1 July - 19 December 1996

**Organisers**:JCR Hunt (*UK Meteorological Office*), ME McIntyre (*Cambridge*), J Norbury (*Oxford*), I Roulstone (*UK Meteorological Office*)

Weather forecasts are routinely computed for up to 10 days ahead, based on large quantities of wind, temperature and humidity data that are collected continuously and used to modify the computations. The data are of course insufficient to determine the exact state of the atmosphere. Since they are very expensive to obtain there is a premium on their optimal exploitation. Therefore it is of the highest importance for numerical weather prediction to identify the dominant processes and flow features that determine how the large scale weather patterns develop. By then ensuring that the continuous assimilation of data is consistent with these features the accuracy of the forecasts is greatly increased. Ocean modelling is beginning to develop similar data assimilation techniques. Recent exchanges of ideas between mathematicians and atmosphere-ocean dynamicists has brought a new geometric global viewpoint to these problems, in particular a new appreciation of how fluid-dynamical conservation laws, for example potential vorticity, connect with the symplectic geometric structure of the underlying equations of motion. A major challenge for the programme will be to bring ideas from geometry, analysis and the theory of dynamical systems to bear on the practical and urgent problems of weather forecasting, ocean and climate modelling.