More on Bayesian approaches to detection and attribution

Timely given events all over the place, this new paper by Mann et al has just appeared.  It's a well-aimed jab at the detection and attribution industry which could perhaps be held substantially responsible for the sterile “debate” over the extent to which AGW has influenced extreme events (and/or will do so in the future). I've argued against D&A several times in the past (such as here, here, here and here) and don't intend to rehash the same arguments over and over again. Suffice to say that it doesn't usefully address the questions that matter, and cannot do so by design.

Mann et al argue that the standard frequentist approach to D&A is inappropriate both from a simple example which shows it to generate poor results, and from the ethical argument that “do no harm” is a better starting point than “assume harmless”. The precautionary versus proactionary principles can be argued indefinitely, and neither really works when reduced ad absurdum, so I'm not really convinced that the latter is a strong argument. A clearer demonstration could perhaps have been provided by a rational cost-benefit analysis in which costs of action versus inaction (and the payoffs) could have been explicitly calculated. This would have still supported their argument of course, as the frequentist approach is not a rational basis for decisions. I suppose that's where I tend to part company with the philosophers (check the co-author list) in preferring a more quantitative approach. I'm not saying they are wrong, it's perhaps a matter of taste.

[I find to my surprise I have not written about the precautionary vs proactionary principle before]

Other points that could have been made (and had I been a reviewer, I'd probably have encouraged the authors to include them) are that when data are limited and the statistical power of the analysis is weak, it is not only inevitable that any frequentist-based estimate that achieves statistical significance will be a large overestimate of the true magnitude of the effect, but there's even a substantial chance it will have the wrong sign! A Bayesian prior solves (or at least greatly ameliorates) these problems. Another benefit of the Bayesian approach is the ability to integrate different sources of information. My favourite example of the weakness of traditional D&A here is the way that we can (at least this was the case a few years ago) barely “attribute” any warming of the world's oceans under this methodology. The reason for this is that the internal variability of the oceans is large (and uncertain) enough that we cannot be entirely confident that an unforced ocean would not have warmed up by itself. On the other hand, it is absurd to believe the null hypothesis that we haven't warmed it, as it has been in contact with the atmosphere that we have certainly warmed, and the energy imbalance due to GHGs is significant, and we've even observed a warming very closely in line with what our models predict should have happened. But D&A can't assimilate this information. In the context of Mann et al, we might consider information about warming sea surface temperatures as relevant to diagnosing and predicting hurricanes, for example, rather than relying entirely on storm counts.

It's colder in Boulder

Presently hunkered down at home awaiting the forthcoming -17C minimum temperature overnight, which is something I don't think I have experienced before. Possibly came close to it up a Japanese mountain once or twice. We've been very lucky with the weather for most of our trip, but it's certainly winter now!

At the start of December there is the annual “Colder Boulder” race, a counterpart to the “Bolder Boulder” in May that we've run a couple of times before. The December one is only 5k and has many fewer participants, allowing the event centre and finishing area to fit into an indoor arena which is probably a good idea given the season. In the event it was pretty comfortable, a bit chilly but bright and still. Biggest problem was the altitude, but the modest hills and numerous corners on the course also contributed to a slow run. Oh, and it was a good 100m too long too. 

Taking account of these factors, I was reasonably satisfied to duck under 20 mins (by a whole second) and jules was also relatively fast at 26:05. In fact we were both the fastest in our respective age categories, thanks mainly to the organisers' decision to classify by year rather than the more usual decade (meaning there were only about 10 runners in each category). We are hoping to reap the benefits of the altitude training when we return to sea level shortly!

Heatstroke

We seem to have missed rainy season this year (not all of Japan was so lucky) and moved straight into heatstroke season, which means it's time for the locals to go and run around all day outside in full sun until they collapse.

Apparently, some local idiot thought it might be fun to run 10km to work a couple of mornings ago. But other than losing several pounds of fluid and nearly drowning in his own sweat, he is no worse for it and is threatening to repeat the feat occasionally over the summer. In my defence, it was actually cloudy and relatively unsweltering that morning, and most of the route is under trees. Still broke all previous records for slowness though.

