Thursday, February 28, 2013

Thursday Links

  • The US lost farmland in 2012, mostly due to loss of ranchland in Texas and Montana because of drought.  Kay McDonald has the details (as well as some interesting graphs of the distribution of farmland between farm income categories).
  • Here's some color on how the Texas drought is affecting the economy in one town.
  • I stumbled across this really interesting climate paper when looking for something else.  Essentially it argues that when the climate goes through a major non-linear flip into a different state, that is preceded by a period in which the normal fluctuations in climate get less pronounced To give a loose idea for non-technical folks, it's as though, as the system is being squeezed towards the narrow bottleneck between the old state and the new state, it has less scope to rattle around.  They show this happening to greater or lesser degrees for eight past climate tipping points of very different character.  I was fairly convinced by the paper (but I'm not an expert).  It's less clear what this means for our ability to predict climate tipping points under anthropological global warming, since our carbon emissions signal is basically a step function on geological timeframes, not a gradual change in forcing like all the past ones under study in this paper.  
  • Japan may restart some nuclear plants later this year.
  • US oil production tops 7mbd.


RJMeyers said...

The behavior found in that climate paper--I've seen it elsewhere. Steve Keen is an economist who has developed a dynamic model of the monetary system. His model was able to re-create "The Great Moderation" in the global economy and showed cyclical behavior with smaller and smaller swings... until it abruptly crashed.

According to the paper you linked to, this is a "hallmark of tipping points." If the current climate response ends up doing the same thing, I wonder what this means for the human response to it.

dcomerf said...

I think your interpretation of the Dakos et al paper is a bit off. It's the system dynamics that weaken as a tipping point is approached, not necessarily the variability.

If we're bouncing around a strongly attractive steady state then a year which is warmer than average will cause a response from the system dynamics which can lead us to predict that the next year will be cooler than this year.

If we're bouncing around a weakly attractive steady state (such that we're near the point in the parameter space of the system [e.g. CO2 level] at which this steady state no longer exists) then the system doesn't respond so much to out of steady state behaviour. The fact that this year is warmer than average will not produce the negative response from the system, and our prediction will be much more martingale-like.

The "fluctuations in climate" around this weak steady state can appear more pronounced than those around the strong steady state.

Greg said...

I thought that the notion that feedbacks become less effective near a transition was a well-known result from control systems theory.

I'm concerned about this:

"...we used only data from before the actual transition."

We need to show that slowing down (increased autocorrelation) both happens before transition, and doesn't happen at other times, for it to have predictive value in the case of climate.

Anonymous said...

It appears to me climatologists are chasing a rapidly changing climate. By chasing, I mean they are seeing things occurring and by the time they study, research and come to a conclusion, the climate they studied is long gone. Consequently, it appears to me we are well into an abrupt climate shift. The only prediction that appears true for me is we are in an age of uncertainty where prediction is impossible. Since humans have extreme difficulty coping with uncertainty, we spend endless time discussing things we know little about in an effort to find certainty. If the methane on continental shelves lets go, none of us will be able to make up stories fast enough.

Stuart Staniford said...

Zeke: I'm afraid you have a point.