Nigel Roberts, Expert Scientist, Convective-scale Modelling and Predictability
27 August 2013
Research into high-resolution models has made it increasingly possible to accurately predict extreme rainfall, storms and other events that cause flash floods. Nigel Roberts and his colleagues are at the vanguard of this fast-developing area.
Nigel works in the Mesoscale Modelling Group at the MetOffice@Reading, based at the University of Reading. Met Office staff have been at the university for over twenty years and are actively involved with several collaborative projects, including work undertaken by academic staff and PhD students. It's a long-standing relationship that's mutually beneficial: "The Met Office benefits from access to specialised data and its interpretation along with the insights of world-leading scientists. The university benefits from access to our Unified Model and our own world-leading scientific expertise," explains Nigel.
Looking at Nigel's work over recent years offers an insight into how our high resolution models - and our forecast accuracy - have evolved.
Between 2001 and 2008, Nigel led the Modelling Extreme Rainfall Events and the Storm Scale Modelling Project - both partly funded by the Environment Agency and the Department for Environment, Food and Rural Affairs (Defra). These sought to establish whether a 1-km configuration of the Met Office's Unified Model could improve accuracy in forecasting flood-producing thunderstorms. As Nigel explains, "Previously, this kind of research wasn't possible. It was only when the 'new dynamics' version of the Unified Model was developed that we were able to run simulations at such high resolutions."
The projects did show that a 1 to 1.5 km model could predict heavy or extreme rainfall events - but it was also clear that, paradoxically, the higher the resolution, the more room for uncertainty. So, as a solution, Nigel and his team turned to ensemble forecasting - using multiple forecasts instead of just one to assess certainty.
"We have our state of the art kilometre-scale models now, and they're incredibly realistic. The reason we need an ensemble is that the local fine-scale weather we now represent is more chaotic. Any errors in the forecast - even small ones - will grow. If we run an ensemble, we've got several shots at it. It gives an indication of the areas most at risk - so we can provide more reliable and useful weather warnings."
Breaking new ground
Nigel continues to work on Convective-Scale Ensembles at MetOffice@Reading, in collaboration with the Ensemble Forecasting Group in Exeter. The original aim was to demonstrate the potential value of a high-resolution (1.5 km) ensemble in forecasting the risk of hazardous local weather and to implement such an ensemble after 2012. However, things moved faster than planned.
"The results from initial case studies were very encouraging and it was clear this was something to implement as soon as possible. So a routinely running convective-scale ensemble system called MOGREPS-UK was set up by colleagues in Exeter in time for the Olympics."
There are more exciting developments to come in high-resolution ensemble and probabilistic forecasting. "The whole project is becoming even more collaborative, with researchers at Reading and other universities, as well as national meteorological services all working on the same area."
A lifelong passion
Weather has fascinated Nigel since he was a child. After completing a physics and meteorology degree at the University of Reading, he worked for a private forecasting company and joined the Met Office in 1989. His work has encompassed everything from quality control of satellite observations, to studying mid-latitude cyclones, fronts and thunderstorms, and developing new verification measures.
"It's nice to be able to discover something new about how the atmosphere works. I get a kick out of trying out new ideas and developing things that haven't been done before. That's particularly worthwhile if it leads to improvement in our forecast capability."