Diagnosing errors in our models is the first step towards solving the problems and improving our forecasts.
Comparing the Met Office Unified Model against observations on timescales from hours to decades, combined with feedback from our forecasters and other model users, enables key model biases to be identified. The Model Evaluation and Diagnostics team acts as a focal point for such evaluation work and aims to understand the cause of the biases from a process-based perspective, before working with model developers to fix deficiencies.
- Evaluating model biases which develop on timescales from hours to decades.
- Developing novel diagnostic techniques to understand these biases from a process perspective, utilising a variety of observational data (satellite data, intense observing sites, re-analyses, etc.).
- Develop diagnostic methods to understand the impact of new model science (e.g. stochastic physics).
- Work with model developers to address the biases.
- Process-based comparison of the UM with other GCMs on all timescales.
- Understanding errors in tropical cyclone simulation: Tropical Cyclones require specialist diagnostic techniques to track them and identify the cause of errors in their track and intensity.
- Improving the simulation of processes over Africa: The ability to predict rainfall over Africa, both in terms of intensity and timing of the onset of monsoons, is central to the
DFID Climate Science Research Partnership programme to improve predictions over Africa. We aim to understand the many important meteorological processes operating over Africa and remotely, evaluate the model's ability to simulate them and assess the role of model resolution.
- Errors in mid-latitude synoptic systems: The accurate simulations of storm tracks and blocking is essential for UK forecasts across all timescales. The development of diagnostic techniques which follow synoptic features and identify the origin of errors is key to this project.
- Model error growth in the Madden-Julian Oscillation (MJO): The MJO is an important phenomenon that links weather and climate. This project is to analyse the MJO error in various configurations of the Unified Model to gain understanding of the causes of model biases.
- Development and evaluation of stochastic physics parameterizations: This project involves the development and adaptation of stochastic physics schemes to offset the absence of small scale processes in weather and climate models, and thus include some of the missed variability of subgrid processes.
Transposer-AMIP: is an international project to run climate models in weather forecast mode, initialised from NWP analyses. This allows an assessment of the fast physical processes operating in climate models using detailed observations of particular meteorological events.
Last updated: 29 July 2014