Comments on the proposed revisions to the 1928 Hurricane Season
From the Best Track Supervisory Committee
[Replies to the Committee are indented and boldface. – Chris Landsea – June 2010]
1. Many of the proposed changes in the text do not seem to match what is on the provided track charts. Do we have the correct maps, especially the one showing the track changes?
[I have double-checked to insure that the documents and track changes are consistent.]
2. Please make sure the document is spell-checked. A few examples of problems are noted below.
[The document has now been thoroughly spell-checked.]
Storm #1, 1928:
1. The October 1928 Monthly Weather Review has an article on the two August 1928 hurricanes starting on page 411. Material from this article seems to be included in the discussion for storm #2, but not for storm #1. For storm #1, this article includes detailed pressure data from the eye passage over Fort Pierce, Florida, where the pressure fell to 28.84 in (976.7 mb) about 4 AM local time on 8 August. It also includes a report of a 28.70 in (972 mb) pressure about 30 miles southeast of Jupiter, Florida at 7 PM local time 7 August. This article is likely the source that Barnes used for his landfall central pressure. Please re-write the metadata summary, as well as the daily sections of 7-8 August, to reference this article, and make any best track changes as necessary.
[Thank you for providing this, as the information from the article was not utilized originally. These observations have been added into the daily write-ups and incorporated into the summary. The HURDAT track on the 7th and 8th has been slightly modified to take into account these Fort Pierce observations.]
2. Please justify the removal of 3-4 August from HURDAT. The committee notes an observation of north 20 mph at Santo Domingo on 4 August that (if not orographic) would be consistent with some kind of system approaching the station.
[Three different scenarios were available for the genesis of storm #1: a) The tropical cyclone formed north of Hispaniola on the 5th (MWR – page 336 and HWM), b) The tropical cyclone formed north of the Leeward Islands on the 3rd (MWR – page 441), or c) The tropical cyclone formed near Trinidad and Barbados on the 3rd (HURDAT, McAdie et al.). Examining the data on the 3rd showed that option “b” (forming north of the Leeward Islands) was not likely due to easterly winds throughout the Lesser Antilles and eastern Caribbean. Likewise, data on the 4th indicates that option “c” (formed near Trinidad) was unlikely as well, as moderate easterlies were present in the southern Caribbean on the 4th south of the supposed cyclone’s center. The heavy rain in Puerto Rico and in Dominican Republic along with the 15 kt north wind at Santo Domingo on the 5th do suggest that a strong wave was approaching, but the observations elsewhere in the Caribbean show that no closed low was yet present. Genesis for storm #1 is thus begun as a tropical storm at 12 UTC on August 5th, two and a half days later that suggested in original HURDAT – option “a” looks most realistic.]
3. The 35 kt southwest wind at 12Z 5 August also suggests an earlier genesis, and quite possibly a stronger intensity at 12Z. Note that both 35 kt observations on 5 August have a westerly component. This suggests stronger winds existed elsewhere. Also, both the old and new HURDAT positions cannot be easily reconciled with the positions of the ship reports, which suggest the center was to the northwest of the HURDAT positions.
4. If the 3-4 August data are retained for HURDAT, the intensity at 12Z 5 August may need to be adjusted anyway. The current temporary reduction from 40 kt to 35 kt looks strange.
[Unfortunately, the 12Z 5 August SW 35 kt report was erroneously entered in the metadata. As correctly shown in the spreadsheet database, the observation is actually taken at 21Z. This report – from COADS – may be the same that was reported at 20Z obtained from MWR. It is agreed that the cyclone’s positions did need to be adjusted to the northwest on the 5th and early on the 6th. However, with the existing winds of 40 kt at 18Z on the 5th and 45 kt at 00Z on the 6th, the two (or possibly one) 35 kt reports did not appear to justify indicating a stronger TC at those times.]
5. Are there observations available from Tallahassee and Cedar Key on 9 August that would help refine the track and intensity?
