Hurricane Marilyn devastated portions of the U.S. Virgin
Islands when it hit that area with Category 2 to near Category 3 intensity
on the Saffir/Simpson Hurricane Scale (SSHS).
a. Synoptic History
Marilyn originated from a tropical wave
that crossed from the west coast of Africa to the eastern tropical Atlantic Ocean on 7-8
September. A large circulation of low- and middle-level clouds
accompanied the wave, but little deep convection was generated at
that time. The system moved westward at about 17 knots over the
following few days, under upper-level easterlies on the south side
of a well-defined anticyclone aloft, which also moved westward.
The initial Dvorak technique
T-number intensities of 1.0 were assigned late on the 11th by satellite analysts at the National
Hurricane Center (NHC) and the NESDIS
Synoptic Analysis Branch
(SAB). Although the low-level circulation was rather disorganized then,
deep convection developed and became concentrated near the analyzed center on the 12th.
Based on analysis of satellite pictures, it became the
fifteenth 1995 Atlantic tropical depression
at 1800 UTC on the 12th (Table 1,
Fig. 1 [69K GIF]). The
further, becoming Tropical Storm Marilyn
six hours later. Marilyn reached hurricane
strength 4 hours after that, at 0000 UTC on the 14th, shortly after the
U.S. Air Force Reserves (Hurricane Hunters)
first identified a closed eyewall
during their reconnaissance flight.
Over the following three days, the track gradually became
directed toward the west-northwest and then the northwest while the
hurricane moved toward a weakness in the subtropical ridge over the
central Atlantic Ocean. Marilyn continued to strengthen in an
"embedded center" cloud pattern, but at a slower rate during that
period. It was a Category 1 hurricane on the 14th when the
passed about 45 n mi to the north of Barbados, then just north of
Martinique, over Dominica, to just southwest of Guadeloupe.
Marilyn continued moving northwestward over the northeastern
Caribbean Sea. It hit the U.S. Virgin Island during the afternoon
and night of the 15th as a strengthening Category 2, nearly
Category 3, hurricane. The
reported hail, an unusual occurrence for tropical
cyclones. They noted an eye of 20 n mi diameter.
The strongest part of the hurricane, the eyewall to the east and northeast of the center,
passed over St. Thomas. Maximum one-minute surface winds at that time were close to
After passing just offshore from eastern Puerto Rico early on
the 16th, the center of Marilyn was again over the Atlantic Ocean.
An upper-level low had developed to the west and this could have
enhanced outflow aloft from Marilyn. An eye became distinct on
satellite pictures and Marilyn reached its peak intensity, about
949 mb and 100 knots (Category 3) as it began to turn northward on
the 17th. Flight-level data showed some evidence of a concentric
pair of eyewall wind maxima. Reconnaissance data indicated a
marked weakening later that day. The central pressure rose 20 mb
in about 10 hours and the peak flight-level winds decreased from
121 to 89 knots.
The primary (inner) eyewall disintegrated into a
few fragments. The weakening was likely caused by some combination
of shearing within the system reported by the flight crew, the
impact of nearby waters upwelled not long before by Hurricane Luis
that were 1 to 3C cooler than normal, and the decaying phase of an eyewall cycle.
Marilyn began accelerating toward the north-northeast late on
the 18th and its center passed about 150 n mi to the west of
Bermuda a day later. It had made a brief resurgence, with an eye
reappearing in satellite pictures. However, upper-level westerly
winds then began to shear Marilyn and the low-level cloud center
became partially exposed. Marilyn ceased generating deep
convection late on the 21st and became
on the 22nd. The remnant circulation meandered over the central tropical
Atlantic Ocean for another 10 days before becoming absorbed in a frontal system.
b. Meteorological Statistics
The "best track"
(i.e., post-operational) intensities were obtained from
the data presented in Figs. 2 and 3 (90K GIF),
and in Table 2. Those figures show Marilyn's estimated
central pressure and maximum one-minute wind speed, respectively, versus time.
made numerous flights through Marilyn. Position and intensity estimates
from satellite pictures were provided by SAB, the NHC's
Tropical Analysis and Forecast Branch
(formerly TSAF, as in the figures), and the
Air Force Global Weather Central (AFGWC).
These data were supplemented by surface observations from some
islands and ships and, to the southwest of Bermuda, by data
obtained from a network of drifting NOAA buoys
deployed ahead of Marilyn by the
Surface meteorological data during Marilyn's passage over
Dominica are not available at the NHC.
Over Martinique and Guadeloupe, the maximum wind speed
(presumably, sustained over the WMO-standard of 10-minutes) was 51
knots with gusts between 70 and 75 knots. Guadeloupe had
exceptionally heavy rain, with one station, Saint-Claude, recording
20.00 inches in a 12-hour period. The maximum rainfall reported
from Martinique was about 9 inches.
Part of Marilyn's eye passed over St. Croix. However, owing
to the northwestward motion of the hurricane, Marilyn's strongest
winds were located in the eastern or northeastern eyewall which
passed just offshore. The highest one-minute wind speed (estimated
for open exposure at 10 meters elevation) at St. Croix was likely
a little less than the 85 knots shown in the best track.
