Meteorologist vs. Model
As I have been watching the preliminary ideas on the “blizzard”, “snow storm”, “system” that is supposed to effect the Northeast and Mid-Atlantic come in from a myriad (Accuweather, NWS, Weather Channel, etc.) I have noticed that there is a tremendous deviation on how each are interpreting the same raw data that is being provided by the forecast models and how that deviation has greatly affected what we are being told about the weather.
A forecast model is a complex computer programs that is run on supercomputers and provide predictions on many atmospheric variables such as temperature, pressure, wind, and rainfall. A forecaster examines how the features predicted by the computer will interact to produce the day’s weather. Each model has it adherents based on its known strengths, biases, and track record and no one model is used to predict the weather forecast by any of the agencies listed above. To get an idea of all of the different models, take a look at the Penn State University Weather Wall which provides a pretty good look at many of them and their resultant data.
Depending on your source for weather there are significant differences in what forecast you will receive. I consider myself to be somewhat of a weather zealot (someone who has an above average understanding of meteorology and forecasting and doesn’t necessarily trust weather provided by typical consumer outlets, but is in no way a trained meteorologist) and therefore immediately discount TV weather forecasts and the like. Even for folks with less of a scientific interest in the weather but who would still like an accurate forecast to plan their days activities the source makes all of the difference and trust me, there is a difference.
I digress. So, as I have watched the interpretations of the forecast come in on Sunday and early this morning, I have noticed yet another battle being waged between the models and the meteorologists. As is the case time and time again, the models provide an initial solution and change with every given run (generally forecast models are run on fixed intervals to all the models the ability to factor in new data) and the forecasters are left with a decision: trust the models or trust their instincts.
Saturday (10 February 2007) and yesterday morning (Sunday, 11 Februrary 2007) there was pretty widespread agreement that the “Valentines Day” storm would yield a significant all snow event for the Mid-Atlantic and Northeast. This gave many meteorologists the confidence to issue initial forecasts that cranked up the snow amounts for these areas. However, as the day (and model runs) progressed, many of the forecasters began to change their tune. The new solution developed the primary low pressure system (responsible for bad weather) further west and further south and as a result developed the secondary low pressure system closer to the coast which would allow warm air to infiltrate. The net-result of this shift took an all snow event for much of the area referenced above and changed it to an all snow event inland and a mix along the coast up the I-95 corridor for the big cities. One of the first groups of meteorologists who jumped on this bandwaggon was the NWS and slowly a few others followed.
Yet while the models were saying one thing others stuck to their guns and felt that the model runs didn’t factor in things like how cold the air is coming down from Canada thereby affecting where the actual low-pressure system actually forms which in turn would determine who gets a lot of snow, who gets a mix and ultimately where the line of demarcation is between the two. Rest assured all you snow lovers out there, all of the forecast sources I have been watching still think it will be a good storm, but the deviations between each forecast call into question the amount of snow that will actually fall and who will be impacted the most.
My reason for calling this into question is pretty simple. Who do you trust? Do you trust forecasters and agencies who rely heavily on the forecast models to make their predictions, do you rely on those who use a combination of the models and their judgment, or do you trust those who are willing to throw the models out and rely on their past experience and understanding of all of the dynamics that are at play? For the upcoming storm, this appears to be the $100k question.
Personally, I think I trust the latter as opposed to the former. Sure I am a technologist who relies on the power of computers every given day, but at the same time, being a true technologists also requires an acknogledgement that computers simply don’t know everything. It becomes a case of GIGO (garbage in, garbage out). To turn the situation to something more personal, I would much rather go to a doctor who allows me to self-diagnose and uses his/her knowledge and experience in the medical field, than one that relies completely on hard science to make a diagnosis.
I respect scientists and the work that they do, but somewhere in the mix science has to give way to expert judgment. Science only gets you so far…it provides a key to the door, but ultimately how that key is used and what door the key is used on really do matter. I suspect this is the case with the upcoming forecast (and any forecast for that matter). As with the case with many things in life, only time will tell, but for the upcoming storm, I am sticking with the forecasters who aren’t afraid to deviate from the norm and trust their gut.
