Comments of and replies to the
National Hurricane Center Best-Track Change Committee
Minutes of the best-track change committee meeting, 10 June 2004.
Members Jack Beven, Jim Gross, Brian Jarvinen, Richard Pasch, and chair
(Colin McAdie) present.
< Reply to comments ... CWL - August 2004 >
The committee met to consider the revisions submitted by Chris Landsea et
al. for the period 1911-1914
Reviewing previous actions, it was noted that the final set of changes for
the pre-1911 period had been accepted by the committee.
It was suggested that future research would be made easier if the Partagas
and Diaz publications were made available electronically. They are not
generally available. This series of publications should already exist in
word processor format, which could facilitate the conversion. Richard
will contact Diaz in this regard.
< These are available on-line through efforts of the reanalysis project:
It was also noted that in reviewing the current submission, much emphasis
has been placed on the Historical Weather Map series and COADS. There
appears to be much less mention of newspaper sources than previously.
These should not be neglected.
< The main reason that Partagas utilized newspaper sources (primarily
the New York Times, London Times and Gazetta de Habana) was to
locate ship observations, as these were regularly published in these
papers. Another motivation for using newspaper accounts was to
discover potentially new tropical systems and to further document
existing tropical storms and hurricanes. However, our efforts in
using COADS to provide ship reports preclude the use of newspapers
for the first purpose (plus such reporting in the newspapers was
much less common in the 20th Century). Additionally, communications
were much more rapid in the 20th Century with telephone service and
radio making it much less likely than any completely new tropical
system will be discovered through newspaper searches by this
method. It is noted that the Monthly Weather Review beginning in
the 1910s and 1920s also became much more thorough in it descriptions
and reporting of significant weather events in the Caribbean islands
and the open Atlantic Ocean.
That said, there still is a role for newspaper searches in systems
where there is ambiguous meteorological data to make a confident
determination of the track and/or intensity of a tropical system.
The case of storm 5 in 1913 is an example of a case where additional
local information is being collected to better ascertain the
intensity (ie. was this system a tropical storm or Category 1 hurricane
at landfall). Thus for select cases, additional efforts will be made
to obtain impact information (wind-caused damages, storm tide
values, etc.) to clarify the track and/or intensity. For the years of
1911 through 1914, this was employed for 1911/03, 1913/05 and 1914/01.
But conducting newspaper searches for all systems is beyond the scope
of this project. >
For convenience, the following is organized by year and new storm number.
1911 #1 This additional storm is not accepted for inclusion in the database.
This occurrence is very early climatologically, and seems frontal in nature.
This storm would probably not be considered tropical using current
< Agreed. This system has been removed from consideration for HURDAT,
but retained in the Additional Notes section of 1911. )
1911 #2 Accepted as revised. Semantic note: to our knowledge, oceanfall is
not a word.
< System is hereafter referred to as 1911 #1 with removal of first storm
proposed. "Oceanfall" has been changed to "reaching the ocean". >
1911 #3 Accepted as revised. Change early to earlier (first sentence, last
paragraph). Observations from Mobile were reviewed and found to support a
track slightly to the south of Mobile, as submitted. The one-foot storm
tide is difficult to interpret and seems questionable. This is a case
where local newspaper records could be extremely useful.
< System is hereafter referred to as 1911 #2 with removal of first storm
proposed. Wording changed as suggested. Given the trajectory of the
hurricane toward the west-northwest with landfall about 6 hours before
its closest approach to Mobile, there would be quite limited time for a
surge to be generated into Mobile Bay with southerly winds. Thus the
small storm tide reported in the city of Mobile may be understandable.
It is quite likely that a substantially higher storm tide was produced
at the sparsely populated Florida-Alabama border.
Newspaper accounts from the _New York Times_ and _Miami Herald_ found
at the Florida International University library were reviewed of this
system. The _New York Times_ on the 13th of August, 1911 had headlines
"Pensacola Storm Loss Big; No Lives Lost in City, but Much Damage to
Property" followed by an article on the storm's impact in Florida, but
no description on this (or other day) on what occurred in Alabama.
The _Miami Herald_ on the 13th of August had a similar headline, but
no accompanying article. >
1911 #4 Accepted as revised.
