Wednesday, July 11, 2012

Weather Whys Wednesday: Forecast Confidence


One of our readers, Bobbie, wrote to ask us why heat waves are so predictable, whereas snowstorms and other winter events are less predictable. When meteorologists forecast, they rely on experience, observation, and the models. The models give us different answers to the same question, and our observation and experience helps us fill in the gaps. But certain things are well handled by the models and allow for meteorologists to make good predictions.

Reanalysis map from July 22, 2011, one of the hottest days
ever in Philadelphia. Credit: PSU E-Wall
First, let's ask: How do heat waves differ from snowstorms? Well, generally, heat waves (by definition) have to last three days or longer with temps of 90+ degrees. Snowstorms come and go in as little as 12 hours, but usually no longer than 36 hours or so. So snowstorms are faster moving and heat waves evolve more slowly. While heat waves involve precise temperatures (you need to be above 90 obviously), there's no set rule. If it's 93 degrees or 100 degrees, it's still hot and a prediction of a heat wave can verify with plenty of advance leeway. In winter storms, if you have temps of 32 degrees vs. 33 degrees, that can make a world of difference in what you see. Winter storms involve precipitation forecasting, a weakness of the models in general. Heat waves can involve precip as well, but you can still hit 95+ degrees and get a thunderstorm late in the day. So it requires a little less precision.

So what makes things so predictable in heat? Models tend to see large patterns and trends evolving quite well in the 4-8 day timeframe. You can tell if the pattern is going to transition to heat usually several days in advance. A well established heat dome over the middle of the country might shift east (ala early July). Wind direction may shift southwesterly or westerly with a building ridge in the East and trough in the West. Either way, you can see that the pattern is shifting toward heat. And it's fairly obvious.

The March 1958 Snowstorm: An example of sharp changes
in snow over small areas! Credit: NCDC/NESIS
Snowstorms involve multiple problems. A shift in track by as little as 50-75 miles (or even less!) can mean the difference between rain and a mix or snow. And it can impact much of the "viewing" area with wild shifts in the forecast. For instance, a low tracking 50 miles further offshore could take Philly out of snow completely and hammer the Jersey Shore. The opposite holds true as well. The precip may not all make it to the ground initially because of dry air, which can often occur in winter...and your blizzard quickly starts off as nothing more than a flizzard. Examples like that are part of the problem.

Snowstorms and heat waves can both be classified as synoptic scale events. More info on the synoptic scale and what it means is here. While both snowstorms and heat waves also have finer scale (mesoscale) details that have impacts, it's obvious those in the snowstorm can make HUGE differences on impact to people (bands of snow sitting over one town for a long time, a small area of mid-level warming that changes snow over to sleet for a time, etc.). In heat waves, it's more or less one place checks in at 99-100, while a few towns over it's 95-96 degrees...limited real impact to you and me. So the question isn't so much why meteorologists are so much better at predicting heat than snow. It's more a question of: What would you rather predict? A thousands of square miles area of heat? Or a hundreds of square miles area of precipitation that if it tracks anywhere differently to the east or west means drastically different results? I will take heat waves ten times out of ten as a meteorologist for my own sanity!

As a sidebar, I'd actually argue that while meteorologists are good in general at predicting large scale, long-term shifts, such as heat waves, there's a long way to go in the specifics department. I'm guilty of this too, but there are many a time where forecasts of 92-93 seven days in advance turn into 98-100 degrees in reality. A lot of that is to not overhype things too far in advance. But a good chunk of it is also the state of predictability in our science. Unlocking the key to those specifics is something we need to continue to work out, but needless to say, that impact on people's day to day plans, work, etc. is usually not as impacted by that sort of temperature error in summer. So heat waves by their very nature are just more predictable than snowstorms.