Question: I hear a lot of weather forecasts where they mention the term “attribution science” with respect to some extreme weather event. Supposedly it allows them to say things like “this hurricane was made 50% more likely because of climate change.” How do they figure that out? — GP, Pensacola, FL
Answer: It’s a fairly complex process mathematically. Both statistics and computer climate models are used. You might find it useful to follow that link and read a previous post about how climate models work.
So I’m going to focus on the statistics part here. The graphic shows what mathematicians call a bell curve, specifically a Gaussian distribution. It can be used to describe any set of data that involves random variations. That includes data about coin flips, dice rolls, wind speeds, rainfall amounts, wildfires, tornado frequency, etc.
An important number in this figure is σ (sigma) which is the standard deviation of the data set. It’s a measure of the amount of variation in the data, and determines whether the bell curve is narrow or wide. In a Gaussian distribution, the event will fall into the “average” range 68.2% of the time. It will fall into the “above average” range 13.6% of the time. And it will fall into the “extreme” range 2.1% of the time. That’s the normal variation we would expect in events driven by random inputs.
[Some readers may have wondered why all the probabilities in the chart add up to 99.6% instead of 100%. Two reasons: First, the bell curve doesn’t end at ±3σ. It extends to infinity in both directions, asymptotically approaching the horizontal axis. Second, all probabilities are rounded to the nearest 0.1%.]
And now we can get to how attribution science actually works to analyze extreme weather events. You might want to think of it as a type of forensic science where, after an extreme event, you can “go back in time” and investigate what actually caused it. That’s where the computer models come into play …
Let’s take the Great Texas Freeze back in 2021. It was so cold that natural gas froze in the distribution pipelines, and it overloaded the grid with electrical heating demand. People were left without power for days and hundreds died. To see if climate change was in any part responsible, scientists feed the meteorological data for the week prior to the event into a computer model. But with one big change in the data — they change the percentages of carbon dioxide and methane in the atmosphere to the pre-Industrial Revolution values (280 ppm and 700 ppm, respectively) instead of using the current values (420 ppm and 1900 ppm, respectively). Everything else in the model remains unchanged.
Then they run the computer model(s) thousands of times to get a reliable average result and compare it to what really happened. If it never happened in the models, then clearly carbon dioxide and methane are to blame. If it happened a few times and they have actual data points to place on the Gaussian distribution, then they can say that a given event was made X% more likely and Y% more extreme because of climate change.
Turns out the Great Texas Freeze was not a result of climate change and is within the range predicted using pre-Industrial Revolution data — but it was definitely in the 2–3σ range.
In contrast, the extensive heatwave in the Pacific Northwest in that same year, where 600 died, is indeed attributable to climate change. Using the same method of analysis scientists found the models could not replicate the heat dome, implicating climate change as the cause with high certainty.
Attribution science is a really powerful development in the world of climate science, and it has big political, legal and economic implications. Not just for the insurance industry and property owners, but for the companies that are drilling and selling and burning fossil fuels. With attribution science you can ask, “Who is responsible for putting carbon dioxide and methane into the atmosphere, and what damage has this caused?”
Some critics may argue that climate change is a hoax, but when you look at a graph of carbon dioxide and methane levels since the start of the Industrial Revolution, it’s difficult not to draw a correlation.
If you’d like to learn about attribution science in more detail, I highly recommend this resource:
https://www.carbonbrief.org/qa-the-evolving-science-of-extreme-weather-attribution/
Next Week in Sky Lights ⇒ Controlling Aircraft Yaw with Differential Engine Thrust