Comments on the proposed revisions to the 1926 Hurricane Season

[Reply to comments provided in bold and bracketed. May 2010 - CWL]

[Corrected.]

[Agree to round the RMW values to the nearest 5 nmi. These were primarily obtained from the Ho et al. study. While generally accurate, they do tend to be overly precise.]

[Additional observations were provided for the Caribbean from Mike Chenoweth from his trip to archive in England in 2008. Specifically for this hurricane, observations were obtained from the following locations - Antigua: July 22, 12Z, 1015 mb, SE winds (no speeds), 2.18” rainfall; July 23, 12Z, 1014mb, SE winds, 0.43” rainfall. Montserrat: July 22, 12Z, 1013 mb, ENE winds, 2.17” rainfall; July 22, 18Z, 1010 mb, NE winds; Lowest barometer – 1006 mb at July 23, 0730Z; July 23, 12Z, 1012 mb, S winds, 0.08” rainfall. Dominica – July 22, 12Z, 1015 mb, 5.4” rainfall; July 22, 18Z, 1008 mb; “Storm on night of 22^nd”; July 23, 12Z, 1013 mb, 0.36” rainfall. None of these new observations necessitate any changes to either the reanalyzed track or intensity.]

[Because of confusion of what was being provided in these sentences in the daily descriptions, they are being removed. All significant changes are discussed in the summary paragraphs.]

[See above. The “Analyzed to be a Category 2 hurricane…” statement is summarizing Boose et al.’s findings. This is now clarified.]

[Given that the summary paragraphs are to be focused upon changes to the existing HURDAT, it is preferable to keep these impact descriptions in the daily section. Both entries are now found in the July 23^rd section.]

[There are no data that indicate that the hurricane maintained 90 kt as it brushed Hispaniola. It is agreed that it likely weakened some because of land interactions before reintensifying before hitting the Bahamas. HURDAT is now revised to show weakening between 00 and 18Z on the 24^th.]

[See comment above in #2.]

[An estimate of 106 fatalities and numerous damage descriptions were obtained from Neely (2006) and included in the metadata.]

[The changes introduced just before landfall were very minor – 0.2 degrees southward and 0.4 degrees eastward at 12Z 27^th and 0.3 degrees northward and no change east-west at 12Z 28^th. The change on the 27^th was based upon consideration of all ship and Florida observations and is a compromise between the original HURDAT position and that shown in the Historical Weather Maps. The 28^th position is moved northward to better account for Merritt Island being a peripheral pressure, rather than a central pressure. This change is a compromise between what was originally in HURDAT and that proposed by Ho et al. for their landfall location.]

[This is included in the July 28^th summary section.]

[The 14 nmi RMW at Florida landfall is from the Ho et al. study. They had estimated this from the Jacksonville Weather Bureau wind observations, which are included in the excel database and printed in full from the Original Monthly Record. While likely overly precise, this hurricane does appear to have a somewhat smaller RMW than average for this central pressure (967 mb) and this landfall latitude (29.0N).]

[Mark Guishard of the Bermuda Weather Service provided observations from the Prospect Hill Observatory in Bermuda: Lowest Pressure Observed – 1000 mb on August 6^th, Highest Wind Observed – 47 kt SE on August 6^th. No times available. These observations in combination with a fortuitous set of ship observations and the lack of mention in the Bermuda hurricane history do suggest a slightly weaker system at its closes approach to Bermuda.]

[23Z is correct and is now changed in the excel database.]

[The above mentioned ship report of 975 mb peripheral pressure concurrent with hurricane force winds late on the 7th suggests winds of at least 78 kt from the north of 35N Neumann et al. pressure-wind relationship. As the system was undergoing extratropical transition, winds are boosted slightly from 75 to 80 kt at 00Z on the 8th.]

[Given that the summary paragraphs are to be focused upon changes to the existing HURDAT, it is preferable to keep these impact descriptions in the daily section.]

[Agreed to raise the winds at 00Z to 95 kt. However, the central pressure values in HURDAT – following the convention of Jarvinen et al. (1984) – are actual observed measurements or well-calibrated analyses within 2 hr of the synoptic time. Thus matching how HURDAT has been constructed for every year from 1851 to the late 1970s, the value of 955 mb should be retained in HURDAT at the 00Z entry.]

[Agreed.]

[Agreed.]

