Hurricane Andrew: Category 5, Ten Years Late
Hurricane Andrew crossed South Florida at 5:05 AM EDT on 24 August 1992 and was rated a Category 4 for ten years. In August 2002 the National Hurricane Center raised it to a Category 5, 145 to 165 mph, and the number that moved it had been measured that night: a 162-knot wind at 10,000 feet, an hour before landfall. Every surface instrument that could have confirmed it failed or stopped before the peak, and the Miami radar was blown off the roof. We measured the eye ourselves on the Melbourne Doppler radar, the one recording of that night that survives.
Chapters
- 0:00 South Florida from the air
- 0:17 The radar that kept recording
- 0:49 From storm to Category 5 in two days
- 1:30 Fowey Rocks stops
- 2:04 The Miami radar's last image
- 2:50 The needle that pegged
- 4:12 No official measurement
- 5:03 162 knots at 10,000 feet
- 5:45 The 77% rule
- 6:27 Our measurement: the eye from Melbourne
- 8:43 The dropwindsonde
- 9:43 The 2002 committee
- 11:40 What did not change
- 12:41 The toll
- 13:20 Louisiana, 26 August
- 14:20 What the record supports
- 15:16 Next: San Francisco, 1906
Transcript
South Florida from the air 0:00
Hurricane Andrew was a Category Five, and it took ten years to say so. It crossed South Florida at 5:05 on the morning of August 24th, 1992, and the official record called it a Category Four. In 2002, the Hurricane Center raised it.
The radar that kept recording 0:17
We read the reanalysis, and then we measured the eye ourselves, on the one radar whose recording of that night survives. This is it. Melbourne, Florida, about a hundred and eighty miles to the north, at 5:01 in the morning.
The wind that made Andrew a Category Five was measured that night, an hour before landfall. The number never changed. So this episode follows that number: which instruments failed before the peak, what one aircraft recorded, and the rule that took ten years to change.
From storm to Category 5 in two days 0:49
Andrew had crossed the Atlantic as a weak storm for most of a week. Then, in less than two days, it went from a tropical storm to one of the strongest hurricanes on record.
In the best track the Hurricane Center keeps today, Andrew peaks on the afternoon of August 23rd, near the Bahamas: one hundred and fifty knots, and nine hundred and twenty-two millibars. That evening it crossed the northern Bahamas, and then it turned its eye toward Miami.
This is the satellite at 5 AM Eastern time on the 24th. The eye is about the size of a city, and it is sitting on the coast south of Miami. Everything that tried to measure it on the ground was inside this circle.
Fowey Rocks stops 1:30
The first instrument in the way was an automated station on Fowey Rocks, a lighthouse off Key Biscayne. At 4 AM it reported a two-minute wind of a hundred and twenty-three knots, with a gust to one hundred and forty-seven.
Then it stopped transmitting. When people could reach it, they found the mast that held the anemometer bent ninety degrees from vertical. It had stopped about twenty minutes before the peak of the northern eyewall would likely have reached it. Its last pressure was nine hundred and sixty-eight millibars, and the report estimates it probably fell about twenty more.
The Miami radar's last image 2:04
The Hurricane Center itself was in Miami, with a weather radar on the roof, and the northern edge of the eyewall would pass over it. This is the last image it took, at 4:35 in the morning. The eye is over Elliott Key, and the ring of heaviest rain is almost at the mainland.
Sometime after that scan, the wind blew the radar off the roof. The report says only that it was the last image taken before the radar was lost. It does not give the minute.
The anemometer at the same building recorded a gust of a hundred and forty-two knots at 4:50, about a hundred and sixty-three miles an hour. After that measurement, the equipment stopped working. The report marks the number with a note: a more extreme value may have occurred.
The needle that pegged 2:50
At Tamiami Airport, an observer watched the wind dial. At about 4:45 the needle went a little past the highest mark, a hundred knots, and stayed there. He estimated about a hundred and ten. After three to five minutes it fell back to zero, because the anemometer had failed. When dials of that type were tested later, the needles pegged at about a hundred and five and a hundred and eight knots.
Homestead Air Force Base and Tamiami Airport had both stopped their routine weather observations before the direct hit. On a sailboat in the southern eyewall, the readout showed ninety-nine knots for at least thirteen minutes. Ninety-nine was the highest number it could display.
Miami International Airport, the closest regular station that kept reporting, was outside the eyewall by about five nautical miles. It never felt the strongest wind.
