cc: "Pittock,Barrie" <barrie.pittockatXYZxyz.csiro.au>, email@example.com, firstname.lastname@example.org, "Pittock,Barrie" <barrie.pittockatXYZxyz.csiro.au>, email@example.com, firstname.lastname@example.org, email@example.com, firstname.lastname@example.org, "Whetton, Peter" <peter.whettonatXYZxyz.csiro.au>, email@example.com, firstname.lastname@example.org, shsatXYZxyznford.edu
date: Mon, 23 Oct 2000 22:46:31 +1100
from: "Jones, Roger" <roger.jonesatXYZxyz.csiro.au>
subject: RE: Table 3-10: a third version and some other considerations
to: 'Timothy Carter' <tim.carteratXYZxyz.fi>
Well done to get all this down in the rush!
And all -
The 3rd version sits best with my point of view and I agree with the star
changes suggested so far. WGII really has to stress the importance of
climate extremes in impact, V and A analysis and cannot be seen to endorse
by default a structure that cannot pass on confidences in phenomena (or
create scenarios!) until they have been represented in GCMs. One can still
take information from Table 3-10, regional means for P and E from climate
models, and artificial or historical variability and create valid scenarios
in a better form than a GCM can represent directly.
If confidences can be used to prevent what Steve S. terms as type I analyses
by avoiding 0/100% binary predictions - it is/is not happening - then both
type I and type II errors are more likely to be avoided. Where this approach
has not quite worked is in the representation of confidence in the science
and probability - the two are a little mixed in the presentation and are
likely to remain so for the moment. For instance, in the simple and complex
recurrent extremes, there are three choices: increase, decrease and no
change. Here, columns 2 and 3 are quantifying the confidence in one choice.
By default this almost ends up as a probability, i.e. I read maximum temps
as >95% likely to occur. Peak TC wind intensities may increase, they may not
(this still makes it more likely than either no change or a decrease if they
have equal probability).
This becomes even a little more strained with the singular phenomena. Is a
complete shutdown of the THC >95% not likely to happen this century? Or it
could happen but we have very little confidence in the knowledge surrounding
that possibility (ditto for the collapse of the WAIS). This is the reason
why I think this scale is best used with the supplemental qualitative
uncertainty terms in Moss and Schneider. However, further work has to be
done on how to best represent confidences in terms of probabilities,
particular where several competing alternatives may be present.
Specific comments (quibbles really)
Intense precipitation may not occur in some areas where significant rainfall
In the southern hemisphere projected mid/high latitude winter wet spells are
>40�S (e.g. Tasmania, southern S. America). 20-40�S may experience reduced
winter/spring rains and mid latitude summer drought.
Are the confidences given extended to regionality? Where you have used note
e this works well.
Hope these comments are useful.
Dr. Roger Jones
Climate Risk and Integrated Assessment Project
Climate Impact Group
CSIRO Atmospheric Research
Private Bag No.1, Aspendale
Victoria 3195 Australia
Phone +61 3 9239 4555
Fax +61 3 9239 4688
Email roger.jonesatXYZxyz.csiro.au <mailto:email@example.com>
Climate Change Impacts Network ccin_manatXYZxyz.csiro.au