Thursday, May 10, 2012

4076.txt

date: Sun, 18 Dec 2005 13:53:11 -0700
from: Jonathan Overpeck <jtoatXYZxyzrizona.edu>
subject: more on TS feedback
to: Keith Briffa <k.briffaatXYZxyz.ac.uk>, Bette Otto-Bleisner <ottobliatXYZxyzr.ucar.edu>, Eystein Jansen <eystein.jansenatXYZxyz.uib.no>

<x-flowed>
Dear Keith, Bette and Eystein:

This email should be read after the one to the entire team - it
provides post LA3/TS feedback on figures. Since Bette is going on a
short vacation, she and I emailed about her new LIG fig before I
left, so she's ready to go when she gets home.

Keith (and Tim), on the other hand, have lots to consider, and I just
wanted to reiterate to you (and Bette) that it's a priority for me
and Eystein to help you brainstorm all these figures. Here are a few
more comments I got on Keith/Tim Figs:

For 6.8:

1) removing the oldest portion of the records from the plot is only ok IF:
-we can justify on an obvious and objective basis - for example that
sample depth hits goes down significantly at ca. 700AD or wherever we
want to chop it.
-We don't remove part of the series that will give rise to accusations of bias
Thus, it might be better to leave as was in the FOD, just to be safe,
or to try multiple versions.

2) had a long talk with Martin Manning about the idea of multiple
plots, vs just the existing one (by the way, the TS team WANTS the
instrumental part of the fig as we agreed to modify in Chap 6
sessions). I think the best idea is to keep the bottom panel as is,
with modifications
- keep the error bars as is
- try a version with some sort of annually-resolved volc forcing
placed at the top of the panel, with eruption (sufate) lines sticking
down farther for big eruptions
- try inserting some representation of average (median? or?) sample
depth along the bottom (time) side of the panel. This will thus show,
lots of sample depth back to ca. 1700, then less and less (in
steps?). Martin suggests we go one step farther and color the sample
depth part of the plot with different colors, based on our expert
judgement of confidence. We could have two or three colors - one
color for the interval overwhich we have "very likely" confidence
(e.g., in the exec summary) and another for just "very." perhaps we
want a third for some term reflecting "don't trust inferences
regarding hemispheric temp that much over this interval" - this will
obviously take some thinking/creativity, but this fig will go all the
way to the TSM, so it's worth the effort.

3) linear axis for sure

4) if would still be good to try a density shaded version of this
plot (instead of all the recon lines) for the TS and SPM. When in
doubt, make an extra version. We can then share with our team and
with Susan.

Thanks for doing this!

Also, FYI, Gabe indicated that her regional plots were not scaled
separately. Surprising, but maybe the models are actually better than
we thought.

Best, Peck

--
Jonathan T. Overpeck
Director, Institute for the Study of Planet Earth
Professor, Department of Geosciences
Professor, Department of Atmospheric Sciences

Mail and Fedex Address:

Institute for the Study of Planet Earth
715 N. Park Ave. 2nd Floor
University of Arizona
Tucson, AZ 85721
direct tel: +1 520 622-9065
fax: +1 520 792-8795
http://www.geo.arizona.edu/
http://www.ispe.arizona.edu/
</x-flowed>

No comments:

Post a Comment