On the other hand, St. Thomas was hit by the hurricane's
eastern and northeastern eyewall. In addition, the hurricane
strengthened as it approached and passed St. Thomas. An
uncommissioned FAA Automated Surface Observing System (ASOS) at the
St. Thomas King Airport provided the only continuous "official"
wind record of the event in the U.S. Virgin Islands. Its maximum
two-minute wind was 90 knots at 0352 and again at 0353 UTC
on the 16th. (Around then, the peak 10-second winds in the hurricane at
the 700 mb flight-level were 105 to 110 knots.) The ASOS measured
a gust to 112 knots at 0408 UTC. Based on the ASOS data,
the estimated maximum one-minute wind speed (for open exposure at 10
meters elevation) at that time is 95 knots.
This is 5 knots higher than was estimated operationally. It is likely that
somewhat stronger one-minute winds (perhaps, to Category 3) and gusts above
112 knots occurred on exposed hills. Some unofficial high wind
speed observations remain unconfirmed or have been rejected.
The ASOS measured a minimum pressure of 956.7 mb. This
occurred at 0422 UTC when the airport was still experiencing 60
knot one-minute winds, apparently on the inside edge of the
eyewall. The estimated minimum pressure for Marilyn at that time
is 952 mb. This is lower than implied by the data obtained from
the Hurricane Hunters. They reported extrapolated and dropsonde
pressures of 957 and 960 mb, respectively, at 0305 UTC, and 954 and
958 mb for those techniques at 0600 UTC. This is reminiscent of
Hurricane Andrew's landfall over Florida, where the minimum
pressure obtained from surface observations was lower than analyzed
using aircraft data. The reason for this discrepancy in Marilyn is not obvious.
The storm surge
in the U.S. Virgin islands reached 6 to 7 feet, with an isolated
storm tide of 11.7 feet reported on St. Croix.
Rainfall totals reached about 10 inches in St. Croix and St. Thomas.
An unofficial gust to 109 knots was reported
from the island of Culebra.
The center of Marilyn passed far enough to the east of Puerto
Rico that hurricane conditions were apparently not experienced on
that island. The Naval Base at Roosevelt Roads had maximum one-
minute winds of 36 knots with gusts to 50 knots.
Bermuda experienced sustained winds of 39 knots with a
gust to 52 knots during the passage of Marilyn's outer circulation.
c. Casualty and Damage Statistics
Marilyn was directly responsible for 8 deaths, 5 in St.
Thomas, 1 in St. John, 1 in St. Croix and 1 in Culebra (Puerto
Rico). Most drowned and were on boats at docks or offshore.
Marilyn caused severe damage to the U.S. Virgin Islands, in
particular to St. Thomas. An estimated 80 percent of the homes and
businesses on St. Thomas were destroyed and at least 10,000 people
were left homeless. Some of the damage was reportedly attributable
to lax construction standards and practices. According to
percent of the homes on St. John were destroyed and 60 percent were
roofless. About 20 to 30 percent of homes in St. Croix received
damage. Trees fell and hotel windows broke there. Hillsides were
littered with sheets of metal roofing, wooden planks and household
debris. On Culebra, 250 homes were destroyed or severely damaged
and light planes were overturned.
Large waves crashed over the harbor at Dewey, Culebra,
flooding streets. Flash floods occurred over northern and eastern
Puerto Rico where the La Plata and Manati rivers overflowed.
The American Insurance Services Group estimated insured losses
for the U. S. Virgin Islands and Puerto Rico at $875 million.
Because the overall loss is often estimated to be up to about
double the insured loss, the total U.S. loss is tentatively
estimated at $1.5 billion. The U. S. Virgin Islands Bureau of
Economic Research estimated the economic loss at $3 billion.
placed the cost for their programs at $1 billion in the Virgin
Islands and $50 million in Puerto Rico. The FEMA
totals include losses not traditionally described by the NHC as "damage", such as
FEMA's cost to set up field offices,
inspector's salaries, disaster unemployment compensation, and crisis counseling.
According to The New York Times, the British Virgin Islands
were not seriously affected and some (unquantified) damage occurred
in Antigua. According to the Antigua Meteorological Service, that
island had extensive flooding in low-lying areas, destruction of
banana trees and, otherwise, minimal wind damage.
About 12,000 people went to shelters in Puerto Rico. In the
U.S. Virgin Islands, 2,243 people were sheltered.
d. Forecast and Warning Critique
Media reports in the U.S. Virgin Islands were critical of the
NHC's forecasts and warnings. Some complaints seem almost
unavoidable after U.S. landfall events, even when the advisories
are exemplary. In Marilyn, there was a perception that the
intensity of the hurricane was underestimated. In part, this is
due to the public's unfamiliarity with high wind speeds.
Regardless of the category of hurricane, they express surprise at
the damage and are adamant that winds they experienced were
stronger than indicated by the NHC. In contrast, when there is
disagreement between the NHC and others in the scientific and
engineering communities, analysts within those disciplines suggest
that the wind speeds estimated by the NHC are too high (by about10%).