In Perrine, a man named Randy Fairbank had a home anemometer. He read a gust of one hundred and eighty-four knots, about two hundred and twelve miles an hour, moments before part of a wall failed. The anemometer was destroyed.
That model of instrument was later tested in a wind tunnel. The test suggested the real gust was lower, probably near a hundred and fifty-four knots. And homes about two miles to the south had worse damage than his.
No official measurement 4:12
So the report says something unusual. Neither central pressure nor maximum sustained wind, the two standard measures of a hurricane, was observed at an official station near Andrew's landfall in Florida. The instruments that were in the right place failed before the strongest wind arrived.
The pressure came from people. Observers read their own barometers through the night, in several cases until their instruments and even their homes were damaged. Several of the barometers with the lowest readings were later tested in a pressure chamber by NOAA's Aircraft Operations Center.
From those readings the report estimates the pressure at landfall: nine hundred and twenty-two millibars, near Homestead Air Force Base, at 5:05 in the morning. In 1992, that was the third lowest of the century for a hurricane at landfall in the United States.
162 knots at 10,000 feet 5:03
What remained was the air. The Air Force Reserve's 815th Weather Reconnaissance Squadron, from Keesler Air Force Base in Mississippi, flew into Andrew that night, and a NOAA aircraft collected more data.
At 4:10 in the morning, about an hour before landfall, the aircraft was at the seven-hundred-millibar level, about ten thousand feet. In the eyewall, about ten nautical miles north of the center, it measured a ten-second wind of a hundred and sixty-two knots.
A single spot reading on the same pass was a hundred and seventy. This is the number at the center of this story. It was measured in 1992, and it was never revised.
The 77% rule 5:45
But a hurricane is rated by its wind at the surface, not at ten thousand feet. In 1992, forecasters converted flight-level wind to surface wind with a factor, and for Andrew the factor was seventy-seven percent.
Seventy-seven percent of a hundred and sixty-two is a hundred and twenty-five knots. That is about a hundred and forty-five miles an hour, a strong Category Four. A Category Five begins at a hundred and thirty-six knots.
On the night itself, the Hurricane Center's operational estimate was a little lower still: a hundred and twenty knots. The best track written after the season settled on a hundred and twenty-five, and for ten years, that was Andrew.
Our measurement: the eye from Melbourne 6:27
Now the radar we started with. In 1992 the new Doppler network was just being installed, and Melbourne had one of the first. It recorded Andrew for more than twenty-four hours, from a long way off. Its staff first saw a closed eyewall at about 10 PM on the 23rd, with the eye two hundred and fifteen nautical miles to the southeast.
We took every scan it saved between 11 PM and 9 the next morning, Eastern time, and found the center of the eye in each one, as the clear circle inside the ring of rain. On the scan that began at 5:01, the center sits right at the coast, by Homestead.
A hundred and one scans are in the archive for those ten hours, and one of them cannot be read. The file for the scan that began at 4:37, two minutes after the Miami radar's last image, is cut short.
But look at how high it is looking. At two hundred and ninety-four kilometers, the center of the radar's lowest beam is about seven and a half kilometers above the sea, and almost five kilometers wide. It is seeing the top of the eyewall, not the wind at the ground.
And the Doppler part, the wind, only reached a hundred and twenty-four nautical miles. Only Andrew's outer rainbands were inside that range. So the radar that kept recording could follow the eye, and could not measure its wind.
It still mattered. After the Miami radar was lost, the forecast office in Melbourne wrote, its images filled a critical gap as the eye crossed toward the southwest coast. It was the first hurricane a Weather Service Doppler radar observed in operation.
Here is the eye as we measured it, scan by scan, from 11 PM to 9 in the morning. Between 3 and 6 AM, our centers stay within nine kilometers of the Hurricane Center's best track, and within five around landfall. The radar confirms where the eye went. It cannot say how hard the wind blew.
The same scans give the ring. From 4 to 6 in the morning, the heaviest echo sits about twenty kilometers from the center, scan after scan. The best track puts Andrew's radius of maximum wind at landfall at ten nautical miles, about eighteen and a half kilometers.
The dropwindsonde 8:43
The answer to that came years later, from a different instrument. In 1997, hurricane hunters began dropping GPS dropwindsondes, small probes that fall through the storm and report the wind all the way down to the sea.
For the first time, the wind could be measured from the aircraft's level down to the surface, inside real eyewalls. James Franklin of the Hurricane Center and his colleagues put those profiles together.
Their result: in the eyewall, the mean surface wind was between eighty-six and ninety percent of the wind at seven hundred millibars, and the Hurricane Center adopted ninety. Not seventy-seven. And of the seventeen hurricanes they examined, none had an eyewall factor as low as the one used for Andrew.
