2
Meteorological Aspects
SYNOPTIC HISTORY
On August 23, 1985, a well-organized cloud pattern was first identified on satellite imagery north of the Cape Verde Islands. The unusually fast (34 mph) westward motion, combined with a dry Saharan air mass surrounding the disturbance, apparently inhibited the formation of a tropical cyclone. The rapid motion was the result of a strong high-pressure ridge building westward across the Atlantic, north of the tropical disturbance. The tropical disturbance approached Cuba on August 27. Elena was named on August 28 when the center was over Cuba and reconnaissance aircraft measured winds at 50–55 mph north of the center. It is interesting to note that the central pressure dropped 9 mb while moving over Cuba.
Elena quickly strengthened to a hurricane on August 29. A marked decrease of Elena's forward motion began the next clay as steering currents collapsed with the approach of a frontal trough from the northwest. A deep middle-atmospheric trough had reached maximum intensity west of the U.S. Pacific Coast. Zonal flow was found across the United States at middle-and upper-tropospheric levels. By 0000 GMT on August 30, ridging appeared over the Rockies with a flow from the northwest occurring over the Great Plains. At the same time a middle-and upper-tropospheric trough was forming over the Mississippi Valley. By 1200 GMT on August 30, a middle-tropospheric anticyclone pushing eastward reached the Continental Divide (Figure 2-1). The deepening trough to the east extended into lower latitudes and began to exert influence over the motion of the hurricane, previously on a smooth course toward the north-northwest. Now the motion of the storm was toward the northwest coast of the Florida Peninsula.
By 0000 GMT of August 31, the middle-level trough in the West was
filling rapidly and pushing into the northern Rockies (45° to 50° N). Twenty-four hours later, at 0000 GMT September 1, the ridge at upper levels extending from Oklahoma to the Great Lakes was pushing eastward, obviously set up to cut off the southern part of the middle-level trough. A stalling of the hurricane off the west coast of Florida and eventual reversion to westward motion resulted.
By 1200 GMT September 1, ridging began across the middle-level trough occupying the eastern United States between 30° and 40° N (Figure 2-2). The middle-level anticyclone was over Missouri and moving eastward. From that time on a new trough development over the west coast anchored the
anticyclone over the mid-Atlantic states, steering the hurricane toward the west-northwest and then northwest around it (Figure 2-3).
It should be clear from this sequence of events that the hurricane motion responded closely to the large-scale changes of flow in the mid-and upper-troposphere of midlatitudes. Indeed, the synoptic event could be concisely defined as a short-lived interaction with a midlevel trough, which caused an interruption of what otherwise was a more straightforward track (a "straight mover" track). Some literature exists describing hurricanes or typhoons of exceptional size, where the storm takes on a motion of its own in response to internal forces and where it does not appear merely to drift on the large-
scale atmospheric flow. Hurricane Elena showed no signs of such behavior, presumably because of its modest dimensions. Therefore, at least in principle, the main aspects of the motion of this hurricane should have been predictable within the present state of the art.
NEARSHORE AND LANDFALL STORM CHARACTERISTICS
Wind Speeds
An accurate assessment of the wind speeds is needed to determine whether or not the design wind conditions for buildings and other structures have
been exceeded. Unfortunately, in many hurricanes few, if any, reliable anrecords emometer are available, and forecast wind speeds or those based on damage observations—often exaggerated—eventually become part of the record of the storm. In the past this has led to highly erroneous conclusions regarding the performance of structures.
In this storm a number of anemometer records were available. Unfortunately, none of the records in the critical areas was complete, due to anemometer damage or power failure. Nevertheless, R. D. Marshall at the National Institute of Standards and Technology (formerly the National Bureau of standards) was able to reduce the data to the standard meteorological conditions used as a basis for structural design, that is, as ''fastest-mile'' wind speeds at 33 ft in open country (Marshall, 1985). (For a complete discussion of the method used for data reduction to standard meteorological conditions, see Marshall, 1984.)