[This sentence has now been removed. Perez indicated a tropical storm impact from this system that moved northwestward just west of the western tip of Cuba. Relatively few observations are available for this system. Instead of HURDAT beginning this as an instantaneous 60 kt tropical storm at 00Z on the 22^nd, it is estimated that it became a tropical storm around 12Z on the 21^st and steadily increased in intensity. The intensity values are reduced some on the 22^nd, but not as much as originally proposed.]

[The 87 kt entry for Houma is incorrect. No anemometer measurement, nor visual estimate, from Houma was available. This is now corrected in the writeup.]

[Agreed.]

[Additional wind and pressure observations were obtained for Morgan City. Baton Rouge, Alexandria and Lafayette had only volunteer observers that measured temperature and rainfall (Climatological Data publication provided). Plotting up the available 00Z 27^th data including the 1004 mb 3 kt wind at Lake Charles does suggest a position closer to Lake Charles than previously indicated. However, given the available observations and that there appears to be a general low bias in the Lake Charles anemometer records, it appears that the center of the decaying tropical storm stayed north of the town by about 45 nm.]

[This is now clarified to say: “No change is made to the dissipation of this hurricane after 18Z on the 27^th.”]

[From Jeff Pereira of the NHC Storm Surge group: “I attached two MEOWs and flagged a few grid cells in the vicinity of Houma, LA for a CAT 3 and CAT 4 storm; moving North at 15MPH at High Tide.  The CAT 3 MEOW captured 13.8 ft above the datum (NAVD88) while the CAT 4 MEOW captured 17.0 ft above the datum (NAVD88).  According to these MEOWs, a surge value of 15' feet could very well be an accurate finding for this area considering we are using updated basin data. Please let us know if we can help out in any other way.” Thus the assignment of Category 3 at landfall in Louisiana for this hurricane appears reasonable based upon this one observation and the SLOSH MEOWs.]

[Observations provided for Prospect Hill, Bermuda from Mark Guishard indicate that no tropical storm force winds were observed there in association with this tropical cyclone. However, a minimum sea level pressure (for the month) was observed on the 13^th at 1006 mb on the 13^th and 2.02” maximum daily rainfall (for the month) was observed on the 14^th. These are now included in the writeup.]

[See below.]

[In addition to the text about the S.S. Mayaro on page 399, the Ocean Gales and Storms table on page 392 provides more information from this ship. Here we find that the ship measured a 60 kt S wind at the same time as the minimum 975 mb pressure observed by the Mayaro (at 12Z on the 14th). (Apparently, the ship did not record the pressure while in the eye.) The ship did report a maximum wind of 70 kt NE wind (Beaufort force 12) at an unknown time. While it is clear that the cyclone was still a hurricane, the smooth seas in the eye are suggestive that the 120 kt originally in HURDAT are too high. Winds are reduced to 95 kt, but this is done so with substantial uncertainty. (Because of the lack of knowledge of the ship’s movement, estimating the eye diameter and RMW are problematic.]

[The tropical storm force winds observed late on the 15^th at Cape Hatteras were likely due to the direct effect of the hurricane, but with some enhancement from the substantial cold air advection behind the front.]

[Corrected.]

[Data are sparse over the eastern subtropical North Atlantic before the 10^th of September. Certainly the system did not actually originate as a 60 kt tropical storm initially, but data preclude a more realistic assessment.]

[Observations on the 15th indicate that this system no longer maintained a closed low, as it had opened up into a trough as indicated by numerous observations. (However, the frontal structure shown in HWM on the 15^th does look suspect given the rather warm, isothermal field.)]

[Agreed. We will retain the original HURDAT of the system starting at 00Z on the 11^th.]

[Agreed. Deleted.]

[There were no tropical storm force winds or low pressures observed at either Miami or Key West Weather Bureau station. Key West, however, did measure a peak NW wind of 23 kt at 01Z on the 17^th. This is consistent with a 30 kt TC southeast of Key West.]

[Agreed, the winds on the 15^th are kept at 35 kt.]

[Brian Jarvinen graciously provided the following answer:

“Hi Chris and Jack,

Over the years I did a fair amount of research on the 1926 hurricane,

especially its impact on south Florida, which includes Miami, Lake

Okeechobee and the west coast of Florida. At one point, the storm surge

group created the Biscayne Bay basin as it existed in 1926. Of interest

is that Miami Beach was just being developed and was mostly sand.

