By far the most popular data assimilation techniques in use today are Kalman filtering type schemes, which provide an improved estimate of the state of the system up to the current time level. From a forecasting viewpoint, this corresponds to an updating of the initial conditions. The standard forecasting practice is to then run the model uncorrected into the future, driven by predicted boundary and forcing conditions. The problem with this technique is that the updated initial conditions are quickly washed-out of the model. Thus, after a certain amount time the model predictions are essentially the same as if they had been made using an initially uncorrected model. This thesis demonstrates innovative methods using local models (which are developed in the early chapters) for error forecasting and subsequent assimilation in deterministic models. It is shown in this study that by using error forecasts models can be corrected for extended forecast horizons (i.e. long after updated initial conditions have become washed-out), thus demonstrating significant improvements over conventional methods.
Back to the list of MSc abstracts