Frank Selten
KNMI, Holland

Super modeling, a new multi-model ensemble technique

Time and place
Thu 24 Jan 2013, 15.15
Room C609, Arrhenius Laboratory, 6th floor

(This event has taken place)


When weather and climate models are combined in multi-model ensembles, it is generally found  that the ensemble mean state is often closer to the observed state in some measure than the mean state of any of the individual models. This has motivated the idea of a new multi-model ensemble method that allows information exchange between the models during the simulation. The idea is that the ensemble of interacting models forms a consensus on the best simulation by combining the strengths of each. Observations of the true system are used to optimally implement the information exchange between the models in the ensemble. This so-called super modeling approach (SUMO) is pioneered using simple, intermediate complexity and state-of-the-art climate models.