Isaac Held again. Though nobody cared last time. A reminder that you can do basic interesting things with GCMs; that abstract thinking on problems can be illuminating; and that atmospheric dynamics is more complicated than you thought.
The video is from his blog; it shows "tropical cyclones" via their wind speed over a month of simulation on an [[f plane]] representation. Its a very heavily idealised model: SST is constant, there is no land, and the Coriolis parameter is a constant. And they find that in those circumstances, the model fills up with long-lived "tropical cyclones". The real world doesn't fill up with these things, rather fortunately, because they would either landfall or drift out of the tropics; or because they'd suck up the available energy and cool the SSTs. That aspect wasn't explored; it would be interesting to.
[Update: anyone who really wanted to talk about Sherwood et al. "Spread in model climate sensitivity traced to atmospheric convective mixing" should head off to RC, 'cos Gavin has actually read it. Unlike that lazy Dr Annan.]
- Log in to post comments
Some of the dynamics are interesting by eye -- pairs of cyclones orbiting a common center, for example.
William, this is probably a bit off -topic, but what's your opinion in this new paper (http://www.nature.com/nature/journal/v505/n7481/full/nature12829.html) by Sherwood et al, which argues that climate sensitivity is probably at around 4 °C, instead of 3°C? It is kinda intensively discussed atm
[I can find a brief discussion by Dr Roy, but he's clearly hedging his bets. Or, if you prefer, very sensibly not saying much until he's read and digested it (his I’ll either need to find a copy of the paper or buy it to dig into the details is pretty weird though - at least in my day, and that was only 6 years ago, any reputable scientific institution had a subscription to Nature). Who's talking about it?
As for me: I too will go with "looks interesting, but its too early to say" -W]
William
That's really strange. Even my "small" university grants me access to Nature, but never mind.
Nat Geo is talking 'bout it, as well as the Guardian . The press release announces that they "solved" the cloud mystery, but I think that this might be premature.
[Your href's are empty. There's Graun stuff here (with an awful lot of deleted comments) or here. I don't think either of those add anything new. Ditto NG? -W]
Not for the first time, I was surprised to see such a big piece missing from (some of, still) the models, but OTOH I suppose this was at least part of the basis for the longstanding assessment that they're crap at clouds. That apparently the amended models are high-sens and the unamended ones low-sens does start to seem more like enemy action than happenstance, though.
IIRC SS has been a leading member of the Christy/Spencer bashing clique over the years, generally if not uniformly to the disadvantage of the latter, so Roy may be a little gunshy about this even though it looks like something of a silver bullet for all the denialist cloud speculation.
Re the post, as JBL notes the Fujiwara effect in the vid is striking (no surprise, really, since rotation is specified in order to get the TCs to form at all, but basically realistic since in the real world TCs that get close enough to each other exhibit the behavior). I suppose that while this and the related results Isaac discusses are consistent with KE's warm-climate TC hypothesis, as yet the simulations are too idealized to be called support. Succeeding model generations will be interesting to see.
They also don't appear to turn north. One of the metrics in hurricanes is ACE but it seems the storms northward velocity would affect ACE substantially as the quadrant where wind speed + storm velocity would dominate over a static storm. We've seen storms where the eastern quadrant winds seem to be offset by storm velocity while western winds are additive. A static 10 knot storm is 10^2 + 10^2 is less energy than the same storm moving north at 10 knots 0^2 + 20^2. it would be interesting to compare the "pull" of warm SSTs vs. the push of atmospheric pressure to see if there is a difference of when (northern hemisphere) storms turn north. Is there an SST gradient or value that offsets the turn?
[They don't turn north because there is no gradient in the Coriolis parameter -W]
Patrick F,
The Sherwood paper looks similar, in both method and conclusion, to Fasullo and Trenberth 2012. They've identified an aspect of the climate system which appears to explain a large proportion of the spread in model sensitivity, and then noted that observations appear to be most similar to the higher sensitivity models. RealClimate hosted a nice write up on that paper by Karen Shell.
I think the main issue is that it's not really enough to show that an emergent quantitative property is most similar to observations (read: least wrong) in higher sensitivity models, unless we can substantially rule out the possibility that it might have emerged in a different way. As a simplistic example, what if the models are generally deficient in some related aspect so need to work extra hard in order to mimic reality. That would tend to guarantee higher sensitivity models should be closer without necessarily implying anything about reality.
I suspect we may well see a number of papers along similar lines, looking at emergent climate variables which correlate with sensitivity spread with comparison to observations. Some will probably point to lower sensitivity, some to higher.
Would van Gogh have been a fluid dynamics modeller?
A paper on modeling the Eocene "equable climate problem"
http://rsta.royalsocietypublishing.org/content/371/2001/20130123.full
by Sagoo et all, which perturbs climate model parameters to find a better fit with Eocene data. As Paul S suggests, the paper finds that the best fit model has a slightly lower Charney sensitivity of 2.7C. And the paper states: "we may be obtaining a good comparison to data for the wrong reasons."
Strikes Eli that letting the cyclones drift out of the tropics would be pretty easy, just let anything that hits the right latitude disappear. Same with landfall, just, for example, open the box on the west/east side as appropriate.