Comments of and replies to the
National Hurricane Center Best-Track Change Committee
January 2000
Chris Landsea, Craig Anderson, Gil Clark, Charlie Neumann, and Mark Zimmer
Comments from the NHC best-track change committee are given in
offset brackets. <...>
< Minutes of the best-track change committee meeting, 23 March 1999.
All members (Jack Beven, Jim Gross, Richard Pasch, Ed Rappaport) and
chair (Colin McAdie) present. >
< The committee met to consider the first installment of track
additions (1851-1860) received from Chris Landsea et al. Several
questions were posed prior to the meeting, and were dealt with
first. >
< 1) The committee agrees with Landsea et al. that 'single point' storms
should be included in the database. Among the benefits are that,
given credible evidence, the storm count becomes more accurate and the
single observation should be represented. The point was made that
some of the single points may be less valid than others. The
committee felt that these should be questioned on a case-by-case
basis. This will remain part of the committee's function - new data
or interpretation is always possible. >
Agreed. The "single point" storms are included. Users of the new
database will need to be cautioned (for programming purposes) that a
few of the storms only have one point and no track.
< 2) Names for historical storms are important, but should be included in
the hypertext addendum. The committee considers the addendum (i.e.
the text file linked to the data file itself) to be a critical part of
this work, and will urge its strengthening in several comments
below. The naming of historical storms opens a Pandora's box of
potential problems, e.g. disputed names, names used in other references
but not used here and vice-versa, multiple names, and regional
names. It might also be argued that storms in the existing
database be named, for example, the 'Labor Day' storm of 1935.
These problems may be side-stepped without loss of information by
inclusion in the text file. In the case of 1852/02, for
example, 'San Lorenzo' would appear in the text, but not the data
file. Advantages are that brief discussion is possible (such as
that already given in the materials provided to the committee),
statements such as 'also known as...' or 'referred to XXX in
Ludlum, etc. are possible. The text data base is also
searchable by users looking for a particular storm, or an AKA. >
Agreed. We have removed the the included names from the best track
file (all of the 1851-85 storms are listed as "NOT NAMED"), but as
requested have put any available informal names in the metadata text file.
< The committee then took up a list of questions brought in by the
members. Those not resolved are given below in consolidated form,
for response. >
< 3) It was questioned why the extratropical portion of tracks had been
dropped. This may have been for the sake of consistency with the
existing data file. However, after discussion, the consensus was
that the extratropical portion represents valid data and should be
included in the data file, with extratropical entries so
noted. A specific question on this issue is why was 1856/6
made extratropical at 36N 52W on 20 September? >
Agreed. We had originally made the decision to not include extratropical
stages of tropical storms and hurricanes to be consistent with the existing
best track and Neumann et al. track book which first included extratropical
stages in 1899. We have now gone back and included extratropical stages
(based upon observed wind and temperature shifts consistent with
extratropical storm features). Doing so implies that an update of the
years 1886-1898 to include extratropical stages is also needed in the
coming year.
After a reexamination of the 1856/06 hurricane, we concur that the
evidence for calling this one extratropical was quite slim. We have
added back the 21st and 22nd of September as a 60 kt tropical storm.
< 4) It was not clear to the committee how descriptive terms "heavy
gale" etc. had been used in providing wind estimates. The
Beaufort scale provides a relationship, for example. A table or
reference should be provided, or a note on unique or subjective
interpretations given in the text file. >
The conversion from descriptive measures of shipboard winds to quantitative
wind speeds - while quite subjective - is assisted by the usage of the Beaufort
Scale, which was introduced in 1805.
