Why are some things easy to simulate but some difficult? (Speaker: William Ingram, Oxford University)

Speaker: William Ingram, Oxford University, Atmospheric, Oceanic and Planetary Physics, Clarendon Laboratory

Over the last half-century computers have revolutionized atmospheric and climate science.
As soon as numerically well-behaved discretizations of the basic equations
for each process (if they exist - if not, equations that give sensible
approximate descriptions) were derived and and combined, they gave simulations that accurately
represented many emergent properties - some, storm tracks for example, long before
the physics was fully understood.  Yet despite all the complexity and detail of
state-of-the-art GCMs, and their vast range of tunable parameters, we still
struggle to achieve useful simulations of some important features of the
atmosphere and climate system, ENSO for example.

Of course this is not repeated bad luck, nor pure incompetence.  Simulating
ENSO really is much more difficult than simulating storm tracks.  But
we do not usually think or talk much about why.  I think we should,
because it is interesting, and because it may give useful
insights - into the difficult cases, but also, perhaps, cases that we have found easy but have no reason to
expect to be.  As well as storm tracks and ENSO, I shall discuss the internal
variability of forecast and climate models, and two quantities that may seem
physically closely related but vary greatly in "simulability", the cloud &
water vapour feedbacks on climate change.