For much of human history, disasters were caused by surly gods who demanded tribute.
Ga-Oh, the Iroquois god of the winds, was a mean old cannibal who hurled boulders and uprooted trees. In Chile, the Araucanians trembled at the sight of shooting stars. It meant the wicked god Pillan was preparing to assail them with earthquakes, volcanoes and violent storms. Masaya, the Nicaraguan goddess of volcanoes, was even more fearsome. After an earthquake, she could be appeased only by tossing live humans into her crater.
Happily, relations with the disaster gods have improved.
The 20th century has given us scientific instruments to dissect their mysteries and foretell the course of their violent behavior.
In turn we have come to expect much of our modern medicine men, our Ph.D.s of storms and eruptions. We want no surprises, and no lives sacrificed. We expect our natural disasters to be safely predicted.
Usually they are. Volcanologists were at Mount St. Helens in 1980, forewarning its eruption. Meteorologists tracked the storm that grew into Hurricane Andrew all across the Atlantic, then told us when and where it would strike. Weather forecasters have made visible the movement of air currents. They alert us to thunderstorms and predict when rivers will overflow, and often, all who heed their warnings survive.
So when meteorologists fail to foresee a group of tornadoes, and four people in Pinellas County die, naturally we ask: How could this happen? We want an investigation. We want better tornado-spotting radar at our local weather station. Now.
We have cause to be upset. Ten locations, including two in Florida, have Doppler radar equipment that can detect the rotating wind pattern in a cloud forming into a tornado _ and improve hurricane warnings.
Our local weather station does not, and the equipment is not that costly. What our government spends in a month for Star Wars defenses against unknown foes would pay for nationwide Doppler radar protection against the undisputably real threat of 700 or so tornadoes striking the United States each year.
Yet in this craving for predictability, we should remember two things. The most experienced scientist, using the best instruments, cannot say just where a tornado will land or where it will veer from one moment to the next. That being the case, surviving such disasters depends on effective warning plans and sturdy buildings as much as forecasters' accuracy.
Major advances in meteorology
In the century since the National Weather Service started keeping daily temperature records, few sciences have evolved as rapidly as meteorology.
Weather fronts and jet streams, two staples of the nightly TV forecast, were not known to exist until this century. Nor did many of the instruments we now employ to analyze storms _ space satellites that keep watch over oceans, radar systems that can see varying wind speeds within a single cloud, computers that calculate with lightning speed which brew of temperature, wind, air pressure and moisture augurs trouble.
As a result, meteorologists say their five-day forecasts are now as good as their three-day forecasts were in the mid-1980s, and their three-day forecasts are better than their one-day forecasts were 15 years ago.
As forecasts improve, storms are losing much of their deadly menace. Fatality statistics tell the story. While the U.S. population has doubled in the last 50 years, only one-fourth as many people are dying now in tornadoes, hurricanes and lightning storms. Among weather disasters, only flash floods claim as many victims today as in the 1940s.
The property damage Hurricane Andrew caused in Dade County may exceed the losses from the next three worst hurricanes in U.S. history. Yet only 15 people were killed directly by the storm, and of those, only three were in houses outside the evacuation area.
In a typical year, your odds of being killed in a hurricane are about 1 in 20-million now. You're more likely to die in a freak kitchen accident.
For meteorologists, large slow-moving hurricanes are the easiest disasters to track. Tornadoes are trickier. They can form rapidly and wander across a town as erratically as a spinning top. While most last only minutes, they carry a capacity for destruction unlike any other storm on earth. Their whirling winds have been known to drive straws into trees and pluck every feather off a chicken.
Nobody knows a tornado's unpredictability better than the researchers who tried to put TOTO in front of one.
TOTO is not Dorothy's dog. It was the Total Tornado Observatory, a portable weather lab shaped like a garbage can and equipped to measure the wind speed, temperature and air pressure inside a tornado.
Except it never got inside one.
Several times, scientists in Oklahoma left TOTO in the path of approaching tornadoes, but "it proved to be very difficult to do. The tornado didn't have to veer very much to miss it," said Robert Davies-Jones, a meteorologist who has studied tornado dynamics for 20 years.
Finally TOTO was retired to a museum, to be replaced by Doppler radar.
Of the natural disasters that erupt underground, volcanoes are the most predictable.
Jim Riehle, the U.S. Geological Survey's deputy chief for volcanoes, says volcanologists can put seismometers on a mountain and detect an imminent eruption because rising magma causes minor earthquakes. In cases where volcanic mountains are not remote and "you can put geophysical instruments of them, we're batting close to a thousand," he said.
What they cannot do yet is predict the time or ferocity of an eruption. The explosion of Mount St. Helens, which unexpectedly blew a hole in the mountainside, killed a Geological Survey employee along with sightseers who ignored volcano warnings.
Believing that further eruptions are possible, volcanologists now have instruments on Mount St. Helens 24 hours a day. They are also keeping a round-the-clock watch on the massive hulk of nearby Mount Rainier, which overlooks the Seattle area.
