date: Fri, 11 Dec 1998 11:22:06 +0000
from: John Shepherd <John.G.Shepherd@soc.soton.ac.uk>
subject: Multi-millennial ESM
to: gjjenkins@meto.gov.uk, k.briffa@uea.ac.uk

Message also sent to : modellers, sci div heads, centre/survey directors,
a.watson, bill, j.gould, m.fasham, p.liss, p.williamson, r.dickson,
t.tyrrell, tom anderson, j.gash


Dear Colleague
	
	Herewith as body text and as a Word 6 attachment an updated version of an
outline proposal (which some of you have seen before) for a project which I
am planning to develop. I think it's pretty much self-explanatory. I hope
to convene a workshop to kick the idea around early in the New Year, with a
view to submitting a full proposal in summer 1999... 

	All expressions of interest, comments and suggestions are invited.

		With best wishes

			John

----------------------------------------------------------------------------
----------------------------------------------------------------

Third Draft
Straw-man Proposal for a NERC Cross-Board Initiative
 Consortium Project
Multimillennial Models of the Earth Climate System

A) 	Aims

1) To build an integrated model of the Earth's climate system, including
the ocean, atmosphere, cryosphere and biosphere, capable of integration for
about one million years.
 
2) To test its ability to reproduce natural variations of Earth climate
through several major glacial/interglacial cycles.
 
3) To improve the model to the point where it has significant predictive
utility in assessing the combined effects of natural and man-made climate
forcing.

B) Background and Rationale
 
 	We cannot be confident of any long-term predictions of climate change
because we do not yet understand Earth's climate system  :  in particular
the glacial/interglacial cycles which sometime occur.  These are believed
to be due to an enhanced response (positive feedback) of the
atmosphere/ocean/cryosphere/biosphere coupled system to weak solar forcing
(Milankovitch cycles), but a full explanation is lacking.  It is difficult
to evaluate proposed mechanisms quantitatively, because integrated dynamic
models of the complete system are not generally available.  The most fully
developed model is possibly that of the Potsdam Institute for Climate
Research, which is planned to embrace social and economic factors too
(which may be extremely difficult to do well). [note:  the status of other
integrated  models needs to be checked]
 
 It is suggested that the scientific community in the UK is uniquely well
placed to create an integrated model of the Earth Climate system, drawing
on expertise from NERC Centres & Surveys, the Hadley Centre, and other
supported groups.  Since we do not believe that we yet understand the
system, the model would be exploratory rather than predictive, at the
outset, and involve a high risk of failing to simulate the natural system
for some time.  Nevertheless, use of such models is a necessary part of
developing our understanding and evaluating the credibility and magnitude
of proposed mechanisms, and the failures should be instructive.  Testing
and validation will require comparison with palaeodata on temperature,
ocean circulation, productivity, etc.  This may probably best be done by
attempting to simulate the occurrence (or not) of ice-ages as a function of
solar forcing, and  continental and oceanic configurations, for which data
exist in the geological record.
 
C) Method & Model Structure

The model will need to incorporate suitable representations of the oceans,
the atmosphere, the cryosphere (ice-caps and sea-ice), the biosphere
(marine and terrestrial) and certain aspects of the lithosphere (variable
continental configurations, erosion and calcium run-off, at least).

At present GCM-based integrated climate models such as those of the Hadley
Centre are just capable of integration for about one millennium.  To permit
multiple realisations of integrations one thousand times longer will
require extremely simplified representations of most components of the
system, with very carefully selected parameterisations of many important
processes (e.g. clouds, deep water formation, and atmospheric transport).
The longest times scales involved are those for ice-sheet dynamics and
ocean chemistry (the calcium/carbonate system), and these will need to be
represented with most detail and realism.  Other aspects may be modelled
using highly simplified (and possibly transient equilibrium)
representations.  Modest geographical resolution will be required to enable
latitudinal gradients (important for ice dynamics) and continental
configurations, which are important in determining oceanic heat transport,
to be represented.  The model parameterisations should be derived from
theoretical considerations, and calibrated using comparison with field data
and/or the results of more detailed models of subsystems, run for shorter
time periods, such as those of the Hadley Centre and those to be developed
under the Prescient Programme.

The model should almost certainly be modular, to enable alternative
representations of subsystems to be incorporated, evaluated and compared
easily.  Careful attention to the definition and implementation of
interface schemes will therefore be required.  Such a structure will also
permit and facilitate involvement of a broad (and probably distributed)
community in the project.  However, since the goal is the creation of an
integrated tool, the project needs to be a rather highly integrated and
managed effort - a consortium project rather than a conventional thematic
programme. It would therefore be similar to the early phase of the UGAMP
programme (now designated as core-strategic).

