Why Is the Weather So Hard to Predict?
Last week, “The Blizzard That Never Was” left us all wondering how the meteorologists got it so wrong. Tech Talker explains how the technology that meterologists use to make their predictions works and why we should cut them some slack.
Listen
Why Is the Weather So Hard to Predict?
Hey everyone, this week I’ll be covering the technology behind weather prediction.
If you live on the East Coast of the United States, you were probably expecting a massive, life-threatening blizzard last week, but the much-anticipated “storm of century” for the most part turned out to be a dud.
Instead of dumping feet of snow across the northeast, the storm in many places dropped merely a few inches of the fluffy stuff.
That raised a lot of questions about why meteorologists’ predictions were so off in some areas. To that end, there is a combination of factors that led to this less-than-perfect prediction. So, let’s get started by learning exactly how meteorologists predict the weather..
How Meterologists Gather Data
Meteorologists try to gather as much data as they can. By looking at all of the weather patterns in the area and satellites circling overhead, they gather data such as temperature, pressure, wind speed, and rainfall. They also collect the recorded data from previous years.
Doppler radar is a useful tool that meteorologists use to gather more data. Based on the Doppler effect, Doppler radar is very similar to the type of radar used in the military to detect incoming planes and missiles. However, instead of detecting incoming aircrafts, Doppler radar detects incoming storms.
Meteorologists also gather data from the terrain, such as elevation and types of ground cover (asphalt, grass, city, or farmland).
Number-Crunching and Predictive Weather Models
From here, they input all of this data into a huge modeling computer. This computer takes all of the information that the meteorologists collect, organizes it, and uses it to create a computer simulation of what it predicts will happen.
Each of these simulations takes a huge amount of computing power. Some models may even require large supercomputers to crunch all the numbers. This depends on the complexity of the simulation being run and the amount of data being fed into the computer. I worked quite a bit with these models while I was in college, and a number of these simulations took days to finish.
See also: The World of Big Data Part 1 and Part 2
When using computers to simulate weather patterns, there are a lot of assumptions and estimates the program must make. For example, a meteorologist may only take 10 temperature readings from a city. However, there are way more variations in temperature throughout a city. So the computer uses these 10 temperatures as an average for the model.
These averages and assumptions often cause the model to be imperfect when predicting what is going to happen and where. Inputting more temperatures, however, may mean the model will take much longer to run. In some circumstances, this data may not even be available.
A Weather Simulation’s Margin of Error
In weather simulations, all of these imperfections can have a profound effect on the overall result of the model.
Weather simulations also have a margin of error due to the type of data collected and the assumptions made in the program. For example, a model may predict that the temperature for the next day is going to be 70°, plus or minus 3°. The margin of error in the prediction occurs because the model is either lacking the data to make a more accurate prediction, or there was something unknown in the simulation that the computer had to estimate.
Meteorologists are also provided with imperfect terrain maps. The missing data is often estimated, which has an effect on how air moves around it. Under most circumstances, this estimate wouldn’t affect very much, but in weather simulations, all of these imperfections can have a profound effect on the overall result of the model.
To make things even more difficult, there are very strict time constraints on weather forecasts. Every time additional data is made available, the simulation needs to be tweaked. The simulations are constantly checked for more precise results and then updated.
Ultimately, there is no way to know every aspect about a weather system or storm. No matter how much time you spend running a computer simulation, it will still not be 100% accurate.
The Blizzard That Never Was
Last week’s blizzard blunder is a perfect example of the nature of this sort of predictive modeling. Given the conditions provided by the data gathered and the satellite imagery, the meteorologists did their best to predict how much snow was going to fall and where.
This is similar to watching an action movie. The guy typically gets the girl, and the hero triumphs over evil. How do you predict this is going to happen? Well, you’ve probably seen a lot of movies that have a similar theme. None are exactly alike, but they all follow a certain archetype that leads to a similar outcome.
Movies, however, can take an unexpected direction, much like real-life storms. Although meteorology is a very sophisticated science that uses massive amounts of data, computing power, and human intuition, there’s absolutely no way for meteorologists to always get it right.
So the next time you’re watching the news and the weather comes on, keep in mind that they are not just guessing at the highs for the week. They’re taking the data that’s available to them and running it through a very complicated scientific model. This model then produces a possible range of conditions. If accurate data is unavailable or input incorrectly, their predictions could be totally wrong!
Just know that it’s never going to be 100% right, and you should never expect it to be.
Well, that’s it for today! Be sure to check out all my earlier episodes at quickanddirtytips tech-talker. And if you have further questions about this podcast or want to make a suggestion for a future episode, post them on Facebook QDTtechtalker.
Until next time, I’m the Tech Talker, keeping technology simple!
