Saturday, January 07, 2006

2005: second or third warmest year

Do you remember 1998? Was it a special year? Even though Al Gore was a vice-president, 1998 is still the "globally" warmest year on record.

Since that time, Bush was elected twice. The U.S. economy grew by 25 percent or so. The oil companies have been doing their best to produce as much carbon dioxide as they can.

But it does not work. No one can defeat Al Gore's 1998. It is not even clear whether 2005 was the second warmest year on record. It could be the third year after 2002.
As we discussed many times, the dependence of the annual average temperature on the year behaves as something in between
  • a random fluctuation from the temperature that you may want to call the "long-term average" in which the different years are independent
  • and the Brownian motion where the temperature of the previous year is the starting point for the new one, and the step is random
This fact is what you can derive from the exponent in the autocorrelation scaling laws that is between the values that you would expect for these two idealized cases. You can easily see that neither of the idealized extreme descriptions can be correct because the first one essentially allows the temperature to be discontinuous while the latter one allows the temperature to drift away arbitrarily far - as sqrt(t) - in a time interval t which seems rather unrealistic for extremely long time scales.

People usually think about the first approximation as a better description. Every year is independent, they think, and every year fluctuates around some "fundamental" value "T_{Gaia}", and therefore it is shocking that the last 5 years were among the 6 hottest years during the period of the last 27 years, they think.

But in reality, 2004 was the starting point for the temperatures in 2005. You just can't change the temperature abruptly. It is all but guaranteed that the years with similar temperatures will be found in clusters. There is always some inertia, and the amount of inertia (or "persistence") at all possible time scales should be studied scientifically and without prejudices. Let me now take the second approximation, the Brownian motion, as my starting point.

From this Brownian viewpoint, it is shocking that the Earth was not able to improve the record temperature for 7 years. Try to do the calculation. Assume that every year, the temperature either goes up by +dT, or goes down by -dT. Take a year in which you achieved the maximum temperature and call it 1998. What is the probability that the Brownian motion will give you all seven years 1999-2005 to be strictly cooler than 1998? Be sure that even if the Brownian motion is unbiased, you will get a number that is much much smaller than 50%.

And now imagine that the global warming guys argue that the Brownian motion should even be biased towards the increasing temperatures. Try to do your calculation with the assumption that every year, the temperature either increases by 0.015 degrees or decreases by 0.01 degrees. You will find out that the probability that you won't defeat the warmest year for 7 years is tiny.

Nevertheless, it is exactly what we observe. Of course, we can make smart comments about this observation. 1998 was the warmest year because of the El Nino of the century - and we had no El Nino in 2005. But the global warming evangelists whose goals are purely political should think twice before they make such arguments. The more often they make these arguments, the more clear it is that the climate is actually dictated by natural factors, not the anthropogenic ones.

The data simply don't support the global warming speculations too well. Despite Al Gore's vice-presidency, 1998 was certainly not a disastrously hot year. And the planet was cooler for 7 years that followed. Does not this fact itself mean that there is obviously no observable threat to talk about?


  1. This comment has been removed by a blog administrator.

  2. I wonder what quantoken wrote to have his comment removed twice!

    Oh, well: I'll venture into criticizing a little what you write. First of all climate is not a simple thing you can model with brownian motion: it is a complex dynamical system which has a "chaotic" behaviour in the mathematical sense, which is very far from being random!
    This means that if you try to correlate the state of the system at two times separated by, say, two weeks (like when you want to do a two weeks forecast) you will see that there is absolutely no correlation (your forecast is meaningless). On the other hand it is possible to make yearlong global and approximate forecasts thanks to the solar cycle and some kind of inertia.

    Now, what the global warming really is about is not just an average rise of temperature, but more importantly it is a growth in the likeliness of extreme (i.e. catastrophic) events, like very hot summers, glacial winters, droughts, hurricanes etc.

    It is the rapidity of the change that is a problem, not the change in itself, since our climate has been in a constant evolution.

    Next time plese get informed.

  3. Dear nic,

    the second comment was not removed. It was just a trick in which Quantoken wrote "This post has been removed..." himself, and if you click at the link, you will get to his "blog" - whose name tries to steal the trademark of Cosmic Variance.

    (A reader would not be a careful one if she could not distinguish the real Cosmic Variance from the faked one.) ;-)

    I erased the first comment because it was a spam promoting his retarded blog with a stolen name, his retarded ideas about the energy crisis, and moreover it was personally insulting me and writing a lot of stupidities as a bonus. Too many reasons.

    Nevertheless, you can click at Quantoken's second link to get some of his "wisdom" - everything he wanted to publish here and much more.


  4. Incidentally, don't get fooled by the users "mathjunkie" - it is Quantoken's sockpuppet much like seven other sockpuppets here:

  5. You ask "What is the probability that the Brownian motion will give you all seven years 1999-2005 to be cooler than 1998? Be sure that even if the Brownian motion is unbiased, you will get a number that is much much smaller than 50%." My calculation (by monte-carlo; I guess I should be able to do it exactly but I've forgotten how to if I ever knew) is that the chances are about 1/4 for equal up-down increments; and if the up-downs are 1.5-to-1, then the chances are about 1/10. I notice you don't give exact values and I doubt you've checked - but certainly 1/4 isn't "much much" smaller than 50%; and even 1/10 isn't outside a 95% conf interval and couldn't be called "tiny".

    If you look at, then the interannual jumps are perhaps 0.1 oC? With a trend (let us say) of 0.2 oC/decade, then the ups are 0.12 and the downs -0.08; this gets you about 1/9.

    But... thats just stats. I don't think your premise is valid (it would be somewhat if 1998 was just a random year). 1998 was unusually warm, for a reason: the large El Nino of that year, combined with the GW trend.