[In 1928, only cooperative observers were in Tallahassee and Cedar Key with no anemometers and no barometers.]
6. What is the basis for the 10 n mi radius of maximum wind in Florida?
[The 10 nmi RMW value is an estimate based upon both the duration of the lull experienced at Fort Pierce (~ 75 minutes with 180 degree wind shift from NE to SW) relative to the speed of the hurricane (~5 kt) as well as the small radius of outer closed isobar (~140 nm). This 10 nmi RMW could be off by as much as 50%. However, even if the RMW size were 50% larger, this would still suggest a small system with a commensurate boost to the winds from the pressure-wind relationship.]
7. There are several typographical errors for this storm. On 8 August, replace “th 10th” with “the 10th” and “souther parts of Georgia” with “southern parts of Georgia”. In the second summary paragraph, “90 kt is chosen for retained for HURDAT” needs to be re-written.
Storm #2, 1928:
1. All mention of this system suggests that it had a small core when it moved through the Florida Keys on 12-13 August, so observations outside it (like Key West’s) do not help with the intensity analysis. How far were the ship reports on 12-13 August from the estimated track? (A quick estimate suggests the two 45-kt reports were 60 n mi or more from the center, while the 70 kt report was much closer.) Have local sources been checked to see if more details are available?
[It is agreed that the Key West Weather Bureau observations were not of much assistance with the intensity of this system because of the small size. The “lack of significant impact in the Keys” was referring to the statement in MWR about “gales were also experienced along the coast and over the Florida Keys”. This is now clarified. The two 45 kt ship reports were about 70 and 60 nm from the center of the cyclone, while the 70 kt report is about 30 nm (but is more uncertain as no time was given for this observation). We’ve boosted HURDAT from 45 and 50 kt at 00 and 06 UC on the 13th to 60 kt at both time periods and it is possible that this was a minimal hurricane to impact the Florida Keys. This is now so stated in the metadata writeup.]
2. The metadata summary starts with “Genesis for storm 2 began at 12 UTC on August 7th near the Lesser Antilles as a tropical depression rather than a tropical storm.” This does not match either the new or old HURDAT entries. Please correct whichever of these is wrong.
[The text was correct. The HURDAT entries now reflect this.]
3. The metadata summary states “Wind are reduced at landfall in Cuba from a 70 kt hurricane…”. This is confusing, as the intensity was actually increased for the period 10-13 August when the system was near Cuba. Please re-phrase this.
[This is now clarified: “Winds are boosted slightly on the 10th (from 70 to 80 kt, which is the new peak intensity for the cyclone) to account for description of impact in Haiti, though the intensity is highly uncertain. Hurricane intensity for this cyclone was also observed from the U.S.S. Arkansas observations, while south of Cuba early on the 11th and hurricane intensity is maintained until 18 UTC on the 11th, 12 hours longer than previously indicated in HURDAT. Because of the track change introduced, landfall in central Cuba is analyzed to occur around 12 UTC on the 12th as a 50 kt tropical storm rather than just before 12 UTC on the 11th in eastern Cuba as a 70 kt hurricane. This is consistent with the analysis of Perez et al., the Key West OMR, and MWR.”]
4. Perez is mentioned as a data source in the metadata summary, yet there is no reference to him in the daily sections. What data does Perez have on this system? Please list it in the appropriate daily section.
[Perez et al. in the catalogue of Cuban tropical storms and hurricanes indicated that this system made landfall in central Cuba as a tropical storm, rather than a Category 1 hurricane in eastern Cuba. Perez provided a suggested revised track, which was weighted heavily in the reanalysis.]
Storm 3#, 1928:
1. The committee concurs with starting this system as a depression. However, it looks a little strange that the system took more than 24 hours thereafter to become a tropical storm. Do the “numerous” ship reports for 2 September definitively rule out tropical storm strength on that day? If so, please state the key reports in the daily section. If not, it is recommended that the system be made a tropical storm sometime that day.
[Agreed that the system likely became a tropical storm on the 2nd.]
