A method for determining optimal observations for prediction

This is the seminar given in Trento on May 30th by Henk Dijkstra (GS). Henk is mainly an oceanographer but the methods he illustrates, especially the Bayesian tools he develops towards the end of his presentation can be useful also in hydrological cases, so I am very happy to host his talk here.

The discussion that followed is here:

The slides of the talk are here. And here is the paper by Kramer et al. (JPO 2012), Measuring the Impact of Observations on the Predictability of the Kuroshio Extension in a Shallow-Water Model.