Comments of and replies to the
National Hurricane Center Best-Track Change Committee
June 2003
Replies to the NHC best track change committee comments are given
in offset brackets. <...> - C. W. Landsea
Minutes of the best-track change committee meeting, 29 April, 2003
Members Jack Beven, Jim Gross, Brian Jarvinen, Richard Pasch, Ed Rappaport,
and chair (Colin McAdie) present.
---------------------------------------------------------------------------
The committee met to consider the second-revision changes submitted by
Chris Landsea et al. for the period 1851-1910. This set of changes
discussed below are the Landsea et al. response to the "consolidated
minutes" of 01 May 2002.
For convenience, the following is organized by page number (added during
discussion) of the 29 page response section of this installment. Items
without page number are additional comments discussed at that juncture
during the meeting.
General comments:
A significant issue discussed below is changing the track file with
insufficient, or no, data. The committee is unanimous in its opinion that
changes should not be made without the support of observations. As noted
below, the committee is aware of the origin and shortcomings of this dataset,
hence our support for the reanalysis. However, for any particular case
(that is, one lacking evidence) it is not clear that the assignment of 85 kt
is unreasonable. The histogram simply reflects the number of cases,
decreasing in time, for which we have some indication of a hurricane, but
nothing else.
The committee has sought to be consistent in its request for data supporting
any change. Therefore, if a change from an 85 kt maximum (based on the
absence of data) has made and overlooked on our part, please do not change
the existing data set. This applies to the post-1886 era. A specific
example, also mentioned below, is 1892 #3.
< It is agreed that changes should not be made to the pre-existing
HURDAT (1886 onward) based upon lack of observational data. An
example where such a strategy would be inadvisable would be storm #3
of 1911. (We've been working on the 1910s recently, so this system
immediately came to mind.) For this hurricane, we could find no direct
observations of hurricane intensity or even tropical storm intensity
from the Historical Weather Maps, COADS ship database, Monthly Weather
Review or other sources. One might be tempted to alter the intensity
of this system downward significantly. However, its hurricane status
is not in doubt because of its well-recorded impacts at landfall in
Nicaragua. Thus no changes should be made to its intensity as this
particular system was archived with a reasonable lifecycle in HURDAT to
match impacting the Central American coast with a Category 2 (85 kt)
intensity.
Other systems, unfortunately, have more ambiguity as to what their
intensity was for a large proportion of their lifetime. For storms
of the 19th Century and early 20th Century over the open ocean, one
almost always does not have enough data to ascertain with certainty
what the true intensity is. It comes down to evaluating the
available observations versus the educated first-guesses that
Neumann and Hope came up with in their digitization of Cry's track
plots. (The previous reply provided details by Neumann about how
they came up with the rough estimates of intensity for many of the
systems for the late 1800s. In that writeup, he describes how for
the large majority of the systems that they simply quantified the
peak intensity of systems as either 50 kt (tropical storm), 85 kt
(minor hurricane) or 105 kt (major hurricane) and then developed
reasonable lifecycles to match this somewhat arbitrary peak.)
However, as stated earlier, it is agreed that it would not be
advisable to make changes to the intensity of existing storms and
hurricanes when there is insufficient data available. Therefore, we
have re-evaluated the systems mentioned in this set of minutes below
as well as all of the ones identified as problematic because of a
lack of data in the first round of comments from the committee in
2002. These systems are the following: 1887 #16 becoming #18,
1889 #1, 1891 #2, 1891 #10, 1892 #2, 1892 #3, 1893 #5, 1895 #4,
1896 #6, 1898 #1, and 1899 #4. For some of these tropical cyclones,
observations were found to substantiate alterations made to the
intensity. However, a few (i.e., 1892 #3, 1893 #5, and 1896 #6)
were so lacking in data that no changes are now made to the intensity.
In either case, all available observations of gale force or greater
winds (or the equivalent in sea level pressure) are listed in full
in the metadata file.
Specific comments:
page 1. It was noted in the consolidated minutes that the ATCF fix file
uses a 4-digit year. If this format were used by Landsea et al. it would
allow TPC to import the file directly into the ATCF. Is it possible for
Landsea et al. to make this change?