The Beaufort Wind Scale (Byers 1974) as follows:
Beaufort Knots Description Specifications at Sea
Number
-------- ------ --------------- -------------------------------------------
0 < 1 Calm Sea like a mirror
1 1-3 Light air Ripples with the appearance of scales are
formed, but without foam crests
2 4-6 Light breeze Small wavelets, still short but more
pronounced; crests have a glassy appearance
and do not break
3 7-10 Gentle breeze Large wavelets; crests begin to break; foam
of glassy appearance; perhaps scattered
white horses
4 11-16 Moderate breeze Small waves, becoming longer; fairly
frequent white horses
5 17-21 Fresh breeze Moderate waves, taking a more pronounced
long form; many white horses are formed
(chance of some spray)
6 22-27 Strong breeze Large waves begin to form; the white foam
crests are more extensive everywhere
(probably some spray)
7 28-33 Near gale Sea heaps up and white foam from breaking
waves begins to be blown in streaks along
the direction of the wind
8 34-40 Gale Moderately high waves of greater length;
edges of crests begin to break into
spindrift; foam is blown in well-marked
streaks along the direction of the wind
9 41-47 Strong gale High waves; dense streaks of foam along the
direction of the wind; crests of waves
begin to topple, tumble, and roll over;
spray may affect visibility
10 48-55 Storm Very high waves with long overhanging
crests; the resulting foam, in great
patches, is blown in dense white streaks
along the direction of the wind; on the
whole, the surface of the sea takes on a
white appearance; the tumbling of the sea
becomes heavy and shock-like; visibility
affected
11 56-63 Violent storm Exceptionally high waves (small and medium-
sized ships might be for a time lost to
view behind the waves); the sea is
completely covered with long white patches
of foam lying along the direction of the
wind; everywhere the edges of wave crests
are blown into froth; visibility affected
12 > 63 Hurricane The air is filled with foam and spray; sea
completely white with driving spray;
visibility very seriously affected
Thus due to the limitation at the top end of the Beaufort Scale, we only
generally assigned winds in the best track of 70 kt for ship reports of
"hurricane" winds. We did boost this assignment to 90 kt for descriptive
terms such as "severe hurricane", "violent hurricane", "terrific hurricane"
or "great hurricane". We did not assign any best track intensity values
of major hurricane status (96 kt or greater) for hurricanes out at sea
without corresponding central pressure data for support.
< 5) How was the Kaplan and DeMaria model used? i.e. started from a
'known' landfall wind or pressure? A brief sentence in the text
file would be extremely helpful. Committee was uncertain that the
model was appropriate for use over Hispanola, or other rugged
terrain. Please provide a reference for the model (1995?) and
clarify this point. Future users should be able to reproduce
model-generated winds. >
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 for 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. (This
discussion will be incorporated into the upcoming technical memorandum
to accompany the update.)
< 6) The pressure-wind relationships generated considerable
discussion. Some questions may be easily resolved. Could the
rationale for using the new relationships be given? While the use
of the new relationships seems appropriate in historical reconstruction
(for example, to derive a wind from a measured pressure), there is
considerable concern among the committee that the new relationships will
be used in a revisionary sense on more recent data. The consensus
is that there is scatter among the dependent data, and that derived
relationships should not be used to modify observations. Discussion
about these relationships, their use and derivation, should appear in
future tech memo. The committee would appreciate some discussion
about the 'over-weighting' of higher pressures. >
We agree completely that these new wind-pressure relationships are
to be utilized to help derive winds from an observed central pressures
only in the absence of reliable wind data. These relationships are NOT
intended to give best track wind estimates for recent years (i.e. the last
few decades) where accurate flight level wind measurement are
available from reconnaissance aircraft. Where we have reliable, direct
wind observations, we will utilize these and not use the pressure to
modify the observations.
With regards to the development of the regression to obtain the wind-
pressure relationships, we first attempted to develop the equations with
all of the available data. We found, however, that this overweighted
the observations of the tropical storms and Category 1 hurricanes at the
expense of the major hurricanes due to the overwhelming numbers of
observations at the low wind speed ranges. When we compared
the derived equations against the observations of wind and pressure at
the very high wind values (100 kt and greater), the fit was quite
poor. We were able to overcome this by binning the observations into
5 mb groups and then performing the regression. This way the observations
at the 981-985 mb range, for example, were weighted equally to those of
931-935 mb. After doing the regression this way, we were able to get a
much better match of the regression equations versus the observations for
the Category 3, 4 and 5 hurricane ranges.
