There May Be ‘A Wall In Terms Of Tornado Predictability.’ An Atmospheric Scientist Explains Why

This story originally appeared in the Morning Brief email newsletter. Sign up here to get more stories like this and weekday weather updates from The Weather Channel and our meteorologists.

It’s incredible just how advanced our ability to predict the weather has become. However, tornadoes remain difficult to predict with certainty. To find out why, I talked with Dr. Christopher Nowotarski, a professor of Atmospheric Science at Texas A&M University.

Although forecasters have come a long way in predicting when conditions likely to make tornadoes occur will happen, it's still difficult to predict which thunderstorms will spawn tornadoes, and when. Why is that?

The most prolific producers of tornadoes are called supercell thunderstorms-- they produce the majority of significant and violent tornadoes (EF2+). However, current research estimates that only about a quarter of supercells produce tornadoes, even if they have strong low-level rotation on radar. Even on days when the storm environment is highly conducive to tornadoes and multiple supercells exist in this environment, one storm may produce a tornado while the next one down the line does not.

Predicting which supercells will produce tornadoes more than about 10 minutes in advance can be challenging because of small-scale processes that occur within storms that are often unobservable by our current observing networks of radar and surface observations. Storm sensitivities to small differences in the environment such as land use (forrests vs. fields), topography, or unobservable meteorological variations may be the deciding factor between whether or when a supercell will produce a tornado.

Even if we were able to improve our knowledge and understanding of these small-scale sensitivities, recent modeling work suggests that the nearly random nature of perturbations within storms and their environments may ultimately lead to a wall in terms of tornado predictability. Linear convective systems can also produce tornadoes when they develop areas of rotation embedded within them, and these are even less understood and possibly less predictable.

How far in advance is it possible to know if a particular set of thunderstorms and possible tornadoes will be powerful, and how is this determined?

The Storm Prediction Center (SPC) issues convective outlooks as far as 8 days in advance of potential severe weather events and tornado outbreaks. The strongest tornado outbreaks often have strong precursor signals, allowing us to predict regions that will likely experience tornadoes, and potentially even strong tornadoes in this 4-8 day window.

Such multi-day outlooks are issued using forecaster knowledge of how current conditions will evolve with time as well as guidance from regional and global weather forecast models. When tornado warnings are issued, this is an indication that a tornado is ongoing or is imminent. Such warnings usually have lead times on the order of 10-15 minutes, and are narrowed down to about a county-sized area. Tornado warnings are usually issued using radar observations and storm-spotter reports, although active research is ongoing to try to improve tornado warning lead time with short-fuse, high-resolution model forecasts.

What is some of the most interesting research being done at the moment on tornadoes, and how might it help us in the future?

Research on tornadoes focuses on two areas: (1) improving our knowledge and observations of the storm-scale processes that lead to tornadoes and their dependence on the near-storm environment (which is more easily predictable), and (2) improving probabilistic forecasting of tornadoes using weather forecast models and machine learning techniques applied to model forecast data and/or observations.

In the first area, research often relies on high-resolution "cloud" models that can simulate small scale processes in idealized storms, and the sensitivity to changes in the storm environment on these processes can be examined in controlled studies (this is the sort of work that my research group focuses on). We also conduct a variety of storm chasing field projects to collect observations in and around storms. Technology has really helped us here, as we have a considerable number of different mobile radar platforms that allow us to determine precipitation distributions and three dimensional wind fields within storms and tornadoes. There has also been growth in getting weather observations above the ground, using either neutrally buoyant weather balloons that float around at various locations within storms or by using unmanned aerial vehicles and drones outfitted with observing equipment.

Operational forecasting has also seen a large advancement in the number of high-resolution simulations we can perform. Using multiple models with slight changes in each simulation provides an ensemble of possible outcomes that can be leveraged to really improve probabilistic forecasts of tornadoes. This has been made much more feasible by the rapid advancement in computer processing power and data storage and transfer rates.

How do the geographies of the Southeast and Plains create different types of tornado activity?

Recent research looking at historical trends and future predictions from climate models suggest a potential shift in tornado frequencies from the historical "tornado alley" of the Great Plains towards the Southeast United States. The largest difference in tornadoes in these regions is the proximity to the Gulf of Mexico. Tornado outbreaks in the Southeast often occur earlier in the year (late winter and early spring), and are often more evenly distributed throughout the day (with more tornadoes occurring overnight). This is due to the fact that ample water vapor close to the Gulf of Mexico prevents surface temperatures from cooling as rapidly at night, allowing the atmosphere to stay at least marginally unstable during these events. There is also often less warm air aloft in the Southeast, which is farther removed from the desert southwest), which promotes greater storm coverage, particularly when storms are triggered by strong early season cold fronts. Both factors lead to more linear convective storms, such that a greater fraction of tornadoes occurs in them in the Southeast US. The Southeast also has greater coverage of trees and more variable terrain. These changes in the surface roughness may also affect tornadoes in storms, though research is still ongoing here.

The Weather Company’s primary journalistic mission is to report on breaking weather news, the environment and the importance of science to our lives. This story does not necessarily represent the position of our parent company, IBM.