< Yes, however, the center fix file needs to have substantial revisions
before it is complete. This will be focussed upon after the primary
changes to HURDAT, the metadata file, the reply to comments and the
track maps are finished. >
page 4. 1855 #1 was discussed, and is now accepted as revised.
page 5. There was further discussion about the applicability of the Fujita
scale in assessing damage caused by tropical cyclones, especially in this
era (1860's). If Landsea could provide a copy of Boose to Jim Gross, it
would be appreciated. Jim will examine the methodology and summarize for the
committee.
< These papers will be provided. Additionally, Boose maintains a web site
that contains all of the wind-caused damage descriptions as well as
his analyses at: .
See the "Landscape and Regional Impacts of Hurricanes in New England"
(HF011) and "Landscape and Regional Impacts of Hurricanes in Puerto
Rico" (HF012) sites in particular. >
page 6. 1870 #6 A question was raised about the availability of the data
used by Perez. Is Perez (2000) in final form? (Note to committee: we do
have a manuscript copy of Perez. In the Perez summary for 1870 #6, for
example, other sources are referenced - these contain the actual data, or
accounting of events. Landsea's response on the bottom of his page 7
concerning 1882 #7 also addresses this issue)
page 8. 1885 # 2. This is a typo in the consolidated minutes, and should be
1895 # 2. Explanation for shifting this track into Mexico is accepted.
Additional note: committee needs to obtain a copy of Ellis.
< A copy of the report by Ellis will be provided. >
pages 8 - 11 There was further discussion about the "lack of evidence"
issue, and the response of Landsea et al. There has been a valuable
addition of background material to the record, especially concerning the
digitization of Cry. The committee is aware of the highly stratified nature
of the early data, the "ballpark" intensities, and the reasons for their
being so. The essential point remains, given all of this, that changes must
be made based upon observations, or reinterpretation of observations, using
documented methodology. (see, for example, 1887 #16 becoming #18, 1889 #1,
1892 #2, #3)
< Agreed. As described above, all of these systems and several more are
now re-analyzed accordingly. >
page 11. 1886 #4 Accepted as revised.
page 12. 1887 #3 Thank you for clarifying the source of observations as
Partagas, unless otherwise noted. The committee suggests that this
convention be highlighted early in the document for the benefit of future
researchers.
< This is now so stated in the introduction of the metadata file. >
page 14 1887 #16 becoming #18. In this case, P+D did not introduce any
changes from Neumann et al. Landsea et al., however, have "reduced the winds
for whole lifecycle of the storm since the best available observations
indicate that the system peaked as a minimal hurricane (65 kt chosen as peak
winds) rather than the standard category 2 (85 kt) intensity originally
suggested in HURDAT."
Based on the conventions mentioned above, we can assume these observations
are in P+D. It can thus be assumed that 65 kt was chosen as the peak
intensity based on a ship observation, but this should be explicitly stated.
The change must be tied to an observation. It would not serve any purpose to
duplicate at length P+D or any other source, and this is not requested, but a
key observation upon which the peak is determined should be given. It is
noted that other storms with changes of comparable magnitude are accompanied
by fairly significant data listings.
< Agreed that additional details and evidence needs to be provided to
substantiate changes made. This is now so done for this system and some
alterations from original re-analysis have been made. >
1887 Additional systems #4 Given a surface pressure of 29.68 inches
(apparently a good observation) in [the data uncovered by Michael
Chenoweth], this system was a tropical storm and should be included as a
single point storm.
< Additional information provided by Michael Chenoweth indicates that
these values are surface pressures, not sea level pressures. The
height of the observations was about 30 m. Thus the true sea level
pressure value was closer to 1008 mb, rather than 1005 mb. It
still appears that the evidence for this system is inconclusive for
tropical storm intensity, though it certainly was of at least
a tropical depression. >
page 14 1888 #3 Based on discussion and comment from BRJ, increase peak
winds to 110 kt. This would also be more consistent with keeping system as a
"high end Category 3" as stated.
