Sorry to be slow to the party with the news about the release of CSIRO’s climate change report, which you can read in all its glory here. While even a tory cabinet minister should be able to grasp the gist, the detailed stuff is rather heavy going - lots of stuff about beta distributions and model weightings if you really want to get into the nitty-gritty.
One point that struck me is the lengths the report goes to in dealing with the misconception that when climate scientists predict a range of outcomes, that any result within that range is equally likely. That’s not how it works, as the graph (below the fold) I’ve taken from the report illustrates nicely:
This graph, illustrating the predicted rainfall in Melbourne in 2030, demonstrates four different ways of modeling the range of outcomes predicted. The x-axis of the graph is the change in rainfall. The y-axis of the graph is the probability density function. Put simply, if you want to know the chances of the rainfall change falling between two values according to a particular model, you draw two vertical lines on the graph at those two values, and calculate the area of the shape bounded by the x-axis, the two lines, and the graph line of your chosen model. The y-value of the graph indicates the relative likelihood of two rainfall levels; so, roughly, if you use the “sum” model, the chances of the rainfall decreasing by 5% is roughly twice that of the rainfall staying the same.
So, how are these four lines on the graph obtained? The “sum” model is the most straightforward. Each individual climate model (that is, the computer programs derived by various teams around the world), can be used to calculate an individual probability . The sum model is derived by, in essence, adding the graphs together and scaling appropriately (so the total area under the curve is 1). On this particular graph, the results seem quite reasonable; on others, this results in a highly “lumpy” graph with peaks and troughs all over the place. That’s usually not how things actually happen in the real world; so the statisticians seek alternative ways to model the likely range.
One approach to getting a more realistic prediction range is to calculate the mean (average) and standard deviation (how much it varies around the mean), and fit some standard probability distributions to it. The first, the green line, is to assume a uniform probability distribution - that is, that everything within a certain range is equally likely, and there is zero probability of anything outside the range. The second is to assume the normal distribution - the “bell curve” that appears in a huge variety of contexts. That’s the red line.
Both these distributions have a problem, though it’s not so much in evidence in this graph. A better example of the limitations of the normal distribution (and indeed, the need for smoothing) can be seen in the following graph of global mean temperature estimates:
The normal graph (in red) probably overestimates the chances of very low temperature changes, so they use an alternative model, the “Beta” distribution, where they also calculate the skewness - in short, how lop-sided the graph is, and use this as an additional parameter to draw the smooth line. In this case, there is a long tail extending out to the right, and a lump to the left. The beta distribution, the blue line, represents this more accurately. For this reason, the statisticians like to use the beta distribution here, and indeed have used it internally as the basis for calculating some of the information in the report.
But getting back to the Melbourne rainfall graph for a moment, it’s striking how different a picture assuming the uniform distribution gives than the other three approaches. But I bet most non-statisticians, when they hear it said that “climate change predicts within the range of..” implicitly assume something like the uniform distribution, when that’s not the one the statisticians think most representative.
the thing I noticed was a spokesperson saying the highest temperature increase was down 1 degree but the degree of certainty over the ‘most likely’ point was higher. That is, recent research has helped focus the likely results - which are really bad. Especially under business as usual!
The report also shows the 2 degree limit is going to be hard to stay under without deep emission cuts. Sadly, I don’t trust either party to actually listen to the scientists on this one, what with people’s plasma TV needs foremost in their minds.
Plasma TV? Who can afford a plasma television.
I have trouble just paying rising electricity and water costs - even though my household uses less than 8kwh per day of electricity in summer, under 12kwh per day of electricity in the three months of winter and, less than 1kl of water per week all year round.
Thanks for the magical mystery tour, Robert. I’m still reading it and your point about the detailed stuff being heavy going is definitely true.
Juz, concerning the 2 degree limit, the report is looking forward from a 1990 (actually 1980-2000 average) standpoint. Hence the 2 degree in this report is well past the 2 degrees normally given out as the threshold for dangerous climate change, which is referenced to pre-industrial levels.
I heard author Penny Whitton interviewed and she repeatedly assured us that the current drought would pass, that the rains would come, but the average precipitation would decline over time in most parts of Australia. So we are meant to believe that it will be well into the century before things get to be as bad as they are now. I’m interested to know why she believes this other than that it is in the models, but I’m not sure I’ll find it in the report.
On the ‘now’, jt author Scott Power from the BOM said that this year to September had been the hottest ever. He also said that the spring rain in the Murray Darling so far had been the lowest ever. So near term we are meant to expect things to get better.
