«Examples of analogs method in data assimilation: climate model attribution, uncertainties calibration»
In this talk, we will shortly introduce an analogs approximation algorithm. It will then be combined with standard sequential data assimilation (DA) methods (EnKF, particle filter). We will show that the analogs approach allows to run DA without running the meteorological model. It will be illustrated on two examples. The first one is a model attribution problem applied to the simplified SPEEDY climate model. The second is a problem of uncertainty calibration applied to a Lorenz 63 model.
