Bad Science: Meteorology's "Onesie" Problem

[this posting is addressed to meteorologists and weather geeks]

Union City Tornado

Meteorology in general, and the National Weather Service in particular, have a problem that has been around for at least 49 years: the "onesie." My definition of the onesie problem: to make major changes in procedures based on a single example. 

To my knowledge, the first person to point this out was Dr. Chuck Doswell over the influence the Union City, Oklahoma, Tornado had on storm warning technology. That 1973 tornado was the first-ever success for the storm chasing program (begun in 1972) combined with the experimental Doppler program. Unfortunately, because of that onesie, it was assumed all tornadoes formed that way. Problem? They don't. Some of that incorrect science is still being practiced today. 

Although there are other examples prior to 1999, let's start there. On May 3, 1999, the NWS in Oklahoma City issued the first-ever "tornado emergency" warning. It spread like wildfire even though Dr. Patrick Marsh's (NWS SPC) research later showed only 16% of them turn out to be correct. Even though they are terribly inaccurate scientifically, they continue to be issued. Onesie.

In 2004, Hurricane Charley struck southwest Florida. The inland wind speeds were underforecasted. So, rather than directly address the underforecasts, the NWS created the Inland Extreme Wind Warning for 115 mph winds. Onesie.

Fast forward 2011, the NWS issued a terribly misleading report on the Joplin Tornado that primarily blamed the deaths on false alarms rather than the major flaws in its own tornado warnings that day. So, the NWS created "particularly dangerous" tornado warnings (not to be confused with regular and  "tornado emergency" warnings which are still around) to hopefully scare people jaded by "regular" tornado warnings into acting. Onesie. 

And, August, 2020, a dangerous derecho swept across five states and produced incredible damage with gusts above 120 mph. In spite of all the damage, the death toll was very low with just four (at least one was attributed to a traffic accident during the storm). From the time the derecho reached central and eastern Iowa -- when it was strongest -- the warnings were quite good. In spite of that success, the derecho was the justification for creating even more thunderstorm warning criteria (listed below). Onesie. 

Let's review where we stand. For tornadoes, we now have: 

• Severe Thunderstorm Warning with "tornado possible"
• Tornado Warning 
• Particularly Dangerous Tornado Warning 
• Tornado Emergency 

For winds, we now have, 

• High wind warning (40 mph for one hour or more).
• Severe thunderstorm warning (58 mph)
• Considerable damage severe thunderstorm warning (70 mph) 
• Hurricane warning (75 mph)
• Destructive severe thunderstorm warning (80 mph)
• Extreme wind warning (115 mph)

For hail, we now have,

• Severe thunderstorm warning (1")
• Considerable damage severe thunderstorm warning (1.75 inches)
• Destructive severe thunderstorm warning (2.75 inches)

Of course, no member of the public will be able to keep these straight. And, they shouldn't have to. This is our misguided thinking. I cannot find any a priori scientific papers published before these changes. 

And, without any sense of irony, the NWS's press release for all of this includes this laugh,

The addition of [the new criteria] are part of the broader Hazard Simplification Project to improve communication of watches and warnings to the public.

There are lots of smart meteorologists in the NWS and, in the last decade, social scientists have been added for guidance on these issues. Yet, the onesie problem continues to complicate the storm warning system at an accelerated rate. 

We meteorologists are supposed to be scientists. Changes to the warning system should be made only based on solid science from multiple instances and even then only after balancing the issue of public education.