Cast ne'er a clout till May's out

I mentioned some time ago that Piers Corbyn had been up to his old tricks with an apocalyptic prediction of doom for the coldest May ever in the UK, right in time for our recent trip there.

May is long since out, and it's time to get clouting. May turned out to be slightly warmer than average. I wonder who actually pays for his "forecasts"?

Ask them again in August!

There is a faction arguing for re-opening of the nuclear power stations that were shut down after the tsunami. I don't know to what extent safety measures have been improved in the interim, but there is obviously a large cost to the ongoing hiatus.

However, there has also been opposition, at least according to opinion polls. Also some fairly feeble and poorly-attended marches in Tokyo, though being in Japan that is probably about as revolting as the peasants ever get.

However.

Summer has barely started - in fact officially it hasn't started at all, but it clearly has. Following on from the hottest June day ever, we have also had the hottest late June in 50 years, and now we are into the summer power-saving/heatstroke season and air cons are being switched off, and there are various other power-saving measures, including the shutdown of some of our supercomputers at work.

It's all very spirit-of-the-blitz so far but I wonder how long it will last...

Weather forecaster in forecasting weather shock

Well, it had to happen. The meeja demands a forecast, but the Met Office won't give one any more (basically because they realise it's hard to be both skilful and useful within the constraints of newspaper soundbites). Journalists abhor a vacuum but unlike nature they can simply invent something to fill it up, and in that vein the Torygraph has given one of these "independent long-term forecast experts", one Jonathan Powell, a free plug. I didn't find anything useful on his website (and certainly no evidence of demonstrated ability) but according to the Torygraph Article he has basically suggested that there will be weather this summer:

“Summer does not look as good as the last two years – it will be more of a ‘brolly and sunblock’ summer, with a mixture of good and bad weather.

“It will be mix of dry, sunny and warm days plus some cooler, cloudier and wetter conditions.

“It’s not going to be a poor summer, and will encouraging at times – but we will have to take the rough with the smooth and enjoy the good weather while it lasts,” he said.

To be fair, there are some hints at quantifiable outcomes too, such as no prolonged hot periods in June (but what is prolonged and hot?) and temps above 30C in early August. We'll see...

Talking of falling down snowy slopes...

These famous "gassho-zukuri" (praying hands style) steeply thatched buildings (which we visited last year, incidentally) are supposed to be designed to shed snow effectively, so I'm a bit surprised that people still have to climb on the roofs to clear them:

14 fall off Gifu temple's thatched roof

The news report doesn't mention whether they were successful in clearing the roof, but I suppose it's a safe bet that they took a fair chunk down with them. And I bet they didn't have ice axes either (not that they would have been much use).

Hot....and not

So the surface temperature analyses for 2010 are in, and depending who you believe, it was either the hottest year (NCDC), or it wasn't (HadCRUT), or it was basically a dead heat (GISTEMP). Even if it wasn't a clear record, it was certainly close. I wonder whether the sceptics will start talking about "no global warming since 2010"?

Realclimate has a nice round-up of various indicators in comparison to model outputs. I don't have much to add to that. Back in 2007, Smith et al gave a 50% probability of 2010 (and every subsequent year) exceeding 1998 by the HadCRUT measure. Given the current La Nina situation, it looks like at least 2011 will stay stubbornly below that threshold. It might be interesting to consider how long it will take before half the years post 2009 will actually exceed the 1998 value. Just for fun, I've roughly updated the plot in their paper with more recent obs (annual means, blue dots in the below), but only by hand and I'm not confident that it is very accurate. I would say it is too early to be sure they are wrong, the obs are persistently in the low end of their predicted range but it's only been a few years. Even with the La Nina, there probably isn't a great chance of next year (or subsequently) falling outside of their 90% range, but I can't see temperatures reaching the upper half of their predicted range in the next few years either.

Keenlyside et al, on the other hand, looks substantially worse, given that the error bars are for the decadal average - only one single year has just about reached that level, and their initialised forecast is substantially worse than the free run with no assimilation! Of course it was clear that their forecast was wrong as soon as it was published. But hey, it got headlines and 140 (google scholar) citations so far, which of course is what really matters these days...