2. Given how close the cyclone passed to Jamaica and the Cayman Islands, is any information available from those places that would help refine the track and intensity?
[The only observations available from Jamaica were the Kingston 12 UTC measurements shown in the Historical Weather Map series. No observations were available from the Cayman Islands.]
3. The 5 September daily entry begins “HWM analyzed an inverted trough near 15.5N 64.5W”, which looks like an error. Please correct this.
[This has been corrected to state: “HWM analyzed an inverted trough along 85W.”]
Storm #4, 1928:
1. The 50 kt observation on 7 September is a south wind (presumably east of the center), which suggest the possibility that the winds were stronger north of the center. Please consider revising the intensities for this and subsequent days.
2. Does the available data justify the small kinks shown in the track on 12 and 14 September (especially the one that brought the center close to the north coast of Hispaniola?) If not, please consider smoothing the kinks out.
[Observations do confirm a modest turn of the hurricane toward the west-northwest beginning around 00 UTC 13th. Agreed to smooth the kink around 12 UTC on the 14th, as there is limited inner core data.]
3. Is the discussion of the storm rainfall in the 14 September daily description relevant? If not, please remove it. The same applies to the other descriptions of rain and flooding caused by this system. They are relevant to the storm impacts, but they are not relevant to the best track or revising it.
[Agreed to remove.]
4. In the metadata summary, the sea level pressure in the “calm” eye at Guayama was 936 mb, yet in the next sentence there is an assumption that Guayama may not have recorded the central pressure. How calm was the eye at Guayama? The same issue is apparent with the ship report used for the central pressure on 15 September. The account of the wind shift from north-northeast to south-southeast suggests that the ship was not in the geometrical center of the eye, yet the 941 mb is accepted as the central pressure. How calm was the eye over the ship? Please resolve this apparent contradiction.
[Upon further investigation, it appears likely that Guayama was not in the calm eye. Guayama did report a minimum sea level pressure of 936 mb, but Perez’ Puerto Rico hurricane history documentation lists only Aibonito, Cayey, and Adjuntas as experiencing the “central calm”. Thus it appears likely that Guayama did not record the central pressure. This is now clarified.
For the ship observation on the 15th, MWR had the following information: “The German steamer August Leonhardt … was hove to in latitude 23 deg 10 min W., longitude 74 deg 10 min W, when the center of the hurricane passed over it about 3 p. m. of the 15th, the lowest barometer reading being 27.80 inches. Just previous to the arrival of the center the wind was north-northeast force 12 (and more). After the barometer had remained stationary and the wind had calmed down for a short time, the hurricane started again at 3:10 p. m., this time blowing from the south-southeast”. Thus while the ship was inside the radius of maximum wind, it may not have measured the central pressure – as suggested by the peak winds only shifting from NNE to SSE. The original 135 kt in HURDAT is retained on the 15th.]
5. The metadata summary section describes the difficulties in resolving the intensity of the hurricane over Puerto Rico. How certain is the 25 n mi radius of maximum winds? The description of the eye passage over Guadeloupe suggests the eye lasted about an hour, which from the forward speed suggests the eye was about 13 n mi in diameter. This would produce a smaller RMW than 25 n mi. Is there any other potential evidence that the hurricane underwent an eyewall replacement between Guadeloupe and Puerto Rico?
[An RMW value of 25 nm is quite well established between the timing of the peak estimated winds at the San Juan Weather Bureau Office at 19 UTC and landfall of the hurricane’s center in southeast Puerto Rico just an hour earlier. The change in character of the hurricane between its passage over Guadeloupe and over Puerto Rico is consistent with an eyewall replacement.]
6. The committee has some issues with the intensity over the Bahamas. First, the 15 September daily section mentions that the center passed about 9 miles south of Grand Turk. How far is Grand Turk from the proposed track? If it was 9 miles from the center, then Grand Turk was likely inside the RMW and the 965 mb pressure suggests a rise in the central pressure after the hurricane left Puerto Rico. If Grand Turk was not inside the RMW, the central pressure could have been considerably lower than 965 mb.
