On a warm July morning over the Great Plains, the sky is a meteorological nursery dotted with bundles of infant clouds.
Nourished with warmth and water vapor, some of these infants will grow into towering giants that can unleash hail or heavy rain, or spawn tornadoes and flash floods. Yet estimating when and where such warm-weather storms will appear and the type and intensity of what they drop is one of the toughest challenges in weather forecasting.
Now, however, an international team of atmospheric scientists is poring over data that could help uncover atmospheric triggers for warm-weather storms and determine which high-tech tools would be most effective in detecting them.
The hope is that by improving forecasts for these storms, the nation can reduce the deaths and wreckage from events such as flash floods, which inflict more than $5 billion a year in damage and kill more people than hurricanes, tornadoes,7 windstorms, or lightning.
Where today's flash-flood forecasts have a lead time of less than an hour, for example, the team hopes its results will help extend that to at least several hours.
The research effort, known as the International H2O Project (IHOP 2002), is part of a national program aimed at improving precipitation forecasts.
Over the years, these efforts have led to more-accurate forecasts of snowfall amounts and rainfall estimates from storms along broad weather fronts. But the capability to forecast clearly the location, amount, and intensity of rain or hail from warm-weather storms that seem to pop up out of nowhere lags far behind.
"It's really miserable. We are failing," laments Steven Koch, an atmospheric scientist at the National Oceanic and Atmospheric Administration's Forecast Systems Laboratory in Boulder, Colo.
The reasons are many, researchers say. For example, the conditions that trigger warm-weather, or convective, storms can occur in volumes of the atmosphere too small to be picked up by computerized forecasting models. Moreover, while researchers can cite a number of conditions necessary for triggering warm-weather storms, they are less certain about which condition or combination of conditions are the most important to spot and track.
But at the most fundamental level, researchers still have a difficult time measuring and tracking changes in the amount of water vapor in the atmosphere the vapor the fuels these storms.
From mid-May to the end of June, some 100 researchers and technicians from the US, Canada, France, and Germany marshaled a small squadron of research aircraft, mobile radars, radar-like laser detection systems known as lidar, and instrument-laden cars and vans in an attempt to fill this scientific void.
In the past, scientists have chased roiling masses of clouds to study the formation and evolution of storms. For IHOP 2002, however, the scientists opted to chase the invisible.
They focused "on the moisture content in preconvective situations, when there's nothing out there" in terms of storms, explains Tammy Weckwerth, an atmospheric scientist at the National Center for Atmospheric Research (NCAR) in Boulder, Colo., and one of the project's leaders.
Although it will take several years for the scientists to digest the data the $7 million research project has gathered, they note that the field measurements already have yielded intriguing observations.
For example, researchers had held that convection typically begins on the boundaries between contrasting air masses. These boundaries could stretch along the leading edge of cold fronts, between vast parcels of humid and dry air, or along the leading edge of cold-air outflows from earlier thunderstorms.
Yet IHOP 2002 experiments showed that convection typically begins from 5 to 20 kilometers (about 3 to 10 miles) ahead of a boundary. Moreover, the boundaries were not "simple fine lines," Dr. Weckwerth says, but had a more complicated structure than previously believed.
Even knowing where to intercept a boundary to measure its properties and activity could be frustrating.
"Boundaries are tricky and unpredictable," says Kevin Knupp, an atmospheric scientist at the University of Alabama at Huntsville.
He notes one instance where his small squadron of car-borne weather stations found itself perfectly positioned to record a passing boundary. Once it passed, drivers sped down country roads to get ahead of it and prepare for more measurements. While they were heading in one direction, however, the boundary reversed itself.
In addition, new software for weather radars may help detect the convergence of air masses, which can trigger convection, long before traditional techniques can detect it.
"If you have convergence, the air has to go up," explains Frederic Fabry, an atmospheric scientist at McGill University in Montreal, whose new detection techniques proved to be one of the highlights of the experiment.
Another group tried to get a handle on the moisture content within the boundary layer (the layer of atmosphere very close to the earth), on the layer's structure, and on the role terrain plays in boundary-layer circulation patterns.
Using special airborne radar to track changes in the topography of the boundary layer, researchers said they were struck by how widely its height varies over the course of a day in one case shifting by a factor of two.
"The boundary layer can show some really complicated behavior," says Margaret LeMone, an atmospheric scientist at the NCAR.
While IHOP 2002 scientists have been focusing on these areas, other scientists are detecting continental-scale patterns that might aid in boosting the lead time for forecasting summer precipitation.
In a paper published earlier this month in the Journal of the Atmospheric Sciences, a team led by NCAR researcher Richard Carbone reported that it has detected a clear pattern that allowed it to trace thunderstorms in the east to atmospheric triggers pulled by thunderstorms to the west up to two days earlier.
The physical mechanisms driving the pattern are not at all clear, he says. He hopes to identify those processes and include them in forecast models.
But Mr. Carbone adds that even now, weather-radar and satellite data can be used in a purely statistical way to observe the early development of convection to the west of a given forecast area, then issue a storm forecast for a local region.
"Based on time and place where convection begins, and the atmospheric conditions to the east, you can identify the corridors along which this stuff is likely to propagate," he says. His team's initial work suggests that forecasters could determine the probability of storminess to the east 6, 12, or 18 hours later.
Carbone's group is testing the potential accuracy of such forecasts, he says.