< System is hereafter referred to as 1911 #3 with removal of first storm
1911 #5 Grounds for revision are somewhat sketchy, revision is not accepted.
Observations from Barbados are extremely important for this revision. In
this regard please clarify interpretation of plotting convention used in ship
observation of 3 Sept in this vicinity. The arrow may indicate two
observations from the same ship (implying ship movement). Was this the basis
for shifting the track southward?
< System is hereafter referred to as 1911 #4 with removal of first storm
proposed. The plotting convention of the Historical Weather Map
analysts was to write a ship observation that was very close in space
to another one off to the side with an arrow indicating the correct
position. So on the 3rd of September, there were two separate ships
both at approximately 11.5N, 57.5W - one reporting WNW winds at 20 mph,
with the other one reporting 25 mph winds out of the north and a
pressure of 1021 mb. While the pressure seems suspectly too high,
we have used these observations to analyze this system as a weak
tropical cyclone (tropical depression ~30 kt) with a small circulation
and a position east and substantially farther south than originally
indicated. Even if one discards the N wind report completely, a
position farther south and east than that originally in HURDAT is
reasonable. Very little observational data was available for several
days after the 3rd near the center of the system, so a combination of
persistence of the southerly position and climatology were used
to adjust the track. Retaining a track similar to HURDAT would only be
obtained if both of these observations on the 3rd are decided to be
erroneous. It is our suggestion to use these available observations
and keep the system as currently revised. >
1911 #6 Additional system accepted. Is there any evidence to support a few
earlier positions (12-14 September)?
< System is hereafter referred to as 1911 #5 with removal of first storm
proposed. Earlier data were inconclusive in identifying a closed
circulation that is consistent with the rather well-defined vortex on
the 15th. While the system likely was a tropical depression/tropical
storm on the 14th (and maybe earlier), we are unable to extend the track
back further in time. >
1911 #7 This revision needs more work. There are several issues. 1) Where
did the system form? The earliest evidence seems to be as a wave on the
29th. 2) Data is scarce in the northwest Caribbean during the 27-30th. In
spite of this, the track is pushed quite far over towards Cozumel. Is there
any evidence from Cuba that would support the existing track? 3) In the
final stage it seems likely that the storm would turn towards the northeast
as it was absorbed by the front. Remnants could have swept across Florida.
Were Florida sources checked for evidence?
< System is hereafter referred to as 1911 #6 with removal of first storm
proposed. The Cuban data available is primarily from Cienfuegos and
Havana for this time period. These suggest a wave passage to the
south of Cuba during the 24th to the 27th. It is agreed that the
genesis of this tropical storm most likely occurred much latter than
the 23rd. We have revised this now to begin late on the 28th in the
northwestern Caribbean. For the decay of the system, in response to
this request, we did obtain the Original Monthly Records for Tampa,
Pensacola and Jacksonville. While no gales (or equivalent in pressure)
were obtained, the data did reveal that the system did turn to the
northeast ahead of the front. In particular, Tampa's observations
showed a minimum pressure (for the whole month) of 1009 mb at 17 UTC
on the 31st with S winds at 9 kt. At 00 UTC on the 1st, Pensacola
and Jacksonville reported NE winds around 10 kt and Tampa had SW
winds around 10 kt with pressure up to 1011 mb in Tampa. By 12 UTC
on the 1st, all three stations reported N winds, rising pressure and
substantially cooler temperatures than the day before. >
Committee concurs with decision to not add system appearing in 1911
1912 #1 It is noted that no central pressures are given during the overland
phase. Could not these be estimated from available surface observations?
The system was likely extratropical on the 14th. MWR states that a system
passed over Florida at this time. Please rule out any connection.
< The system made closest passes to the Weather Bureau's stations of
Montgomery, AL; Atlanta, GA; Charlotte, NC; and Cape Hatteras, NC.
Unfortunately, none of these provided any non-00 or 12 UTC pressure
measurements. The central pressure would have been lower than 1002 mb
on 00 UTC on the 14th as Montgomery recorded 1002 mb and 29 kt SW
at that time. However, the 1005 mb measured at Cape Hatteras at
00 UTC on the 15th may indeed have been a central pressure as the
winds at that time were nearly calm (4 kt) and were preceeded by
SW winds that peaked at 40 kt and followed by N winds that peaked
at 31 kt. This is now added to HURDAT.
