Data Assimilation and Model Improvement

Data assimilation refers to the process of combining model simulations of a natural system such as the atmosphere or ocean with observations to obtain an estimate of the actual trajectory of that system. It is vitally important to weather and climate prediction. Of all the improvements made to the Bureau of Meteorology’s global forecasting system since 2011, the top 5 were all from improvements to the data assimilation system. It is data assimilation that produces the multi-decadal reanalyses from which details of climate change and climate model error can be deduced. A wide range of industries such as finance, mining and medicine now regularly use data assimilation tools that were originally developed for atmosphere/ocean data assimilation applications. The course will introduce and explain the data assimilation systems now used at the world’s leading weather and climate forecasting centres. These systems include 4DVar and various flavours of the Ensemble Kalman filter. In addition, a brief introduction will be given to more accurate but more computationally expensive methods such as the particle filter and Monte-Carlo-Markov chain approaches.