torsdag 8 oktober 2009

Is El Nino an Oscillation?

Roger Pielke Sr. asks if El Nino Southern Oscillation ENSO is an oscillation, and answers that it probably is not, since it is not sufficiently regular to qualify. This connects to my blog posts Short-Time vs Long-Time Accuracy 1 and 2 pointing to the fact that long-time mean-value accuracy without short-time accuracy, may require cancellation by regular oscillation. 

Again we are led to suspect that short-time inaccurate climate models cannot be long-time mean-value accurate, since the variations of the global climate does not consist of regular oscillation, but is more complex and "chaotic".  

Compare with Birth of a Climate Crock discussing if long cooling periods can be consistent with
overall global warming? Are we now heading for 20-30 years of cooling, as an evidence of overall warming? The arguments, associations and also the tones of the different voices are interesting to analyze. Listen and think.

See also the BBC article What happened to Global Warming? discussing the possible role of the Pacific Decadal Oscillation PDO for the cooling period 1945 - 1977 and the possibly beginning
period the last 10 years, in 30 year oscillations.

The written discussion between pro-AGW Kevin Trenberth och konta-AGW Bill Gray summarizes the main arguments used in the climate debate.

4 kommentarer:

  1. Look here for a simple example of a model of a chaotic system (guess which) that is short time inaccurate but long time accurate. :-)

  2. I think the absolute error is the relevant measure, and then a chaotic oscillation on top of an increasing function stays there even after long time. Of course, you can argue that since a dissipative system can forget
    its initial value after long time, or damp out an initial oscillation, it can exhibit long-time accuracy without being short-time accurate. But I don't think this is relevant either, in the setting of climate models.

  3. It would be easy to modify the equations to get

    |X(t) - x(t)| < g(t)

    where g(t) is some arbitrary decreasing function, X is the output from the model and x is the solution of the chaotic system.

    So one can have long time accuracy but short time inaccuracy in a model of a chaotic system.

    I don't claim this is relevant for climate models but it is interesting.