Both of these issues will be discussed in depth in the accompanying
technical report.
< 7) Concerning 1852/01, it was felt that the 961 mb pressure should be
given in the data file. If the wind is derived from a pressure
which has itself been derived, why not give both, with appropriate
explanation in the text file? It was also pointed out that
many pressure observations were not central pressures, and should not be
used to derive winds. Specific cases are attached. >
For the cases of estimated or derived central pressures, we are following
the convention of the original best track which only puts in central pressure
values if they are observed. (This was the case from 1886 until 1978 in
HURDAT.) We would prefer to continue this methodology. In some of the
landfalling hurricanes, a central pressure estimate was obtained from the
work of Ho (1989) (which is so noted in the metadata file), which were then
used to provide the winds from the new pressure-wind relationship. If no
estimated central pressure is described in the metadata file, then the winds
were determined from ships immediately offshore, destruction at the coast
and/or storm surge values. These estimated winds were then used via the wind-
pressure relationship to obtain a reasonable central pressure. In either case,
the objective was to provide both a peak wind at landfall and a central
pressure.
As the committee noted, in many cases in these early years, we have surface
pressure measurements obtained which are not central pressure readings.
Such data are noted accordingly in the metadata file and are used as a
minimum estimate of what the best track winds were at the time. In most of
these cases, the best track winds chosen are substantially higher.
< 8) Are all of the storms in Rappaport and Fernandez-Partagas accounted
for, e.g. the November, 1852 storm? >
It was assumed that the work from Rappaport and Partagas (1995) were
completely contained in the Partagas and Diaz (1995a,b,1996) reports
because of the common author. This was true for 99% of the cases.
However, the following storms in Rappaport and Partagas were not contained
(for some unknown reason) in the Partagas and Diaz reports:
#214. Northeast Gulf of Mexico - 26 May 1863 - (page 24)
#402. Jamaica - 6-8 Nov. 1852 - (page 33)
#411. W Atlantic - 19-20 Oct. 1873 - (page 33)
#417. SE TX coast - 12-13 Oct. 1880 - (page 33)
Due to time constraints, we will not include them in the current best
track, but will take up the case for these four storms in the coming year.
< 9) 1854/04. What does 'original' mean in this case? It is stated in
the text file that 'evidence' suggests that this was not a separate
storm. Although all of the details do not need to be given here, a
future researcher must be able to find this information. >
"Original" here meant from the work of Partagas and Diaz. We have now
included an expanded discussion of this particular system into additional
notes at the end of year 1854. (Other similar cases are to be handled
similarly.)
< 10) Related to 2 and 9 above. The structure of the text file could
be streamlined by providing numbered references, which should always be
given. For example, Ho could be reference 1, Partagas and Diaz (a)
could mentioned as reference 6, etc. Unreferenced
changes (e.g. to Partagas) will be impossible to trace in future years.
If Landsea et al. intend to include such evidence in a future tech memo,
please reference it now. >
Unless the committee feels strongly about this, we would prefer to keep the
full references in the individual writeups of the metafile. This would
allow the users to immediately know the authors of referenced material
rather than having to go the the Bibliography for this information every
time.
We have now attempted to more fully document any significant change to
the analysis of Partagas and Diaz in each of the metafiles.
< 11) As potential users, the committee found the use of 'new,' 'original,'
or 'old' to be somewhat confusing. An improvement would be to
adhere to the order of storms in the data file. At the end
of that year, itemized storms deleted from Partagas (or elsewhere) should
be given, with appropriate comments. >
Agreed. We have now restructured this exactly as requested.
< 12) 1856/03. Ludlum states that the 'Charter Oak' storm (see
comment 2) 'crossed the neck of Cape Cod,' yet the plotted track provided
here clearly does not. If cause exists to alter the track, please
document in the text file. >
We suggest a track which keeps this tropical storm's center offshore
of New England, which does indeed disagree with Ludlum's assessment.
