Tim HolloJul 25, 2023

I'm finding it a little hard to work today with this in my head.

Antarctic ice extent is now 6.4 standard deviations below the mean. That is, I'm reliably told, a one in 13 billion year event.

We're about to see a lot of shit hit a lot of fans. And we are far from ready.

Business as usual is over. Politics as usual is over. We need to be putting our effort into building systems that can help us survive what greed and power and wilful blindness have wrought.

#ClimateCrisis #Antarctica

Show threadTom Quinn
@timhollo should be million, not billion. 13 billion is almost 3 times the age of our solar system.
Jul 25, 2023 at 7:45amWeb
Show threadTim HolloJul 25, 2023

@TomQuinn yeah, it's a probability factor. The number is correct, and mind boggling.

IF all else were equal, the chances of this happening would be once in the age of the universe.

The point is, all else is not equal. As someone said on Bluesky, it's measuring the same ice in the same Antarctica, but the whole world is different.

Show threadTom QuinnJul 25, 2023
@timhollo gotcha. Either way, a lot of shit to hit the fan!
Show threadAral BalkanJul 25, 2023

@timhollo @TomQuinn Decimal place error?

6σ ~= once every 1.38 million years (given the time axis is daily events)

https://en.wikipedia.org/wiki/68%E2%80%9395%E2%80%9399.7_rule#Table_of_numerical_values

68–95–99.7 rule - Wikipedia

Show threadderleJul 25, 2023
@aral This is not a daily but a yearly event, according to your table with 6.5\sigma and translating daily into yearly the approx of 1 in 13 billion does not seem of by much @timhollo @TomQuinn
Show threadAral BalkanJul 25, 2023

@derle The chart is showing a daily mean not a yearly mean, no?

@timhollo @TomQuinn

Show threadderleJul 25, 2023
@aral That depends how you see the problem. This is one in number of independent events. Ice extend from one day to the next are probably not independent event. @timhollo @TomQuinn
Show threadDoug WadeJul 25, 2023

@aral @derle @timhollo @TomQuinn your argument is that the amount of Arctic sea ice is randomly determined on a daily basis? That each individual day is a separate event? You think that tomorrow, we could randomly be at +2 std deviations?

Yearly makes much more sense, imo.

Show threadOldTurk🗽Focus: #FreePeltierJul 26, 2023
@derle @aral @timhollo @TomQuinn 13billion would be a double sided distribution, single sided applies here
Show threadKevin LeecasterJul 25, 2023
@timhollo @TomQuinn
The worst aspect of the measurements of ice coverage in Antarctica is that it falls right within the range of models we've been trying to get people to take seriously for decades.
Show threadJames 🦉 #FBPE Jul 25, 2023
@TomQuinn @timhollo The great thing about Mastodon is you can edit posts...
Show threadlunchyJul 25, 2023

@TomQuinn @timhollo using standard deviations to talk about this is wrong because the "probability distribution" we're measuring from is constantly changing, shifting in one direction

we'd have to have a model that corrects for the average change in average sea ice we already know about and looks at the extent with that effect already factored out in order for stddev "1 in x years" statements to make sense

Show threadDoc Edward Morbius ⭕​Jul 25, 2023

@TomQuinn Based on a normal distribution and standard deviations, ~13 billion years is indeed what the maths say, if the trend is taken to be annual. (Divide by 12 for months, or by 365.25 for days.)

The problem as I've pointed out here https://toot.cat/@dredmorbius/110773687361166325, is that the data series is far too short to be drawing conclusions at this span. We're seeing 34 years of highly-correlated time-series data.

It absolutely is alarming enough as it stands. But @timhollo's post greatly oversells it.

Doc Edward Morbius ⭕​ (@[email protected])

@[email protected] Not discounting the severity of this data at all ... Keep in mind that using standard deviations to measure for expected occurrence rate relies on both a normal distribution and some defined sampling basis. Both of these assumptions are ... challenged here. Should we be considering this as an *annual* event or a *monthly* one? And we *do* know that Antarctica, all of it, not merely the seas surrounding it, has been ice-free in the past. And that Earth has likely been entirely entombed in ice ("Snowball Earth"), probably multiple times. So the data record we *have* is at best partial. I'd be comfortable saying that this is far outside all previous measured data. I'd caution on drawing inferences *based on statistics alone*, and would strongly urge that previous geological / climactic *evidence* (and the associated atmospheric and other conditions influencing climate) be considered. Stats get really hairy on the thin margins, especially with comparatively small samples. We're looking at *34 years* of data here, not hundreds, thousands, millions, or billions. In statistics, n ~=30 is just where "large sample" methods start becoming applicable, but typically *not* for drawing inferences at the 1:1,000,000,000 scale.

Toot.Cat