Casino-21: Public Participation in Climate
Simulation of the 21St Century
David Stainforth Department of Atmospheric, Oceanic and Planetary
Physics, University of Oxford. United Kingdom Casino 21 is a project
aimed at quantifying uncertainty in forecasts of human-induced climate
change. It will use a massive Monte Carlo approach which will involve
running multiple versions, known as an ensemble, of a full-scale climate
model. These runs will be carried out using idle CPU on personal computers.
The project has two main themes. These are a) the execution of a very
large, perhaps multi-million member, ensemble forecast of the climate
of the 21st century, and b) the raising of public awareness of the difficulties
and possibilities for climate prediction. An ensemble forecast involves
comparing the predictions of many slightly different versions of a climate
model. However, an ensemble of this size requires so much computing
capacity that it would be infeasible using the facilities available
to the research community. Therefore, following the lead of the SETI@home
project, the plan is to utilize the spare CPU available on PCs in homes
and businesses. To do this the project must engage the interest of many
PC owners and their interest must be maintained for at least the length
of the model run, which could be a year or more. Consequently the public
awareness issues will be intrinsic to the project's success. The need
for such a large number of model runs comes from the fact that climate
models involve the parameterisation of many atmospheric and oceanic
processes. Unfortunately the details of many of these parameters are
only known within certain relatively large ranges. The goal is therefore
to perturb selected forcing parameterisations within their range of
uncertainty and perform many repeat simulations of the period 1950-2050.
Each simulation will use a different combination of the chosen parameters.
The model-versions which are inconsistent with observed climate change
over the 1950-2000 period, i.e. up to the present day, will be discarded.
The surviving simulations will provide an estimate of uncertainty in
decadal-timescale forecasts of climate change under a range of future
scenarios. Further details can be found at http://www.climate-dynamics.rl.ac.uk
|