A tentative outline of a possible structure and scope of the model for each
major subsystem is given below. The initial spatial resolution might be of
the order of 10 deg latitude by 30 deg longitude, with (say) 5 levels in
both the atmosphere and the ocean.

Atmosphere
A global spatially resolved radiative/convective or energy balance model
(?), including parameterised representations of meridional heat (and water)
transport, the effects of water vapour and green-house gases on
transmission, emissivity and radiation balance, and exchanges with the
land, oceans, and cryosphere, including precipitation and evaporation.

Ocean
Multiple meridional-vertical representations (two or three per ocean,
linked via the Southern Ocean (i.e. similar to that of Stocker (1997)).
CO2/carbonate/calcium chemistry, including calcite formation/dissolution,
needs to be incorporated, as do biological production, organic & inorganic
carbon burial, and interchanges with the atmosphere and cryosphere.
Parameterisation of deep-water formation (and therefore the meridional
circulation) will be a crucial feature (to be derived from process models?).

Cryosphere
A model of the thickness and latitudinal extent of sea-ice and ice-sheets,
and their dynamics, accounting for the presence/absence of land, will be
needed.  This module will be closest to the state-of-the-art
representations in this field.  Effects on sea-level and ocean temperature
and salinity must be included because of possible transient (?) effects on
ocean circulation.

Biosphere
(a)  Terrestrial  :  to be determined  :  a simple (transient-equilibrium?)
representation of biomass and its growth/decay for latitudinal zones on
each continent (or a coarse subdivision thereof) may be adequate.  Effects
on albedo and interaction with the hydrosphere should be included (as
simple parameterisations)

(b)  Marine
A simple (NPZD?) representation of marine production needs to be
incorporated and linked to the carbonate chemistry, to represent the carbon
cycle and the interactions with atmospheric CO2, so that both natural and
anthropogenic effects of CO2 variation can be assessed.
	
Hydrosphere
[to be inserted (CEH)]

Lithosphere
A static (but variable) representation of continental configuration (and
elevation ?) and ocean bathymetry should be adequate.  Variable effects of
precipitation on erosion and Calcium  dissolution need to be included to
close the Calcium/Carbonate/sediment accumulation cycle. 
NB:	The acquisition of suitable palaeo-data from the geological record is a
vital part of the overall project, but not part of the model construction
per se.

Overall
In all of the above modules, it should (so far as possible) be possible to
switch key processes on and off, in order that (numerous) alternative
simulations can be run to determine the relative importance of each process.. 

	The model needs to be capable of running for several hundred thousand
years at least, for each of a large number of scenarios. With the degree of
spatial resolution outlined above, there would be around one thousand grid
points in the atmosphere and the ocean, compared to several million in
GCM-based climate models.  Great care will be needed to parameterise
sub-grid scale processes adequately, and this represents a major challenge.

It is not clear whether or not seasonal processes will need to be
represented explicitly. If they do, then time-steps of less than one year
will be needed, and computational speed is likely to be a serious limiting
factor. If they can be parameterised effectively, much longer time-steps
mat be achievable, allowing considerable elaboration of other aspects of
the model.

D)	Resources
It is suggested that the model should be highly modular, (possibly with a
NERC Centre/Survey or University group taking lead responsibility for each
module). As a first guess, it should be possible for prototypes of each
module to be constructed by a few people (mostly post-docs or equivalent)
working for a few years.  The scale of the initial development project,
including a synthesis and interface teams, co-ordination and project
management and hardware (workstations, etc.) is therefore of the order of
1M per annum for (say) 5 years.  Access to supercomputer facilties may
also be required during the later stages, especially for sensitivity
testing, estimation of confidence limits, etc. It is possible that some
external funding (or at least co-funding) might be secured (DoE ?).  The
project does not easily fit into any of the categories of NERC's  present
Funding Model.  If successful, it should probably evolve in due course into
 a joint collaborative core-strategic programme between the Centres/Surveys
and the other institutions involved.  These would include SOC, BAS, CEH,
CCMS, BGS, UEA, UGAMP/CGAM (Reading) and the Hadley Centre. The development
work can be carried out in a distributed but highly co-ordinated manner,
and may probably thus best be treated as a "consortium" project, rather
than a thematic programme. A core team (including probably the synthesis
and interfacing teams) will be required at one location, and the
possibility of secondments and joint appointments of staff at other sites
should be considered as a possible mode of operation.

	The people involved would all require good mathematical/computational
skills, and appropriate appointments might therefore be to fellowships
similar to those available under the existing conversion scheme for
physicists and mathematicians.
 John Shepherd
Southampton Oceanography Centre
10 December 1998


Attachment Converted: "c:\eudora\attach\Mmesm.doc"
