IÕll talk about the last one: most similar to NWP data assimilation -- fewest assumptions about the time variability of the fluxes