Background to “Seasonal statistical-dynamical prediction of the North Atlantic Oscillation by probabilistic post-processing and its evaluation”

Recently my paper “Seasonal statistical-dynamical prediction of the North Atlantic Oscillation by probabilistic post-processing and its evaluation” was accepted in “Nonlinear Processes in Geophysics” and as it is now published, I will use this blog as it is a tradition (see here, here and here) to explain in more detail what it is about and what the problems are. So I will highlight some background to the paper in the upcoming week and will show why some of the points I raise therein will be important for the development of the community around seasonal- and decadal prediction as well as the one for the wider climate science.

So my background stories will take a look at the following topics:

  1. What is sub-sampling?
  2. Why does sub-sampling work?
  3. Why do we need verification with uncertain observations?
  4. EMD and IQD? What is it about?
  5. Do we need new approaches in verification?

As always, these topics are of course just an addition to the regular paper and are all just personal view. As it is my first (and I somehow hope only) single-author paper, the manuscript reflects of course mostly my view. Anyway, there are limits of things you can do in scientific literature, and that’s what those blog posts are about. And of course, it is statistical literature, so explaining it in more detail for those who are no fans of equations, is certainly a surplus.