< Agreed. So changed. >
page 15 1889 #1 Another case similar to 1887 #16 becoming #18. Is the
reduced peak intensity based on one observation, or many? Is the remainder
of the track determined by this peak, or are there other observations at
other times? Are there observations on 5/18-19 to justify the reduction in
winds? There is strong consensus that this should be explicitly stated.
< Specific observations are now explicitly included into the metadata
writeup to support changes made to HURDAT. For this case in
particular, the 1889 Monthly Weather Review was very helpful in
providing guidance as to the intensity for much of the lifetime of
the hurricane. >
page 15. 1889 #3 Although accepted as revised, the consensus is that the
end of the track still seems odd, given the populated coastline. There is
no readily apparent physical explanation for the sudden demise, and thus
this portion of the track may need further work in the future.
< Agreed. >
page 16. 1891 #7 #8 #9. The committee appreciates the additional time and
effort that has gone into this proposed reduction in systems from 3 to 1, and
notes that COADS is a valuable additional source of data. In this case,
however, the committee finds that COADS is not compelling. The committee
considered keeping revised track #7 and also keeping #8 and #9, but revised
#7 incorporates pieces of #8 and #9, thus this would not be a good solution.
There are multiple scenarios that could explain the data. Further, there are
no observations in Florida that support any tropical cyclones, complicating
the issue. The committee continues to acknowledge uncertainty in the
disposition of these systems, but for now retains Neumann et al. (Additional
notes: There is no mention of Perez in the meta data. Can he provide
anything here?)
< It is certainly true that the situation remains ambiguous at many times
for these three systems. However, portions of the existing tracks
cannot be correct given the information uncovered from COADS, such as
storm 8 on the 6th and storm 9 on the 10th and 11th.) However, it is
agreed that the most prudent course of action at this point is to
table any action on these three storms, until additional station data
can be obtained. Thus efforts will be taken to obtain observations
from Cuba as well as to search the archives for U.S. stations at the
National Climate Data Center. This will be conducted over the next
few months. >
1891 additional system #1 (July) MWR mentions 'gale' winds. This system
may warrant further research. Is there any COADS?
< Agreed that this system deserves further investigation. COADS and
other data will be analyzed over the next few months. >
1891 additional system #4 (Sept) MWR gives TS force winds at coastal
stations. This system is also given a high probability by P+D, and bears
further investigation.
< Agreed that this system deserves further investigation. COADS and
other data will be analyzed over the next few months. >
page 17 1892 #2 Extratropical transition accepted, however, reduction to
65 kt does not appear to be supported.
< All gale force and greater observations available for this system
are now listed in the metadata files. These observations support
the system peaking near minimal hurricane status, rather than the
standard Category 2 (85 kt) originally found in HURDAT. >
page 17 1892 #3 There appears to be minimal evidence, at best, for
changing anything here. If an observation cannot be referenced, do not
change.
< It is agreed that there should be no changes made to the original
intensity estimates found in HURDAT for this system, due to lack
of observations. >
Note caption error in 1892 plot for system #5.
< So corrected. >
page 18 1893 #6 Committee previously requested opinion of BRJ on this
storm. His opinion did appear in the previous minutes, however, he has
continued to work on this storm and now has a substantial revision. He has
agreed to present this to the committee when complete.
page 18 1893 #6 If manual changes to the plotting program are required,
please ensure that the resulting plot depicts the points as given in the
dataset.
< Agreed. This has always been the policy. The manual changes to the
intensity indicated has been to correct the hourly interpolation
produced by our software for some of the cases near or after landfall. >
page 18 1893 #10 There appears to be a timing error in this track.
JLB will provide revised points.
< J. Beven further clarified that this was a timing error in the central
pressure listed at 12 UTC on the 2nd of October. This value was
measured at 1530 UTC on the 2nd. Given that HURDAT includes central
pressures within +/-2 hours of the synoptic time, this value is now
removed from HURDAT. However, it is still included in the metadata
file as well as the U.S. landfalling hurricane table. >
1898 - note possible additional system. Significant rainfall in Jamaica
found in MWR. No wind data, but a pressure is given of 28.66. (May 23-27).
< Agreed that this system deserves further investigation. COADS and
other data will be analyzed over the next few months. >
page 22 1899 #6, becoming #8 - wording - system formed to the south of
Jamaica, not the west.
