Monday, January 17, 2022

1.5°C by 2033

Berkeley Earth has global warming at 1.2°C (a little higher than NOAA and GISS by 0.1°C) and is projecting 1.5°C by 2033, relative to 1850-1900, which is the baseline I think everyone should be using now if they can. (Or the oldest baseline they can obtain.)

Reminder: this tweet won't display properly in the Edge browser. Google's little joke, I guess.

12 comments:

Entropic man said...

Oh dear.

Layzej said...

Tweet looks fine to me in Edge...

David Appell said...

Hmm.... I wonder why the Tweet doesn't appear alright in my Edge. Thanks for letting me know.

Layzej said...

In 2014, Mann had suggested:

"To my wonder, I found that for an ECS of three degrees C, our planet would cross the dangerous warming threshold of two degrees C in 2036, only 22 years from now. When I considered the lower ECS value of 2.5 degrees C, the world would cross the threshold in 2046, just 10 years later...."

It seems to have continued with linear warming since 2014, continuing on a target of 2C by ~2060. Should we presume ECS is closer to (or below) 2.5?

Entropic man said...

Layzej

I notice that five of the last seven years are above the trend line from which the dates were derived. They should probably be considered optimistic.

Layzej said...

Yeah. Pray he's wrong on the first estimate through. ~0.5C in 16 years (to hit the 2C by 2036) would be a really nasty trajectory.

JoeT said...

Layzej, I think your last comment is on the money. But, to answer your question about presuming ECS to be closer or below 2.5, I think the answer is no. If anything, the recent IPCC report says now that ECS is 2.5 - 4.0, narrower than the previous 1.5 - 4.5

I wasn't too thrilled with Mann's Scientific American article when it first came out. For starters, if you look closely at what he's doing he is considering Northern Hemisphere temperatures, not global, which is about 25% higher (I got that from looking quickly at RSS data). (He says he's using NH data, so he's not hiding it. Except a lot of people wouldn't pick that up so readily.) Berkeley Earth has 2013 temperature at 1.1 C, but Mann I think was using HadCrut4 NH, so maybe it was around 1.0 C. To get to 2.0C in 23 years, that's over 0.4C per decade. Roughly twice what it is now. Not very likely at all.

But using NH data is not the biggest problem; it's what Mann is calling "business as usual", which has changed drastically since he wrote his article. Just to get a back-of-the-envelope estimate, the IPCC says that for CO2 only, there is about a 0.45C increase per 1000 GT. But that excludes other greenhouse gases as well as aerosols. If I look at Berkeley Earth temperatures versus cumulative emission, I get roughly 0.57C per 1000 GT. To get a 0.9 to 1.0 C increase takes something like 1600 - 1800 GT. Back in 2013 annual CO2 emission was around 40 GT (including land use). If I take a 3% increase per year, then by 2036 I get a cumulative emission of 2200 (from 2013). By 2036 annual CO2 emission is over 100 GT/year.

However, roughly around the time that Mann published his article CO2 emission started turning over, so that now it's roughly flat. The high emission scenarios like SPP5-8.5 have pretty much been ruled out. If I recall, Mann's Matlab programs to do these calculations are available online. You can check them out and see what assumptions he used (I never got around to looking at them, but it would be interesting in retrospect.)

The other comment I would make is that the original tweet that David is commenting on is a bit misleading. Sure, Berkeley Earth data for 2021 shows that for THAT particular year, it was a bit over 1.2 C higher than the 1850-1900 average. However, picking out one year doesn't give a good idea of where the climate is going; any more than if I picked out 2020 and said the Earth had warmed 1.36C. If I do a linear extrapolation from 1980, I get that 2021 was at 1.27C, whereas a loess smooth puts it around 1.33C. BUT, there is no physics reason why the temperature should go up linearly over 4 decades. In fact, a closer look shows that it's accelerating, but for shorter intervals the error increases considerably. However, there is a good physics reason why the temperature should go up linearly with cumulative CO2 emissions. In short, I think Mann's analysis is just out of date.

One last point: when the IPCC said in the last report that the Earth warmed 1.1C since the 1850 - 1900 baseline, that's the AVERAGE over the years 2011 - 2020. And it's an average of around 7 (or more, I forgot) different databases. It's neither a linear fit nor a smoothed result. The warming now is at least 1.2 C for the average of the databases, 1.3 C for Berkeley Earth.

Layzej said...

Good info. Thanks Joe.

JoeT said...

BTW, I learn a lot from reading Zeke Hausfather's articles. Here's one that might interest you:

https://www.carbonbrief.org/analysis-what-the-new-ipcc-report-says-about-when-world-may-pass-1-5c-and-2c

If you look at the SSP5-8.5 scenario, you can see that it predicts 2.0C will be reached around 2041, which is pretty close to Mann's prediction. And of course, the IPCC report came about 7 years after Mann's article.

Layzej said...

Even with the SSP2-4.5 scenario, "the world is likely to exceed 2C around 2052, with a range of 2037 to 2084."

Not too long afterward.

That's pretty close to the linear growth scenario that gets us to 2C by 2059.

JoeT said...

Yeah, not too far off, but like I said there is no intrinsic reason to think that linear extrapolation should work. It certainly doesn't work for SSP5-8.5. or for Michael Mann's scenario.

I had been looking for a quick and dirty estimation of temperature given an emission trajectory so that I don't have to resort to a climate model in making estimates. If I take my 0.57C/1000 Gt CO2, I can get to within 0.2C of the temperature time series for the SSP2-4.5 scenario. Using Berkeley Earth, we're at 1.3 now, need 0.7C more to get to 2.0C. That takes roughly 12280 Gt. This scenario, which is probably the closest to the trajectory we are on now, is roughly flat (bounces between 40 and 45 Gt, lets call it 43 Gt to make it easy) out to 2060 and then comes down. That puts the year we cross 2.0C at roughly 2051.

JoeT said...

That's 1228 --- not 12280.