Comments of and replies to the
National Hurricane Center Best-Track Change Committee
June 2000
< Minutes of the best-track change committee meeting, 24 February, 2000.
Members Jack Beven, Jim Gross, Richard Pasch, Ed Rappaport, and chair
(Colin McAdie) present. >
< At the close of year 1, the committee would first like to
acknowledge the work that Chris, Craig Anderson, Gil Clark, Mark Zimmer,
and Charlie Neumann have done on this project. We recognize the
magnitude of the task, and congratulate them on the results. We
also appreciate all of the work that has gone into responding to these
comments. >
< The committee met to reconsider the all three revised installments of
track changes for the period 1851-1885, received from Chris Landsea et
al. The emphasis was placed on the final installment (1871-1885). >
< The following points were discussed during the meeting: >
< 1) It was noted that the commentary (the future 'hypertext' or metadata
document) was much improved and much more 'user-friendly.'
There was discussion about the benefits of linking the whole database
together using web-based technology. >
Agreed that it is ideal to utilize the Web to provide access to the
datasets being revised and expanded. A preliminary version of a Web
based display system was presented to the NHC Committee on 23 March, 2000.
< 2) The responses to last set of minutes (10 August, 1999) were then
reviewed. >
< On item 4, concerning use of the inland wind model, thank you for
clarification on the use of the model. In cases affected by
topography, the consensus was that if a decay rate other than the
standard one was used, that it be specified in the comments for that
storm, along with the reference to the tech memo. For example, the
comments for 1873/05, should include a brief statement, such as
"accelerated decay rate used to account for topography, see tech
memo". The accelerated rate of decay used should be quantified
somewhere for each storm. This will easily distinguish these cases
from those using the standard model. At your discretion, a
table might be included in the tech memo of all non-standard cases.
The motivation is to leave the clearest possible record of
changes. >
Agreed. The storms for which an accelerated rate of decay was
utilized are now specified in the metadata files. In the documentation
for all of the storms, we are incorporating this discussion: "The Kaplan
and DeMaria (1995) model was utilized to provide guidance as to the wind
speeds for the best track after landfall of a tropical cyclone - but only
in the absence of observed inland winds. The model from their paper starts
with a sustained wind at landfall and provides winds up to 48 hours after
landfall. Since it was derived for landfalling tropical cyclones over the
primarily topography-free southeast United States, decay of the winds over
higher terrain areas such as Hispanola and Mexico is chosen to occur at a
faster rate than that given from the model (on the order of 50% accelerated
rate of decay)."
< On item 5, concerning inclusion of a derived pressure (931 mb) in the
best-track file along with the wind, the committee recognizes the
convention, and the reasoning behind it, but the consensus is that the
pressure be included. This is because almost all of the pressures
in recent years are in fact either derived, or estimated. As long as the
pressure is clearly noted as derived in the comments, which it is, it
seems more straightforward to include it. This is not the same as
trying to obtain a pressure with little or no data just for the sake of
filling in a blank. In this regard, the point was raised that to
get the full benefit of the project, some conventions may need to be
reconsidered. Documentation seems to be the key. >
Agreed. We are now including into the HURDAT central pressures that
were derived from peripheral pressure measurements in addition to those
that were directly measured.
< On item 9, concerning inclusion of the depression stage, the committee
wishes to emphasize that we were not trying to unnecessarily increase the
scope of the project, or to place an undue burden on the
participants. The intent here is to include the depression stage
_only to the extent that it is already_ known.
We do not expect additional time to be spent in creating a depression
stage. This point became somewhat more important as the meeting
progressed. We noted in a number of examples during
1871-1885, that existing tracks (i.e. as specified in Partagas and Diaz)
had been truncated to exclude the depression stage. The committee
was unanimous in its opinion that these portions of the tracks be
included. This cannot be stated strongly enough. There is great
utility in future work on inland flooding, for example, that will be lost
otherwise. Again, we do not in any way expect the construction of a
depression phase if not known, but strongly believe in its inclusion,
_if known_. >
Agreed. We have now gone back into the years of 1871-1885 and added
back in a dissipating tropical depression stage for those tropical cyclones
that decayed over land. This primarily is the case for those storms making
landfall over the United States or Mexico. We have utilized the Kaplan
and DeMaria inland decay model to assist in providing windspeed estimates.