I spotted that Vicky Pope described the 2010 temperature as being a "dead heat" for 1998 despite their analysis coming in about 0.05C colder. I would guess that when the temperature actually does beat the 1998 value by a similar margin, she won't put it in quite the same terms :-) I don't think there is any need to talk things up, the evidence for ongoing warming is quite clear enough as it is. Whatever the details of 2010 beating the HadCRUT record or not, it doesn't change the fact that the world continues to warm at more or less the expected rate.

One place where I do have a nit to pick in RC's post is in their explanation of the multimodel mean not providing a perfect fit to the data. They excuse this (as if it needs excusing) through the models being merely an "ensemble of opportunity". However, even if it was a carefully designed ensemble that perfectly described our uncertainty, the ensemble mean would still not match the truth, instead inevitably being "biased" relative to some future observation (although we would not yet know in which direction the bias will be). My point being, the mere fact that the ensemble mean does not match incoming observations of reality - even ignoring the fact that these observations are themselves imperfect - in itself tells us nothing about the design and adequacy of the ensemble as an indication of uncertainty. It is not possible in principle to design an ensemble which would have this "truth-centred" property, even in simplistic idealised "perfect model" cases. It is therefore not meaningful to try to use it as a means of assessing (or criticising) the ensemble.

You know it's too cold...

...when your Mac laptop won't even charge. Apparently the minimum operating temperature for these things is 10C.


Fortunately, other than charging, they seem to work fine at lower temperatures (at least down to 4C which is as cold as we've been here), and the CPU soon warms things up sufficiently for things to get back to normal. But jules wasted some time on various tasks like resetting the power management before we worked out what was going on.

This is what the conditions are inside our lounge, of course, both when we get up in the morning and when we come home from work. It's been a bit chilly recently, and with the fairly feeble hot air blower (air conditioner working in reverse, as a heat pump) it takes more than an hour to get the room - which is the only one we even attempt to heat - up to a comfortable temperature. Which isn't really that helpful, especially in the mornings. Why don't these things have a decent timer system so they can be set to switch on every day before we get up in the morning? (There is a timer of sorts, but it has to be reset every time in a "switch on in x hours" way which seems a bit primitive.) It's not even merely a case of getting a more powerful heater, as things are our fuse box blows when we dare to use the a/c, microwave and kettle at the same time thanks to the 4KW max rating of the WHOLE HOUSE. Yes, that's a fraction more than the equivalent of one 13A socket in the UK.

Rumour has it that in some colder parts of Japan, they have actually heard of insulation and heating. Lucky them.

First cherry

While some have been grumbling about the cold winter, we've been having fairly mild weather. In fact Tokyo dropping as low as 1.8C made the news last week, with the first frost being 27 days later than average. Out in the sticks here it's been a bit colder, but still fairly mild (the problem with winters here is not that they are cold, but that it's as cold inside the house as out). In fact I spotted the first cherry when walking back from town today. Didn't have a real camera (and jules didn't notice it when she passed by later) so you'll have to make do with what my iPod can manage:


Technically, it's not really the first cherry of the year for us, as there is one species that flowers rather feebly right through the winter (the Japanese name means "October cherry") and we saw a few flowers on the tree in Zuisenji at about 12:05 am on the 1st. But this one (Kawazuzakura) is one of the earliest of the spring species. The Kawazu cherry festival down on the Izu peninsula runs from early February, and though it's hardly in full bloom yet, I see that our equivalent photo from 2007 was taken more than a month later. I therefore deduce that we are having an unusually warm winter here.

Big wind

And no, that's not a reference to the spicy noodles.

I thought we might have missed typhoon season by being away over the late summer, but I shouldn't have worried, there is one coming along right now. They are generally more "big rain" than big wind really, at least in our sheltered neighbourhood, and this one is no exception, with 8-10 inches of rain falling in a day. Closest approach is due in a few hours. Must go up on the roof and fix that leaking guttering :-)


That's us somewhere under there...

Barbecue summer - the sequel!

Oh, I suppose I shouldn't poke fun. Having got their fingers burnt last year - or perhaps I should say, having had their parade thoroughly rained on - the UK Met Office is declining to offer public forecasts, but is still doing "experimental" research on seasonal prediction and has obviously given a nod and a wink to the author of this Times article.

