Hurricane Andrew (1992) in South Florida: CAT 4 or CAT5?

Peter G. Black
NOAA/Hurricane Research Division
Miami, FL 33149

26 July, 2002

Regarding the issue of whether Hurricane Andrew was a CAT 4 or 5 storm, i.e. whether it's 1-min sustained, 10-in open terrain winds exceeded 135 kt, there are 5 sources of evidence, among others, that should be- addressed. These are as follows:

  1. New estimates have been derived since Hurricane Andrew of average 'reduction factors' by which a 10-m, 1-min sustained wind is estimated from aircraft flight level observations at 700 mb (approximately 10,000 ft or 3 km). These new reduction factor estimates have been obtained from emerging new technology such as GPS dropsonde observations of the boundary layer wind profile over water and Stepped Frequency Microwave Radiometer (SFMR) observations. The former task is being address by Powell and Franklin. The latter consists of remotely sensed observations of the microwave emissivity of the sea which has been found to be directly proportional to surface wind speed and used together with concurrent flight level winds to compute surface wind reduction factors for use in cases, such as Hurricane Andrew where only flight level wind observations exist.

  2. An assessment of gradient wind radial profiles based on surface pressure observations compared with radial profile of damage survey estimated winds.

  3. Examination of radar echo core motion observations.

  4. Re-examination of the Hurricane Andrew damage survey wind estimates in the north eyewall.

  5. An assessment of evidence for eyewall intense convection and mesovortex occurrence from Tampa radar, NOAA satellite photos, public minimum surface pressure reports and anecdotal observations of the timing of eye and eyewall events. An assessment of a I-minute sustained wind from fine scale, unpublished analyses of damage patterns and wind speed estimates in the most heavily damaged area of South Miami-Dade County, i.e. the Naranja Lakes subdivision.

Maximum winds from Reduction Factors in the north eyewall

Uhlhorn recently computed reduction factors based on SFMR and flight level wind observations for 7 flights in 5 storms from 1997-2000. Ratios of the peak 1 -minute average surface wind anywhere along a radial leg to the peak 1-minute average flight level wind anywhere in the leg were computed. Powell shows how this average represents a 10-minute average over-water wind at 10 m at a fixed point. To estimate a 10-minute average surface wind over land at 10 m, the average ratio of the 10-min overland (open terrain) wind to the 10-min over water wind first needs to be multiplied. This value is 0.92 ± 0.08 according to Black. Then the ratio of 1-minute to 10-minute overland wind for open terrain would have to be multiplied. This ratio is 1.53 ± 0.08 for over-land, open terrain conditions according to Black, to Krayer and Marshall and more recently to Schroeder.

Uhlhorn finds that when all 49 flight legs in his study are considered, the ratio of surface to flight level mean winds is 0.93 ± 0.11. For only the two CAT 4 storms in the data set, Floyd and Bret, this ratio for 32 flight legs is 0.88 ± 0.07. The ratio tends to decrease slightly as the wind speed increased. The surface peak wind is found radially inward from the 700 mb flight level wind by 2-3 km on average for the CAT 4 storms and 10 km for the weaker storms. For the CAT 4 storms, the ratio of the peak surface to flight level wind anywhere in a complete 'figure 4' or 'alpha' pattern is also computed. For only 4 patterns, this ratio is 0.84.

If we confine our discussion to only the north-south flight leg in Andrew, in the right quadrant of the storm, the relevant ratio to use would be 0.07 figure for intense storms relating the maximum mean surface wind over a single pass to the maximum mean flight level wind. Multiplying this by the mean over land to over water ratio of 0.92 ± .08, we obtain 0.81 ± .08. Multiplying again by the mean ratio of the overland I -min wind to the over land 1 0-min wind of 1. 16 we obtain a reduction factor of 0.94 ±.08 to be multiplied by the flight level mean wind to obtain the surface over land 1-min wind estimate. The maximum 1-min wind for Andrew at flight level was 159 kt, only slightly less than the maximum 10-s average wind of 162 kt, which when multiplied by the appropriate 'reduction factor' is roughly equivalent to the surface 1-min sustained wind. Applying the ratio to the 1-min mean, we obtain a maximum 1-min sustained overland wind estimate of 149 ± 13 kt. Using this argument, Andrew's maximum 1-min sustained overland wind estimate would exceed 135 kt and be considered a CAT 5 storm with more than 75 % probability. This estimate is slightly higher than the value of 145 kt, derived from the 10-s average flight level wind without an overland correction, and almost identical to the 150 kt estimate derived from the improved HWIND analysis scheme (Dunion and Powell, 2002).

It should be noted that Andrew's peak flight level wind was observed offshore, and subsequently decreased along two flight legs made overland, one just inland from the coast and another along Krome Ave. (US 27). The nominal central pressure near the geometric eye center also filled by about 6 mb. However, what appears to be the strongest of several mesovortices formed as Andrew's center was passing over Homestead resulting in a lower pressure than the center by about 10 mb, but offset close to the western eyewall.

Maximum winds from surface pressure reports

Over 67 reports of minimum surface pressure made by the public were compiled by Rappaport. The barometers for the lowest 5 or 6 of these were borrowed from their owners and calibrated at the NOAA Aircraft Operations Center Calibration Lab to insure accuracy. All of these observations were composited relative to the flight level center as
it moved from the coast to Krome Avenue (US 27). A cubic spline fit to the observations as a function of radial distance was computed. The derivative of this fit was computed every 2 nm to estimate the gradient wind. The gradient wind was assumed to be the mean wind at the top of the boundary layer. The 0.88 ± 0.07 figure was then used to estimate a
mean and standard deviation for over water mean surface winds. The above reduction factor was then used to obtain a mean and standard deviation for the over land 1-min sustained wind. The peak gradient wind from the north-south pressure profile was 160 mph 7-8 nm from the center, or roughly along 152nd St, or Coral Reef Drive. This corresponded to the location of the maximum damage swath north of the center and to the location of maximum storm surge near the Burger King headquarters building found by Jarvinen. However, using the pressure profile from the pressure center to the west of the center, the gradient was stronger with a maximum gradient wind of 190 kt located only 5 nm from the pressure center. Applying the above reduction factors, we arrive at an estimated 1-min sustained overland surface wind of 150kt north of the center and 178kt west of the center at 0930 UTC, the time of the minimum surface pressure observation near Krome Avenue. This analysis would thus support a conclusion that there was a 85% chance that winds exceeded 135kt and should be classified as a CAT 5 storm.