This is due to all wind reports from New York, Connecticut, Rhode Island
and - especially - Massachusetts which remained at an easterly
component for the duration of this storm. If, as Ludlum suggested,
the storm crossed the neck of Cape Cod, there would have been a wind
shift to a westerly component at Cape Cod and/or Nantucket. Both remained
easterly, thus suggesting a just-offshore track. (This writeup is now
included in the metadata file.)
< 13) As to the ATCF fix file and the plots, it should be noted that some
plotting irregularities have occurred because of 'blank' dates in the fix
file. How are blank dates used and interpreted? >
These ship observations with no specific dates were primarily utilized by
Partagas and Diaz to provide information about the track (e.g. a SW gale
noted by a ship captain would indicate a tropical storm off to the
northwest of the ship) as long as other ship/land observations could help
pinpoint the timing of the track. We have also utilized ship observations
that had no specific date to assist in the intensity estimates. This
writeup will also be included forthcoming technical memorandum.
< 14) In the future, the plots should be scaled to accommodate maximum wind
speed. >
This will be done for plots from 1871 onward.
< 15) In the future, utility of the track plots and fixes would be greatly
enhanced by plotting the fixes (or obs) on the track maps. >
We have been plotting the fixes on the track maps, but the actual number
of center fixes is quite small. In the majority of tropical storms and
hurricanes from 1851-1885 there simply do not exist any "center fix" data
points.
< Due to time constraints, not all of the specifics could be addressed at
this initial meeting. The following additional comments were provided by
Jack Beven after the meeting, and are attached for response. >
< 16) The work involving observed pressures does not seem to have been
careful enough in regards to whether the report was in the eye or not. A
notable example is the Last Island storm (1856/01), where Ludlum's
description clearly shows the 955 mb pressure was not measured in the
eye. Yet, it was treated as an eye observation in the best track. Please
check all available surface observations to determine whether they
were truly central pressures and adjust the best tracks if necessary.
(Other examples of this follow.) >
Agreed. We should have been more careful in our evaluation of this
observation. This has now been changed as requested. We have also
gone through all available surface observations and only accepted the
few that clearly were central pressure measurements.
< 17) Ludlum describes many surface observations, including Smithsonian
weather observers. Are these observations available from the National
Climatic Data Center or elsewhere? If so, they should be obtained and
used for additional information. >
< 18) On a related note, Ludlum mentions the Redfield Collection at the
Yale University Library. Has this potentially valuable data source been
accessed? >
Both of these are excellent suggestions, but go far beyond the original
approved NOAA/NASA grant that is funding this work. We will attempt to obtain
additional resources that would allow us to do exactly what is encouraged.
However, for this first effort in providing best track position and
intensity estimates for 1851-1885, we do not have the time or resources
to pursue this currently.
< 19) Storm 1854/02: Ludlum's information shows the observer in Savannah,
Georgia recorded a pressure of 973.1 mb at 1600 (local?) 8 September.
Please add this to the fix file. >
As we are placing only central pressure data into the fix file, this value
has not been included. However, it is now utilized more explicitly in the
metadata writeup as shown below:
"Peripheral pressure reading of 973 mb (at 20 UTC on the 8th of September
in Savannah, Georgia) suggest winds of at least 83 kt utilizing the same
subtropical wind-pressure relationship. Ho used this value with other
information to estimate a 950 mb central pressure at landfall which gives
103 kt again from the subtropical wind-pressure relationship - here we are
choosing 100 kt for the best track."
< 20) Storm 1855/06: Ludlum describes this storm as a hurricane, complete
with a 10 to 15 ft storms surge. The best track file has this as a 60 kt
tropical storm. Please review the data and make appropriate
corrections. >
We agree that this was an important omission. This has now been boosted
to a Category 3 at landfall based upon storm surge and destruction along
Louisiana and Mississippi from the information presented in Ludlum's book.
