coast of the Yucatan Peninsula, Cuba, and south Florida and the Florida Keys. As the effects of Mitch on Nicaragua were confined to freshwater flooding, there were no hurricane warnings there.
Official track forecasts, consistent with normally reliable guidance models, called for a slow mostly northwestward motion when Mitch was in the northwestern Caribbean. In fact, Mitch moved westward and then southward; the forecast turn toward the northwest did not take place until well after the hurricane had moved inland. In retrospect, the slow southward, then southwestward, motion that began early on 27 October, was likely due to a weak midlevel anticyclone over the western Gulf of Mexico. However, the absence of radiosonde data from Mexico and Central America appears to have hindered prediction models from resolving this feature. Also, the intensity of Mitch was underpredicted by as much as 75–80 kt in the NHC forecasts at 72 h. This underscores the difficulty in forecasting rapid strengthening with the current state of the science.
n. Hurricane Nicole, 24 November–1 December
Nicole developed from a nearly stationary and strong frontal low that persisted for several days over the northeast Atlantic, a few hundred miles south of the Canary Islands. Satellite imagery suggested that the frontal low acquired tropical characteristics when a tightly wrapped convective band developed around the center of circulation. The system reached tropical storm status around 0600 UTC 24 November. Later on, a ship (call sign PFSJ) confirmed that the system had acquired tropical characteristics, with a report of 36-kt winds at 1200 UTC 24 November just to the north of the center of the tropical storm. The TC was located near the center of a larger upper-level low where the vertical wind shear was relatively weak. This is typical for late-season developments in the subtropics. Nicole continued to become organized later on 24 November; an intermittent eye was observed in satellite images. Maximum winds increased to 60 kt as indicated by reports from the same ship.
Nicole moved toward the west-southwest for the next few days south of a strong midlevel high pressure ridge. An upper-level trough moved rapidly eastward over the system, producing a strong wind shear. The shear removed most of the convection associated with the TC, which weakened to a tropical depression on 26 November. In fact, the system became so weak that it practically dissipated. However, the ridge that followed the upper-level trough became superimposed over the system, decreasing the shear. Deep convection regenerated, and, unexpectedly, the system reacquired tropical storm strength by 27 November. Nicole's regeneration is yet another illustration of the uncertainties in intensity forecasting, particularly at higher latitudes where extratropical influences play a role. The rejuvenated TC moved on a west-northwest track. Thereafter, it turned toward the northeast ahead of another strong approaching cold front. Nicole intensified further and reached hurricane status with peak winds of 75 kt and a minimum pressure of 979 mb at 0000 UTC 1 December. These estimates were based on satellite images that revealed the formation of an eye, resulting in objective T numbers near 4.5. In addition, data from the Defense Military Satellite Program 85-GHz microwave sensor showed an almost complete eyewall. During that period, Nicole was moving over sea surface temperatures about 2°–3°C warmer than normal, and these anomalously warm waters probably contributed to the intensification of the system. Nicole moved rapidly northward and north-northwestward around the periphery of a large deep-layer cyclonic circulation and became extratropical by 1800 UTC 1 December.
3. Forecast verification
The NHC verifies their forecasts of tropical and subtropical storms and hurricanes by comparing the “official forecast” positions and intensities to the “best track” data for each cyclone. For all TCs that are identified operationally in the Atlantic basin, the NHC issues 6-hourly official forecasts of the center position and maximum 1-min average wind speed. These official forecasts are valid 12, 24, 36, 48, and 72 h from the initial time.
Table 8 is a listing of the average track forecast errors for the tropical storms and hurricanes of 1998 for the official forecasts and for the climatology and persistence model (CLIPER; Neumann 1972). Also shown in this table are the 1988–97 average official and CLIPER track forecast errors, and the departures of the 1998 average official and CLIPER track forecast errors from the 1988–97 averages. Official track forecast errors for 1998 were slightly lower than the recent 10-yr averages at 12–72 h. A comparison with the corresponding CLIPER errors shows that, except at 0 and 12 h, the official forecasts for 1998 were slightly more improved from the 10-yr average than was CLIPER. This is of interest, since the CLIPER model is often used as a measure of forecast difficulty. Thus, the 1998 tracks were just slightly easier than average to forecast, and the official track forecasts for 1998 were, on average, quite good in comparison to other years.
