Back to the programme printer.gif Print

Delegates are invited to meet and discuss with the poster presenters in this topic directly after the session 'The model chain: First steps towards tomorrow's technology' taking place on Thursday, 13 March 2014 at 09:00-10:30. The meet-the-authors will take place in the poster area.

Karen Walter Met Office, United Kingdom
Karen Walter (1) F P
(1) Met Office, , United Kingdom

Printer friendly version: printer.gif Print

Presenter's biography

Biographies are supplied directly by presenters at EWEA 2014 and are published here unedited

Karen Walter has been an Applied Scientist in the Renewable Applications team at the Met Office for 11 months. During this time, she has worked on evaluating the performance of the Met Office advanced wind-energy site-screening and planning tool based on downscaled mesoscale Unified Model (UM) data. Karen holds a BSc in Mathematics from the University of Strathclyde and an MSc in Weather, Climate and Modelling from the University of Reading.


The model chain pushing the limits of numerical models


It has become common practice to use nested suites of fully dynamic Numerical Weather Prediction (NWP) models to produce virtual wind fields at wind sites. As the resolution of the inner domain of the nested suite increases, progressively more detail of the boundary layer flow is resolved. However, NWP models are computationally expensive, physical processes exhibit scale dependency, high resolution land use and orography data are not always available and there are limits to predictability that restrict how finely the flow can be resolved. This study will illustrate and discuss these issues using results from the Unified model.


The Unified Model (UM) is the Met Office NWP model which runs at scales from 100m to 100km and from nowcast to centennial climate projection timescales. Using a nested suite of the UM, comparisons are made between a blended turbulence parameterisation designed to operate at various resolutions with conventional boundary layer and sub-grid turbulence parameterisations to illustrate the impacts and limits of resolution increases. A 2km ensemble configuration shows the limits of predictability and, using compute costs on the Met Office supercomputer, along with current upgrade plans, conclusions are drawn on the current limits on resolution in NWP models.

Main body of abstract

The UM been has configured as a nested suite, using an initial 60km Global model driven by the ERA interim reanalysis dataset, which provides boundary conditions for successive higher resolution domains nested within this. The inner 100m resolution domain is 10km by 10km.
The suite currently has both boundary layer and sub-grid scale turbulence parameterisations which are active at different resolutions. Comparisons are made with a new scheme which blends a conventional parameterisation suitable for large grid scales with a sub-grid turbulence scheme that is suitable for large eddy simulation. The key parameter for the blending is the ratio of grid length to boundary-layer depth and it has been shown to improve the transition from resolved to unresolved turbulence as grid length is reduced. The resulting flow simulations at different grid lengths are verified against mast observations and we use the results to illustrate the performance of NWP models at high resolutions.
Forecasts from the 2km UK operational ensemble members demonstrate the limits of predictability in boundary layer flow. Using compute costs of the nested suite on the Met Office supercomputer and the anticipated upgrades within the next five years, simulation costs and elapsed times for configurations of the UM at high resolutions for a variety of domain sizes and run lengths are determined. The implications for the limits of achievable fully dynamic NWP wind fields are discussed.


Nested suite NWP models provide a useful tool for simulating wind fields at sites of interest at resolutions down to order 100m. Current NWP models use a mix of physical schemes and parameterisations which are not always scale aware, which places limits on the resolutions at which NWP models provide skilful predictions. Anticipated increases in supercomputer resource available to the renewables industry make sub kilometre scale modelled fields for large domains and long periods achievable, but further research is required to ensure that skilful predictions are generated at these scales.

Learning objectives
Current generation NWP models have limits on their resolution determined by their physics and dynamical cores.

High performance computing advances make practical high resolution configurations for large domains and long timescales.