Sea-level spoil areas existed where Dodge Island (i.e.the cruise port)

exists today. The MacArthur Causeway, with an elevation of about 5-feet

above sea level, was the most significant east-west barrier with

openings at each end. The Miami Weather Bureau Office obs initially

suggested that the hurricane passed directly over that location, but

closer inspection of the original record obtained from NCDC shows that

the strongest measured gust of 80 mph from the northeast occurred at

4:59 AM as the northwestern part of the eyewall passed over. The next

highest gust of 60 mph from the southeast was measured at 8:56 AM when

the southeastern portion of the eyewall passed over. One question, are

these gust maxima at or near the RMW in that particular quadrant? If

so, this suggests the Miami office was in the eye for about 3-4 hours.

The wind data from an anemometer on top of the Allison Hospital on Miami

Beach suggested that that site was at or near the RMW. When we made a

SLOSH model run with the track directly over the Miami office and an RMW

of 6 nm and 935 mb central pressure we did not even come close to the

observed storm tide elevations along Miami Beach, Miami and Ft.

Lauderdale. Further investigation suggested that the southern portion

of the eyewall passed south of Homestead and Florida City. I estimated

that the northern RMW passed just north of the Allison Hospital on Miami

Beach and south of Homestead or a distance of approximately 38 nm. The

half-way point is just south of the Charles Deering Estate. A SLOSH

model run was made with the track just south of the Charles Deering

(25.6N, 80.3W) with an RMW of 19 nm and a central pressure of 930 mb.

The SLOSH model results were much closer to the observed high water

marks at all locations, including the 7.4 foot value that we measured at

the Charles Deering location based upon photos of the debris line near

the house, that were taken right after the hurricane. Note: The

animation of this SLOSH run, but with the modern day elevations, is

available in the SLOSH display program that Will Shaffer's group

produces. In the modern basin, the Miami Beach barriers are higher than

in 1926 and the Port of Miami is a major east-west barrier with

elevations of 13 feet at the east end and 10 feet at the west end.

There was a really great eye-witness account by a gentleman who was

walking with his friend to view the damage and get some breakfast in

downtown Miami when the rain stopped at 7 AM. They were walking in the

eye along Biscayne Boulevard. They had difficulty standing because of

the strong wind, and had to walk at a 45 degree angle, but made it to a

dinner that was open. They did not observe any flooding at Bayfront

Park, where many ships were docked, during their walk. They would not

have been very far from the Miami Weather Bureau Office. Thus, they were

likely walking into the wind which was blowing from the east or

southeast. While eating, "the second storm started, was worse than the

first storm and was accompanied by a tidal wave, which washed boats up

two blocks from the waterfront". The Charles Deering track SLOSH

model run confirms, during the first phase of the hurricane, that while

Miami Beach was being over topped by the storm tide with a wind from the

northeast, the water inside of northern Biscayne Bay was being driven

south and not piling up near Bayfront Park. An eye-witness account at

Dinner Key also confirms this. The flooding at Biscayne Park and Dinner

Key, which pushed and floated the ships inland, occurred when the

southeastern part of the eyewall passed over and the water was driven

toward the northwest and north. Water piling up in northern Biscayne

Bay over-topped Miami Beach from the bay side and created channels

across the island. I used 930 mb because of the likely pressure

gradient between the center and the Miami office (935 mb). It could

have been a little lower. I have tried over the years to find

eye-witness accounts south of the track without success. Obviously,

wind speeds would have been considerably less on the south side and

damage also less.

As the hurricane crossed southern Florida the RMW likely expanded and

the pressure began rising such that it was 950 mb when the center passed

over the "Barracks" on the west coast at Punta Rassa. The Barracks was

an old Army Signal Corps Station. Since it took approximately 9 hours

to cross the state, the average filling rate was about 2 mb/hr. Using

the Lake Okeechobee SLOSH model and inputting the levee locations and

elevations that existed in 1926, plus the lake elevation at the time, we

were able to replicated the flooding that took place as a result of the

levee being over-topped. As you know there were many drowning

fatalities in and around Moore Haven due to the over-topping. I think

we estimated the maximum sustained wind speed on the lake at 95 to 100

mph. The Miami WFO made a copy of a DVD that we made showing an

animated loop of the flooding on Lake Okeechobee for the 1926 and 1928

hurricanes and what would happen with the current levee systems if those

hurricanes were to repeat. I believe we labeled the SLOSH input values,

at the closest point of approach, in the headers before each animation.

I don't remember exactly what we did for our estimates for SLOSH input

for Pensacola Bay, but I think your values are reasonable for the

Pensacola landfall. The track direction, RMW and 955 mb would likely

have produced the 7.2 foot storm tide that was observed at the Pensacola

tide gage.