< 21) Storm 1856/05: Both Partagas and Diaz and Ludlum indicate the
"Daniel Wester" (Webster in Ludlum) measured a 968.5 mb pressure
outside the eye at noon on 30 August with the wind "blowing a perfect
tornado". This does not sound like the 50 kt observation shown in the fix
file. Please correct both the fix file and the best track to account for
this observation. >
We agree that the quantitative wind estimate from the "Daniel Webster"
was too low, but difficult to pinpoint as the term "perfect tornado" does
not adhere to the Beaufort Scale terminology. We are suggesting that
this may refer to winds of hurricane force around 90 kt.
The 968 mb peripheral pressure does suggest winds in the best track of
at least 91 kt. We have chosen 100 kt for the 30th of August, but believe
that the hurricane weakened to a Category 2 at landfall in Florida based
upon the observed (6') storm surge and moderate damage in Apalachicola.
< 22) Storm 1858/03: Ludlum's description of the observations at Sag Harbor
suggest the station was not in the eye when the lowest pressure was
recorded. Please review this and make appropriate corrections to the fix
and best track files. >
We agree that this observation of pressure was a peripheral value, not
a center fix. Based upon the central pressure reading of Providence of
979 mb as the hurricane was moving inland and this peripheral pressure
value of 978 mb at Sag Harbor, we estimate that at landfall along New York
that the central pressure was 976 mb.
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
< Minutes of the best-track change committee meeting, 18 June, 1999.
Members Jim Gross, Richard Pasch, and chair (Colin McAdie) present, with
Richard Pasch presenting comments supplied by Jack Beven. The committee
met to consider the second installment of track additions (1861-1870)
received from Chris Landsea et al. Prior to the consideration of specific
points, it was noted that Chris will supply the third installment using the
suggestions appearing in the minutes of the first meeting, and then take up
the task of reconsidering or amending the first two installments, as
necessary. >
< The following points were discussed: >
< 1) It was first noted that in the track file itself, lines 01875, 02875,
and 02965, hurricanes are coded as tropical storms, i.e. "TS" should be
"HR". These are easily overlooked and should be machine or manually
checked. >
All of these lines have been checked and are now believed to be correct.
< 2) It was also noted that in the track map supplied for 1868, that storm
number 4 was labeled storm 2, and vice-versa. >
This is now corrected.
< 3) There was considerable discussion about the duration of stationary
storms. The following were contrasted: In the materials presented, it is
assumed that 1864/4 was stationary for 4 days, but it was assumed
unreasonable that 1866/4 was stationary for 3 days (becoming single-point
storm), and in the case of 1869/4, a stationary storm became a 48 hr
track. The consensus was that while these _may_ be reasonable solutions, it
was at least physically unlikely that a storm could remain stationary for
4 days, especially at a constant intensity. It should be stated in
the hyper-text that the data, such as they are, do in fact support this
(assuming they do). The conclusion would be strengthened by a brief mention
of possible dynamical situations that might create these observations (i.e.
a tight loop might create the same observations as a truly stationary
system). >
Agreed. Additional discussion of an alternative solution (i.e. a tight but
slow loop) has been added for storm 1864/4.
< 4) 1861/2 Given a shipwreck at Cape Florida, it seems likely that this was
a Keys landfall. Note that Ludlum refers to this as the "Key West
hurricane". Track supplied shown as going through the Straits. Please
review. >
After reviewing this hurricane, we agree that it should be considered as
producing hurricane conditions in the Florida Keys and is now listed in
the "U.S. landfalling hurricanes" file as well as with an "SSS=1" for the
Keys ("BFL1") in the HURDAT. However, doing so required no track change
since the original track had taken the center about 40 n mi south of the
Florida Keys chain of islands. Thus while it is still our suggestion that
the center did not make landfall, this hurricane should be counted as one
which produced hurricane force winds over U.S. land.
< 5) 1861/8 At least northern portion of the track is probably extratropical,
given time of year (November). Please review and code as extratropical.