< Changed wording as suggested. >
page 22 1900 #1 It is noted that some additional information may be
gleaned from a publication of Isaac Cline's brother. (JLB for details)
page 23 1901 #3 This is a typo in the consolidated minutes. It pertains
to 1903 # 3. Although accepted as revised, the issue was that any
statistical mean, or fit of data, obviously results from a population of
points. Thus, some outliers may be acceptable (and realistic) if the
data seems to indicate that such is the case. Also, the committee wishes
to warn against a false sense of precision in using Ho's RMW estimates.
< Agreed. >
page 24 1903 #2 Discussed and accepted.
page 24 1903 #3 Previous minutes had requested the opinion of BRJ, and
this is still pending. Decision pending on this system. Committee notes
that center should be over Nassau at 00 UTC, so work is required on the
track.
< Agreed that the center needs to be moved to be close to Nassau at
00 UTC on the 10th. However, the observations of east winds shifting
to increased south winds at the time of lowest pressure suggests that
the center went just to the west of Nassau, not over. Track changed
accordingly. >
page 24 1903 #8 Discussed and accepted.
1903 Additional system #1 Block Island and Nantucket observations
suggest tropical storm force winds, although the structure of the system
is not clear. Needs further research.
< Agreed that this system deserves further investigation. COADS and
other data will be analyzed over the next few months. >
1903 Additional system for 10-11 Sept. A possible depression in the
Gulf of Mexico. Isaac Cline states that warnings were issued. System not
noted in P+D. Needs further research.
< Agreed that this system deserves further investigation. COADS and
other data will be analyzed over the next few months. >
page 26. 1906 #5 becoming #6. Provisionally accepted, although it may
need more work.
page 26 1906 #7 becoming #8 By analogy with H. Andrew, why would we
reject the unofficial surface pressures given in the Miami Daily
Metropolis? This is how observations were solicited after Andrew, and
the unofficial Fairbank's observation was considered a key piece of
information. What is the difference here? Also, the Ho RMW values are
estimates, and are not always correct. Note that Barnes suggests a small
core (~10 n mi may be a better estimate.)
< Agreed that these unofficial pressures should be utilized and are
now incorporated, along with a revised estimate of RMW from Barnes'
description, into the revised HURDAT. Note that the changes now
bring the hurricane up to a Category 3 (105 kt) at landfall in south
Florida, which does agree with Partagas' assessment as well. >
1907 additional system #1 This system bears re-evaluation in light of
Ana (2003). There is no doubt of 35 kt winds, but the structure is not
known.
< Agreed that this system deserves further investigation. COADS and
other data will be analyzed over the next few months. >
1908 additional system #2 It is noted that Isaac Cline felt that
warnings in the Gulf were justified for this system.
< Agreed that this system deserves further investigation. COADS and
coastal station data will be analyzed over the next few months. >
page 27 1908 #6 becoming #7. After much further discussion, the
consensus is that it remains extremely unlikely that this system could
have re-strengthened to Category 2 prior to landfall in eastern Cuba. If
anything, weakening would be expected, rather than strengthening, so this
is not accepted as a plausible solution.
< Agreed. The system is reduced back to a Category 1 hurricane at
landfall in Cuba. >
page 28 1909 #5 becoming #6 The committee cannot accept this landfall
intensity in Cuba, given topographic interference. A rule-of-thumb
would be to weaken the system two categories, rather than strengthening.
< Agreed. The system is reduced back to a Category 1 hurricane at
landfall in Cuba. >
page 28 1909 #7 becoming #8. Please check the plot of this system.
< Plot of this hurricane has been checked and small changes have
been made to its track near landfall in the United States. >
page 28 1909 #9 becoming #10. The crucial question here is whether the
middle Keys were in some portion of the eye. Given the close call, the
decision is to leave this hurricane as a Category 3. Implied precision in
reducing to a Category 2 is not warranted. Also note while some changes
in intensity are small, there are some very large reductions in intensity
(30 kt) in the early part of the track. The text does not make this
distinction.
< Agreed and so revised on both points. >
Return to Overview