The locations of the anemometer stations used are shown in Figure 2-4 together with the adjusted wind speeds. Table 2-1 gives details of the data used and compares the adjusted fastest-mile wind speeds with those currently used for design in the area. It should be noted that in no case did the adjusted wind speed exceed the basic design wind speed. However, the adjusted wind speeds were based on rather meager data and could be subject to some error. Nevertheless, it seems reasonable to conclude that, at worst, the conditions were within 10 percent of the design conditions and in most instances were
TABLE 2-1
Wind Speed Data
Map location number (see Figure 2-6) |
Type of data recorded |
Recorded peak gust at height h |
Time GMT (hrs) |
Anemometer height h (m) |
Wind azimuth (degrees) |
Maximum Estimateda roughness length (m) |
ANSIb terrain exposure |
Adjusted fastest-mile speed (mph) |
Design fastest-mile wind speedc |
|
1 |
Pensacola Naval Air Station, Florida |
Strip chart |
73 kts (84 mph) |
0800 |
23.77 |
100–110 |
0.20f |
C-B |
59c |
101 |
2 |
Pensacola Regional Airport, Florida |
Strip chart |
56 kts (64 mph) |
0730 |
6.71 |
90–105 |
0.05 |
C |
52c |
101 |
3 |
Mobile Regional Airport, Alabama |
Strip chart |
52 kts (60 mph) |
1230 |
6.71 |
70–130 |
0.05 |
C |
52c |
98 |
4 |
Dauphin Island, Alabama |
Partial strip chartg |
106 kts (122 mph) |
0920 |
10.0 |
0–90 |
0.01 |
D |
96d |
105 |
5 |
Bay St. Louis OTP 42007, Mississippi |
Periodic observationsi |
43 m/s (96 mph) |
1100 1200 |
10.7 |
330–220 |
0.005 |
Ocean |
71d |
106 |
Map location number (see Figure 2-6) |
Type of data recorded |
Recorded peak gust at height h |
Time GMT (hrs) |
Anemometer height h (m) |
Wind azimuth (degrees) |
Maximum Estimateda roughness length m) |
ANSIb terrain exposure |
Adjusted fastest-mile speed (mph) |
Design fastest-mile wind speedc |
|
6 |
Point Biloxi, Mississippi |
Partial strip chartg |
122 mph |
1130 |
7.71 |
340 |
0.005 |
Ocean |
96d |
102 |
7 |
Keesler Air Force Base, Mississippi |
Periodic observationsh |
41 kts (47 mph) |
1055 |
3.96 |
340 |
— |
— |
— |
102 |
8 |
Harrison County Civil Defense, Mississippi |
Peak gust |
100 kts (115 mph) |
1300 |
27.13 |
— |
0.25 |
B |
92d |
102 |
9 |
Gulfport Sea Bee Base, Mississippi |
Peak gust |
75 kts (86 mph) |
— |
10 |
|
0.25 |
B |
79d |
102 |
10 |
NSTL, Canal Buoy 42009, Mississippi |
Periodic observationsi |
19 m/s (43 mph) |
1400 |
7.0 |
300–240 |
0.10 |
C-B |
36d |
100 |
a Estimated roughness length used to normalize fastest-mile speeds to standard exposure. b Approximate correspondence. c Based on hourly means. d Based on observed peak gusts. e Based on ANSI A58.1 (1982) including hurricane importance factor. f Roughness based on influence of trees for given direction, see Reinhold (1975). g Record interrupted. h Record terminated 1055 GMT on September 2, 1985. i Based on maximum gust in 8-sec window. Source; Dale Perry. |
less than the design conditions (see Figure 2-5). In other words, Hurricane Elena was not a storm that significantly exceeded the normally accepted design conditions in use for at least the last 20 years in the applicable building codes and standards (see Chapter 4). Thus, extensive damage observed cannot be blamed on the use of inappropriate wind-speed criteria.
Tides
Tides were generally 3 to 6 ft above normal along the coast from Grand Isle, Louisiana, to Sarasota, Florida. The maximum estimated storm surge was 10 ft above mean sea level (MSL) near Apalachicola, Florida. Along the Alabama and Mississippi coastal areas, surge averaged 6 to 8 ft above normal. The actual magnitudes of storm surge were somewhat less than those that could have occurred had the angle of impact on the coast been larger.
Rainfall
Elena was a rather dry storm, with rainfall amounts averaging less than 5 inches. The maximum reported rainfall was 11 inches at Apalachicola, Florida; the storm was close to this location for an extended time.
Tornadoes
Several tornadoes occurred in central Florida when Elena was stalled off the west coast of Florida. Widespread damage to a number of mobile-home parks was noted to the northeast of Tampa. Several injuries resulted, but there were no reported fatalities.
At least a dozen tornadoes were reported in the coastal counties of Mississippi as the eyewall was crossing the coast. Only in one instance did the survey team find evidence of possible tornado damage in this area.