Table 9 contains a listing of the average intensity (maximum one-min average wind speed) forecast errors for the 1998 tropical storms and hurricanes, and for the 1990–97 time period. Two measures of intensity forecast errors are used: the average error or bias, that is, the average forecast minus observed maximum 1-min wind speed, and the average absolute error, that is, the average absolute value of the forecast minus observed maximum 1-min wind speed. Also listed in this table are the corresponding intensity forecast errors for the Statistical Hurricane Intensity Forecast Model (SHIFOR; Jarvinen and Neumann 1979). There is a slight negative bias in the average official intensity forecasts through 36 h, but the average official errors are quite small at 48 and 72 h, indicating little bias for these later forecast time periods. For the corresponding average SHIFOR intensity forecast errors, there is a substantial negative bias at 48 and 72 h. A small average error, or bias, in intensity forecasts may be misleading since it could be the result of averaging under- and overpredictions of the maximum wind speed. So, the average absolute intensity forecast errors are a helpful measure of forecast skill. For 1998, the average absolute official intensity forecast errors ranged from 7.5 kt at 12 h to 21.2 kt at 72 h. Moreover, for 12–72 h, these average absolute errors are lower than the corresponding average SHIFOR errors. Since an improvement over SHIFOR is an indication of skill, we conclude that the 1998 average official intensity forecasts had some skill out to 72 h. It is also of interest to compare the 1998 average absolute intensity forecast errors to those for a longer-term period, 1990–97. One can see that the official errors were higher than the recent 8-yr average but that the SHIFOR errors were even higher. Therefore, since SHIFOR provides a measure of intensity forecast difficulty, 1998 featured storms whose intensity changes were harder to predict than usual. In summary, the 1998 official intensity forecast errors were slightly larger than average because of higher than normal forecast difficulty, but these forecasts exhibited skill from 1–3 days.
4. Weaker tropical systems
In this section, we will briefly discuss the tropical depressions and tropical waves that occurred during the 1998 season. The tracking of tropical waves goes back at least as far as the work of Dunn (1940). These waves play a crucial role as precursors to TC formation over the Atlantic and eastern Pacific Oceans. In fact, an average of 10% of the tropical waves develop into named TCs each year (Table 10). Normally, the strongest hurricanes originate from tropical waves. Moreover, tropical waves are one of the principal modulators of rainfall in the Caribbean basin (Riehl 1954). References to publications about tropical wave structure and properties are included in Pasch et al. (1998). A summary of how tropical waves are tracked operationally at the TPC/NHC can be found in Avila et al. (2000). Here, the statistics of depressions and waves, which began in 1967, are updated; hereafter, the compilation will be included in the Atlantic hurricane season annual summary article.
a. Tropical depressions
There were 14 tropical depression in 1998, a number that is below the 1967–97 average of 19. As pointed out by Pasch and Avila (1994), notably fewer depressions per year were counted after the early 1980s, so this average is likely not representative. Because all of the tropical depressions reached tropical storm status, their histories have been included in section 2.
b. Tropical waves during 1998
Figure 10 summarizes the tropical wave activity during 1998 and highlights the TCs that formed from the waves. For the period of May to November, 60 tropical waves crossed Dakar and moved westward over the tropical Atlantic, the Caribbean Sea, and Central America. Most of them appeared to continue into the eastern North Pacific. There was approximately one wave crossing at Dakar every 3.5 days. The long-term average number of waves observed for the same period is 61. Note that, except for Hurricanes Karl and Nicole in the Atlantic and Tropical Storm Kay in the eastern North Pacific, the rest of the TCs developed in the “envelope” region of genesis attributed to tropical waves.
Figure 11 displays vertical time sections of the wind for Dakar and Guadeloupe from 15 August to 15 September 1998. Figure 12 shows sequences of twice per day satellite images for nearly the same period. This sample represents the climatologically most active period when tropical waves are usually the strongest and most convectively active.