Regarding its extratropical versus tropical characterization, it is
agreed that it went extatropical for a time over the SE United States.
However, it appears from the data available that it did once again
regain tropical cyclone features on the 16th and 17th as warm,
isothermal air masses were present around and in the system. Thus
the system is declared extratropical around 06 UTC on the 14th and
then regaining tropical status around 00 UTC on the 16th.
A review of the Historical Weather Maps for this time period does
not indicate any significant closed lows impacting Florida during
this time period. It is not clear what the Monthly Weather Review
was referring to. However, it can be ruled out that what was
mentioned had any direct connection to storm #1. >
1912 #2 Accepted as revised.
1912 #3 Additional system accepted.
1912 #4 There may be a systematic difference between the observations at
Mobile and Pensacola. If so, the large RMW would not be necessary.
Given the closeness in time, it is possible that additional system #3 was the
origin of #4. This possibility should be examined further. Otherwise accepted.
< It is agreed that there may be unknown biases in the wind records
(e.g., poor exposure, funnelling of winds) that makes direct evidence
of RMW problematic. However, due to impacts that were about as
damaging by surge (and to a lesser extent by wind) in both Mobile
and Pensacola, an RMW between the two location appears reasonable.
But to accomodate the uncertainty, a range of 40-60 nmi for RMW for
this system at landfall is now indicated.
Beginning on the 6th of September as storm #3 was inland over Georgia,
a trough of low pressure formed in the eastern Gulf of Mexico. This
trough appears to be a separate entity to storm #3 or its remnants.
The trough remained in the eastern Gulf of Mexico from the 6th through
the 9th, though it did not seem possible to close off a well-defined
center of circulation. By the 10th, this system had developed
enough to characterize it as a tropical depression, one day earlier
than originally shown in HURDAT. >
1912 #5 Accepted as revised.
1912 #6 The reported 6 ft storm surge at Port Isabel would support a
category 3 landfall. Intensity at landfall is not resolved at this time.
Are there any Mexican observations to support the intensity over the Yucatan
< It is to be noted that the 6' storm TIDE (not surge) value comes from
an unofficial source (Ellis) and thus the reliability may be somewhat
suspect. Additionally, wave impacts may also have been counted as part
of this storm tide observations. However, it may be prudent to
upgrade this to a Category 2 at landfall, which would be more
consistent with the storm surge submerging Padre and Brazos Islands.
The system is estimated then to retain about 85 kt (~973 mb) from
late on the 15th (as originally listed in HURDAT) until landfall
around 18 UTC on the 16th. Unfortunately, due to the Mexican
Revolution during the 1910s, very few observations were available
for Mexico. >
1912 #7 It is not clear what the discussion from Trinidad has to do with this
storm. Is this evidence for some other system? The upward revision in
intensity during the initial period (13-14th) is too great. Concerning the
correction for instrument bias, the U.S. was using the 4-cup anemometer
during this period, but was this also true of Jamaica? Please note the
second account of this storm appearing in MWR pp 1756-1757.
(discussion continued during the next meeting - 18 June 2004)
< It was investigated to see whnther either the HWM closed circulation
and/or the squall in Trinidad had any direct association with storm #7.
However, surface observations from HWM, COADS and newspaper accounts
show no closed circulation existed from the 6th through the 9th, though
the data is sparse away from the Lesser Antilles.
Agree to lower the spin up some, but the 992 mb pressure ship on the
13th would suggest minimal hurricane conditions were reached by at
least late in the day.
The Caribbean countries also utilized the same 4-cup Robinson
anemometers during this time frame (up until the early 1930s) and
so they have the same high bias that the U.S. anemometers
We made use of the original account of this storm that appeared
in Hall (1913), which was then abridged for use in Monthly Weather
Review pages 1756-1757. >
Minutes of the best-track change committee meeting, 17 June 2004.
Members Jack Beven, Jim Gross, Ed Rappaport, and chair (Colin McAdie) present.
The committee reconvened to consider the revisions submitted by Chris Landsea
et al. for the period 1911-1914.
It was suggested that when non-developing depressions are found in the
routine course of this work, it might be advisable to place them in a
parallel track file.
< We are providing once daily estimates of non-developing tropical
depressions in the Additional Notes section for each year.