Of course we all know how the last "barbecue summer" turned out:

“Well, let’s put it this way. I’ve put my barbecue in the shed,” Dave says. “I don’t want it to get any rustier.”

I'm going to be in the UK for a chunk of the summer, so I hope they have got it right this time.

I am reminded that earlier this year, the BBC put out its weather contract to tender. Rumour has it that the esteemed Piers Corbyn, in a change of focus from his current work in volcano and earthquake prediction (no I'm not joking on that bit), is putting in a strong bid.

A once in a thousand years weather event

Spotted some sad news in the papers today.

Kamakura has a famous old ginkgo tree in front of the main temple in the middle of town. It is quite a spectacle late in the year when the leaves turn brilliant yellow. It is reputed to date from the era when Kamakura was the capital of Japan, and be the site of a famous murder which wiped out the last heirs of the ruling Minamoto shogun family at the time (the story goes that, following the untimely death of his father and previous shogun Yoriie, Kugyo blamed his uncle Sanetomo for usurping his rightful position, so hid behind the tree and killed Sanetomo when he came to pray one night. Kugyo was caught and killed, leaving no more male heirs.) Actually according to Masuda-san, the tree is not old enough to have been around at this time, but it's a nice story.

Anyway, the tree is no more. On Tuesday night, it was snowing as we walked home through the temple grounds. And at 4:40 am on Wednesday, it was windy....and the tree blew down. Story here. Interesting that the Yomiuri is happy to repeat the myth about the tree's age as if a fact, even though Masuda-san says it has been tested and shown to be false.

Strange really as it did not seem all that windy (neither was the snow heavy), and the tree survived a direct hit from a typhoon a few years ago that even knocked off some of the larger branches and so should perhaps have made it less vulnerable to a gale.

Actually, it occurs to me that the Japanese do a lot of tree transplantation, so I wonder if they will try to save it?

It's the end of the world as we know it

Couldn't help but laugh at this from the Grauniad, complete with caption:


And with this much snow they can't even run trains in tunnels!

UKMO agrees with me

They don't have a great record with their recent annual temperature forecasts, but it's still encouraging to see another set of experts endorsing my prediction, after Jim Hansen did at the start of the year and Gavin Schmidt did more recently.

ENSO continues to grow stronger and the forecasts are getting slightly more confident.

No, global warming hasn't stopped (part 94)

Fashionably late to the party as always, but (in contrast to Stoat's snark) I thought this was a pretty well-done piece by Seth Borenstein:

Have you heard that the world is now cooling instead of warming? You may have seen some news reports on the Internet or heard about it from a provocative new book.

Only one problem: It's not true, according to an analysis of the numbers done by several independent statisticians for The Associated Press.

No prizes for guessing which "provocative book" he is talking about, and of course the "global warming has stopped" meme has been kicking about the delusionosphere pretty well for ever (eg see here). What he did was quite neat - sending off the time series of temperature data to a bunch of statisticians without telling them what it was, and asking them for their analysis. I would have preferred to disguise the data in some way (any linear transformation) but that's a bit of a nit to pick. Of course the statisticians gave the same answer that all competent climate scientists have already given, cos this stuff is hardly rocket science and there is no evidence for a marked drop in trend (mind you, the models suggest the trend should be accelerating, not just continuing). Cutting-edge science it is not, but it is an interesting and well-constructed news feature.

Best comment prize goes to Deep Climate, in the thread to Stoat's post

RP jr didn't write about it, so it must be right.

Sad but true.

Gavin Schmidt can be spotted highlighting the chances of a new temperature record next year. Obviously he's been reading McLean et al's ground-breaking paper where they show for the first time that ENSO affects global temperatures :-)

I've been keeping an eye on the ENSO forecasts recently myself, not because I care about ENSO itself - I don't think it has much of an effect here (cue Japanese scientists telling me how important it is) - but rather due to its likely effects on global temperature. The Hadcru temperature anomaly has been about +0.5C for the last few months, and the old 1998 record was +0.526C. The weekly pdf updates presented on this page give an accessible summary of recent ENSO status and predictions. According to it, the ENSO is "expected to strengthen and last though Northern hemisphere winter". It is almost certainly late now for this year to break any records, but given the lag of about 6 months in ENSO's effect, every extra month of positive ENSO index from here on means additional warmth next year.

