An image from video provided by KWTV-KOTV shows two funnel clouds formed in Crescent, Okla., on May 20. (KWTV-KOTV/AP)

On Monday, more than 20 tornadoes developed across four states in the southern Great Plains, producing pockets of considerable damage throughout Texas, Oklahoma, Kansas and Missouri. Simultaneously, heavy rains from severe thunderstorms flooded roads and houses and triggered water rescues.

But residents across Oklahoma and portions of Texas were led to anticipate the worst, as the National Weather Service Storm Prediction Center placed the region under the highest threat category of tornado outbreak. This was Level 5 out of 5, the high-risk category for a widespread, major severe weather outbreak that included large and destructive hail, destructive winds, and tornadoes, “some potentially long-track and violent.”

Severe storms outlook for May 20. (NOAA)

This is a little-used threat category, and it had been two years since a similar high-level alert had been issued in the United States. The proclamation quickly became a lead story among the major news outlets. Schools and colleges in central Oklahoma were closed for the day, not only out of safety concerns for the sheltering in place but also because of vulnerable bus routes.

Meteorologists seemed awestruck at the tornado outbreak’s potential, using terms such as “nightmare scenario” and “a rare kind of event that may take many lives.” And a horde of storm chasers converged on Oklahoma, with some hoping the day would bring the chase of a lifetime.

While tornadoes and other forms of severe weather did materialize in the threat area, the highly anticipated violence of extremely intense, widespread tornadoes did not. Understandably, those in the region unaffected by the severe weather breathed a collective sigh of relief. But at the same time, it left behind a sense of false alarm, ire that schools were needlessly shut down and a desire to better understand the sometimes fickle nature of weather prediction.

Several days before, there were many indicators of unusually severe potential converging over the southern Plains. In many ways, it seemed like a slam dunk. There was the unusual strength of the upper-level weather system, including jet stream winds. There was the northward surge of highly unstable, tropical air with high moisture off the Gulf of Mexico. Many meteorologists saw this as a rare combination of extremes, by late May standards, that would trigger a swarm of rotating thunderstorms or supercells, unfolding in several rounds from early afternoon through late evening.

Indeed, as I was monitoring weather maps across the threat area, these factors did appear to materialize. The wind shear was very impressive, as was the rotational potential conveyed in those low-level winds. Highly technical indexes such as “supercell composite” and “significant tornado parameter” were approaching extreme values. Robust supercells, such as the one shown below, began developing across the Texas panhandle. For a time, it did seem like a slam dunk.

A tornadic supercell develops over the Texas Panhandle.

But the feared phalanx of violent tornadoes never materialized.

In a number of late-day tweets and discussions, meteorologists and storm chasers developed a consensus that while the air mass was indeed highly unstable, the manner in which all that buoyant energy was distributed in the vertical ended up being problematic. There was an intervening, stable layer about two miles above the ground, which partly suppressed updrafts in developing thunderstorms … and not enough focused, low-level uplift of air to help push nascent updrafts into the deeper layers.

In a nutshell: Having a tremendous amount of buoyancy available for storm updrafts is one thing, but it needs to be released in a manner so that all that upward acceleration is concentrated low down in the atmosphere so it can amplify the developing rotation inside supercells. Some of the attributes pertaining to vertical distribution of instability were not well captured in our state-of-the-art prediction models.

It would seem to be a subtle effect, but the sublime is everything when it comes to the dynamics of tornadic supercells. Drawing on my earlier research days attempting to “grow” mathematically simulated, intense thunderstorms on a Cray supercomputer, I can attest to the “nonlinearity” of the atmosphere: Very slight changes in the thermodynamic environment of a storm can have disproportionately large — profoundly large — impacts on the response of cloud systems, most notably the vigor and depth of cloud updrafts.

These points have certainly been noted by forecasters responsible for issuing Monday’s 45 percent chance of tornadoes within 25 miles of a point. The professionals are in a very tough position, having to carefully evaluate and balance two ends of a spectrum: miss a critical forecast for widespread, severe storms and many lives could be unnecessarily lost, vs. over-forecast an event and face criticism and potential loss of credibility for the next go-round. In the delicate balance of this decision lies the potential for disconnect between forecast models and reality, and the subtle nuances of processes governing convective storms.