Are pressures in the center? >
We would have expected that given the location and time of year that
the system would have been transitioning to or have already become an
extratropical storm when it reached New England. But the available
evidence suggests that this did not happen (see, for example, the
observations from Boston as reported in Ludlum's book).
The pressures listed are kept as central pressure values, though the
evidence that they were is not concrete. However, even if the storm
was not extratropical in New England, one would expect a spreading out
of the pressure gradient and that the lowest pressures would extend
over a larger area than in the tropics (or subtropics). Thus the values
of 1000 and 999 mb are likely to be (or very close to) central pressure
values and are treated as such.
< 6) 1863/3 The point was raised as to whether ships were in or near the
center. Committee finds this is probably OK. >
Agreed.
< 7) 1865/7 Partagas does not support decision to drop. Please leave in data
set as a single-point storm. >
We have added this storm back into the HURDAT database.
< 8) 1865/6 The 975mb central pressure was questioned. However, upon review,
committee finds this appropriate. >
Agreed. (This storm, because of adding back in the previous storm, is
now cataloged as storm 1865/07.)
< 9) 1869/7 Committee finds that wind shifts seem to support location of
pressure observations as near the center. >
Agreed.
< 10) 1869/8 Landsea et at. drop the storm. The committee strongly questions
whether this is advisable. Partagas finds that this _was_ a distinct
system. Reasons for this re-evaluation are not clear. >
We have added this storm back into the HURDAT database.
< 11) 1870/3 and 1870/4 same discussion and conclusion as for 1869/7 above. >
Agreed.
< 12) 1870/6 Committee finds 969mb pressure was based upon synoptic analysis,
and is acceptable. >
Agreed.
< 13) In view of the above comments, the committee recommends that when
pressures are given, that they be accompanied by the statement "believed
to be at or near the center" if in fact this is the conclusion of Landsea
et al. This should avoid confusion with observations which are reliable,
and possibly relevant, but are known or suspected to be well-removed from
the center. >
We have now specified in the metadata text file whether a pressure measurement
is believed to have been a central pressure or whether it was more likely
to be a peripheral pressure. While both can be useful in diagnosing the
estimated maximum sustained windspeed, only those that are central pressure
observations are included directly in the best track file. This should
avoid the ambiguity mentioned above by the committee.
< 14) Please check to be sure the "XING" variable in the track file correctly
records U.S. landfalls. It appears the 1870/6 passes close enough to SE FL
be considered a landfall given coding as "SSS=1".
The hurricane of 1870/6 is counted as one that causes hurricane conditions
on the coast (thus "SSS=1" and "BFL1"), but does not actually have the center
make landfall. Under the current structure of the HURDAT as defined in the
_NOAA Tech Memo_ by Jarvinen et al., this is to be listed as "XING=1". Note
that the new "U.S. landfalling hurricanes: 1851-1885" attempts to clarify
these points explicitly.
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
< Minutes of the best-track change committee meeting, 10 August, 1999.
Members Jack Beven, Jim Gross, Richard Pasch, Ed Rappaport, and chair
(Colin McAdie) present.
The committee met to consider the third installment of track
revisions (1871-1885) received from Chris Landsea et al. (It is
noted here that operational priorities due to the 1999 hurricane season
intervened following the meeting, delaying the preparation of these
minutes.) These comments may not encompass all issues
concerning this set of revised tracks; the committee felt that more
work on their part was needed. However, in the interest of letting the
work proceed, Chris has agreed to respond to the comments available thus
far. The committee will then meet to re-consider the full (revised)
package of all three installments.
Prior to the discussion, McAdie suggested that Brian Jarvinen become a
member of the committee. This suggestion was approved by the
committee. Welcome to Brian.>
< The following points were discussed during the meeting:>
< 1) It is noted in the comment section provided by Landsea et al., that
for storm 1872/05 there were no major changes from Partagas and
Diaz. The committee does however note major changes from Neumann et
al. The consensus was that major changes from Neumann et al. (since
this publication has had the widest distribution) should be clearly
indicated in the comments. It is expected that changes from Neumann
will be a principal focus of the review process from now on. >
Agreed. We now are clearly indicating when there are any changes from
Neumann et al. (1993) for the tracks between 1871-85 and providing
the motivation for such changes.