From the early portion of the season, the tropical waves were clearly identified as they crossed Dakar and, in general, were accompanied by a strong midtropospheric easterly jet. In fact, a strong wave crossed Dakar on 17 June with a 50-kt easterly wind speed maximum at 600 mb. The midlevel easterly jet is considered of great importance in relation to the structure of the tropical waves. Burpee (1972) has shown that the meridional variations of the horizontal wind at the midtropospheric levels are related to wave amplification. Experience has shown that years with either a relatively weak midtropospheric jet or (as is more often the case) with a southward shift of the jet in the Atlantic normally coincide with inactive hurricane seasons.
Tropical waves were very well defined through July and in fact, one of these waves produced Tropical Storm Alex late that month. This earlier than normal TC genesis from waves suggested that the large-scale environment was already becoming favorable for waves to initiate TC formation. The waves continued to be quite strong, and the environment mostly favorable, during August and September. Figure 11 shows distinct tropical waves crossing Dakar between 15 August and 15 September. Note the deep and sharp cyclonic wind shift below 400 mb associated with the waves as they crossed Dakar. These waves were embedded within a deep easterly flow that extended to at least 200 mb throughout the period.
Because in general, waves become more difficult to track as they propagate westward across the Atlantic, the mean wind for the period was removed from the Guadeloupe time section. This way, one can then depict more clearly the passage of the waves by focusing on the sign change of the perturbed meridional component of the wind. The Guadeloupe section reveals the well-defined mid- to lower-tropospheric cyclonic perturbations associated with several waves crossing that site. In contrast with near Dakar, there was a change in the upper-level winds at Guadeloupe from easterlies, during the last 2 weeks of August, to westerlies during the following 15 days. Consequently, waves moving through the area during the first half of September encountered an environment dominated by upper-level westerlies.
Satellite imagery Hovmöller diagrams constructed operationally from May to November reveal the westward-propagating cloudiness associated with the waves during the period. The waves became distinct and convectively active over the tropical Atlantic during the second half of August (Fig. 12). Although waves were well defined in the wind field during early September, their convective activity was rather suppressed between the Antilles and Africa. This period was dominated by upper-level westerlies, alluded to above, and coincided with a relatively inactive 2-week period of no TC activity in the tropical Atlantic. During this period, most of the convection was concentrated in the western Caribbean and the Gulf of Mexico, where TC activity also was concentrated.
c. Comparison with other years
Table 10 summarizes the tropical system statistics from 1967 to 98. The ratio, R, between the number of tropical storms forming from tropical waves to the total number of tropical storms is used to describe the overall character of the hurricane season (Avila and Clark 1989). Low values of R indicate that many tropical storms originated from nontropical “seedling” disturbances such as upper-tropospheric cold lows or perturbations along frontal zones. High values of R indicate that storms mostly develop from tropical waves in the tropical Atlantic, generally south of 20°N. With a few exceptions, years with high values of R are years of strong hurricanes. The 1967–97 average contribution from tropical waves to the total number of storms is 0.62. In 1998, R was 0.86. Thus 1998 had many intense hurricanes in the deep Tropics.
Acknowledgments
Miles B. Lawrence, Max Mayfield, and Edward N. Rappaport of the TPC/NHC contributed to this article. James Franklin of the TPC/NHC and two anonymous reviewers provided useful comments. The authors are grateful to Stephen R. Baig of the TPC for producing the track chart, to Jiann-Gwo Jiing of the TPC for producing the vertical time sections of winds, and to Joan David of the TPC for assistance with the figures. We also thank Christopher W. Landsea of the NOAA Hurricane Research Division (NOAA/HRD) for helpful comments, and for providing the sea surface temperature anomaly map, as well as Stanley B. Goldenberg and Steve Feuer of the NOAA/HRD for the wind shear anomaly map. Finally, we thank Chris Velden and Tim Olander from the Space Science and Engineering Center, University of Wisconsin, for providing the satellite pictures.
REFERENCES
Avila, L. A., and G. B. Clark, 1989: Atlantic tropical systems of 1988. Mon. Wea. Rev, 117 , 2260–2265.
Avila, L. A., R. J. Pasch, and J-G. Jiing, 2000: Atlantic tropical systems of 1996 and 1997: Years of contrasts. Mon. Wea. Rev, 128 , 3695–3706.
Burpee, R. W., 1972: The origin and structure of easterly waves in the lower troposphere of North Africa. J. Atmos. Sci, 29 , 77–90.
