Spatio-temporal Pattern Analysis in the Atmosphere and Oceans

Monday, September 24, 2001 - 9:30am - 10:30am
Keller 3-180
Michael Ghil (University of California, Los Angeles)
The study of large-scale atmospheric and oceanic motions depends critically on a description of these motions that should be as reliable and as precise as possible. There exists thus a trade-off between the wealth of information that one wishes to extract from limited data sets, on the one hand, and the statistical confidence in that information, on the other.

This trade-off is optimized by applications of Karhunen- Loeve theory in the space and time domain, separately, as well as concurrently in the spatio-temporal domain. I shall illustrate this approach by applications to climate variability on the intraseasonal (10--100 days), interannual (1--10 years) and interdecadal (10--1000 years) time scales. The applications will involve phenomena that reside mainly in the atmosphere (so-called low-frequency variability), mainly in the ocean (its wind-driven and thermohaline circulation), and finally in the coupled atmosphere-ocean (the El Nino-Southern Oscillation).

Connections between this parsimonious, optimal description of climate variability and its explanation via dynamical systems theory will be outlined. I shall select one of the examples above to illustrate this connection in greater depth.


Ghil M., R. M. Allen, M. D. Dettinger, K. Ide, D. Kondrashov, M. E. Mann, A. Robertson, A. Saunders, Y. Tian, F. Varadi, and P. Yiou (2001) Advanced spectral methods for climatic time series, Rev. Geophys., accepted.

Pascal Yiou, Didier Sornette and Michael Ghil Data-adaptive wavelets and multi-scale singular-spectrum analysis
Physica D: Nonlinear Phenomena, Volume 142, Issue 3-4 (2000), pp. 254-290
[Abstract] [Full text] (PDF 1.1 Mb) (one of the 8 Hottest Papers in Physica D)

Software (free)

SSA-MTM Toolkit for spectral analysis This software will be illustrated during Workshop #3 by one of the co-organizers, Ferenc Varadi.