Fishel: Expressing uncertainty in weather forecasts is not being wishy-washy
Posted March 7
Undoubtedly, many of you are familiar with the chaos theory. Dr. Edward Lorenz published a paper in 1963 in the Journal of Atmospheric Sciences in which he spoke of the chaos theory. As it relates to weather prediction, his point was that in numerical weather prediction, the atmosphere is very sensitive to initial conditions, and that imperceptible differences could lead to large differences in the forecast.
Since it is impossible to generate a "perfect" analysis of the current atmospheric state, no computer forecast has a prayer of producing a perfect forecast. Until recently, meteorologists were restricted to using what we call deterministic forecasts. In other words, a given model was saying to us "here's my solution-take it or leave it!".
Well, with few if any alternatives, we almost had to follow it, although experience and the school of hard knocks would help many forecasters to learn the biases of a given model, and to try to account for them.
Now, with more computing power than at any time in the history of mankind, we can generate what we call ensemble forecasts. One technique used widely is to "perturb" or alter the initial conditions of the model just slightly to 20, 30, or even 50 different extents. By doing this, in a sense, we are accounting for those errors in the analysis, with the goal being to see how sensitive the atmosphere is to these changes in the starting point of the model forecast.
Ensembles allow meteorologists to assess the degree of sensitivity of the model forecast to these small changes, and to quantify the amount of uncertainty in the solution by looking at the variance in the forecasts. This is 21st century technology at its finest!
Now, I can certainly understand how this "new" verbiage could be misinterpreted as "copping out", or taking the "CYA" approach to forecasting. You might have even had this thought: "Gosh, when I was younger, these TV guys and gals used to take a stand and commit to a forecast. Even if they were wrong, I respected them for committing to a forecast".
Well, guess what? We didn't know what we didn't know! Now we have a much better idea as to how, why, and where things could go very wrong. When my colleague Nate Johnson and I took to the road late in the evening on Jan. 6, we experienced rain at the WRAL studios, slushy roads at the Harrison Avenue exit off Interstate 40, and all snow at the Davis Drive exit of I-40.
Those are big differences over a very small distance. Would you rather I, or any meteorologist, "guess" where that transition zone will be, or be honest with you and inform you of the large variety of conditions possible where you live? I would hope the latter.
When a meteorologist tells you they don't know, or they're not sure, it's not that they haven't done their homework, or that they're just not willing to commit. It does mean that they're being honest and forthright with you when they talk of uncertainty. If you hear "Duh, I don't know", that's not good. But if you hear an honest appraisal of a challenging forecast scenario, and the sharing of the 2 or 3 most likely scenarios, you are the recipient of a forecast based on the most advanced science available to the meteorological community.
Our challenge is to continue to strive for better ways to communicate that level of uncertainty, not just on television, but on all the platforms where weather forecasts are disseminated. Apps on your phone, by their very nature, do not provide much context-you see an icon and a number for each day. That has to change, and WRAL, along with TV stations all across the country, are trying to figure out how to do just that. In the meantime, if you have any ideas as to how that forecast certainty or uncertainty can be better communicated, by all means let us know.