If you go here and activate the animation at the bottom of the page you’ll find that Sydney and Canberra do quite well in the rainfall stakes and are better at the end of the century than at the beginning. This is not good for concentrating the political mind.
It was not that long ago that Howard had any CSIRO Scientists who supported Global Warming sacked or cut their funding so they had to leave the country.
Like this one.
Brian, one hypothesis about the drought mentioned in the report (note: a hypothesis, and that’s all) for the cause of the current drought that includes human influence.
They’re wondering out loud whether it has something to do with the cloud of sulphur dioxide, ozone, particulates, and whatnot emanating from China.
Juz, re revising the upper limit downwards, one of the commenters at GreensBlog made a great point:
Kind of puts it into perspective…
Let’s put this temperature rise in perspective, shall we? To quote from the 2005 report on climate change from the Australian Greenhouse Office, “many climate scientists recognise that the planet has only warmed around 5 degrees centigrade between the last glacial maximum (around 20,000 years ago) top the current interglacial temperatures. This 5 degrees centigrade has brought about a major redistribution of ecosystems, as well as extinctions and the emergence of new species.”
In other words, this 5 to 6 degrees centigrade predicted global average temperature rise is more than this planet has experienced since the last ice age, 20,000 years ago!
And it’s all caused by those *!%$$! that worship the god of the economy, with total disregard for the environment!
And if you delve deep enough into the IPCC reports from the United Nations, you’ll discover that the lag time between emissions and climate change is so great that it’s too bloody late to worry about reducing emissions - by the time reducing them (even to zero, right this very instant) has any effect upon the climate almost all life on the planet will be dead anyway!
Sleep well.
Just a heads-up on the drought-now / rainfall change under greenhouse issue. The projections of change from this report contain only the information on the anthropogenic signal distilled from climate models. In the real world, we are subject to both climate change and variability, so the effects can be additive. The recent decadal low in rainfall with the current drought occurring as a low point within that is likely long-term climate variability operation with some influence of climate change.
We can’t attribute the relative level of influence of natural variability and change on rainfall at the moment, and there is also an evaporative effect with warming that is increasing dryness.
The key issue made by putting the recent climate with climate change in the report is that we need to take a whole of climate approach. I would like my fellow climatologists to make this point more emphatically and obviously and indeed have been saying so all week at the Greenhouse 2007 conference.
What we are most likely experiencing now is a double whammy of long-term climate variability interacting with climate change. If the recent low in rainfall (downward shift from 1997 in southern Aust) is variability, then we may expect a partial recovery in the future. If it is all anthropogenic, we would expect this or worse in the long term. My priors are leaning towards the former, and it is an area of active research to determine what exactly is going on.
Roger, I guess the next 10 years will tell the story, but we’ll be in very poor shape if the current weather continues. It would galvanize the minds of politicians, though.
Jake, if we go up 5 or 6 degrees we are, as far as I can make out, in territory going back about 40 million years in a world with sea level 80 metres higher and a world climate not especially suited to homo sapiens even with our clever technology.
tim, Hansen reckons with 4C plus we have an entirely different world. Of 5 or 6 degrees Mark Lynas reckons it’s like playing Russian roulette with a Luger, the point being I think that it doesn’t (can’t) have an empty chamber.
Jake, I can’t yet be as pessimistic as you are, but personally, being risk-averse, I’d like to see us going for a 350 ppm of CO2e target ASAP on the way to 300. Lynas (following Jones, C et al 2006 in Scellnhuber, H J (ed.) Avoiding Dangerous Climate Change Cambridge UP) reckons we need 350 ppm CO2 (not CO2e) to stay within 1C longer term. Since we are committed to about 1.3C if we stop now I think 350 ppm would have to happen real soon.
Lynas (following Meinshausen also in Schellnhuber and quoted also by Stern)reckons that 400 ppm of CO2 (we are 384 now) gives us a 75% chance of staying within 2C. Those are less good odds than Russian roulette, although a 2-3C world may not be terminal to civilisation as we know it. Then again the tipping points it might trigger could carry us through to a 4C plus world. This is the very essence of why 2C plus is dangerous.
It’s my ambition to do a post spelling all this out a bit more and citing the supporting documentation, partly because I’m sick of being called an alarmist but mostly I’d like people like Roger to tell me whether it’s reasonable or way off beam. Time is my enemy, though.