Pressure
The lowest observed pressure on the coastline was 953 mb at Pascagoula, Mississippi. Elena's minimum surface pressure, extrapolated from the reconnaissance aircraft flights of the National Oceanic and Atmospheric Administration (NOAA), was 951 mb as the storm was south of Apalachicola, Florida. The National Data Buoy Office (NDB) recorded a pressure of 976 mb at a location 35 miles west of the landfall point. This storm filled quite rapidly after crossing the coast. As the storm moved from Pascagoula to Gulfport, Mississippi, a distance of 20 miles, the central pressure rose approximately 17 mb.
FORECAST GUIDANCE
The National Meteorological Center (NMC) produces calculations of atmospheric state and motion around the globe using numerical models of the atmosphere. These calculations are distributed to field stations of the National Weather Service (NWS) and elsewhere as guidance for forecasts. The National Hurricane Center (NHC) in Miami uses this guidance when preparing specific forecasts of hurricane development and motion.
NMC uses several different mathematical models of the atmosphere to produce forecast charts for distribution. Atmospheric models are being continuously improved, and the guidance to the field stations is often in a state of change. The models in use at this time are:
-
global medium-range forecast model,
-
regional model (LFM),
-
new regional model (NGM) designed to replace the LFM model, and
-
high-resolution hurricane model (MFM) designed to run within the framework of the global model.
The forecasters at NHC and at other NWS field stations have specific hurricane center forecasts available to them from the LFM, NGM, and MFM models. NHC takes these forecasts into consideration in preparing the
official hurricane forecasts and warnings. There is as yet no standard predictive model generally accepted for accurate numerical forecasting of hurricane movement. Information from all available sources is used in preparing the final forecast.
The track followed by Elena was unusually complicated. Important changes in course were:
-
an abrupt turn from a north-northwest course toward the east at midday (1200 to 1800 GMT) on August 30,
-
a near-stall at midday (1800 GMT) September 1 as Elena executed a tight loop, and
-
resumption of motion of the hurricane toward the west-northwest after completion of the loop.
These abrupt changes posed some very difficult forecast problems. To be effective, guidance forecasts from NMC should have shown the eastward turn of the hurricane on August 30 in the calculations from 29/1200 GMT, 30/0000 GMT, and 30/1200 GMT. The MFM forecast from 29/1200 GMT did accurately track Elena for the next 24 hours. It then showed a northeastward turn to take place between 24 and 36 hours. Elena turned sharply eastward during that period. To be useful, the 0000 GMT on August 30 would have been the crucial time for a forecast to show the sharp eastward turn that began 16 hours later. None of the forecast models was successful at this—all of them showed Elena crossing northern Florida into southeast Georgia.
The stalling of Elena off the Florida coast should have appeared in the guidance forecasts from 31/0000 GMT and 01/1200 GMT. The LFM and NGM forecasts from 31/0000 GMT did show the stall for the first 36 hours of the forecast, although the forecast stalling position was about 70 miles west of the actual one. Normally, this is considered to be good forecasting, but when Elena arrived only 40 miles off the Florida coast before stalling, the forecasters at NHC must have found themselves in a very tense situation. By this time, the Tampa Bay area had probably been evacuated.
The final west-northwest motion of Elena should have been shown by the forecasts from 31/1200 GMT and 01/1200 GMT. The forecasts from the NGM model did show this motion. The LFM also forecast the westward motion of Elena from these times, but not as well as the NGM.
A summary of the usefulness of predictions from the three forecast models is presented in Table 2-2. The term "useful" indicates that the general trend of the storm's motion was correctly indicated by the forecast, although the precise forecast location might be in error by 50 to 80 miles. However, at times even this much error is unacceptable for determining who must evacuate and when.
Overall, the best numerical forecasts for this hurricane were obtained from the NGM model. The disappointing forecasts from the high-resolution
TABLE 2-2 Usefulness of Hurricane Elena Forecast Information Up To 36 Hours After Initial Data Time
Initial data time (GMT) |
LFMa |
NGMb |
MFMc |
20/00 |
NA |
U |
U |
29/12 |
A |
N |
U |
30/00 |
N |
N |
U |
30/12* |
U |
U |
N |
31/00 |
U |
U |
U |
31/12* |
U |
U |
N |
01/00 |
U |
U |
N |
01/12* |
U,A |
U |
N |
02/00 |
A |
A |
U |
Note: NA—not available; A—serious analysis difficulty at initial time; U—useful forecast; U,A—although there was an analysis problem, the forecast was still useful; N—not sufficiently accurate to be useful; *—time of crucial change in direction of storm movement. a Regional model. b Now regional model. c High-resolution hurricane model. Source: James Belville. |
hurricane model—the MFM model—prompted a reexamination of the model by NMC. The reexamination revealed that an error had been made in conversion of the program from the old to the new NMC computer. The consequence of this error was a significant distortion of the hurricane movement forecast in the test case. Testing of the corrected model is under way.
