This meeting is now fully booked.

Met Office, 16th/17th October 2013

Organisers: Primary contact: This email address is being protected from spambots. You need JavaScript enabled to view it., Ben Booth (Met Office) and Ken Carslaw (University of Leeds).

The meeting will be held over two days at the Met Office, Exeter, from lunch time on 16th until lunch time on 17th October. Lunch on day 1 will be provided. A location for (an unfunded) dinner will be announced nearer the time to enable participants to continue discussions in the evening.

Applications for travel and subsistence bursaries for research students are now CLOSED.

Who: The meeting is aimed at scientists working with components of Earth system models who want to learn about the latest statistical approaches available to evaluate the effect of model uncertainty on predictions and make robust conclusions as a result of understanding the model behaviour throughout the space of its parametric uncertainty.

Why: Parameter uncertainty is rarely thoroughly investigated due to computational limitations. However, a thorough understanding of parameter uncertainty can help to direct research efforts, improve the robustness of model predictions and highlight the need for model structural changes when the model is evaluated against observations. Statistical methods have been developed to provide a framework for model evaluation to help recognise, understand, quantify and even reduce uncertainty in the computer models. These methods can be applied to the process-based models frequently used in climate science. Here we present successful case studies in a didactic style along with a list of relevant links so that participants can learn more and begin to exploit these statistical methods.

What: The 30 minute presentations will explain the concepts of the statistical methods currently being used to quantify and understand parametric uncertainty and show successful case studies in which statistical methods have been applied to components of Earth system models. 15 minutes will be left at the end of every presentation for discussion and questions. The scope and didactic nature of the presentations is designed to enable modellers to leave with a clear picture of how to investigate parametric uncertainty in their own models and to use statistics to learn as much as possible from their models. Speakers will define the type of uncertainty they are dealing with, the science question that is being addressed, the statistical methods being applied, explain the results, and discuss their experience of applying statistics in a practical way – both negative (dealing with statistical assumptions) and positive (the new things learnt by applying statistical methods). The speakers will be asked to provide a reference list before the meeting.

The meeting will cover the following topics: expert elicitation, statistical design and sampling, emulation, history matching and calibration, model reduction, and sensitivity analysis.

Cost: Attendance at the meeting will cost £25. Speakers and are exempt from this fee. Met Office employees who register will be charged internally.

How: This meeting is fully booked. Contact This email address is being protected from spambots. You need JavaScript enabled to view it. about the possiblity of being placed on a waiting list to attend this meeting.

acites banner 193



Wednesday 16th October




Registration and Lunch (including tea and coffee)



Introduction to meeting

Ken Carslaw



Introduction to uncertainty in environmental models

Jonty Rougier



Introduction to statistical design: Statistical design in environmental models

Peter Challenor



Tea and Coffee



Expert Elicitation: Web Based Expert Elicitation of Uncertainties in Environmental Model Inputs.

Lucy Bastin



Emulation: Emulation of a cloud microphysics model

Jill Johnson



 A user guide to carrying out statistical analysis of complicated models: pitfalls, caveats and limitations based on a climate - crop yield investigation.

Dan Cornford








 Thursday 17th October    


Tea and Coffee



Sensitivity Analysis: The magnitude and causes of uncertainty in a global aerosol model

Lindsay Lee



History Matching: History matching for efficient climate model tuning and quantification of parametric uncertainty

Daniel Williamson



Calibration: The potential of an observational data set for calibration of a computationally expensive computer model

Doug McNeall

Met Office


Tea and Coffee



Model Reduction: Over-parameterisation and model reduction applied to soil models.

Neil Crout



Example Number 2: Probabilistic climate projections using observations and ensembles of climate model simulations

David Sexton

Met Office




acites banner 193