Subject: Scenarios Conference - Simple Models
Date: Wed, 14 Oct 1998 13:25:07 +0100
3. Use of simple climate models
3.1 Simple models used only as tools for extrapolationg/interpolationg GCM
results to estimate the effect of different scenarios or sensitivities?
1-D UD/EBMs (upwelling-diffusion energy balance models), such as the Wigley
and Raper (1992) model updated in Raper et al. (1996), in my opinion, come
into this category. I along with Jonathan Gregory and Tim Osborn have
completed a very detailed comparison of this and several alternative 1-D
models with HadCM2 results. With the addition of a sea ice parameter the
Raper et al. model reproduces well the HadCM2 results for global mean
surface temperature and thermal expansion out to 2100, for several scenarios.
However, the distinction between 3.1 and 3.2 below is not clearcut. By the
end of the 900 year 2xCO2 experiment the thermal expansion for the HadCM2
model is nearly 5 times larger than that simulated by the fitted (over
1860-2100) UD/EBM, and unlike the UD/EBM shows no sign of coming to
equilibrium. In our analysis we conclude that it is not immediately obvious
which if either model is correct. The difference serves to highlight the
uncertainty in the thermal expansion commitment. Incidently a fitted pure
diffusion/EBM gives good simulation of the HadCM2 results in both the short
and long term.
3.2 Simple models used to offer independent climate predictions?
It would probably be difficult to use 2+D models for 3.1, so they may belong
I think, 3.1 and 3.2 serve different purposes. Both may be desirable.
3.3 Depending on the answers to 3.1 and 3.2......
Whichever 3.1, 3.2 or both is adopted the results and the attendant simple
model versus A/OGCM comparisons should be given in the projections chapter.
A selection of the results should then carry over to the sea level chapter.
This consistency is very important.
It is a separate question as to whether the simple climate model results
should subsequently be used as scaling factors for regional scenario
development in the scenario chapter.
3.4 How many simple climate models are needed...
For 3.1 in order to fit the A/OGCM results extensive comparisons using
alternative parameter values/models (for example, UD versus pure diffusion)
will be necessary. As well as my HadCM2 comparison mentioned above a
comparison with ECHAM3/LSG results is also well underway. In both cases the
work shows that it is advisable to calculate the effective climate
sensitivity of the A/OGCMs for use in the simple model. We found that the
effective climate sensitivity is non-constant but apparantly varies with the
surface temperature in these models. For this calculation and for
comprehensive model comparisons a specific list of A/OGCM output is
required. This includes decade ocean mean temperature profiles, a measure of
the strength of the thermohaline circulation, the A/OGCM forcing change for
2xCO2 etc. I am keen to continue these comparisons specifically as input to
the new IPCC assessments. Unfortunately, and I think mistakenly, the US DOE
have recently decided to discontinue this line of research. An endorsement
of the need for this work by the IPCC would help my attempts to acquire
For 3.2 there would be no need of tuning to A/OGCM results and many model
results could be used to give a range. This would serve a different purpose
to 3.1 where A/OGCM results are interpolated/extrapolated for different
sensitivities and forcings.
| Dr S. C. B. Raper |
| Climatic Research Unit |
| University of East Anglia |
| Norwich |
| NR4 7TJ |
| Tel. +44 1603 592089 |
| Fax +44 1603 507784 |