< On item 11, concerning the extratropical stage of 1874/03, it does not
seem that a wind shift from southeast to northwest necessarily implies a
transition to the extratropical phase. >
While a single wind shift from a ship itself does not necessarily imply a
transition to extratropical phase, however having extensive coverage of
ships all showing the same southeast to northwest wind shift indicate a
very asymmetric wind field. That in combination with the high
latitude (and cool waters) provide enough evidence that the hurricane was
undergoing extratropical transition around the latitudes of 41-43N.
< As to the additional question posed about the treatment of 'single-point'
storms, the committee opts for position 3) keep single-point storms as
currently configured. We recognize that some reprogramming will be
necessary. Our reasoning is as follows. It is possible, for
example, that a single-point storm was traveling at 10 knots, while an
additional artificial point would imply a speed of zero.
Rather than inventing data by some other means, we prefer the single data
point. The single-point entry emphasizes the fragmentary nature of
the data. >
Agreed. We will keep single-point storms as currently configured.
< The committee then continued to review the track maps provided for the
years 1871-1885. >
< 3) It was noted that 1871/01 provided a good example of the
truncation of the depression phase. Please restore. >
This storm, along with many others from 1871-85, have now had the
depression stage restored.
< 4) Substantial track changes were noted for 1872/05, 1875/01, and
1876/03. Further investigation was deemed necessary. In order
to proceed with the meeting, it was agreed that the work should be shared
equally among the members, assigned post-meeting, and to report back to
the full committee. (see below) >
< 5) 1877/02 Did the use of the inland wind model commence over
Louisiana or the Florida panhandle? >
The inland decay model usage was commenced after the final landfall
along the Florida panhandle.
< 6) 1879/04 (existing track book) was deleted. Partagas and
Diaz, using the same evidence as Landsea et al., chose _not_ to
delete. While some doubt may certainly be justified, and should be
noted, the committee feels that compelling evidence for deletion is
lacking. >
First, it is to be noted that for all of Partagas and Diaz re-analysis
work from 1851-1910, not once did they recommend removing a storm from
the historical database. Perhaps to be safe, they were being very
conservative to be sure that a real tropical storm would not be
discarded. However, this instance does appear to be a case where the
storm should be removed. We went back to the original September issue
of the _Monthly Weather Review_ (available on microfilm at the
National Hurricane Center). No substantial rain occurred at the
following Florida stations available at the time that the tropical storm
supposedly went across the state: Daytona, Gulf Hammock, Okahumpka, Mayport,
Punta Russa, St. Augustine and St. Marks. Of these, Punta Russa
and Daytona should have gotten rain if a tropical storm occurred.
The only significant rain that occurred within a few days of the
supposed storm occurrence was 4.2" at Key West on the 18th and 19th of
September, three days before landfall. This is the same time (Partagas
and Diaz 1995b) that Tabasco, Mexico was reporting the very strong "norther".
Additionally, the September 1879 issue of _Monthly Weather Review_ did
not include this storm in its figure of tropical and extratropical
storm tracks. Other investigators of the era - including Loomis' (1881)
investigation of Atlantic basin tropical cyclones of the 1870s - also
did not show this storm to be occurring. The first report that
showed this track of this storm was Garriott (1900), which has
subsequently (Fassig 1913, Cry 1965, and Neumann et al. 1993) been
kept identically to the track presented in Garriott (1900). Unfortunately,
Garriott (1900) did not provide a source or analysis for how this
track was being arrived at.
To summarize: the 1879 _Monthly Weather Review_ and Loomis' 1881
reports did not identify this as being a storm; there was no rainfall
in Florida at the time of its supposed landfall; there were no
ships that reported gale force winds over the Caribbean or Atlantic
in association with this storm; the only strong winds somewhat in
the vicinity of this storm were in Tabasco, Mexico and were northerlies
the entire time. It is concluded that this storm was likely not tropical
in nature but was instead a strong (especially for this time of year)
cold front that swept through the Gulf of Mexico, causing the
"norther" conditions in Tabasco at the same time as dropping a
substantial amount of rain in the Florida Keys. Thus, we still
recommend removing this system as a storm since there is NO EVIDENCE
in support of the track apparently first provided by Garriott (1900).
