Mike leads a Met Office data assimilation research group and also works on theoretical atmospheric dynamics and nonlinear partial differential equations.
Areas of expertise:
Mike's current role at the Met Office is to lead research in new methods of data assimilation. This involves combining dynamical and statistical knowledge to optimise short-range weather forecasts by exploiting the available observations, which are not sufficient to define the initial state for the model completely.
Mike's main activities within this topic are to make the best of the 4D-Var formulation currently used by the Met Office and several other large operational centres. This formulation is very expensive computationally and it is necessary to see whether its use will still be justified on future supercomputers with more parallel architecture. Mike is also taking a lead in making fuller use of prior dynamical information in the assimilation process, since there is insufficient data available to make purely statistical methods viable.
Mike is also a visiting professor at the University of Reading, and is currently co-supervising five PhD students at Imperial, Bath and Reading, on topics including dynamics, numerical methods and data assimilation. He was a main organiser along with John Ball and Sergei Kuksin of an EPSRC-funded workshop on the Mathematics of weather and climate prediction held at the Met Office in Spring 2009, in collaboration with the Oxford Centre for Nonlinear PDEs and the joint Heriot Watt Edinburgh Centre (CANPDE) sponsored by the Maxwell Institute .
Mike has spent most of his career in the UK Met Office in various roles. He has also spent time on secondment to ECMWF and the University of Reading. In the period up to 1988 he was responsible for developing the numerical methods used in the operational weather forecasting models. In the period 1987-1991 he designed both the software formulation and the integration scheme for the Unified Model. Following this, he designed a new dynamical core for the model, which both exploited his theoretical knowledge to improve treatment of the large scale flow and created a non-hydrostatic capability allowed the application of the model to very small-scale local forecasts and simulations. This model became operational in 2002.
Since the governing equations cannot be solved except in a coarsely averaged sense, Mike also developed simpler sets of equations which explained many mesoscale phenomena in terms of large scale flow. More recently their solutions have been made mathematically rigorous. This is important in understanding predictability of weather systems and informs decisions made in operational modelling. It has led to many collaborative activities, such as Newton Institute programme (Cambridge) on Atmosphere - Ocean Dynamics and membership of an NSF-funded focused research group on Optimal transportation, its geometry and applications.
From 1983-1988 and again from 1991-1998 Mike held senior management positions responsible for the strategic direction of modelling and data assimilation on all scales from UK to global. During this period the Met Office established its position as a leader in global numerical weather prediction.
Mike received the L.F. Richardson prize in 1980 for work on finite element methods. He received, with others, the Esso Energy award in 1983 for developing a new high-resolution global model which improved forecasts for aviation. He received the UK Met Office/Royal Air Force: L.G. Groves memorial award in 1991 for the development of the Unified Model. He received the Royal Meteorological Society: Buchan prize in 2005 for papers on large-scale atmospheric dynamics.
Mike is a visiting professor in the Department of Mathematics at the University of Reading.
Mike published a book on A mathematical theory of large-scale ocean/atmosphere flow in 2006.