< 2) 1873/02. Related to the comments above, would it be better
stated that changes to Neumann et al. (1995) by Partagas and Diaz (1995)
have been accepted. This makes it clear that 1) Neumann has been
changed 2) by whom and 3) (briefly) why. >
Agreed. This is now clearly stated in the metadata file.
< 3) There was discussion about the fact that the track book includes the
tracks now under discussion (1871-1885) but the existing NHC best-track
file does not include 1871-1885. It was felt that if the existing
(Neumann) track plots were digitized, this would create a means for
comparison with Landsea et al. The concern is that, for example,
unchanged tracks will appear in the new digital file as
eye-balled points taken off the track maps by Craig Anderson.
This may introduce a noise factor into the track file. (post-
meeting, Chris Landsea objects to requirement of digitizing
existing tracks. TPC may digitize and plot a few unchanged
tracks to settle the issue.) If two digital files existed, storms,
or years, could be over-plotted to ease comparison. >
While we originally objected to following through with this point
in its entirety, we are suggesting this compromise: digitizing
those tracks from Neumann et al. (1993) that were unaltered by
Partagas and Diaz (1995b, 1996). This has now been done so that
for the years 1871-85, any tracks taken without change are now
digitized from the source (Neumann et al.) instead of contending with
added noise and possible errors due to digitizing a second hand
source (Partagas and Diaz). We believe that this has improved the
final product substantially.
< 4) There was considerable (further) discussion about the use of the
inland wind model, particularly over the Caribbean islands, e.g. 1873/05,
1874/07, 1875/03, others. The committee asked the chair to solicit
an opinion from Mark DeMaria about the use of the model. This was
accomplished. Mark's response was that there are two versions of
the model, one with a distance to land function, one without. The
one including distance to land may not be reproducible, because there are
different land files. His suggestion is that this version not be
used. Beyond this, Mark's opinion is that it better than nothing
over the islands, and does not object to this use. However,
Landsea, et al. should specify which version they are using, or may wish
to pursue the issue further with Mark. >
The Kaplan and DeMaria (1995) model - with the distance to land/ocean
coastline - 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 for 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 are chosen to occur at a
faster rate than that given from the model. (This discussion will be
incorporated into the upcoming technical memorandum to accompany the update.)
< 5) 1880/02. Central pressure of 931 mb in the comments does not
appear in the track file, although the derived wind of 130 kt does
appear. It was noted that an observation given in Partagas and Dias
is not consistent with 130 kt. >
This hurricane, as described in the metadata file for it, has an estimated
931 mb central pressure based upon a inland central pressure measurement
(after landfall) of 943 mb by work from Ho (1989). For the cases of
estimated or derived central pressures, we are following the convention of
the original best track which only puts in central pressure values if they
are directly observed. (This was the case from 1886 until 1978 in HURDAT.)
We would prefer to continue this methodology.
< 6) Are any of the tracks extended inland by extrapolation? i.e. are
tracks extended to supply points for the inland wind model? >
Yes, we did extend one hurricane's (1871/03) track out (for an additional
day) to allow for a typical decay after landfall. The lack of data
inland precludes us from knowing whether it did indeed decay faster
than usual.
< 7) Reasons should be given for departures from the northern,
southern, or other pressure wind relationships. For example,
1874/06, why reduce the winds from 67 kt derived using the subtropical
relationship to 60 kts? A number of other fairly large
changes are found from the derived values. >
Agreed. We have now attempted to clarify when there were substantial
alterations from the suggested wind-pressure relationships. After
further review, the 1874/06 wind at the time of the 987 mb central
pressure is now assigned a 70 kt best track wind. It is noted that,
in general, peripheral pressures readings provided wind speed values
that would be a minimum that could have occurred. In these cases, the
wind speeds chosen are generally either the rounded ten knot value just
larger than the minimum value or ten knots higher than that (e.g. for a
peripheral pressure of 973 mb in the southern latitudes suggesting
87 kt, either 90 or 100 kt for the best track value would be chosen).
