An ensemble probability forecast of rainfall > 10 mm

Regional model evaluation and development

It is the role of the Regional Model Evaluation and Development (RMED) team to deliver regional system at these grid-lengths for weather and climate applications. As the Met Office have many partners worldwide, and provide advice in various regions around the world, the systems must perform well across a range of environments and weather conditions. For example, the system must provide valuable advice over the U.K in winter and summer while also being an effective tool for forecasting the impacts of a tropical cyclones.

The research carried out by the RMED team looks beyond the next release of the system towards the “next generation” of regional systems. This includes systems where the grid-lengths are the order of a hundred metres or less and there is increased importance on the representation of urban processes. RMED research also assesses the effect of feedbacks between the atmosphere, ocean, wave and land surface on predictions of the coupled environmental system, with interests on the value of coupled predictions around the UK and for other regional domains around the world.

RMED teams 

Convective-scale modelling research 

We are looking ahead to next generation modelling systems with gridlengths of order 100m – often known as “city-scale” models. These models are likely to be important for forecasting high impact phenomena such as flooding, urban heat and poor air quality. The work involves understanding the best way to represent partially resolved turbulence, convection and the urban surface in models of this scale. A related focus is work to improve the representation of convection in the current generation of km scale models which is also highly relevant to the above issues. This team is based at MetOffice@Reading and carries out much of this work as collaborative projects with University of Reading scientists. 

Coupled regional system development

This team aims to understand processes associated with air-sea interaction using kilometre scale models of the atmosphere, ocean, waves and land surface, in order to deliver improved regional coupled model capability for weather, ocean and climate applications. The team’s research focussed on the UK is conducted in collaboration with others under the UK Environmental Prediction activity.

Regional systems evaluation 

The Met Office runs a number of operational and real-time regional weather models across the globe, for both weather forecasting and science research. The Regional Systems Evaluation team works to understand and evaluate regional models in different environments. The team focus is on both UK and tropical domains, primarily in Southeast Asia and Africa. Innovative use of new analysis and visualisation techniques is core to creating a holistic picture of model performance, leading to greater understanding of model behaviour and deficiencies to be targeted for improvement by model development teams. The team works closely with model developers and operational meteorologists, both in the UK and with our worldwide partners, to ensure technical advances are fed directly to operations and their feedback on performance is captured and pulled back into Science.

Regional atmosphere delivery 

The RAD activity focuses on delivering an improved kilometre scale Regional Atmosphere Land (RAL) model for weather and climate applications. This is achieved by releases of the science configuration which are described in peer reviewed publications A deeper understanding of model systematic behaviour and errors is gained through the regular assessment of RAL configurations with UM Partners and areas for model improvement are prioritised through engagement with model user groups.

Process Evaluation Groups and working groups address priority areas of research such as convection and ensemble spread. In addition, the team provides evaluation, development and support for the 300m London model and maintains and develops RAD Trialling Suites to make then more user friendly, robust and computationally efficient.