However, to develop a standard 6 hourly HURDAT format database
for these systems is beyond the scope of this project. >
The phrase "at most" is used frequently in reference, we believe, to the
innermost closed isobar on the HWM analysis. This terminology is
somewhat confusing. Does this mean that the pressure was no higher than,
or no lower than, the specicied value? For convenience, the following is
organized by year and new storm number.
< To clarify, "at most" in reference to the pressure analyzed on the
Historical Weather Map series can be the innermost closed isobar,
but it also in some cases is an estimated central pressure listed
for the system. It does mean that the actual central pressure was no
higher than, but could be lower than the specified value. (In much the
same way, one could say that a system had "at least" 35 kt of wind,
meaning that the storm had no lower than, but possibly higher than,
35 kt.) >
1912 #7 (resumption of discussion from previous meeting) The issue of the
track of the system north of Cuba remained unresolved. Although heavy
emphasis is placed on the HWM, other analyses are possible from these data.
In particular, a baroclinic circulation could be analyzed off Cape Canaveral
on the 22nd. It is also possible that a weakened system passed over Cuba.
Also note that the metadata states that observations from Cuba (Perez 2000)
indicate that no tropical storm or hurricane made landfall in Cuba. However,
Perez does list this system in his Table 3.5 (El Huracan de Jamaica) as
< The baroclinic cyclone east of NE Florida was described in the
metadata on the 22nd of November. In the concluding paragraph in
the metadata, it is clarified that "it is to be noted that on the
22nd the HWM had the baroclinic low well to the southeast of its
It is now clarified that the MWR/HWM/HURDAT tracks cannot be correct
as Perez' (2000) observations show that no system made landfall in
CENTRAL Cuba. Perez did suggest that instead the hurricane impacted
Guantanamo as a Category 1 system. But this was analyzed without
any supporting data to somehow reconcile having a strong hurricane
in the Caribbean and a hurricane later off of the U.S. Atlantic
seaboard. Further contacts with Perez confirmed that no observations
were taken in eastern Cuba that actually showed (or even implied)
hurricane impacts in that part of the country (or elsewhere). >
1912 Additional systems
Additional notes #3. Note that MWR has the system peaking over water and
weakening over land, which is not usual for a baroclinic system. A late
September occurrence of this sort is also somewhat against climatology.
Josephine (1996) would be a possible analog.
< This estimated central pressure variation is now noted in the
Additional Notes writeup of this suspect. Howver, re-examination of
it reconfirms the clearly baroclinic nature of the system.
Temperatures in the 40s in Texas on the 22nd and 23rd are indicative
of the rather unseasonal cold air outbreak that occurred in
conjunction with this system. >
1913 #1 Regarding the sharp reduction in intensity during the initial phase
of the storm, the available observations are quite distant from the center,
and do not appear to rule out the existing intensity. It is difficult to
justify this much reduction. How were the wind estimates made (i.e. why
not 45-50 kt, rather than 35-40)? Revised position on the 23rd does not
seem consistent with the observation at San Andres. It is noted that the
metadata contains a mix of units (kt and mph). Are all of the conversions
correct? It is stated that the 12.7 ft storm surge is considered
erroneous. Is the 100 mph wind also erroneous?
< Agreed that the intensity was too drastically reduced from the 22nd
until the 25th. These are now brought up to the level suggested.
The observations near 11N, 80.5W is not the island San Andres, but
is instead a ship that does not appear to be reliable in either its
pressure or winds.
The "100 mph" was a direct quote, so it was kept as is (but conversion
to kt added). The Ellis reference is an unofficial history of
Texas hurricanes that occassionally does provide useful information
not contained in standard sources. However, these storm surge and wind
wind values are erroneous as can be shown from the Weather Bureau
Galveston station Original Monthly Records: "27th - The tide was
about 1.5 feet above normal most of the day...Maximum velocity 39 mph
[34 kt]". >
1913 #2 Additional system accepted.
1913 #3 The segments added to the beginning and end of the track appear
to be justified; however, there are some questions about the intensity, and
the revised intensity of the existing track. In the initial portion, it is
unusual that a system on this track at this time of year would stay a
constant intensity of 40 kt for 9 days. Observations available would
support a 45 kt intensity during the westward portion of the track.