Regards,

Brian”

Based upon Brian’s analysis, the reanalysis for landfall in SE Florida is slightly adjusted (from what was originally proposed) with the latitude moved south by 0.1 deg, the central pressure dropped by 3 mb, and no change to the winds at landfall.]

[All of the Weather Bureau stations were still using the old style 4 cup anemometer until 1 January 1928 (MWR 1927). The Miami Beach reading during the hurricane was with a new 3-cup Robinson anemometer was from Allison Hosptial.]

[Agreed, that a third landfall is required for Mississippi, as the hurricane’s center went back over the Gulf of Mexico near Mobile and moved westward between the barrier islands and the Alabama and Mississippi coast. The third landfall occurred around 16Z on the 21^st near 30.4N 89.1W with maximum winds around 60 kt. Based upon the track, intensity and size of the cyclone, Category 1 conditions were likely observed in Mississippi as well (along the barrier islands), which is new to HURDAT.]

[Agreed. No changes are now introduced for the 11^th to the 14^th.]

[Agreed. Done.]

[Agreed. This has been done for 00Z, 12Z Sep 18^th and 00Z, 12Z Sep 19^th. Other times have too few observations to be useful.]

[It is agreed that the Kaplan-DeMaria model is understating the wind for a Florida peninsula hurricane, which is why the 18Z value selected – 90 kt – is substantially higher than provided by the model. Note that we also have the 950 mb central pressure from Punta Rassa, which suggests winds of 105 kt from the north of 25N pressure-wind wind relationship. However, the maximum winds at this time were still overland. Adjustment from marine exposure to open terrain over land is a 15% reduction (Vickery et al. 2009), or down to about 89 kt. Winds at 18Z are thus estimated to be 90 kt, given that the hurricane was close to the west coast after passing the Everglades (so, higher than Kaplan and DeMaria’s model) but still overland (so, substantially less indicated by the pressure-wind relationship). It is likely that moving back over to marine exposure allowed the hurricane’s winds to increase somewhat back to 100 kt, given that the system still had 950 mb central pressure right before making oceanfall.]

[The daily section is now corrected to show the 5 min 96 kt winds. Values obtained with shorter averaging times with the old style anemometers were not considered to be reliable.]

[Corrected.]

[The combination of this ship at 08Z with NNE 70 kt wind/985 mb and a second ship at 12Z with S 20 kt wind/996 mb allows for a fairly accurate positioning of the hurricane. The first ship is estimated to have been about 30 nm from the center of the hurricane, supporting – along with the data from the second ship – a reduction in the intensity from 105 down to 90 kt.]

[Mike Chenoweth did obtain official records from archives in London, including information from the “Belize Official Record Commissary Building”. However, there was a gap in the recorded archives during 1926 and no additional data (beyond what is available at 12Z in the Historical Weather Maps) are available.]

[A major southward change to the track of the cyclone is introduced on the 3^rd because of easterly ship observations south of the original HURDAT position. Data are somewhat inconclusive, however, whether a closed low was actually present on this date.]

[Intensification to tropical storm is delayed by 18 hours to 00Z on the 15th, based upon a very weak broad low on the 14^th and moderately deeper (~2 mb) pressures recorded in the western Caribbean on the 15th.]

[While the system was slow moving up until the 18^th, the movement of the hurricane early in the 19^th had accelerated to 10 kt, which is only slightly below climatology. The environmental pressures though remained low (outer closed isobar of 1009 mb). Thus it is agreed that the winds need to be boosted. Winds are now assessed in HURDAT as 115 kt at 06Z.]

[The 933 mb report mentioned in the Monthly Weather Review writeup for Havana – which was not in the eye of the hurricane - was not considered to be reliable as it was a momentary value associated with “wind-pumping” of the barometer.]

[Agreed to keep winds provided in original revision: 130 kt Category 4 striking Cuba.]

[There is significant uncertainty about the evolution of this system. In reassessing the case, we continue to analyze the cyclone as remaining tropical on the 24^th, but at a farther south position. This would keep the main baroclinic zone north of the cyclone on that date. As the committee points out, the system clearly appears to be a tropical cyclone later on (on the 25^th and 26^th). Introducing a short-lived extratropical structure would be quite unlikely climatologically.]

[This has been corrected.]

[Agreed. Ship report deleted.]

[Agreed. Track map so altered.]

[The lowest believable pressure was a ship on the 13^th with 1007 mb and 20 kt E winds, which may also support minimal tropical storm intensity. This has been added into the summary section.]

[A search was conducted, but we were unable to find additional high winds or low pressures. It certainly is plausible that this was indeed a tropical storm.]