from: Tom Wigley <wigleyatXYZxyz.ucar.edu>

subject: thanks again

to: Sarah Raper <sraperatXYZxyz-bremerhaven.de>

Sarah,

Thanks again for all that useful material. I will absorb

it into the text.

Although it will not make much difference, I guess

I should use an exponent for V in the final version

of the model. I will stick to n=1 for the analytic

results -- but n=1.2-1.4 seems necessary to be

state of the art. As far as MAGICC goes it makes

no difference since I solve the equation numerically.

My 1880 start year comes from TAR p. 667 which

refers to 'the climate of 1865-1895'. What is confusing

is the T1990 term is relative to 'the late 19th century'

(p. 682, which I take as 1865-1895, equal near enough

to the 1880 value in MAGICC) --- but the g1990

is relative to the glacier steady state. There is a 0.15degC

correction for the warming from this time to 1880, so the

steady state time must be well before 1880. In MAGICC

with DT2x=2.6degC the 1765-1880 warming is only

0.1degC.

All this is accounted for in MAGICC strictly following

the TAR, but there is still an inconsistency problem that

I noted before (and it's worse than I said before). The

AOGCM gu(1990) range is 1.5 to 2.7cm (I use the

average in MAGICC). The corresponding gs(1990)

value is pretty much the same. These must be changes

from the glacier steady state point, which must be well

before 1880 and probably before 1765. Table 11.10

(1910-1990 data) suggests a change larger than this,

2.4cm over 1910-1990

.

The question is, 2.4cm from 1910 equals what from the

glacier steady state point? Even though 1910 was a

cool spot in the global temperature curve, the use of

the 0.15 correction implies that there must be positive

GSIC melt to 1880, and hence almost certainly positive

melt to 1910. So the 'observed' gs(1990) must be more

than 2.4cm.

I guess there is nothing we can do about this. At worse

we will be about 1cm too low with gs(1990) if we stick

to the TAR AOGCM mean, and this won't affect the

longterm behaviour which must tend to V0. What is

needed is a model that runs from the steady state point

rather than 1990. Probably easy to do, but it would make

everything more messy and less directly comparable with

the TAR.

Thwarted by Jonathan's ad hockery!

Tom.

## No comments:

## Post a Comment