cc: sfbtett@meto.gov.uk , k.briffa@uea.ac.uk , t.osborn@uea.ac.uk
date: Tue, 06 Feb 2001 23:47:47 +0000
from: "Tim Osborn" <t.osborn@uea.ac.uk>
subject: sea level in SOAP
to: plao@geo.vu.nl

Orson,

I late and probably incomplete reply to your various questions about the sea
level package in SOAP.

(1) Going back to your e-mail of 1 Feb.

Yes, the global/hemispheric picture is of monotonic cooling from 1000 AD to
1900 AD, then rapid warming.  So you might well ask why glacier melting
began earlier, and why sea levels seemed to rise suddenly with little lag
after 1900 AD.  One answer might be that some records/regions, especially
circum-North Atlantic reconstructions, indicate more inter-century climate
variability, with more marked Little Ice Age and Medieval Warm Period
phases, with perhaps notably cooler conditions during the 1600 and 1800s,
with a warmer 1700s between.  Perhaps the 1600s were coolest of all, then
glacier melting began during the warmer 1700s?  Or is that too early.  Also,
there's seasonality to consider (winters showed an upward trend prior to
1900 I believe), plus (for the glacier question) precipitation variability
(e.g. Scandinavian glaciers have been growing recently rather than losing
mass, due to NAO-related enhanced winter precipitation).

As I understand it, the models are capable of producing regional differences
in sea level change, related to circulation changes.  So the thermal
expansion part of sea level should show regionally different results. 
However, I think the glacier-melt models will be run off-line (i.e., after
completion of the climate simulations).  If this is the case, then the
glacier-melt part of sea level cannot influence ocean salinity and hence
ocean density and circulation.  That would be a disadvantage of course, as
glacier-melt would then have to be added uniformly over the globe.  Unless
Simon tells me otherwise!

(2) Your e-mail of 2 Feb.

Thanks for the rationale text - that'll be useful.

The model to be used (HadCM3) for the past 500 years is also one that is
used for future sea level and climate scenarios.  So its relevant to test
it.  We will have one simulation 1500-2000 forced by natural forcing (solar,
volcanic and maybe the small orbital changes).  Then a second simulation
1750-2000 that will have natural and anthropogenic forcings (greenhouse
gases, sulphate aerosols, and maybe stratospheric ozone and maybe land-use
changes).  For the overall project, a second model (run in Hamburg, ECHAM)
will have a natural forcings run from 1000-2000, plus a
natural+anthropogenic forcings for 1750-2000.  They have offered their model
output for the sea level workpackage too, though I'm not sure how easily it
can be used to force the Hadley Centre's glacier melt model (Simon - you'll
have to advise here...do you just need monthly fields and temp and precip? 
I guess they'll produce thermal expansion component anyway, since there
model probably doesn't have a rigid lid and will automatically predict
regional sea level).

Model adaptation is not really possible within this project, so differences
between paleo and simulated sea level will raise more questions for future
work - but also perhaps provide a range of variation from different sources.
 The HOLSMEER project does not involve such a model-data comparison, so we
should indeed stress that in the SOAP proposal.

(3) Your e-mail of 6 Feb.

Q1. As outlined above, the model has capability of producing regional sea
level (Simon, please correct if I'm wrong, though I'm fairly sure about
this), though how realistic this is I'm not sure, especially if glacier melt
isn't allowed to be added regionally and affect salinity & circulation (not
sure how big a term it would be in the salinity budget, given the fairly
small amounts of climate change in the past 500 years, especially under
natural forcings only.  So we can certainly look at variations within the
Atlantic, at a fairly coarse scale though.
Q2. Links between simulated multi-decadal/century climate patterns and
simulated sea level variability can be looked at within the model data.  If
these are internally-generated climate patterns (e.g., multi-decadal North
Atlantic Oscillation anomalies), then these wouldn't neccessarily coincide
with the timing of such variability in the real world, but would still be
useful to find out how sea level responds to such patterns.  If they're
forced externally, then they may also have happened in the real world (a
number of caveats here), and so could be searched for.
Q3. HadCM3 will be forced as described above.  The difference between the
natural and the natural+anthropogenic runs will be useful to estimate how
much sea level rise we would have got anyway, due to coming out of the
Maunder-minimum-induced cool period, compared to what is observed and
compared to what the anthropogenic forcing adds.

If the model and paleo sea levels show no similarity at all, then I guess we
will just have to present them as two equally possible results (or with
unequal possibilities if we can gauge uncertainty ranges on each approach).

All questions you raise, regarding ocean-climate links etc., can actually be
included in the proposal as things that this workpackage can start to
address.  It sounds like there's scope for lots of work.  If you need to
expand on the budget and number of months previously suggested then that
should be fine, though I remember you didn't want to commit to too much.  If
not, then we'll just have to make it clear that we will just make a first
attempt at this model-data comparison and a first attempt at answering some
questions, but won't achieve them all.

Apologies to all for what's turned into rather a long e-mail.  I'm away
Wednesday, but back Thursday.  You can cc replies to Keith if you need
anything clarified while I'm away on Wednesday.

Regards

Tim

