Tuesday, June 21, 2011

Is Forecasting the Weather Art or Science?

In this technological world that we live in where the resources for weather related information are readily available for ANYONE in seconds, it still surprises me how many of us still hang onto the belief that simply because we have "super-fast computers", weather forecasting should be a science that is an already mastered discipline. In the minds of most, because of advancements in technology, forecasting should be an old science relic that should have been perfected by now. 

I am a firm believer that weather foreacsting is made up of two parts:  Science and Art.  The science encompasses objective observation and analysis. The art deals with formulating your forecast so it best fits everyones "perception" of what is happening--  a poor-man's exercise in psycho-analysis.  More on that in a moment.

First the science:

Contrary to what many believe, forecasting accuracy is MUCH BETTER than it was 10 or 15 years ago.  This is in part to several factors:  The first is the development and implementation of satellite data. The second is faster computers. The third is more advanced weather simulations (called computer models)
Satellites monitor each and every square inch of the planet. Soil moisture, ground temperature, atmospheric temperatures, infrared radiation, wind profiles, pollution, dust, pressure, ice coverage...you name it.  All of this data is compiled and shuttled onto websites each second of each day.  The first decade of the 2000s will be remembered as the time when volumes of satellite data became available on the internet to most everyone. Want to analyze ocean surface temperatures? You can do it.  Want the dataset for ozone over Antarctica?  Its right there!  Our understanding of the oceans and atmosphere is so much greater with this library of satellite data so readily available.

Computers are much faster and getting smaller each year.  A personal computer from 2006 is grossly outdated. In 2015, the best and faster computer in 2011 will be a turtle in comparison.  Faster computers means faster and more detailed computer simulations of thunderstorms, hurricanes, ocean currents and other patterns. Combine the faster computers with more detailed data and you get a much better simulation on what the specific weather will be for a certain area. 

That said, simulations have and will ALWAYS have limitations.  First, simulations by their very nature are not exact answers.  They are approximations. The complex math that describes the atmosphere is one big approximation.  Factor these inexact simulations created by humans with super-fast computers, also made by humans and you get two things: MUCH better forecasts that still have some degree of error.

Now the art (psychology):

Weather accuracy is highly subjective.  If you mention on the air that "90% of the area will get thunderstorms today" and 90% of the area gets rain and storms, your accuracy is pretty darn good.  Yet for the 10% who didn't get any rain, you are the worst forecaster in history!  

Your "weather reality" is what is over your head at any given time.  Much of the public as a whole views the weather around them with a biased lens. Substitute this "weather reality" with politics, the economy, religion, crime, your friends, it doesn't matter. We are all biased to some extent.  That's what makes us human.  Its our built in selective perception of the weather and events around us that gives us perspective.  The art of forecasting is formulating your words based on how people will "perceive" what is going to happen. 

20 years ago, the only sources of weather were:  television, radio, newspaper and word of mouth.  Now, add real-time Internet and smart phone updates, 24 hour cable.  Its and endless bombardment of weather both beautiful and destructive.  We are all conditioned partly with this overflow of information to gauge the weather events as something new.  How many times have we heard "I've never seen anything like this before!" after a weather event?  The most recent weather is overemphasized because its fresh in our minds. This shapes our viewpoint on weather a few days to a few weeks in the future.

Most outside of the field of weather find it very difficult to grasp the fact that the weather is one big approximation. Weather should be exact; we all want a forecast that fits a nice and neat one-size-fits-all package.  I'd like to say that I make a forecast whether short or long term with a cold, rational eye rather than my biased, emotional side.  But I don't.  I take into account how the general public will react to my EVERY word knowing that most selectively perceive the weather to fit their "sphere of reality".  I learned that quickly years ago.  For all of the simulations, super-computers, highly detailed satellite data, it doesn't matter how exact your forecast is or what scientific reasoning you used in coming to your conclusion, people will ignore the facts and the data that disagree with their perceptions and will "rationalize" what they want and react accordingly. More often than not, the reactions are negative.  Worse still, its accumulative.  The more we selectively perceive the weather to fit our negative connotation, the more negative our reaction and the more rigid our bias becomes.  Its a vicious circle that feeds on itself. 

This is called the Disconfirmation Bias. Its the tendency to accept supportive evidence of a belief uncritically, but to discount evidence that challenges that belief.

Here is an example:  Go back to last winter.  The first big snowstorm forecast calls for "4-8 inches" for the following day.  In your backyard, you get 3 inches yet 80% of the area received 6 inches.  You say "what happened to the 4-8 inches?"  Second snowstorm, the forecast calls for "6 to 10 inches".  You get 6 inches.  The majority of the area received 9 inches.  You say "what happened to the "6 to 10 inches?".  By the third snowstorm, you are already frustrated by the "lack of accuracy" so you are primed to think negatively. The forecast calls for 5 to 10 inches.  You get 5 inches and you say "what happened to the other 5 inches?" 

The forecast was accurate given the range in the snowfall prediction.  Since you were preconditioned to believe the forecast would be way off because of your established bias, no amount of evidence in the world will break you of that bias!

Weather is an old science. Weather forecasting is a field that continues to grow.  But the heart of a forecast lies in molding the science to fit the psychology.  Human nature is tough and stubborn.  I'll try to be better sculptor.


Dennis Boylan said...

couple of thoughts. I agree that there are those who do react the way you state; you said 5 inches of snow and I only got 3, your forecast is wrong! And I am sure these are the people that you hear from the most. But on the other side of the coin, I think the vast majority of people know and understand the probabilities associated with a forecast. Having lived in the snowbelt for over 30 years, well acquainted with the probabilities and now a recent westsider, I see it even more ( and am enjoying that the deviation on the west side is more to the low side of a snow event than the high side!). So I think you need to give your audience some credit that there are many, the majority I believe, that understand the probabilities in a forecast and that an 8 day forecast is a guide and not an actual.

Second, you started this post off with some information regarding satellites and technology, stating "Satellites monitor each and every square inch of the planet" and provide a wealth of climatological information. Given that, is it not feasible to get a highly accurate set of conditions for a given location at a given point in time. With that information would it not be feasible to take that information and use it to produce a highly accurate abet a short term forecast?

Would it also not be possible to use all that data in combination with some education on weather to make us smarter about weather implications. I am thinking now of people who, like me are boaters and need to understand the weather especially on the lake. There are others two who may be event planners like with the RNC coming up that would need to understand very short term weather and its implications for that day.

Lets build the app to do this!

Bottom line, we appreciate the forecasts, understand the issues and look forward to the continuing sharing of information and perspectives!


Scott Sabol said...


You are correct however there is a significant % of the population who don't understand probability and that is why I address it here and in other blog posts.

Re Satellite data: Its not as simple as this (I wish it were). Satellites don't give us as much information as you might think that is useful here on an hourly or daily basis in, say northern Ohio. That information has to be fed into a computer model and run VERY often. That type of high resolution forecast you speak of is still a long way off. The HRRR model is close because it assimilates radar data and is run each hour. But it has its faults and biases. So something as hi res as what we would need for a county basis forecast is not feasible...yet.

I write about the psychology because its a HUGE element in how we create forecasts. The better we understand how we think and what drives our thought process, the better people we will become. Unfortunately social media doesn't allow this too much