During the existing portion of the track during which peak intensity
occurred, the committee is not convinced that the system should be reduced
this drastically (from a hurricane to a weak tropical storm) based on the
distribution of the available ship data and uncertainties in the vortex
position. Please comment on what became of the pre-existing extratropical
low. It should be noted that the HWM analysis is not consistent during the
latter stages of this system, and does not attempt to depict two systems.
Is there any support, or other reference, for the final westward movement
shown in the existing track?
< It is agreed that having a constant intensity for nine days is not very
realistic. However, the system traversed the very data sparse region
of the tropical North Atlantic. Winds are increased to 45 kt on the
westward portion of the track as requested. It is also agreed that
the winds were reduced too drastically to a weak tropical storm.
The winds are now increased back to a peak at the boundary between a
tropical storm and hurricane at 60 kt on the 8th through the 10th.
The interaction with the non-tropical low is now better described
for the dates of the 10th through the 12th. The tropical cyclone
stayed separate from this low and dissipated to its northeast on
the 12th. Given the close proximity, it is understandable that
HWM did not depict two separate systems on the 11th and 12th and
that HURDAT originally mistook the non-tropical low for the tropical
cyclone on the 12th. >
1913 #4 The 10 ft storm surge and 92 fastest mile wind suggest that the
winds may have been stronger than 75 kt at the time of landfall. Is the
Ho RMW accepted at face value? It appears to have been a factor in the
reduction of the wind estimate at landfall. Otherwise accepted.
< While we utilized the 5 min "maximum velocity" to get (after
bias-removal and conversion to 1 min) 64 kt, one could use the
92 mph (80 kt) 1 min "extreme velocity" to obtain 61 kt after bias-
removal. (In general, 5 min winds were the most reliable to use
in that era of the 4-cup Robinson anemometer as shorter observation
intervals were not well monitored by that instrument.) Ho's estimate
of the RMW came from an average of the pressure profile in Raleigh
as well as the wind observations in Cape Hatteras. This somewhat
larger than usual value was the main reason for reducing the subtropical
pressure-wind relationship of 80 kt for 976 mb central pressure down
to 75 kt. >
1913 #5 Multiple 50 kt ship observations (e.g. 10/4) would seem to support
an intensity of greater than 50 kt. In the same sense, the WSW 60 kt ship
observation would support a 65 kt hurricane. Is it possible that the RMW
passed over Georgetown? Please revise to reflect category 1 status at the
time of landfall.
< Agreed to boost the intensity late on the 3rd to the 5th to 55 kt.
The landfall intensity is somewhat problematic. While the 60 kt
W ship with 992 mb pressure would suggest minimal hurricane
conditions at landfall, other evidence indicates tropical storm
impact instead. We asked Prof. Cary Mock to provide excerpts of
newspaper accounts in South Carolina at the time. (Relevant portions
of these are now included in the metadata file. These accounts
show that no signficant structural damage was reported at all in
Charleston, Georgetown or surrounding communities. It would seem
easier to discount the ship observation as being somewhat too high
in its subjective Beaufort scale estimation of Force 10 (~60 kt),
rather than not taking into account the lack of significant
wind-caused damages at the coast. Thus we would argue to keep
the designation of a 60 kt tropical storm, rather than a hurricane
at landfall in South Carolina. >
1913 Additions to the additional systems.
Sept 12-15. A system is mentioned in the MWR near the mouth of the Rio
Grande. Storm warnings were issued. Was it tropical?
< A review of the Historical Weather Maps for September 12-15, 1913
shows a vigorous cold front pushing through Texas on the 12th and
13th, cyclogenesis forming along the front late on the 13th and
14th off of Texas, then the low moving northward through Texas
and Louisiana on the 15th. The system clearly was of extratropical
nature throughout its lifetime. This is now added in the
Additional Notes section of 1913. )
Sept 24-28. A stationary system is mentioned in the MWR off the Texas coast.
Storm warnings were also issued for this system. Was it tropical?
< A review of the Historical Weather Maps for September 24-28, 1913
reveals a strong cold front moving through Texas on the 24th and 25th,
cyclogenesis just off the Texas coast late on the 25th and 26th along
the front, and the low weakening into an open trough on the 27th and
28th near the Louisiana/Texas border. While the HWM does analyze a
small closed low ahead of the cold front on the morning of the 25th,
available observations do not confirm that the system had a
closed circulation - though it may have been a tropical depression
briefly before the front arrived. Thus despite the heavy rains that
accompanied the front/low (over 8" in Brownsville), the system was
baroclinic for the duration that it retained a closed circulation.
