Weather Two (Forecast Accuracy? Twitter for NHC Updates)

So, as forecast, today dawned gloomy and rainy, just as yesterday had ended.  I was up at dawn, actually, so I know the weatherman was right.  You might assume I'd be encouraged by the accuracy of what the past 48 hours forecast has produced, but I'm not.  You see my little sojourn onto the world of weather forecast accuracy has shown me too much.  This is just an example of what is called the "persistence method of forecasting."  You see, forecasting the weather is a highly complex problem, riddled with opportunities for things to go wrong.  Being accurate in your forecasting requires employment of highly technical methods to make it all come together.  The most-utilized of these methods is the "persistence method."  The weather forecaster looks at what yesterday's weather was and assumes it will "persist."  If yesterday was hot and muggy, today probably will be, too.  If it was rainy and gloomy (wait for it, wait for it, here it is) today probably will be, too.  The "persistence" method is a great aid to today's meteorologists.

If you read yesterday's discussion of weather accuracy, you know that this issue has become very important to me.  Living 35 yards from water that eventually connects to the Atlantic Ocean means this is oh so personal, my behind is on the line, so to speak.  If you think of a day when rain was forecast and it was instead a sun-filled day, you are pleasantly surprised and you think nothing else about it.  On the other hand, if the day is forecast to be sunny and instead it pours, you are looking for a phone to call the forecaster and complain, or you are tweeting some insult in their direction.  Either way, in a few weeks it's all forgotten, presumably.

But put yourself in my shoes, if they get it wrong about coastal flooding or tropical storms (or even the h-word, we don't say that word given that whole water-connectedness thing I explained), I'm washed away.  So, I'm getting a lot more interested in accuracy.   Believe it or not, there is no arm of the National Weather Service that works on measuring the accuracy of past forecasts.  The same, as I mentioned in my earlier commentary on this subject, is true for the private sector, home to all those professional meteorologists who don't even keep up with the accuracy of their own past forecasts for bragging rights, resumes, salary reviews, nothing.  

Ah, but the private sector to the rescue.  I found the father of a 7th grader who, on discovering there was little or no local information available began a test in his area by collecting his own data.  It began as his son's science project, collecting data for a week, Dad went a little overboard, and was still going after 7 months when he wrote about it for the Freakonomics website.  Here are some of the comments he heard from TV people: 

“We have no idea what’s going to happen [in the weather] beyond three days out.”

“There’s not an evaluation of accuracy in hiring meteorologists. Presentation takes precedence over accuracy"  "All that viewers care about is the next day. Accuracy is not a big deal to viewers.”  He concluded that few, if any forecasts were accurate consistently beyond the next day.  

It turns out, however, that there are a few more nerds like him who have seriously created studies and are assembling a large body of work.  One began doing it for money on a part time basis in 2004.  Today, he has started his own business full-time and is growing by leaps and bounds.

Today, his conclusions are based on forecasts from around 800 cities in the US, or about 275,0000 forecasts each year.  That's more than 2 million forecasts in his data base.  His name is Eric Fluehr and he has a commercial web site from which he sells data to businesses including commodity futures traders and media companies.  He also has a free consumer web site that allows you to look up accuracy ratings nearest to your zip code and read his blogs, etc.  It's called ForecastAdvisor.com and it's worth a look.  It's nice to know who is the most accurate provider of forecasts in your area.    

Nice-to-know material, but the forecasts I have to watch are the tropical storm activity from the National H----cane Center, which is federal government.  With a little digging, I learned that this summer, at the start of the six month h----cane season, they doubled their computing power, after-effects of Sandy most likely, but I am thrilled to learn it.  Maybe I can stay here after all.  I tried to sign up again for email notification of all their updates on storms as they are identified, only to learn that due to its cost the system has been discontinued, however, you can follow them on Twitter at @NHC_Atlantic.  This is my return to Twitter after a four year absence, i think.