So apply the new factor to the old number. Ninety percent of a hundred and sixty-two is a hundred and forty-six knots. The other readings on the same pass, at 4:09 and 4:11, come out at a hundred and forty and a hundred and forty-one.
The 2002 committee 9:43
In August 2002, the Hurricane Center's Best Track Change Committee met to review Andrew. Herbert Saffir, the engineer who co-created the scale, sat in as an observer. Chris Landsea's submission proposed a hundred and forty knots.
On one point the room agreed unanimously: Andrew had been underestimated over open water. The long discussion was about whether those winds reached the coastline and the land.
The committee also looked at the Miami radar's own last hours. Before it was destroyed, the motion of features inside the eyewall, tracked from scan to scan, pointed to about a hundred and forty-eight knots at the surface.
And the damage surveys? The submission's conclusion was that the damage could be read either way: either Category Four or Five is consistent. The paper that followed called that evidence of secondary importance.
One survey read the damage as F3 conditions, gusts of a hundred and thirty-five to a hundred and seventy-five knots. Converted to sustained wind, that becomes a hundred and five to a hundred and thirty-five. Category Three or Four.
The committee did not adopt the proposal of a hundred and forty. It set Andrew's landfall at a hundred and forty-five knots, with a likely range of a hundred and thirty-six to a hundred and fifty-five. Category Five. In the paper that followed, the authors wrote that it is quite unlikely Andrew was a Category Four at landfall, as was originally thought.
One presenter suggested the wind may have weakened by five to ten percent at the coastline. The committee considered that too speculative to apply.
On August 21st, 2002, NOAA announced it. After ten years, a hundred and forty-five miles an hour became a hundred and sixty-five. Andrew became only the third Category Five on record to strike the continental United States, after 1935 and Camille.
What did not change 11:40
Notice what did not change. The position of the eye, hour by hour, was not adjusted. The pressure at landfall, nine hundred and twenty-two millibars, was not adjusted. The flight-level wind was not adjusted.
What changed was the rule that turns a measurement at ten thousand feet into a wind at the ground. The rule was built from instruments that did not exist in 1992.
There is a limit in the decision itself. The committee wrote that the winds at specific places over land in Miami-Dade County are unknown. Category Five conditions likely occurred on land only in a small area by the coast, near Cutler Ridge. There, the paper estimates, peak gusts were likely on the order of a hundred and sixty to a hundred and seventy knots.
Most of the county south of Kendall Drive got Category Three or Four conditions, the reanalysis says. So the Category Five is the hurricane's rating at landfall, not a measurement of every street.
The toll 12:41
On those streets, the toll does not depend on the category. The Hurricane Center's report counts twenty-six deaths directly from Andrew, and sixty-five in all. About twenty-five thousand homes destroyed, and more than a hundred thousand damaged.
In Homestead, eleven hundred and sixty-seven of eleven hundred and seventy-six mobile homes were destroyed. Damage was estimated at twenty-five billion dollars, at the time the most expensive natural disaster in American history.
And the water. The storm tide reached sixteen point nine feet on the western shore of Biscayne Bay, at the Burger King headquarters, the highest in the report.
Louisiana, 26 August 13:20
Andrew was not finished. It crossed the Everglades in about four hours, regained strength over the Gulf, and on August 26th struck a lightly populated stretch of the Louisiana coast, west of Morgan City, as a Category Three.
This is the Coast Guard over south Louisiana that day. The reanalysis also touched this landfall, and in the other direction: it lowered it by five knots, to one hundred.
Sixteen years later, Hurricane Michael was also raised to a Category Five after the season, and we told that story in episode four. In wind at landfall, the Hurricane Center ties Michael for fourth with the San Felipe hurricane of 1928 in Puerto Rico, behind 1935, Camille, and Andrew.
Michael's pressure at landfall, nine hundred and nineteen millibars, also moved Andrew down a place. By pressure, Andrew is now fourth, behind 1935, Camille, and Michael.
What the record supports 14:20
Here is what the record supports. The strongest wind in Andrew's northern eyewall was measured, at ten thousand feet, an hour before landfall. Every instrument at the surface that could have confirmed it failed or stopped before the peak arrived.
Andrew went from Four to Five without a single new measurement from 1992. It changed because later hurricanes, measured with better instruments, showed what the old rule was missing.
And here is what the record cannot say. Exactly how fast the wind was at any one house. Exactly when the Miami radar left the roof. And whether the strongest gust was ever measured at all.