The European Center for Medium-Range Forecasting produces global forecasts once daily. These are generally of excellent quality. Examination of its forecasts for Elena showed an uncharacteristically poor performance from starting times of 1200 GMT on August 29, 30, and 31, most probably because of their inadequate analyses of the initial positions of Elena. Reasons for this miscalculation are unknown, but the matter is under investigation. The center's forecasts from 1200 GMT September 1 onward were generally accurate.
The varying guidance from the three U.S. numerical forecast models was a problem for the NHC forecasters who had to contend with sudden changes in the speed and direction of Elena. While many of the forecasts are indicated as "useful" in Table 2-2, they still require improvement in accuracy. It will take a long time to determine how much of the difficulty in hurricane forecasting is due to sparse data coverage in the Caribbean. It is clear,
however, that more upper-air data from Cuba would be useful in improving hurricane forecasts for the Caribbean, Gulf of Mexico, and the United States. The only upper-air observations received from Cuba came from the U.S. Naval Base at Guantanamo Bay. It is also clear that more research and development is needed in numerical analysis and modeling of tropical storms and hurricanes.
Under these circumstances, the forecasters at NHC had to be rather cautious in their interpretation of the numerical guidance. Neil Frank, NHC director, used radio and television broadcasts extensively to explain to the public the uncertainties of the hurricane's motion, and he restricted his projections to times and distances consistent with the given certainties and uncertainties. Interviews by the survey team revealed that Frank's approach to the problem was successful. People seemed to have understood him very well. The survey team encountered no serious criticism of the forecasts or warnings and heard much praise. This confidence by officials and by the public at large was fundamental to the successful evacuations that took place and to the minimal loss of life.
STORM SURGE AND THE SLOSH MODEL
Storm surges are frequently the major cause of death during a hurricane episode. NWS has developed two computer models to predict surges: the Special Program to List the Amplitudes of Surges from Hurricanes (SPLASH) and the Sea, Lake, and Overland Surges from Hurricanes (SLOSH) model. SPLASH looks at storm surge only up to the coastline. SLOSH goes further and computes surges over inland water bodies and gives inland inundation. Both models require information about the hurricane's track, intensity, and size, which are difficult to predict. Experience with the SLOSH model indicates that, given an adequate description of the hurricane and its movement, the SLOSH surge forecast is in good agreement with observed high-water data.
SLOSH hurricane simulation studies have been performed in several coastal areas, including the New Orleans and Lake Pontchartrain area, to determine the vulnerable coastal areas. This work has frequently been jointly funded by the Federal Emergency Management Agency (FEMA) and the U.S. Army Corps of Engineers as part of a comprehensive hurricane evacuation plan, which is tested locally. The plan also includes studies of the population, road capacity, and evacuation psychology.
Several hundred SLOSH computer runs are made to simulate hurricanes encountering a basin. Hurricane track direction, landfall location, forward speed, and Saffir-Simpson hurricane category are varied. The result is a comprehensive picture of the flooding possible for an area.
Because of the large amount of data from such a study, composites are
made from the individual runs. For example, in the New Orleans basin where a comprehensive hurricane evacuation plan is under development, all of the hypothetical hurricanes of a given intensity category and approaching from the south have been combined and the highest computed water elevation has been recorded for each SLOSH grid square. This type of composite is referred to as the Maximum Envelope of Water (MEOW).
For the New Orleans area, a total of 50 MEOWs were generated for the 5 categories of hurricane, 4 general track directions, and 2 forward speeds for each hurricane. Local evacuation planners found that this number of composites was unacceptably large. They examined the various MEOWs for similar flooding patterns and reduced the data to 12 scenarios. Evacuation planning for the New Orleans area is based on these 12 hurricane scenarios and the results of detailed studies of the population, evacuation routes, and road capacity in vulnerable locations.
New Orleans is one of the nation's most vulnerable cities to hurricane storm surge. Parts of the city are between 5 and 10 ft below sea level. The city is ringed by a variety of levees, which range from roughly 12 ft to over 17 ft in height. The levees are generally earthen, with little reinforcement and vegetation to protect them. In the past, hurricanes have caused breaks in the levee system. Most recently, Hurricane Betsy in 1965 breached parts of the eastern levee in St. Bernard Parish.