Sea Ice forecasts and outcome

I've not really paid that much attention to the sea ice area myself, but I did take a punt on Lab Lemming's betting pool, mainly because (at the time of my bet) there seemed to be a pretty big hole in the bets close to where the sea ice graph seemed to be heading (based on nothing more than eyeballing the graph). So without any detailed calculations, I chose a value in the middle of the hole, to maximise my chances of winning. I didn't quite win, but was one of the closest few.

More interesting than that, is the performance of the "pros". I use the inverted commas because as far as I know, there is no real history of producing annual forecasts - and neither is there any customer - so these people aren't really experienced forecasters, rather they are sea ice modellers and experts of some sort or another. Note that weather forecasting certainly makes heavy use of both numerical models and experts in atmospheric science, but on top of that it involves experts in prediction. Anyway, here is the final sea ice prediction of the experts, issued in August and based on July data:

Clicking on the pic should get a bigger version, or you can go here for more details.

SEARCH have now published the final outcome, including a comparison with the July forecasts. That earlier set of forecasts happens to be the only month that contains a prediction which exceeds 5, which I'm sure is entirely unrelated to their decision to cherry pick that month. Note also that SEARCH are using a different observational analysis which gives a slightly higher observed value (5.36 versus the 5.25 of Lab Lemming). Also they are using monthly mean rather than daily min. None of this, however, obscures the fact that the forecasts are all wrong, and most of them are very very badly wrong.



Looking at the IARC-JAXA graph, it seemed very obvious through July that the sea ice extent was towards the low end of the pack, but well above the 2007 track and not clearly lower than the other years (the ticks mark the start of the month):

So I don't know what basis the researchers had for expecting such low values. SEARCH helpfully adds that many of the forecasts had their own uncertainty estimates of about 0.4, which means that almost all of them failed dismally to even include the observed value in their predicted range, and one might have been as far as 6 sigma out (interpreting the 0.4 value as 2 sigma). Back to the drawing board for them!

Turning briefly to the "semi-pros", Stoat has a bit of a gloat, although he still managed to lose a bet with Bob Grumbine. However, both of them were a lot further off than me...(but to be fair, they made their bets much earlier too).

Listen to me (being vaguely alluded to) on the radio

The latest More or Less has Andrew Gelman being politely dismissive of some extremely dodgy work on sex ratios that he has blogged about a number of times (and written some papers).

There was also a bit about climate prediction. Mojib Latif has got himself in the news recently with some comments about his prediction of short-term cooling which of course the denialosphere loved. He's part of the Keenlyside et al team, but I'm far as I'm aware there is no new research behind his statements which were made at a big political meeting. I don't think anyone believed the Keenlyside et al work when it appeared, and the Pope (ha) put the boot in ever so gently, describing the work the Hadley Centre had done (that I mentioned some time ago) to show that the method that Keenlyside et al used doesn't work.

Oh yes, as for the vague allusion to me, if the promised El Nino lasts over the winter then I should have a good chance of collecting on the bet I made on More or Less last year. Not that it's a sure thing or anything. I'm in with a reasonably good chance here too - if the current melt continues for another week I'll be in the money...

They don't make 'em like they used to

The Tokyo Daibutsu may vanquish the forces of evil on a regular basis (or whatever it did, I'm happy to say I've never seen the program) but it was brought to its knees by yesterday's wimpy little typhoon:



Rumour has it that the Kamakura Gundam is still sitting comfortably, having seen much worse in his 750+ years...including regular typhoons, a notable tsunami that destroyed his house, and major earthquakes roughly every 80 years.

Apocalpse then

We've recently had three big quakes and a typhoon, all in the space of less than a week. Fortunately they were all centred some way off the coast, and only managed some limited damage (quake 1, 2, 3 and typhoon). An expressway was partially collapsed by the second one, but it was repaired in 2 days! It obviously takes a lot more than a piddling earthquake to get between a Japanese salaryman and his annual Obon holiday.

It's just as well we don't have too many religious nutters round here (at least not vocal ones) or we'd be hearing how it was all the fault of sinners (random link).

Whoops, there's another one.