In the case of Marilyn, the NHC issued:
with about 33 hours lead time for St. Croix
and 40.5 hours for St. Thomas (Table 3
and Table 4).
with about 24 hours lead time for St. Croix and 31.5 hours for St. Thomas.
Track forecasts whose errors were, on average, about
two-thirds of the usual magnitude. For the period near the northeastern
Caribbean, the forecasts were even better. The 16 track forecasts
from 0600 UTC on the 14th through 0000 UTC on the 18th were, on
average, in error by about one-third of the usual magnitude
In fact, the 36-hour forecast made when the watch was issued
(around 1200 UTC on 14th) was off by 46 n mi (about 150 n mi is the
long-term average). The 24-hour forecast made when the warning was
issued (2100 UTC on 14th) was off by 8 n mi (100 n mi is the long-
term average). The corresponding intensity forecasts were about 15
knots too low (equivalent to about 1 category on the
Those intensity errors are close to normal magnitude.
Beginning at 1500 UTC on the 14th, corresponding to 13.5 hours
prior to Marilyn's center passing by St. Thomas, NHC public
advisories carried the headline, "Marilyn approaching the U.S.
Virgin Islands as a Category 2 hurricane on the Saffir-Simpson
Scale." The advisory 6 hours later also noted "...and Marilyn
could intensify from a Category 2 to Category 3 hurricane on the
Saffir-Simpson Scale tonight or Saturday." The
corresponding tropical cyclone discussion stated
"The strongest winds are in the N and NE part and could spread over the St. Thomas area."
Watches and warnings were smoothly coordinated with most areas
of the Caribbean, the Bahamas, and Bermuda. An exception was the
series of issuances by French officials which sometimes differed
from the coordinated advice disseminated by neighboring Caribbean islands.
Although Marilyn passed well offshore from the U.S. mainland,
some NWS offices along the U.S. east coast issued heavy surf
advisories for swells emanating from the hurricane.
Some of the data in this report were made available by Rafael Mojica of the
NWS San Juan
office and by the meteorological services of Martinique, Guadeloupe, Antigua
and the Netherlands Antilles.
Tropical Cyclone watch and warning summary, Hurricane Marilyn
||Tropical Storm Warning issued||Barbados|
|Tropical Storm Watch issued
||St. Vincent, Grenadines, St. Lucia, Grenada|
|13/0300||Tropical Storm Watch issued
||Tobago and Trinidad|
|13/0900||Tropical Storm Warning issued
||St. Lucia, St. Vincent, Grenadines, Grenada, Tobago|
||Hurricane Warning issued
||Barbados, St. Vincent, Grenadines, St. Lucia|
|Hurricane Watch issued||Dominica|
||Hurricane Warning issued
|Tropical Storm Watch discontinued||Trinidad|
|14/ N/A||Hurricane Watch issued
||Hurricane Warning extended
||Grenadines through St. Martin, except Guadeloupe, St. Barthelemy and French portion
of St. Martin|
|Hurricane Watch issued||British and U.S. Virgin Islands|
|Hurricane Warning discontinued||Barbados|
|Tropical Storm Warning discontinued
||Grenada and Tobago|
|14/ N/A||Tropical Storm Warning issued
|14/1500||Hurricane Watch issued
|Hurricane Warning discontinued||Grenadines|
|14/1700||Hurricane Watch issued
||Guadeloupe, St. Barthelemy, and French portion of St. Martin|
|14/2100||Hurricane Warning issued
||Puerto Rico, U.S. and British Virgin Islands, and Guadeloupe|
|Hurricane Warning discontinued
||St. Vincent and St. Lucia|
|14/????||Hurricane Warning discontinued
|15/0300||Hurricane Warning discontinued
|15/1000||Tropical Storm Warning replacing
||St. Barthelemy and French portion of St. Martin|
|15/1200||Hurricane Warning discontinued
||Guadeloupe, Antigua, Barbuda, and Montserrat|
|15/1500||Hurricane Watch issued
||Dominican Republic from Cabrera to Cabo Engano|
|Hurricane Warnings discontinued
||Guadeloupe, Nevis, and St. Kitts|
|15/2200||Tropical Storm Warning discontinued
||St. Barthelemy and French portion of St. Martin|
|16/0000||Hurricane Warning discontinued
||St. Martin and Anguilla southward|
|16/1500||Tropical Storm Warning replacing
||Puerto Rico and U.S. and British Virgin Islands|
||All Hurricane and Tropical Storm Warnings and Watches discontinued
|16-17/ N/A||Hurricane Watch issued
||Turks and Caicos and Mayaguana, Acklins, Crooked Islands of the southeastern Bahamas|
|17/2100||All Hurricane Watches discontinued
|18/1500||Tropical Storm Watch issued
|18/2100||Tropical Storm Warning issued
||Tropical Storm Warning discontinued
Lead time refers to time lapsed from issuance to closest approach of center.