Data Assimilation and Ensemble Forecasting

The largest improvements in weather forecasting over the last decade have been attributable to data assimilation improvements. Data assimilation schemes generate initial conditions for forecasts by using observations to correct short-range forecasts. The corrected states provide initial conditions for subsequent forecasts.

However, more research is needed. For example, clouds are known to be a leading cause of prediction error. Cloud observations could help fix these errors, but few cloud observations are assimilated. Furthermore, the dream of allowing observations to teach model parameterisations of clouds has yet to be achieved.