Our Dilemma

Tuesday, January 11, 2011
8:26 P.M.

 

Thankfully I don't have this dilemma. Who would possibly want to go to college?

I have to make this quick, as additional homework warrants priority, but I just felt like I needed to have a quick note to explain what is going on and all the hype that we received earlier this week and how none of it is coming true. Before, the models were predicting a historic snow event. The biggest snow event in our lifetimes, possibly even in our parents lifetimes if everything panned out correctly. It could have been the snowiest week in Seattle history. This is not an exaggeration. That possibility seemed very real, and models were continuing to forecast it only 4-5 days out, which meant that it was well within the believable range of probability, though admittedly far from a slam dunk. 

Then, all of the sudden, the low pressure system was forecast to go to our north, and here I am at my house, distraught with "what could have been" with a measly .10 inches of snow on the ground, forecasting a changeover to rain in an hour or so. 

Here is the dilemma - should forecasters tell the truth to the general public about what they see in the models? Or should they wait before claiming there could be a gigantic snowstorm? Is it better to give the public advance notice of a serious event developing? Or is it better to wait until you are sure to prevent confusion from occurring and your tv ratings/reputation from going down if you are wrong?

I supported what the media did on this one - they went ahead and forecast a giant snowstorm and then changed when the models changed. Sure, it got super hyped up and nothing really happened, but if the original forecast was right and the media didn't let people know, people could become stranded. People could die. And I think that we can all agree that massive disappointment is better than death. 

What I feel like the general public and even some professional forecasters don't realize about forecasting is that is a field of probabilities. There are ensembles of models, and they all show something slightly different. There is a feature called "probcast" at the uw that gives a forecast of a probability of what things might occur based on averages of different models. It can be found at http://www.probcast.com/. Although it is not very specific and still has a lot of work to do with regards to development, it is a great start, and it is often very accurate for short range forecasting, particularly with weather events where the models vary wildly and you can't really pick a solution. When meteorologists make a forecast, they generally look at a model and interpret what it is saying. But all the models show different things, so wouldn't it make more sense to give a probability of what might happen? I think it would.

But the question is if this is friendly to the consumer. A lot of times, I'll be asked if we will get snow or not, and I'll start explaining convergence zones and wet bulb cooling and adiabatic warming/Chinook winds, and people will just interrupt me and say, "look, do I have to do my hw for tomorrow?" Do people want a probability forecast that is more accurate and gives a better glimpse of what truly goes on in the profession I plan to enter, or do people just want a simple number? I'm guessing they would just want a simple number with an icon (smiley face sun optional). There are meteorologists who are opposed to forecasting with probabilities because in their eyes, one thing is going to happen, and we are going to forecast the one thing we think is going to happen. Those opposed to probabilistic forecasting also point out the contradictions in National Weather Service forecasts. They might say "How can you have a 50 percent chance of snow, and still have it accumulate to 2 inches? Does that mean it will snow 2 inches if it doesn't snow at all?" Obviously, it means more like it will so 2 inches if it does snow, but the wording can be a bit ambiguous at best in NWS forecasts. Even though the NWS forecasts can be confusing sometimes, I still think that giving PoP (probability of precipitation) forecasts along with quantitative forecasts make the most sense.

So what should the commercial sector do in regards to these storms? Should they give a big heads up? Should they play it safe? My answer is that they should give a big heads up if it occurs in the models more than twice, and mention it if it has occurred at least once. But they absolutely NEED to stress the probability aspect of it. I actually wrote an essay about computer models for my computer science class (maybe I'll post it on here sometime for ya'll to read), but the bottom line is that forecasts are chances of what might happen, they are not what will happen. Commercial stations need to give larger warning times (as a safety measure) but STRESS UNCERTAINTY. They glaze over it, but when snow-crazy kids like you Garfield students read the forecasts, "there is a slight possibility of two feet of snow" becomes "Jeff Renner says there could be two feet snow," which becomes "Jeff Renner says it is going to snow two feet" which becomes "OMG!!!!!!!!!! IT'S A NEW ICE AGE!!!" Of course, King5 contradicts itself often because the forecasters explain that it might happen, and then all of the sudden Parka Boy (Jim Foreman) comes in and tells us how we could get devastated with two feet of snow, but nobody listens to the "could" part of it.

When I write these forecasts, I try to show you a little bit of my world in forecasting - I give models, I give probabilities, and I generally take more risks in forecasting and mention things further in advance than the tv stations and even the NWS or Cliff Mass. But I cannot stress enough that this is a field of probabilities, and while it is fun to look at models of crazy events happening (and beneficial to those who could be affected), nothing is certain and things are bound to change with each model run.

I would like to hear from you guys on this! Is it better to play it safe, or to give lead advance time? Would tv weather be more interesting if they showed computer models? Do you have a favorite tv station for weather? Do meteorologists stress enough that every forecast is a forecast of something that has only a chance of occurring? Post on this blog! I'd love to hear your opinions. Cliff Mass routinely gets 20 comments per post, often times more. Let's bring the ruckus and knock him off his top spot for once!!!

Charlie

0:50