9
$\begingroup$

When examining the potential for large-scale electricity grids, and when identifying the lowest-cost ways to generate clean electricity, a recurring theme is of spreading wind generation across a wide-enough area that it evens out.

One way to bring down costs a lot, is to combine on the same grid, regions where the wind over a period of weeks to years, are negatively correlated: that is to say, it helps if we can easily identify pairs of regions where low winds in one region tend to happen at the same time as high winds in the other.

At a scale of days, sufficiently distant regions are uncorrelated. At a scale of weeks to years, a set of correlations of reanalysis data (e.g. ECMWF or CFSR) suggests that there are pairs of regions which do exhibit negatively correlated wind, either seasonally or annually.

Is there a reliable way to identify such combinations of regions?

$\endgroup$
6
$\begingroup$

My suggestion would be to do a Principal Component Analysis (PCA) or Empirical Orthogonal Functions (EOF) analysis on the wind data. The result of the analysis would be a set of modes of variability. You would be looking for modes that show areas that are large in magnitude but out of phase. As for the time scale, you need to check the eigenvectors and eigenvalues of the analysis to determine the scale of temporal variability.

You can find appropriate code in many places, for instance here.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.