date: Tue, 02 Nov 1999 14:54:55 +0000
from: Tim Osborn <t.osbornatXYZxyz.ac.uk>
subject: Summary of HC-CRU discussions 29/10/99
to: Simon Tett <firstname.lastname@example.org>,email@example.com,k.briffa@uea
We're having e-mail (and other computer-related) problems, so please reply
if/when you receive this. Thanks.
An aside for Simon:
At 11:55 AM 11/1/99 +0000, Simon wrote:
> Can I trouble you further -- Phil tells me that you have somewhere
> data on observational error on decadal timescales. Can I have
> access to it and can you tell me where it lives and format details.
These disappeared from our ftp-server during an earlier crash. I was just
re-extracting them when our current computer failure occurred. If I don't
get back to you in a day or two, then please remind me.
Summary of discussions between HC & CRU (Norwich 29/10/99)
Mat Collins, Simon Tett, Keith Briffa, Tim Osborn
(1) Mat summarised his comparisons between regional summer temperatures
simulated by HadCM3 and those estimated from tree-ring density chronologies.
Two versions of the tree-ring reconstructions were used. Tim and Keith
explained the difference between these two versions. The first had the
age-effect on trees removed by standardisation (fitting and removing a
Hugershoff function from each individual tree core), which can also remove
low-frequency variability. The second had the age-effect circumvented by
banding the tree density series according to discrete age-bands, before
later combining normalised age-banded series together. The age-banding
retains the low frequencies, but may add noise/uncertainty early in the
reconstructions when there are fewer tree cores.
The Hugershoff-standardised reconstructions matched the HadCM3 variability
reasonably well, in terms of spatial patterns and temporal variability.
Bigger differences were found when using the age-banded reconstructions,
which showed greater temporal variability. It is clearly important to
identify, if possible, which version is closer to the true level of
low-frequency climate variability.
(2) Tim explained parts of the calibration process. The 'NH' age-banded
reconstruction is now calibrated using principal components regression, and
Tim will send a revised series with revised calibration statistics to Mat.
(3) It was agreed that the age-banded reconstructions were now finalised
and could be used for this particular comparison, and presented in Venice.
(4) The objectives and focus of possible proposals to EU and to
NERC-PRESCIENT were discussed. The following items were considered:
(a) updating of tree-ring network needed to investigate recent changes in
growth and in growth-climate relationships.
(b) should the implications for carbon cycle modelling be followed up?
(c) proxies other than tree-rings should be used.
(d) 500-yr simulation with a lower-resolution version of HadCM3 would be
cheaper, and such a model would also be attractive to the NERC community
(PRESCIENT, also COAPEC?, and wider). A low-res HadCM3 1500AD-2000AD
simulation plus staff might bring costs down to 200K euros (with 50%
contribution from EU, 50% from HC budget).
(e) HadCM3 now contains the TRIFFID vegetation model (from Peter Cox),
offering the possibility of growing trees and comparing with tree-ring
(f) should proxy reconstructions be used for detection, or to assess model
simulations that are then used in detection, or should the focus be on
providing uncertainty ranges for future projections?
(g) can precipitation variability be assessed (less important for
detection, but important from an impacts point of view)? This can be done
for limited regions using drought-sensitive tree-rings and ice core
(h) what rationale could be given for further improvement/extension of the
handful of multi-millenial chronologies?
The three questions that Simon most wanted an answer to and that proxy data
could help to solve are:
(a) Are model esimates of natural climate variability realistic (with a
focus on the 10-100 yr timescale)?
(b) What is the contribution to natural variability from external forcings
(solar, volcanic, natural CO2 changes)?
(c) What is the contribution to post-Maunder Minimum sea level rise from
natural (internally-generated & externally-forced) climate variability?
So, a brief rationale for an EU proposal might be (paraphrased from Simon):
"All climate change signal detection and attribution studies to date have
assumed a model-based estimate of natural climate variability. This is a
major and relatively untested assumption to make, and is potentially a
major source of criticism that could be used to detract from all such
studies. The relative shortness, and possible contamination by
anthropogenic signals, of the instrumental climate record hinders an
assessment of levels of simulated variability on decadal to century time
scales. Accurately dated and calibrated climate proxy records with dense
and widespread coverage provide an independent record of climate
variability covering a number of centuries. Existing proxies must be
synthesised, additional samples and sites must be taken to extend the
temporal and spatial coverage, and objective calibration together with an
assessment of (possibly time-dependent) error/uncertainty must be
completed. The resulting climate reconstructions can then be used to
assess the variability simulated by appropriately-forced coupled climate
system models. Assessment and comparison methods must be developed to
compare forward/inverse transfer between climate and proxy variables, to
incorporate uncertainty in reconstructions and model forcings, and to
quantify differences in terms of their influence on climate change
detection results. While widespread coverage is required to provide an
assessment of climate variability at the large spatial scales that are
important for climate signal detection, the spatial scale associated with
climate variability increases with time scale. Multi-millennial
reconstructions of small-area climate also provide, therefore, important
information for comparison with model simulations, particularly at the
century time scale. Such reconstructions need to be further improved and
extended, and used to assess how unusual the twentieth century climate is
in relation to records that are longer than the existing natural
variability simulations from state-of-the-art coupled climate models."
Well it's something to build on. Not sure that we covered any *details* of
a potential CRU proposal to PRESCIENT.
Let me know if I've omitted anything major.
Dr Timothy J Osborn | phone: +44 1603 592089
Senior Research Associate | fax: +44 1603 507784
Climatic Research Unit | e-mail: t.osbornatXYZxyz.ac.uk
School of Environmental Sciences | web-site:
University of East Anglia __________| http://www.cru.uea.ac.uk/~timo/
Norwich NR4 7TJ | sunclock:
UK | http://www.cru.uea.ac.uk/~timo/sunclock.htm