This is now added in the Additional Notes section of 1913. >
1914 #1 Note spelling: Florida Straits.
Given three 35 kt ships (9/15) the intensity should be higher. Suggest
40 kt. The 60 kt ship (9/17) is quite far from the center position. Please
reconcile with 60 kt intensity. Is the ship representative? Also,
committee notes that newspaper reports could be extremely useful in gauging
the intensity at landfall.
< Spelling corrected. Agree that intensity on the 15th should be
boosted to 40 kt. It should be noted, as can be seen from the
provided spreadsheets of all available data, that ship locations
are often apparently given in 1 degree latitude-longitude
increments. In the case of the 60 kt ship report, it was
located at 32.5N, 78.5W at 05 UTC (also listed there at 01 UTC).
Its 09 UTC position four hours later was given as 31.5N, 79.5W.
Given these rather coarse estimates of position, immediately
there is an error inherent of about 0.5 degree latitude-longitude.
This assumes that the estimate of position does not have additional
errors introduced by an incorrect recconing of the ship's
position by the navigator/captain. Occassionally, there are also
errors found in the time and/or date of ships in the COADS
database. So for this case, it is somewhat perplexing about the
distance of the ship to the storm's center and the winds reported.
It may be that the ship's location is in error, either to a minor
degree or worse. However, given the relative wealth of other
ship and coastal station data, it does not appear that we should
move the track of the storm closer to this single observation.
It is also possible that the winds reported from this COADS report
were too high given the somewhat modest pressure (1004 mb) given at
the same time. However, this higher wind report does match
statements found in _Monthly Weather Review_, regarding "vessel
reports indicate that it was even more severe off the Georgia
Finally, a search of the _New York Times_ and _Miami Herald_ was
conducted for additional clues about the intensity of this system.
The _New York Times_ reported in the "Weather" section on Sept. 18th
"The southern storm passed inland during Wednesday [16th] night and
Thursday [17th] night its center was over Alabama. This disturbance
has diminished greatly in intensity but during the last twenty-four
hours it caused general showers in the South Atlantic and East Gulf
States, and during Wednesday night it caused strong shifting winds
on the South Atlantic Coast". The _Miami Herald_ reported also on
the 18th: "The northeast of Wednesday [16th] raised some water
around St. Augustine, causing the tide to come in so high that it
ran over the South Street Causeway, and tons of dead grass were
washed away from the marshes about the city. No damage was done as
the boatmen had plenty of warning of the blow." No reports in either
the _New York Times_ or _Miami Herald" discuss that any substantial
wind caused damage occurred, though the storm tide does suggest
a stronger system than originally indicated. Thus a boost for this
system to a high end tropical storm (60 kt) at landfall from the
original minimal tropical storm (40 kt) looks reasonable. >
1914 Additional systems.
It should be noted in the metadata that additional system #2 appears both in
Connor and Tannehill. Given this historical reference, are we completely
confident in not including this system? Are there any Mexican observations?
< It is now noted in the metadata that Connor and Tannehill considered
this system as a tropical storm. However, after further re-review
of this complex system, it appears to be rather clear that the
main low was baroclinic throughout its lifetime and that the
short-lived system in the northwestern Caribbean did not obtain
tropical storm intensity. Naturally, this is all dependent upon
observations available and that the short-lived system could have
achieved gale force winds. Unfortunately, Mexican observations
are rather sparse throughout the 1910s because of the ongoing
Mexican Revolution. Without substantiating observations, we can
not offer this system up as a new entry to HURDAT.
Note: We are as surprised as anyone that no additional new
systems could be added to HURDAT for 1914. Of all the years that
one would suspect that there were tropical storms or hurricanes
that occurred (of course, HURDAT currently lists this season as
the only one with a single tropical storm), but were simply not
monitored. Despite our searches, no suitable candidates
arose. We still feel that out in the open Atlantic there were
likely a storm or two additional, but that these were not
monitored by the rather sparse shipping traffic. >
Return to Overview