So the question stays open. If you have the Miami radar's last scans after 4:35, or the Fowey Rocks data after 4 AM, we would like to read them. The record of Andrew's peak is one aircraft pass, and a rule written later.
Next: San Francisco, 1906 15:16
Next, an earthquake read from the ground itself. San Francisco, April 1906. From the way the ground moved along the San Andreas fault, a geologist from Johns Hopkins worked out the theory of elastic rebound. That is the next episode. San Francisco.
Description and sources
At 0810 UTC an Air Force Reserve aircraft (815th Weather Reconnaissance Squadron, Keesler AFB) measured a 10-second wind of 162 kt at 700 mb in Andrew's northern eyewall. In 1992 that was reduced to the surface with a 77% factor: 125 kt, Category 4. GPS dropwindsondes, used from 1997, showed eyewall surface winds closer to 90% of the 700-mb wind (Franklin et al. 2003); 90% of 162 is 146 kt, and the Best Track Change Committee set the landfall at 145 kt in August 2002 (Landsea et al. 2004). The position and the 922 mb pressure were not changed. On the ground, Fowey Rocks stopped after 0800 UTC with its mast bent 90 degrees, the NHC anemometer stopped after a 142-kt gust at 0850, Tamiami's needle pegged near 110 kt and fell to zero, and the NHC's WSR-57 radar was blown off the roof after its last image at 0835. Our measurement on the KMLB Level II archive: the eye's centre, scan by scan, from 294 km away, with the lowest beam about 7.6 km above the sea and no Doppler velocity at that range.
PRIMARY SOURCES
- NOAA 02-107, After 10 Years, Hurricane Andrew Gains Strength (21 August 2002). https://www.nhc.noaa.gov/news/NOAA_pr_8-21-02.html
- Landsea, C. W., et al. (2004), A Reanalysis of Hurricane Andrew's Intensity, Bulletin of the American Meteorological Society 85(11), 1699-1712. https://www.nhc.noaa.gov/pdf/04landsea.pdf
- Rappaport, E. N., Preliminary Report, Hurricane Andrew, 16-28 August 1992 (NHC, updated 1993) and Addendum (2005). https://www.nhc.noaa.gov/1992andrew.html
- Landsea, C. W. (2002), The Re-analysis of Hurricane Andrew (1992), submission to the NHC Best Track Change Committee. https://www.aoml.noaa.gov/hrd/hurdat/clanalysis.html
- Hagemeyer, B. C. (1992), A Look at Hurricane Andrew Through the Eyes of a WSR-88D, NWS Technical Attachment 92-29. https://www.weather.gov/media/wrh/online_publications/TAs/ta9229.pdf
- NWS Melbourne, Hurricane Andrew. https://weather.gov/mlb/andrew
- Beven, J. L., Berg, R., Hagen, A. (2019), Tropical Cyclone Report, Hurricane Michael (AL142018). https://www.nhc.noaa.gov/data/tcr/AL142018_Michael.pdf
- NOAA NHC HURDAT2 best track, AL041992. https://www.nhc.noaa.gov/data/hurdat/
- NOAA NWS NEXRAD Level II, KMLB, 24 August 1992: https://unidata-nexrad-level2.s3.amazonaws.com/?prefix=1992/08/24/KMLB/
ARCHIVE PLATES — public-domain documents, with author, date, archive and licence.
- Infrared image of Andrew over Dade county at 0900 UTC August 24, 1992 | NOAA National Hurricane Center | public-domain
- The last radar image taken from NHC before the WSR-57 radar was blown | NOAA National Hurricane Center | public-domain
- NWS Melbourne WSR-88D image of Andrew over South Florida, 1011 UTC Aug | NOAA National Hurricane Center | public-domain
Hook Echo reads a disaster through the instruments that recorded it: radar, sensor, survey and report, with the primary sources above.
About the material. Video: U.S. Coast Guard aerial footage of South Florida (24 August 1992) and south Louisiana (26 August 1992), released through DVIDS (Defense Visual Information Records Center, 758420), public domain. Every radar image rendered by Hook Echo comes from NOAA's public archive of KMLB Level II data (NEXRAD, unidata-nexrad-level2 on AWS), in the National Weather Service colour scale; the eye positions and the beam geometry are our own measurements from those files. The last Miami radar image, the satellite image and the Melbourne radar image are NOAA works reproduced from the National Hurricane Center's report. Every diagram is drawn in code from the sources listed. Nothing in this video is AI-generated imagery.