[The passage of the hurricane 9 miles south of Grand Turk estimated by the observer was likely incorrect. Instead, our distance at the time of the minimum pressure was about 30 nm. Thus the 965 mb pressure value is likely a peripheral one, not a central pressure. As discussed earlier, also on the 15th, a ship had a peripheral pressure of 941 mb, so the 965 mb is likely much larger than the true central pressure.]
Second, the same arguments can be made for the Nassau observation. How far was it from the center and was it inside the RWM? If it was inside the RMW, then the 951 mb pressure and the estimated winds suggest a pressure close to the 941 mb ship report. If it was not inside the RMW, then the central pressure could have been considerably lower. Note that the Monthly Weather Review description of the hurricane at Nassau does not include anything that looks like an eye passage.
Please clarify these issues if possible and adjust the intensity if necessary.
[Reviewing the account of the hurricane in Nassau does suggest a close bypass (winds NE-NW-SW), but without any mention of a lull. Thus the central pressure could have been close to the 929 mb central pressure at landfall in Florida. Winds of 135 kt – high end Category 4 - retained in HURDAT during passage across the Bahamas.]
7. Please state the basis for the 28 n mi RMW at landfall in Florida. Various accounts of this hurricane suggest the eye was anywhere from 25 to 40 miles wide, with only the larger of those supporting this RMW. The Monthly Weather Review description of the eye passage at Canal Point indicates that the strongest winds were separated from the minimum pressure by 45-60 minutes for a storm moving about 13 kt. While not conclusive, this also suggests a smaller RMW. Given the importance of the RMW in determining the landfall intensity, there needs to be more information about how the 28 n mi figure was arrived at.
[The 28 nm RMW is provided by the Ho et al. study. Reviewing all of the data available, it is agreed that the 28 nm is somewhat too high. (Of course, unless a particular location that experiences the eye has the exact center go over it, the implied radius will be a smaller distance than the true radius.) Thus our best analysis is that the RMW was 25 nm, plus/minus 5 nm.]
If it turns out that the RMW is smaller than 28 n mi, the committee suggests the intensity for the Florida landfall remain 130 kt.
[Given that the hurricane’s RMW is still fairly large compared to climatology, the environmental pressures were low, and the translational velocity was near average, this would still call for a modest reduction from the pressure-wind relationship. Now that the 15th and 16th intensity is retained at 135 kt, we are constrained to not use the intensifying subset of the Brown et al. pressure-wind relationship. Taking an average of the pressure-wind relationship for south and north of 25N gives 128 kt. Accounting for the large size and low environmental pressures would indicate a maximum sustained wind of about 125 kt at landfall in southeast Florida, making this a US Category 4 hurricane. This is a 5 kt reduction from what was in HURDAT originally.]
8. Is there any time series of data available from Jacksonville to show where the center passed in relation to that station?
[Yes, this was already shown in the spreadsheet database and plotted previously as a time series on the 18th of September map. Based upon this, the center passed just east of Jacksonville.]
9. While the center may have been just offshore, the new proposed track is close enough to the Georgia coast so that the left side RMW may have been onshore. Given that, is removing the GA1 status justified? Observations from Jacksonville could help resolve this as well.
[Agreed to keep Georgia in as a Category 1 hurricane impact from this system.]
10. In the 18 September daily section, is “26 N and 979 mb at 13 UTC at Savannah” missing a kt abbreviation?
11. What is the relevance of the data from the Kaplan/DeMaria inland decay model for 19 September in the last paragraph of the metadata summary? The revised HURDAT does not seem to follow it, and its use during extratropical transition appears dubious. Please clarify why it is there or remove it.
[Agreed to remove the run of the Kaplan/DeMaria model for the 19th of September.]
Storm #5, 1928:
1. On 10 September, the 0000 UTC position is not consistent with the ship report of 60-kt winds from the east-southeast, as the ship at 53.5W suggests a position west of the best track longitude of 52.5W. Some blending of the old HURDAT track and the proposed new HURDAT track might be in order.
