
photo: nbc4i.com
Jym Ganahl made a fairly profound comment in the past week or so that we’d all be smart to remember. It stems from his experience as a forecaster and observer and because he’s a man with a keen eye for patterns. As a fellow forecaster, it was a great lesson for me… and for our viewers.
Computer weather-prediction models and meteorologists struggled mightily with the path of this year’s Christmas Storm. Small shifts in the storm’s track made enormous differences in the extended forecast. And, as it turned out, the long-range projections couldn’t have been more wrong.
It was just as frustrating for the weather guys as it was for viewers and travelers. We had all said there was a huge snow storm on the way. A few days before it was supposed to pummel us, all the forecasters began backtracking. From the National Weather Service, to the broadcast mets to the presenters on national television, everyone was pulling back on the Snowmageddon predictions.

FOX8 Extended Forecast: fox8.com
Six to ten days into the future, the storm’s projected track took it right through parts of Ohio and dumped half-a-foot of snow on the region, guaranteeing a very White Christmas. As the days wore on and the models kept churning out new solutions, the storm track shifted to the south. Then the models diverged on both the timing and the track. They were different by hundreds of miles and nearly 24 hours. They produced radically different forecasts.
Part of the problem is that small differences in the early part of the number-crunching can produce larger and larger differences later on.
Even if you despise math, bear with me for a moment:
If you multiply 5 x 2, you get 10… then 20… then 40… then 80…
Continue multiplying by 2 and after about 20 repetitions you’ll get to 2,621,440.
However, change it to 2.01, which is a relatively small modification, and in the same number of multiplications, you’ll get to 2,882,007.425.
That’s a significantly different result from a very small change… and very quickly.
Run it out 50 repetitions and look at the difference:
2814749767106560.000
vs.
3593989615656160.000
The very small difference compounds and, eventually, turns into a very large difference.
That’s pretty much the concept of the “Butterfly Effect.” When a butterfly flaps its wings, it creates a very small disturbance in the atmosphere. That initial ripple then sets off a chain of events, eddies and waves that result in a massive impact down the line; in one example, the end result is a tornado in Texas.
As Ben Gelber mentioned in an earlier post, it’s all part of Lorenz’s Chaos Theory.
That leads us back to the computer models.
Garbage in, garbage out. Remember that from the 1980s? It still holds true today. If the initial conditions in the computer models aren’t the same, the results won’t be either. In the extended forecasts, the models are projecting the formation and evolution of storms and systems that don’t yet exist. If they don’t start the process with reliable or identical data, they will produce wildly different results. They didn’t start with the same butterflies, so they don’t end with the same tornadoes.
The shorter the forecast period, the less opportunity there is for the errors to compound.
In other words, the farther you go out in the forecast, the less accurate it can be… inherently… because the errors just feed on one another.

Graphic: National Hurricane Center
This is one of the reasons the National Hurricane Center’s forecasts come out as cones instead of just lines. There’s a larger and larger margin of error as the forecast moves farther into the future.
Which, finally, brings us to what I’m going to start calling “Ganahl’s Law.”
During the week, as the models and the forecasts continued to change, Jym Ganahl reminded us that we can’t really know what a system is going to do until it (a) forms in real time and (b) moves east of the Rocky Mountains. Once it’s past the Rockies, we can get a better handle on its projected track because then the system actually exists, it has real properties and characteristics and it is finally interacting with known atmospheric dynamics.
Sometimes, that’s less than 48-hours into the future. Before that, it’s some pretty serious guesswork.
And Jym was absolutely right.
Some of us, myself included, occasionally put way too much faith in the computer models.
Therefore, the newly-dubbed Ganahl’s Law states that, “The path of a synoptic scale cyclone may not be accurately forecast in the eastern two-thirds of the United States prior to the physical storm moving east of the Rocky Mountains.”
Sure, I’m willing to accept some minor modifications to the new Law. But you get the idea.
Our meteorological crystal balls may be good, but they have very real limits. It takes a storm like this Christmas’ to remind us that the computer models are not infallible and that extended forecasts, even five to seven days out, are merely a well-educated guess.