ABSTRACT
It was noted in Chapter 1 that the problem of forecasting the near future is different from the calibration of a simulation model of the system. In the forecasting case it is not necessary to simulate all aspects of the system. The requirement is to provide predictions at the lead time of interest with minimum uncertainty. The forecasts need to be available with sufficient lead time to be useful in making decisions, and data assimilation can be used to improve the predictions and reduce the forecast uncertainty as a particular situation or event evolves.