[Agreed. Position shifted between the two tracks.]
2. If the system was absorbed by a larger baroclinic low rather than continuing toward Iceland as stated in the Monthly Weather Review, please include the position of the larger low in the daily sections of 12 or 13 September.
Storm #6, 1928:
1. The 45-kt ship report at 2200 UTC 10 October appears to be on the order of 100 n mi from the proposed track. Based on this, is it possible the system was stronger than 50-55 kt at that time, or does other data preclude this?
[Agreed that the system was likely stronger. Winds adjusted upward an additional 5 kt.]
2. The daily discussion for 15 October suggests a center was present at 1200 UTC that day, but there is no HURDAT position given at that time. What is the reason for having no position in HURDAT?
[The daily discussion was in error, as a closed low was not observed with this system by 12 UTC. The system is judged to have been absorbed the larger extratropical cyclone after 06 UTC on the 15th.]
3. The 11 October daily section starts with “HWM indicates a warm front extending east-southeast from 35N 44W to 29N 34W”. Is this relevant to the analysis of the tropical cyclone? If not, please modify or remove this.
System #1: The committee concurs with leaving the system out of HURDAT.
Other comments: The Monthly Weather Review indicates several systems that should be investigated to see if they were tropical cyclones, or for possible inclusion in the Additional notes section.
1. The May Monthly Weather Review and its chart of Centers of Cyclones shows a low pressure system that formed over the Bahamas on 4 May, then followed a recurving track offshore of the U. S. east coast. The HWM shows this as a frontal low and there are no accounts of strong winds with it. However, given the unusually far south origin the system should be checked to see if developed subtropical cyclone characteristics.
The May MWR Tracks of Centers of Cyclones and the Historical Weather Maps show a low developing south of 25N over the Bahamas in early May. This system deepened slightly as it moved north-northwest from the 5th to the 7th and then north-northeast until passing through the Canadian Maritime provinces on the 12th. No gale force winds were observed and the system was extratropical throughout its lifetime with significant frontal boundaries present even while south of 35N. Because of the low latitude formation of this cyclone, this system is included in Additional Notes, but likely never had tropical (or subtropical) characteristics during any portion of its lifetime.
2. The September Monthly Weather Review reported a “depression” over the western Gulf of Mexico (page 383) with no winds above Force 6. The HWM show an eastward moving frontal low. This is likely not a tropical storm but it should be examined.
The September MWR Tracks of Centers of Cyclones and the Historical Weather Map series show a low that developed in late September in the Gulf of Mexico. It moved to the east-northeast from late on the 22nd to the 24th. The system did not have any observed gale force winds, though it did have well-defined frontal boundaries during its brief lifetime. Because of its occurrence during the peak of the hurricane season, the system is included here in Additional Notes, even though it was very likely extratropical throughout its short lifetime.
Additional comments received in December 2010:
We've discovered what appears to be a major problem with theearly analysis of storm #1 of 1928. The analyses for 3-4 Augustin the binder do not match the Historical Weather Maps for thosedays. Indeed, it looks like the binder has the scanned blow-upsfrom August 3-4 1927! The HWM for 3 August 1928 shows San Juanwith west wind 15 kt (or 15 mph), which is not on the chart inthe binder. The analyses for 5 August on look OK.
The maps were indeed scanned in erroneously. The correct maps have now been utilized and COADS observations included. The revised analysis now shows that a closed circulation was present on the 3rd (with SW winds at San Juan), but the observations are ambiguous on the 4th. The data are fairly conclusive, however, that a closed circulation was not present in the eastern Caribbean on the 4th (like what was shown in the original HURDAT). The revised HURDAT positions have the system forming north of the Leeward Islands on the 3rd, most similar to what was described in MWR (page 441). Thus major track changes are introduced on the 3rd and 4th. No changes are made to the intensity on these dates as the data – while not showing any observed gales – are ambiguous as to the intensity.