The second paper of the DADA project (Data assimilation, Detection and Attribution), has been accepted by the Quarterly Journal of the Royal Meteorological Society: Data assimilation for model evidence. The online early access paper is available here and the preprint is available here. Alberto Carrassi, Alexis Hannart and Michael Ghil are co-authors. The paper generally addresses the computation with data assimilation techniques of the evidence, which is nothing but the normalization factor that appears in Bayes' formula. The evidence has a lot of attractive features that have not been exploited much in geophysical data assimilation. However, its computation is not easy. The paper does apply to detection and attribution in climate sciences but is actually meant for any type of application in the geosciences. See also the blog post by Colin Cotter from the Imperial College about the topic.