In the late afternoon of May 22, 2011, the National Weather Service issued a tornado warning for the Joplin, Missouri, area. Although the warning triggered civil defense sirens, it was a mere 17 minutes before one of the worst tornadoes in U.S. history touched down in the small midwestern city.
As the tornado intensified, it tore a path of destruction through the southernmost part of Joplin, eventually reaching EF-5 strength (the highest tornado rating) with sustained winds of more than 200 mph. In its wake, the catastrophic tornado left 161 people dead and caused more than $3 billion in property damage, including the destruction of nearly 7,000 homes.
Although tornadoes are triggered by thunderstorms, it takes a unique mixture of atmospheric conditions to produce the kind of severe storm that spawns a tornado. Thunderstorms develop when warm, moist air rises over a cooler mass of air, a process known as convection. As the air rises, some pockets of air—or air parcels—become cooler or warmer than their surrounding environment, and the atmosphere becomes unstable.
“These air parcels are all converging with their payloads of water vapor and temperature,” says Peter Kalmus, a climate scientist at NASA’s Jet Propulsion Laboratory (JPL). “There’s something special about the specific combination of air parcels coming in—the directions they’re coming from and the energy and moisture they have—and they all merge together to make that tornado,” says Kalmus.
Looking to satellite data for clues
Scientists can evaluate the complex thermodynamics that trigger a tornado by looking at atmospheric temperature and water vapor profiles. One way to do this is by sending up radiosondes on weather balloons; as the balloon moves up through the atmosphere, it measures temperature and water vapor at different altitudes.
These in situ measurements are very precise, and the measurements are taken at an exact location and instant in time. In other words, if a weather balloon had been sent up over Joplin 30 minutes before the tornado struck, those measurements would tell us the conditions that led to the tornado. However, most weather balloons are launched over land, twice daily and simultaneously around the globe, and at limited locations. In the Midwestern U.S., the launch times are late morning and late evening, and the launch locations are typically only at a few airports per state. As a result, most Midwestern tornadoes occur hours after and far away from most weather balloons.
To obtain complete and global coverage, scientists turn to satellites. The Atmospheric Infrared Sounder (AIRS) instrument, which flies on the Aqua satellite, provides high-quality temperature and water vapor profiles by looking down at the atmosphere from space, globally. But there’s a catch, says Kalmus. AIRS flies over each day at 1:30pm local time, but tornadoes typically happen in late afternoon or early evening.
“You have this big problem hanging over your head: the temperature and water vapor profiles from 1:30pm aren’t necessarily going to be characteristic of the environment when the storm actually happened,” says Kalmus. “It doesn’t make a lot of sense to try and understand a tornado that happened in the evening by looking at satellite data from lunchtime. Those four hours are critical in terms of how the environment changes. It could be a bright sunny day when AIRS flies over, and then several hours later there’s a tornado right at that same spot.”
We now have a way to unlock the potential of these AIRS datasets, and that could also be useful for understanding how severe storms might change in our warming world.
"Back-tracing" to diagnose a storm
To overcome this time gap, Kalmus and a research team added a new piece to the puzzle: a model developed by NOAA called HYSPLIT (Hybrid Single Particle Lagrangian Integrated Trajectory). HYSPLIT uses meteorological data, such as winds, to calculate the movement of air parcels either forward or backward in time, at any altitude in the atmosphere. Once they know the direction and speed of winds during that four-hour time gap, they can trace the air parcels back from the storm’s location to where they were located at the time of the AIRS overpass.
“With HYSPLIT, we can basically move air parcels around in time and space. You can take an air parcel that started out at one spot and see where it’s going to go in the coming hours,” says Kalmus. This backtracing method enables the researchers to effectively “run a video in reverse” to see where the air parcels that caused the storm came from. Then, using the temperature and water vapor profiles from AIRS, they can estimate the characteristics of the air parcels that caused the storm.
“If you have complete satellite coverage of the area at the time of a tornado, then you have everything you need to see what was going on when the storm happened. But we only have a twice-daily snapshot from AIRS,” says Eric Fetzer, AIRS project scientist at JPL. “Now, we can take this new information and create the temperature and water vapor profiles four or five hours in the future.”
One of the key factors that makes an environment prime for a tornado is high Convective Available Potential Energy, or CAPE. If you know where air parcels are going to be and what their temperature and water vapor profiles are, then you can predict the CAPE in that location. For example, “You can look at CAPE at 1:30 in Joplin and there’s nothing going on, but then a few hours later a hotspot develops,” says Kalmus. “You would never see that from the AIRS data alone.”
Kalmus and his colleagues now hope to evolve their concept into an actual forecasting tool. “The key thing here is that you can take information from different sources and produce a short-term forecast,” says Fetzer. “We want to go from all these satellite observations every day over the Midwest to making some projection of where things are going to get nasty over places like Joplin.” And with an average warning time of only 13 minutes for tornadoes, these projections could save lives.
“We now have a way to unlock the potential of these AIRS datasets, and that could also be useful for understanding how severe storms might change in our warming world,” says Kalmus. “I think we’re on our way to succeeding at better forecasting these storms.”
Trajectory-Enhanced AIRS Observations of Environmental Factors Driving Severe Convective Storms
Peter Kalmus and Brian H. Kahn
Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California
Sean W. Freeman and Susan C. van den Heever
Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado
Monthly Weather Review, 147(5), pp.1633-1653