One of the things you learn as a scientist is the ability to look at a plot and think, "that just doesn't look right." That's the feeling I got when I saw this plot that Lomborg is currently pushing.
This just doesn't look right. The 1930s were hot in the U.S., but not that hot. And the 2010s barely show up. The first clue something is amiss was that the quantity plotted is "Heat Wave Index". That's mysteriously vague, so I decided to figure out exactly what this was.
If you go the EPA website, it references a 2008 CCSP report, which references a 1999 BAMS paper:…
Here we find that the "heat wave index" counts the occurrence of 4-day heatwaves of temperatures exceeding a 1-in-10 year recurrence.
There's nothing intrinsically wrong with this definition, but it sure seems arbitrary. Why 4 days? Why 1-in-10 events?

When doing science, you should always be worried that arbitrary decisions (e.g., thresholds in an index) will give you arbitrary results.
So the question is: if you change your definition of "heat wave", would you get a different answer?

Let's find out. The most obvious thing to do is to just count the number of hot days.
Let's take Berkeley Earth's gridded daily maximum land temperature over the (approx.) continental US.

For each grid cell, find the 95th-percentile temperature from the entire time series (1880-2020) and then count the number of days each year that exceed that.
Then, for each year, we count the number of exceedances in all grid cells in the domain. This is what the time series looks like for the U.S. Yes, the 1930s were hot, but hot days occur more frequently in the 2010s.
So what's going on? This is what we call cherry picking. Cherry picking is when you analyze a small amount of data to reach a conclusion that the full data set does not support.
There are really two cherry picks here. First is the choice of an obscure (and frankly weird) metric. This metric almost certainly gives an answer that is opposite to what a more exhaustive set of metrics would show.
The other cherry pick is only looking at the U.S. Let's look at the plot for the entire northern hemisphere mid-latitudes (29N-60N). The occurrence of hot days explodes after about 1980.
There are lots of papers that correctly estimate heatwaves. E.g., check out this paper by @sarahinscience and S. Lewis. This is where I got the idea to use the Berkeley daily data.…
tl;dr: This is classic cherry picking: the plot cherry picks an unusual metric and also cherry picks a particular region. Examining the full data set shows the opposite of what's claimed.
Temperatures and heatwaves are certainly worse today than in the 20th century.
In addition, other types of extreme weather are also definitely getting worse. Zero debate about that.
Next time someone shows this plot, please respond with this thread.
Important additional info by the expert, @RARohde:
.@BjornLomborg asked about heat waves instead of just the raw number of hot days.

Here's the times series of occurrence heat waves warmer than the 95th percentile of heat waves for heat waves of 2-5 days (where the temperature of a heat wave is the minimum daily temperature).
Here's the occurrence heat waves warmer than the 95th percentile, where the temperature of a heat wave is the average temperature.

For the last two tweets, the y-axis is the sum of days over all grid boxes over the (approx.) continental US.
It's clear that Lomborg's defense will be "I was just using a plot off the EPA website." But there are lots of plots on that page and he picked the *one* that presented the message he likes.
That's a weak response. If you know anything about climate then you know the plot looks off and you owe your audience a minimal amount of due diligence before tweeting it out.
just for completeness: here are the multi-day heat waves for the entire NH mid-lats. This plot uses the minimum Tmax during a time span as the temp of the heat wave. Plotted here is the occurrence of heat waves > 95th percentile.
This plot uses the average temperature of the time span as the temp of the heat wave. Again, plotting time series of occurrence of heat waves > 95th percentile. Y axis is grid cells times days.

• • •

Missing some Tweet in this thread? You can try to force a refresh

Keep Current with Andrew Dessler

Andrew Dessler Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!


Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @AndrewDessler

12 Sep
In case you’re wondering why 2 feet of sea level rise over the coming century matters, it’s because it turns a 2–4 foot storm surge into a 4–6 foot storm surge. That will increase the damage exponentially.
Sea-level rise impacts are non-linear so that going from. 3 ft storm surge to a 5 ft storm surge could increase the damage by orders of magnitude. It depends on local thresholds.
Ugh. Either Pielke is an idiot or he's intentionally misreading what I said. The data support both hypotheses, so I won't speculate on which is correct.

What I'm saying is this: if you add 2 ft of SLR to a 2-4 foot storm surge, you get the damage of a 4-6 ft storm surge.
Read 5 tweets
7 Sep
A lot of wrong takes in this thread. If you want to provide energy to people at the lowest cost, solar and wind ARE the best choice.
You want evidence? Let's look at what the Texas energy market is doing: It's installing wind and solar. Why? They are the cheapest energy source.
This is true *despite* the enormous subsidies that fossil fuels get.
Read 6 tweets
4 Sep
Climate change has gotten me thinking about the Drake equation and the future of humanity ...

What is the Drake equation, you ask? A 🧵:
The Drake equation is an example of order-of-magnitude estimation. There's some quantity you want to know (in this case, how many intelligent civilizations there are that Earth can communicate with), so you break it down into the terms that would constrain the value.
From the wikipedia page, here are the terms.…
Read 10 tweets
1 Sep
About 2/3rds of global warming comes not from direct heating by CO2, but from feedbacks. The most powerful feedback is water vapor. As CO2 warms the climate, the mass of water vapor in the atmosphere increases. WV is itself a greenhouse gas, so this creates more warming.
This process, known as the water vapor feedback, can double the warming you get from CO2 alone. As such, it is one of the most important processes in the climate system.
It has long been speculated, and recently been well documented, that relative humidity (RH; the amount of water vapor in the air relative to saturation) in our atmosphere remains relatively fixed as the climate warms.
Read 6 tweets
31 Aug
One of the great mis-directions form climate deniers is to focus on *deaths* in a disaster. Obviously, deaths are important. But in the rich world, natural disasters don't kill very many people.
For example, Ida's death toll is (right now) less than 10, which is amazing considering how intense of a storm it was.

But while people survived, the damage and suffering is extreme.
People may be out of power for weeks.…
Read 4 tweets
31 Aug
In case you're wondering how well it's going in Texas with no mask mandate in schools, here's @CSISD's number of COVID cases for 2020-21 school year (in blue) compared to the first two weeks of 2021-22 school year (in red).
This is a still from this video. In it, the spokesperson says they'll keep the school open for as long as they can, but at some point so many teachers/administrators are sick that they'll have to shut down.

.@GovAbbott says he's relying on "personal responsibility". But at what point does keeping the schools open become more important than letting people make their own decisions?
Read 4 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!