Dr Jonathan Flowerdew
Jonathan is a research scientist focusing on the verification of raw and post-processed ensemble weather forecasts.
Areas of expertise
- Generation of initial perturbations for ensemble forecasting
- Ensemble data assimilation
- Ensemble verification
- Ensemble calibration and combination
Recent increases in computing power have made it possible to run local weather forecasts with horizontal grid spacings of one or two kilometres. This allows the explicit simulation of convection, improving predictions of severe weather such as heavy rainfall. However, forecasts at these scales can be very uncertain, and ensembles are needed to predict the range of outcomes which might occur in any given situation. Even state-of-the-art forecast models involve approximations, limited spatial resolution, and simplified representation of key processes, such that they can exhibit systematic errors in both forecast climatology and ensemble spread.
Post-processing techniques aim to improve the issued forecasts by measuring and correcting these systematic errors. Both the underlying forecast models and the post-processing systems require comprehensive verification to measure their performance and suggest potential improvements.
Jonathan’s research focuses on the development of novel verification techniques, particularly for high-resolution ensemble forecasts. He is currently developing the verification component of the new post-processing system, to measure and optimise the benefit from each post-processing step, and enable alternative post-processing techniques to be effectively compared.
Jonathan's previous work at the Met Office covers several areas of ensemble forecasting. He developed an ensemble forecasting system for coastal storm surges, which has been extended to provide forecasts up to seven days ahead. He worked on the initial condition perturbations for the Met Office global ensemble, in particular the system which matches the ensemble spread to recently-measured forecast error. This in turn required constructing accurate estimates of the error variance of ATOVS brightness temperature observations, with requirements that differ from standard data assimilation. As part of the THORPEX and GEOWOW collaborations, he developed a scheme for calibrating the statistical reliability of ensemble forecasts whilst preserving their spatial structure, and evaluated the benefit of combining 15-day ensemble forecasts from different operational centres. More recently, he conducted the first Met Office research into the synergy between ensembles and data assimilation for convection-permitting models.
Jonathan joined the Met Office in 2006, after completing a DPhil at the University of Oxford on the use of nudging and feature tracking techniques to evaluate climate model cloud. Prior to that, he spent two years working on graphics, audio and system device driver software for digital TV. Jonathan studied physics as an undergraduate, again at the University of Oxford.