Carrier, J., 2001: The Ship and the Storm. McGraw-Hill, 266 pp.
Dunn, G. E., 1940: Cyclogenesis in the tropical Atlantic. Bull. Amer. Meteor. Soc, 21 , 215–229.
Dvorak, V. F., 1984: Tropical cyclone intensity analysis using satellite data. NOAA Tech. Rep. NESDIS 11, National Oceanic and Atmospheric Administration, U.S. Department of Commerce, Washington, DC, 47 pp.
Gray, W. M., 1968: Global view of the origin of tropical disturbances and storms. Mon. Wea. Rev, 96 , 669–700.
Hagemeyer, R. H., 1998: Super Typhoon Paka, December 2 through 21, 1997. DOC/NOAA Service Assessment, National Weather Service, Honolulu, HI, 28 pp.
Hock, T. F., and J. L. Franklin, 1999: The NCAR GPS dropsonde. Bull. Amer. Meteor. Soc, 80 , 407–420.
Jarvinen, B. R., and C. J. Neumann, 1979: Statistical forecasts of tropical cyclone intensity for the North Atlantic basin. NOAA Tech. Memo. NWS NHC-10, 22 pp.
Landsea, C. W., R. A. Pielke Jr., A. M. Mestas-Nunez, and J. A. Knaff, 1999: Atlantic basin hurricanes indices of climatic changes. Climatic Changes, 42 , 89–129.
Neumann, C. J., 1972: An alternate to the HURRAN (hurricane analog) tropical cyclone forecast system. NOAA Tech. Memo. NWS SR-62, 24 pp.
Pasch, R. J., and L. A. Avila, 1994: Atlantic tropical systems of 1992. Mon. Wea Rev, 122 , 540–548.
Pasch, R. J., L. A. Avila, and J-G. Jiing, 1998: Atlantic tropical systems of 1994 and 1995: A comparison of a quiet season to a near-record-breaking one. Mon. Wea. Rev, 126 , 1106–1123.
Rappaport, E. N., and J. Fernandez-Partagas, 1995: The deadliest Atlantic tropical cyclones, 1492–1994. NOAA Tech. Memo. NWS-NHC-47, 41 pp.
Riehl, H., 1954: Tropical Meteorology. McGraw-Hill, 392 pp.
Riehl, H., and R. J. Shafer, 1944: The recurvature of tropical storms. J. Meteor, 1 , 42–54.
Simpson, R. H., 1974: The hurricane disaster potential scale. Weatherwise, 27 , 169186.
Tracks of tropical storms and hurricanes in the Atlantic basin during 1998
Citation: Monthly Weather Review 129, 12; 10.1175/1520-0493(2001)129<3085:AHSO>2.0.CO;2
Sea surface temperature departures from normal (average from 1961 to 1990) for the period 1 Aug–31 Oct 1998. Contour interval is 0.25°C. Shaded areas denote warm anomalies
Citation: Monthly Weather Review 129, 12; 10.1175/1520-0493(2001)129<3085:AHSO>2.0.CO;2
Anomalies of the magnitude of the vertical shear (850 mb minus 200 mb) of the wind from normal (1958–98 mean) for Aug–Oct of 1998. Contour interval is 1 m s−1. Dots show locations where TCs developed (tropical depression formed) and crosses show locations where the cyclones reached tropical storm strength during these three months. Shaded areas denote wind shears higher than the long-term mean
Citation: Monthly Weather Review 129, 12; 10.1175/1520-0493(2001)129<3085:AHSO>2.0.CO;2
(a) Mean 500-mb-height contours for Aug–Sep of 1998, with tracks of TCs that occurred during this period superimposed. Contour interval is 10 m. (b) Anomalies of the mean 500-mb height field for Aug–Sep 1998 from normal. Contour interval is 5 m
Citation: Monthly Weather Review 129, 12; 10.1175/1520-0493(2001)129<3085:AHSO>2.0.CO;2
Visible Geostationary Operational Environmental Satellite-8 (GOES-8) satellite image of Hurricane Bonnie at 1615 UTC 23 Aug 1998, near the time of peak intensity
Citation: Monthly Weather Review 129, 12; 10.1175/1520-0493(2001)129<3085:AHSO>2.0.CO;2
Visible GOES-8 satellite image of Hurricane Georges at 1545 UTC 19 Sep 1998, showing a well-defined eye
Citation: Monthly Weather Review 129, 12; 10.1175/1520-0493(2001)129<3085:AHSO>2.0.CO;2