> 7) 1881/02 (existing track book) was deleted. Partagas and Diaz,
> introduce some doubt, but include it. The case for deleting it
> seems much weaker than the case to retain it. As above, the
> committee feels that uncertainty can and should be documented, but does
> not find compelling evidence for deletion.
Agreed. We will add the storm back into the database as depicted in
Neumann et al. (1993).
< Regarding 4) above: >
< 1872/05 Jack Beven investigated and reviewed the October 1872
MWR. The U.S. Army Signal Service describes 1872/05 as (of the
thirteen storms noted for the month) "the most severe of these
which, from the 21st to the 27th traveled from the
Gulf northeastward over the south Atlantic and Middle States, and thence
eastward over southern New England into the Atlantic......brisk to high
winds, increasing at times to gales....nearly 7 1/2 inches of rain at
Norfolk." Given this assessment, and the fact that the center passed
directly over their office in Washington D.C. it is hard to believe that
they got it this wrong (i.e. compared to Partagas). Committee
refers this storm back to Landsea et al for reconsideration. >
Agreed that the evidence is strong that the storm went over the middle
Atlantic states, rather than out to sea after passing Florida as Partagas
and Diaz (1995b) had recommended. (The new evidence is primarily the
7 1/2 inches of rainfall in Norfolk and the lack of an extratropical storm
track drawn by the Signal Corp that would have matched what Partagas and
Diaz suggested.) However, consideration does need to be given to the ships
south of the Carolinas that apparently did get directly impacted by the
storm. Thus we are advocating a compromise track that incorporates portions
from both Neumann et al. (1993) and Partagas and Diaz (1995b). This new
track takes it across Florida a bit farther south than both previous tracks
to allow for a turn northward near the ship "Cardenas", then the new track
brings the system ashore as a tropical storm in North Carolina just a bit
farther east of Neumann et al.'s landfall. The new track then stays east of
Neumann et al.'s track while over the middle Atlantic states to correspond
with the the northeast to north winds over Washington and the low pressures
measured in New York City. After leaving New England, the new track rejoins
the original Neumann et al. track.
< 1875/01 [in Neumann et al. (1993) - 1875/02 in the current re-analysis
dataset] Jim Gross investigated. Even with the issue of the Spanish ships
reporting longitudes off by 6 degrees, there does not seem to be adequate
reason to chop off the early part of this track. Certainly climatology
suggests a Cape Verde type storm. In addition, the truncated track begins
at 70 kts, Committee supports extending this track back 2 or 3 days at least
to comply more closely to existing track book. The hook on the northern end
of the track is also questioned as unrealistic. >
Agreed that the track needs to be extended back to the 1st of September.
We have extrapolated the track back 48 hours based upon the subsequent
motion of the storm and climatology of other Cape Verde hurricanes. The
intensity also is extrapolated backward in time with reasonable (lower)
intensities that match the typical development of tropical cyclones in
the region. The hook on the northern end of the track was a typographical
error in the digitization of the best track positions, which has now also
been corrected.
< 1876/03 [in Neumann et al. (1993) - 1876/05 in current re-analysis
dataset] Brian Jarvinen investigated. Although removal of the
"Caribbean" portion of the track is supported by Cuban
observations, storm surge and swells on the southern coast of Cuba
supports backing up the track farther south, and increasing the winds to
50 kts on the 12th. Committee supports this
finding and refers this storm back to Landsea et al for reconsideration. >
Agreed that the track needs to be extended back to the 12th of October
to better take into account heavy swells at Tunas de Zaza, Cuba. The
track is extrapolated back to the south to start in the climatologically
favored genesis region in the southwest Caribbean Sea. This also
necessitated a slight alteration on the positions on the 15th to the 17th
as well for continuity. Intensity values are begun at 40 kt and brought
to 50 kt by 12 UTC on the 12th and then held steady until the 16th.
Return to Overview