The science of predicting earthquakes, by contrast, is in its infancy.
The crack in the ground across California, a phenomenon whose significance escaped its settlers, has been mapped. The earthquakes occurring near it are now understood to be products of a collision between two drifting continents, and North American and Pacific plates.
Geophysicists also know that earthquakes occur periodically in places, just as Old Faithful regularly blows steam. Beyond that, their powers of prediction are scant.
They are now studying whether sudden changes in well water levels and the radon gas content in water signify a buildup of underground stress preceding an earthquake.
Willis Jacobs, a National Earthquake Information Center geophysicist, thinks detectable phenomena such as these eventually will be identified as precursors of earthquakes and used to issue warnings.
"I doubt that it will happen within our lifetimes," he added. "There's too many unknowns about what's going on in the interior of the Earth. The statistical approach is about the best we can do now."
That approach suggests that another "big one," comparable to the 1906 San Francisco earthquake, could occur in the next 30 years somewhere in California _ not a very practical warning.
For residents of Parkfield, Calif., the statistical approach suggests more immediate danger. In the time records have been kept, earthquakes have occurred there in a time "window" of every 20 to 22 years. If Parkfield's next earthquake erupts during the same window, it's due within weeks.
The 20th century opened with the most deadly natural disaster in U.S. history. A hurricane roared through the Galveston area, trapped residents on an island and killed more than 6,000 people.
In Florida, a 1928 hurricane killed nearly 2,000 people in the Palm Beach area, and in 1935, another hurricane killed hundreds on the thinly populated Florida Keys.
In the 21st century, such calamities will be unthinkable.
Today, meteorologists are able to warn of approaching hurricanes days in advance, to forecast with increasing accuracy where they will land and to develop evacuation plans based on the estimated strength of the storm.
Tomorrow's technology _ Doppler radar, new computer models, better-equipped reconnaissance aircraft _ is expected to yield "mesoscale" forecasts based on local atmospheric activity and 10-day forecasts in temperate zones reliable enough to use for vacation planning. It is also expected to produce better estimates of a hurricane's strength, speed and direction, and better information about the bands of wind and rain within a single storm.
For instance, the weather service could warn people in Clearwater to expect two hours of 100 mile-per-hour winds and eight inches of rain, followed by winds gusting to 60 miles per hour, and warn people in Zephyrhills of torna-does forming as the hurricane moves inland.
With Doppler radar, "you can do warnings with warnings _ and do that with some precision," said National Hurricane Center director Bob Sheets. "We call it "nowcasting' _ just telling you what's going on right now, and what's likely to happen within the next two or three hours."
Warnings and reactions
No matter how accurate such warnings become, however, the death toll from disastrous storms will continue to depend on the communication of such warnings and the public's response.
"Flash flood deaths remain our number one killer," said Linda Kremkau, the keeper of weather service fatality records. That's because "people are living in more flood-prone areas," she said, and "are not heeding our advice." That advice: Get to high ground immediately, and don't try to drive through flooded spots.
Lightning kills about 70 people in the United States each year, with Florida usually leading the list, mainly because men didn't know enough to get out of a storm. Year after year, almost every lightning victim is male. In 1990, for example, lightning killed seven women and 67 men in the United States.
Mass evacuation plans, combined with better forecasts, have greatly reduced hurricane deaths. The number of Americans killed by all hurricanes since 1950 is smaller than the number who perished in the 1928 Florida hurricane alone.
And in tornado-prone areas, new radar systems and quick evacuation plans are similarly saving lives.
Ron Alberty, director of the NEXRAD (Next Generation Weather Radar) facility in Norman, Okla., says its new system finds about 80 to 90 percent of all tornadoes before they form. But it finds them, on average, only 20 minutes before they touch down.
That isn't much time, but it is enough to alert radio and television stations and police departments. In some communities, local officers drive into mobile home parks, "sirens blaring and loudspeakers blaring," Alberty said.
In one tornado, "something like 800 people were warned and took shelter, and none of them were scratched. Of the 30 that didn't, 12 were killed," he added. "It's important to do the warning, and it's important to do the preparation."
David Bilodeau, Pinellas County's emergency management director, considers such evacuation plans impractical in a county with 45,000 mobile home residents and few fatal tornadoes.
He would advise people in mobile homes to seek safer shelter when the weather service issues a tornado watch. But in the case of a tornado warning, which means tornadoes are imminent or occurring, "it's too late," he said. At that point he suggests scanning the sky, "and if you see one, you move at right angles to it."
In Alberty's opinion, we should also be looking at how mobile homes are constructed and anchored. As a former mobile home salesman, he has some knowledge of their construction, and what he knows scares him.
"Some are actually held down to the frame by a dozen medium-size screws," he said, and "once they tip over, they just crumble like a matchbox."
David Olinger is a staff writer for the Times.