SLOSH for Hurricane Elena
Generally, forecasters at NHC have SLOSH computations available to them in two forms: MEOW data (in basins where a simulation study has been performed) and real-time SLOSH model runs. There has been no consensus among forecasters about the usefulness of the SLOSH MEOW data versus the SLOSH real-time forecast computations. MEOW data have the advantage of indicating specific problem areas that may not be encountered in a particular run. Such areas include long, narrow bays where flooding will be excessive if the storm passes to the left of the bay, but slight whenever a hurricane passes over or to the right of the bay.
MEOWs also help to show local officials the uncertainty of hurricane track forecasting. A disadvantage of MEOWs is that they may not be representative of specific storms. A problem arose in New Orleans when local officials expected flooding as depicted by the MEOW, for the lower Mississippi Delta Region, even though this area was many miles to the left of the storm track.
During Hurricane Elena's earliest phases, the storm was intensifying and heading on a northerly track toward New Orleans. Forecasters at NHC needed an estimate of storm surge for their advisories. The SLOSH model had not been run yet at NHC, since the storm was well out in the Gulf of
Mexico. Personnel at NWS Headquarters examined an experimental display of the MEOW data to make the initial surge estimates. The MEOW data were available in computer-graphic form through an experimental display being developed at the Techniques Development Laboratory of NWS. The first MEOW data entered into the system, fortunately, were for the Lake Pontchartrain basin. From this display, surges of 8 to 12 ft appeared likely from a category 2 hurricane moving rapidly in a northerly direction. MEOW showed that the most vulnerable locations were in the area between Slidell and Gulfport.
As the storm later turned eastward with a projected landfall in the Cedar Key, Florida, area, surge forecasts remained in the 8-to 12-ft range, this time based on real-time SLOSH forecasts. When the storm circled and headed westward, the same 8-to 12-ft surge value was retained in the forecast.
As Elena moved closer to landfall in the Biloxi-Gulfport area, two SLOSH runs were made—one with a projected track about 5 miles south of Slidell and a second with a track that crossed New Orleans. The first track was considered by forecasters to be the more likely of the two. However, the second would be more critical for the east New Orleans area.
Figure 2-5 shows the SLOSH output generated from the forecast hurricane, with a track 5 miles south of Slidell. This track, it turns out, produces some of the highest possible flooding along the coast for this category of westward-moving storms. Notice that in the Bay St. Louis area, surge values reach 16 ft. Surge values along the outer coast ranged from about 12 ft at the Rigolets to 15 ft just south of Bay St. Louis and back down to 10 ft east of Pascagoula. The highest water to impact the city of New Orleans was computed in this SLOSH run to be 6 to 8 ft.
The SLOSH run over the New Orleans area indicated moderate flooding in the areas surrounding New Orleans. However, no surges of the magnitude predicted by these SLOSH runs were observed. The reason can be found in a SLOSH run produced with the "best-fit" track and with best-fit storm parameters. These, of course, were not available until after Elena made landfall. Values of best fit were those available just after the storm. As more data are analyzed, an even better fit can be expected.
On the best-fit track, SLOSH surges ranged to 8 ft, with most of the Gulfport-Biloxi area experiencing 7 ft along the outer coast. Bay St. Louis surges were calculated to be only about 3 ft—well below the values computed along the original forecast track. The highest surge noted in this run occurred in the area of Pascagoula Bay.
A SLOSH run done for the Mobile Bay basin, using the best-fit track, is shown in Figure 2-6. The storm-surge values for Gulfport and Biloxi were computed to be 7 ft. Had the hurricane passed on its predicted path, surges of about 15 ft would have been experienced (see Figure 2-5). Because
Elena's actual track passed just 15 to 20 miles closer to the coast, the winds from Elena were mostly offshore winds that blew away from the coastline, generating negative surge values. Local observations indicated that this was, indeed, the case. Figures 2-7 and 2-8 show the time profile of surges at Gulfport and Biloxi. Only after the hurricane's center passed a given location dido, the wind shift to onshore, giving the potential for generating water level at the time of the wind shift was depressed; water had to be brought back in to counter the depressed state before positive surges could be generated. This, coupled with the weakening of the hurricane and the weaker winds behind the storm, led to very modest storm surge values. Figures 2-9 and 2-10 summarize measured values of surges in the region.
Local officials were unhappy with the SLOSH data used in making evacuation decisions. They state that water heights indicated by the SLOSH model would never occur within the levee system because of the levee heights. It was found that, indeed, levee heights were in error in the Pontchartrain SLOSH basin because of recent levee improvements. However, the increased levee heights produced a larger amount of tidal flooding outside the hurricane-protection levees. This information was not made available to the NWS until after the storm.