Radar eyewall reflectivity core velocities

Radar echo core velocities and reflectivities were analyzed for the north eyewall of Andrew as it made landfall. Cores were tracked using the Tampa WSR-57 radar in Tampa with a mean beam altitude over Miami-Dade county of 27,000 ft. seven cores were observed to develop at the coast as the eyewall transited across the coast. The lifetime of these cores averaged approximately 12 minutes and developed approximately every 10 minutes. The initial growing cores are considered to be accompanied by intense updrafts carrying low level momentum. Hence core velocities are assumed to be proportional to the winds near the top of the boundary layer at about cloud base. This assumption is supported by the results of Shenk who tracked growing convective towers in the Caribbean with geostationary satellite imagery and found their translation speed to be nearly equal to the winds observed by aircraft at cloud base. This conclusion is also supported by radar studies undertaken by Senn.

The maximum echo core velocities in Andrew were 180 kt for 4 cores that developed in the north eyewall as it crossed the coast, and 170 kt for two other cores. As reflectivities approached a maximum near 48 dBZ in the southwest eyewall, the cores declerated rapidly to 50-60 kt. As the cores rotated around to the south (left) sector of the eyewall, reflectivities decreased rapidly as the cores became elongated patches of stratiform precipitation. Applying the same argument as the previous section, i.e. using the .94 reduction factor, we arrive at an estimate of 169 ± 10 kt for the estimated 1-min, over land maximum wind. This too would support the conclusion of CAT 5 status with more than 85% confidence factor.

F-scale estimates of eyewall maximum winds from damage surveys

The key result from the Andrew damage survey done by Wakimoto, Black, Forbes and Fujita is that it revealed an intense F3 damage area in the Naranja Lakes region caused by southeast winds following northwest peak winds in the leading edge of the eyewall. Intense southeast winds following eye passage cannot be explained by any wind analysis produced thus far. The question is what happened here, what was the time scale of the wind events and what could have caused such winds?

The north-south profile of F-scale derived surface winds agrees very well with the gradient wind north-south profile, yielding a maximum wind radius of approximately 6-7 nm. The peak gust F-scale derived estimate (based on CBS construction) for the north eyewall was 170 kt, which corresponds to a 1-min over-land sustained wind of 130 kt, if the damage is indeed caused by peak gust winds. However, from the gradient wind analysis, the strongest winds appear to have been in the west quadrant. The F2-3 damage in that area suggests a peak gust wind in that area of 150 kt, or a sustained wind of 115 kt, somewhat lower than the gradient wind estimate of sustained wind.

However, the damage survey showed the highest winds in the southeast quadrant of the storm, somewhat over F3 corresponding to 180 kt peak gust, or 135 kt sustained 1-min wind. However, the unpublished analysis of Fujita shows the width of the peak damage swaths being 400 ft (120 m) and the length being 1,800-2,000 ft (600 m) in length. A wind gust or transient traveling with a mean wind of 130 kt over these length scales would require a life time of 12-15 s. Thus, the sustained wind representative of the swath area would likely be of order 156 kt. This would be the same as averaging the wind estimates over the 3.5 by 3.5 nm. area of Naranja Lakes.

The most important discovery concerning the Naranja Lakes area is that the peak southeast winds arrived in the middle of the clear eye, some 10 minutes before the eastern eyewall area moved over Naranja Lakes. According eyewitnesses, such as Mike Shoemaker, an oceanographer at AOML who lived in the area, the winds on the 'back side' arrive only 15 min after the front eyewall winds ceased, well before the radar 'back' eyewall arrived. The roar of the wind in the distance could be heard approaching while winds were still calm. There was an almost instantaneous increase of the winds from near calm to peak values, which lasted intermittently for less than 10 minutes, whereupon the winds decreased abruptly as the 'back' eyewall approached.

The damage survey showed the peak winds in both the front and back eyewalls to be at nearly right angles to the circular isobars. Thus a transient associated with convection that might cause increased inflow into the convective cells would cause the air to cool and parcels to be forced downward in much the same manner as a down burst, which would enhance the inward acceleration. It is thus hypothesized that periodic surges of surface super-gradient flow penetrated into the clear eye producing several momentary wind transients, the strongest of which occurred over Naranja Lakes. These periodic transients appear to be associated with eyewall supercell convection that was triggered at the coast by enhance frictional convergence as the eyewall moved across the shoreline. Analysis of the Tampa radar data showed that 7 convective cells were triggered at the coast line about every 12 minutes and lasted about 10 minutes as they traveled around the
eyewall, reaching peak reflectivities in the southwest quadrant. Each eyewall cell appears to have a had a wind transient associated with it, with the one passing over Naranja Lakes being the strongest.

This supposition is supported by the NOAA satellite image which showed the warmest spot in the eye to be skewed to the west side of the eye, immediately adjacent to the west eyewall, and further showed a finger of low clouds protruding into the center of the eye and emanating from the south eyewall. The Tampa radar data showed this finger to be at the leading edge of the intense convective core which was reaching peak reflectivity at this time.