Warren manages the Global Data Assimilation and Ensembles Group with a research interest in understanding and estimating model forecast error.
Warren is a scientific manager of the Global Data Assimilation and Ensembles Group. His research background is developing stochastic physics schemes in the Met Office Ensemble Prediction System MOGREPS. Accurate estimates of forecast uncertainty are important products of an ensemble system. This is taken from the ensemble spread which is generated through initial condition perturbations and stochastic physics perturbations, which control the rate of growth of spread within the ensemble.
The recent focus of Warren's work in stochastic physics includes the SKEB scheme, SST and SMC perturbations, and the Random Parameters scheme. The SKEB scheme injects wind increments into the model at every time-step, and are based on a random stream-function forcing pattern that is modulated by locally diagnosed energy dissipation fields. In this way the model forecasts are forced to diverge more rapidly in regions where there is a perceived drain of energy from the model. SST and SMC perturbations are provided at the start of the model forecast, while Random Parameters perturbs selected physics parametrization parameters. These vary during the course of the run, and differ between each ensemble member.
Warren is currently responsible for maintaining the data assimilation and forecast model components that feed into the operational global suite. This includes implementing new developments like SKEB and testing other changes to the system (like resolution, number of ensemble members and initialisation).
Warren joined the Met Office in 2008 after having worked at the South African Weather Service since 1993. He was involved in Numerical Weather Prediction (NWP) for most of that time and was responsible for implementing and maintaining various modelling systems including the COLA T30 GCM, NCEP T126 GFS, NCEP Eta and Met Office UM (regional 12km and 4km). He also developed a forecast verification system and managed the NWP group for several years.
Warren completed his PhD at the University of Cape Town in 2002. This work involved a new approach to understanding large-scale atmospheric variability on a seasonal time-scale through atmospheric energetics.
Warren was a member of the (THORPEX) GIFS-TIGGE working group from its inception in 2005 until 2008 and he helped introduce the successful concept of local forecast processing centres into the plan of a Global Interactive Forecast System (GIFS). This has been a big step forward in transferring NWP output from global producing centres into products with local value. More recently, until December 2013, he was a member of the EUMETNET Short-Range Numerical Weather Prediction Expert Team on Ensemble Prediction Systems, which addresses predictability issues and uncertainties in the NWP models, LAMEPS and ensemble data assimilation.
Last updated: 30 June 2014