< 8) Grammatical notes and corrections: northern seems to be the correct
usage, instead of northerly. Also some spelling errors in other
entries. 1878/05 stray "i" near end of text. >
These are now corrected.
< 9) In the same sense that the committee recommended inclusion of the
extratropical portion of a track (if sufficiently known), if there is
reasonable information suggesting a tropical depression stage, it should
also be included. The idea is to reflect current practice, to the
extent possible. >
While including the extratropical portion of the tracks of these
19th century storms was possible in some occasions because of the
common occurrence of gale force winds (and thus reports from ships
and inland stations), adding in the tropical depression stages for
these earlier years would force us to go back with a huge effort
in collecting not just gale force reports, but ALL reports of
winds within the circulation of the forming/decaying system. Such
an effort is not within the scope of this work and, frankly, our
efforts would be best placed not in trying to narrow down, for
example, when a system went from a 25 kt tropical depression to
dissipation, but rather the much more important climatologically
and societally information on when gale force and hurricane force
conditions prevailed. The current standard of having the "Tropical
depression (development) stage" and "Tropical depression (dissipation)
stage" as beginning in 1951 and 1886, respectively, will - in our
preference - remain that way.
< 10) 1873/02. Landsea, et al. should contact Alan Ruffman on
proximity to Nova Scotia. Ed Rappaport has contact information. >
Our group has already been in contact with Mr. Ruffman will be
incorporating his work into the re-analysis effort when it is
completed. This will likely have to wait for the coming year to
make these changes.
< 11) 1874/03 How was it determined that the storm was already
extratropical south of Newfoundland? >
Partagas and Diaz (1996) describe several ships with winds nearly
all beginning from the S.E. and veering to the N.W., quite consistent
with a system that has made the transition to an extratropical storm.
< 12) 1878/11 Not clear on the meaning of 'yet' in 'yet, fast
moving.' Do obs (MWR) support hurricane force winds for such an
extended period? Does the inland model apply to extratropical
systems? >
To clarify: the "storm was inland by this point suggesting lower winds
than 78 kt, yet fast moving suggesting higher winds than 78 kt".
Agreed that the winds of hurricane force were carried too far
inland. The 00 UTC winds on the 24th of October are now reduced
from 70 kt to 60 kt at latitude 42.2 N. However, the winds observed in
Philadelphia and large damage does support keeping the system at
70 kt up to 18 UTC on the 23rd at latitude 40.6 N.
Finally, the inland decay model does not apply explicitly to
extratropical storms. However, it can provide a windspeed value
that can generally be reduced downward (because of the tendency
of extratropical-transitioned storms to have a broader pressure
gradient field and thus reduced winds for a given central pressure).
< 13) 1880/10 If winds were modified 'partially due to cooler SSTs', what
were the other reasons? If nothing else if available, suggest
deleting 'partially.' >
Agreed, "partially" is removed.
< 14) In footnotes for table of U.S. landfalling hurricanes, should the
sentence beginning "Maximum winds refer" appear twice? >
This is now corrected.
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Additional question for the NHC Best-track Change Committee
(January 2000):
For the single-point storms, Charlie Neumann brought up the point that
in trying to plot these storms as currently configured, changes are
needed to existing HURDAT programs. There are two other options
available to us that would avoid forcing users to re-work their
tracking programs. The options are the following:
1) Assume the storm was stationary for 12 hours. (This would allow
existing programs to plot these storms, but may incorrectly imply
to the users that the storm was not moving.)
2) Utilizing a model similar to CLIPER to provide a reasonable 12 hour track
centered on the known single position. (Again this would allow existing
programs to work, but may impart an incorrect track where in reality
the track is not known.)
3) Keeping the single-point storms as currently configured.
Please let us know which would be preferred.
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Return to Overview