Clear Weather

Uncertainty in the models and record-breaking cold

One of the first weather puns I learned when I was a kid in answer to the question "what's the weather forecast in Mexico?" was "chili today, hot tamale." Little did I know then that I'd be answering serious questions about the weather in my adult life. I love doing it, but sometimes trying to answer can be a bit stressful. Let me explain.

Any forecaster will tell you that predicting the weather more than two or three days in advance is a bit of a gamble. The forecast models are only as good as the initial data with which each run begins. So, in building the various numerical weather prediction computer models, scientists do their best to account for every piece of data they can get their hands on including current temperatures, barometric pressure, wind direction and speed, and water vapor among other variables from as many places as possible. The more data and the more accurate data we have, the better. The models then take this data, assumptions based on theory, and proven mathematical equations and spit out a forecast. Each model is slightly different in how it works, which is why meteorologists have so many to use in determining our personal predictions.

When the models are in agreement, our job is easy, or should be in theory. Sometimes the models all agree in the morning run, but then in the afternoon, a few flip-flop and show much different results. Occasionally, they all flip-flop. One way meteorologists can hedge their bets is by taking an average of all the model results and using that as the basis for their forecasts. This process is used in other sciences as well. When given results from multiple, independent, well-informed resources, the average is usually more accurate than any one educated guess.

Even with the best results available, there will still be uncertainty in weather forecasting. The farther out the forecast is, the more uncertainty there will be, partly because we don't know everything, and we may not - ever. Meteorologists do the best we can with the information we have. Some will tell you they have high confidence in every forecast. Good for them! Most will admit that there are some forecasts that we have less confidence in than others. Intellectual honesty requires us to do our best to communicate this uncertainty with the public. The hard part for us is figuring out the best, clearest way to do that.

Take our current, real life example of Tuesday's precipitation. On Thursday morning, the models showed that the temperature would remain well above freezing and it would fall as rain throughout the period. Thursday afternoon, the models did a 180 degree turn and showed temperatures at or below freezing for long enough during the period to consider a wintry mix. Our forecasts had to change dramatically in less than a twelve hour period. With a wild swing like that one, it is hard to say right now that the models won't swing in the opposite direction this evening. We really are too far out time-wise to be certain.

The closer Tuesday gets, the more certainty we should have. Until then, it is always the safer bet to prepare for the worst. Is your car ready? Do you have rock salt for your porch steps? What if schools are delayed or closed? Are you prepared?

Sunday's forecast, on the other hand has been much more consistent and could even be record-breaking. As of this writing, the National Weather Service is forecasting a high of 28 degrees. Our record coolest high temperature for February 15th, occurred in 1943, and it was 29 degrees. If the forecast 48 hours in advance proves accurate, Sunday could become our coldest day on record for that date. Bundle up!