Are Probabilities the Answer?
Once everyone started to calm down this morning, many of my colleagues started proclaiming forecast probabilities are the answer to the Manhattan 'bust.' For the time being, I mildly disagree.
The concept is fine, especially for specialized users. The issue is the implementation. The snow amounts across the NYC Metro area ranged from 7 to 28 inches.
So, let's consider my friend and NYC meteorologist Janice Huff. Keep in mind that objective research demonstrates, over and over, that television is the #1 source of information in critical situations.
Imagine her weathercast last night under the proposed probability technique:
The probability of 2" of snow at Newark is 80%. The probability of 8 inches is 60%. The probability of 15 inches is 40%. The probability of 20 inches is 20%.
The probability of 2" of snow in mid-town Manhattan is 90%. The probability of 8 inches is 70%. The probability of 15 inches is 50%. The probability of 20 inches is 30%.
The probability of 2" of snow in Queens is 90%. The probability of 8 inches is 70%. The probability of 15 inches is 50%. The probability of 20 inches is 30%.
The probability of 2" of snow at Islip is 95%. The probability of 8 inches is 80%. The probability of 15 inches is 70%. The probability of 20 inches is 40%.
…and on and on and on across the Tri-State area!!!
Do you want to watch that weathercast!? I doubt it.
Now, with a better range of calibrated forecast tools, better dissemination technology (GPS+GIS, etc.) and a huge education effort, it will likely be possible to make probabilities useful. But, that is not the end of the story. As Chris Matthews said this evening, "The politicians still have to make a decision!" Some of my academic colleagues don't realize that, in the end, a decision has to be made and the decision usually comes down to a yes or a no.
So, can probabilities be useful? Yes. But, for the public-at-large, I do not believe they are ready for prime time.
The concept is fine, especially for specialized users. The issue is the implementation. The snow amounts across the NYC Metro area ranged from 7 to 28 inches.
So, let's consider my friend and NYC meteorologist Janice Huff. Keep in mind that objective research demonstrates, over and over, that television is the #1 source of information in critical situations.
The probability of 2" of snow at Newark is 80%. The probability of 8 inches is 60%. The probability of 15 inches is 40%. The probability of 20 inches is 20%.
The probability of 2" of snow in mid-town Manhattan is 90%. The probability of 8 inches is 70%. The probability of 15 inches is 50%. The probability of 20 inches is 30%.
The probability of 2" of snow in Queens is 90%. The probability of 8 inches is 70%. The probability of 15 inches is 50%. The probability of 20 inches is 30%.
The probability of 2" of snow at Islip is 95%. The probability of 8 inches is 80%. The probability of 15 inches is 70%. The probability of 20 inches is 40%.
…and on and on and on across the Tri-State area!!!
Do you want to watch that weathercast!? I doubt it.
Now, with a better range of calibrated forecast tools, better dissemination technology (GPS+GIS, etc.) and a huge education effort, it will likely be possible to make probabilities useful. But, that is not the end of the story. As Chris Matthews said this evening, "The politicians still have to make a decision!" Some of my academic colleagues don't realize that, in the end, a decision has to be made and the decision usually comes down to a yes or a no.
So, can probabilities be useful? Yes. But, for the public-at-large, I do not believe they are ready for prime time.
Sounds like a challenge. Time to figure out how to present this information in graphical form. Weather forecasting is all about probabilities. Time to educate the public and politicians on that subject. Otherwise, we are doomed to fail over and over again. Calling for specific snow amounts in location X does not work all that well.
ReplyDeleteIt's clear that weather forecasting has come a long way. You (NWS and AccuWeather) are very good at "emergency" situations such as tornado and flash flood warnings. NWS and AccuWeather also got the forecast correct (more or less) for LI and coastal NE. Well done.
ReplyDeleteBut at the same time National Whoops Service and InAccuWeather (play on words) need to answer for the miss in Philly and NYC. Billions of dollars of commerce is at stake now. Based on your information, politicians and businesses make critical decisions that impact a tremendous amount of commerce. Like it or not that is the world you live in.
Keep up the great work Mike, both at your professional life and the efforts you put into this blog.
Brian: I don't insult your employer and I wish you would not insult mine. We put tremendous effort into our forecasts. Here are my thoughts on your comment: http://meteorologicalmusings.blogspot.com/2015/01/a-meteorologists-thoughts-on-blizzard.html
ReplyDelete