Communicating Uncertainty to the Public

People want answers. How much is it going to snow in Orchard Park? Could the sleet make it up to Wheatfield by 9am tomorrow? Those kinds of answers. Sometimes we have those answers but many times we do not.

To start with, meteorology is a physical science which is--to say the least--inexact. We are on a spinning globe, 3/4 of which is covered in water and ice, being heated by a thermonuclear furnace 93 million miles away. We're dealing with fluids and gases and how they interact, along with differing land topography, plant cover, snow cover, & different reflectivity of the land (dark soil absorbs far more of the sun's heat, while ice and snow reflect much of it back toward space), to name a few of the changing variables. Sea surface temperatures, and where warmer or colder temperatures are located in the Pacific have a HUGE effect on weather as well. So, weather is not something akin to Leggo blocks where everything snaps neatly into place.

As for the issue of uncertainty, not all forecasts are created equally. I'm not referring to the training and experience of various Buffalo tv weathercasters and meteorologists--I'm referring to forecast confidence. Verification scores of forecasts generally show that competent meteorologist produce forecasts that are rather accurate 80% of the time for the first 36 hours. But that includes some relatively easy calls, such as when we're in a stagnant, stable weather pattern in which conditions will vary little for a few days. For the tougher calls, I'd venture a guess those probably register closer to 60% accuracy, which would still be pretty good, and getting better.

When meteorologists occur scientific conferences (for example, WIVB sends me to an American Meteorological Society once each year for continuing education), a great deal of time is spent by lecturing scientists on the necessity to communicate UNcertainty, in order that our public understands a particular forecast carries certain probabilities of not working out. The NWS uses probability percentages of precipitation as one way of communicating uncertainty. For myself, I've chosen not to use them in our on-air forecasts because my experience tells me many people don't understand the meaning of those percentages. A 30% chance of snow, for example, seems to mean to many that 30% of the area will have snow. That's not what it means, though. A 30% probability means that any given location in a forecast area has a 3 in 10 chance of having measurable snow. I've even seen a few meteorologists make this error. So, I use more informal means of communicating uncertainty, as you've seen on this blog prior to the weekend storm and on the air--letting folks know that computer models are at odds with one another. Sometimes it's in my inflection pattern and, I suppose, facial expression. If I had a paying client who wanted the most precise information possible, I would assign probability percentages to that forecast.

In any case, it's very important for you to know the uncertainty in a forecast or, for that matter, in environmental concerns and scientific controversies. Let's take it to a level where the stakes are higher; suppose you go to your internist with a worsening dull ache in your abdomen. After ordering tests, I think you'd want your doctor to give you the complete perspective on how certain or uncertain he or she is about your diagnosis before he begins treatment.

So, we're not going to present each and every forecast to you as if they were all created equally. Some are high confidence forecasts, and some are low, and you need to know that. Speaking of which, as of today, a forecast for a White Christmas with new snow is a lower confidence forecast.