Radar image, from the Meteorological Service of the Dominican Republic in Santo Domingo, of Hurricane Georges near landfall in the Dominican Republic at 1332 UTC 22 Sep 1998
Citation: Monthly Weather Review 129, 12; 10.1175/1520-0493(2001)129<3085:AHSO>2.0.CO;2
Visible GOES-8 satellite image of Hurricane Georges at 1315 UTC 22 Sep 1998, near the time of landfall in the Dominican Republic
Citation: Monthly Weather Review 129, 12; 10.1175/1520-0493(2001)129<3085:AHSO>2.0.CO;2
Visible GOES-8 satellite image of Hurricane Mitch near its peak intensity of 155 kt
Citation: Monthly Weather Review 129, 12; 10.1175/1520-0493(2001)129<3085:AHSO>2.0.CO;2
Total number of waves that maintained their identities while traveling the Atlantic, Caribbean, Gulf of Mexico, and the eastern Pacific during 1998. The figure highlights the (shaded) envelope in which tropical cyclones developed, as shown by the lettered symbols (named cyclones) or numbered symbols (tropical depressions). Symbols outside the envelope are the approximate locations of tropical cyclone formation from disturbances other than tropical waves
Citation: Monthly Weather Review 129, 12; 10.1175/1520-0493(2001)129<3085:AHSO>2.0.CO;2
Vertical time section of the wind at (a) Dakar, Senegal, and (b) Guadeloupe from 15 Aug to 15 Sep. Winds are plotted every 12 h according to convention with each half barb and full barb denoting 5 and 10 kt, respectively, and the solid flag denoting 50 kt. The mean (average for the period) wind was removed from the Guadeloupe section. Shaded areas correspond to a positive (southerly) perturbation wind component
Citation: Monthly Weather Review 129, 12; 10.1175/1520-0493(2001)129<3085:AHSO>2.0.CO;2
Longitude vs time (Hovmöller) diagram of GOES-8 infrared images taken twice a day at 2345 and 1145 UTC from (a) 15 to 29 Aug, and (b) 30 Aug to 12 Sep 1998. The latitude belt is roughly 5°–25°N. Latitude–longitude lines are at 10° intervals
Citation: Monthly Weather Review 129, 12; 10.1175/1520-0493(2001)129<3085:AHSO>2.0.CO;2
(Continued)
Citation: Monthly Weather Review 129, 12; 10.1175/1520-0493(2001)129<3085:AHSO>2.0.CO;2
Atlantic hurricane season statistics of 1998
Hurricane Bonnie selected surface observations, Aug 1998
Tropical Storm Charley selected surface observations, Aug 1998
Hurricane Earl selected surface observations, Sep 1998
(Continued)
Tropical Storm Frances selected surface observations, Sep 1998
(Continued)
Table 6a. Hurricane Georges selected surface observations, Sep 1998
Table 6a. (Continued)
Table 6a. (Continued)
Table 6a. (Continued)
Table 6b. Hurricane Georges selected National Buoy Data Center (NBDC) observations, Sep 1998
Table 7a. Hurricane Mitch selected Honduras rainfall totals, 25–31 Oct 1998
Table 7b. Hurricane Mitch selected surface observations, Oct 1998
Comparison of 1998 Atlantic official and CLIPER average track forecast errors (rounded to the nearest n mi) for a homogeneous sample (excluding extratropical and tropical depression stages) with 1988–97 10-yr average. A forecast error is defined as the great circle distance between a forecast and a postanalysis best-track position for the same time. Cases include all subtropical storms, tropical storms, and hurricanes. Also shown is the range of the track forecast errors (n mi) for each forecast period
Homogeneous comparison of official and SHIFOR maximum 1-min wind speed forecast errors (rounded to the nearest 0.1 kt) for subtropical storms, tropical storms, and hurricanes in the Atlantic basin for 1998 with 1990–97 8-yr average. Error = forecast − observed.
Atlantic tropical system statistics for 1967–98
