27 September 2001
Data files for Ruby
B06.22 2m Air Temperature in Kelvin Mar 1870 thru Nov 1999
B06.22 Total Precipitation in cm (per month) Mar 1870 thru Nov 1999
B06.22 Surface Zonal Wind in m/s Mar 1870 thru Nov 1999
B06.22 Surface Meridional Wind in m/s Mar 1870 thru Nov 1999
**********************************************************************
15 May 2001
Data files for Andrew Wood at UW
Ascii data on Andy's grid:
The pcm data is on the T42 grid so the data cells look like:
|
233.44 |
236.25 |
239.06 |
241.88 |
244.69 |
247.50 |
250.31 |
253.13 |
|
| 54.41620 |
Land |
Land |
Land |
Land |
Land |
Land |
Land |
Land |
52 |
| 51.62573 |
Land |
Land |
Land |
Land |
Land |
Land |
Land |
Land |
51 |
| 48.83524 |
Land |
Land |
Land |
Land |
Land |
Land |
Land |
Land |
50 |
| 46.04472 |
Sea |
Land |
Land |
Land |
Land |
Land |
Land |
Land |
49 |
| 43.25419 |
Sea |
Land |
Land |
Land |
Land |
Land |
Land |
Land |
48 |
| 40.46365 |
Sea |
Land |
Land |
Land |
Land |
Land |
Land |
Land |
47 |
| 37.67309 |
Sea |
Land |
Land |
Land |
Land |
Land |
Land |
Land |
46 |
| 34.88252 |
Sea |
Sea |
Land |
Land |
Land |
Land |
Land |
Land |
45 |
| 32.09194 |
Sea |
Sea |
Sea |
Land |
Land |
Land |
Land |
Land |
45 |
|
84 |
85 |
86 |
87 |
88 |
89 |
90 |
91 |
|
Four variables are available here for 2 simulations (B06.44 and B06.45):
B06.44 2m Air Temperature in Kelvin Feb 1995 thru Nov 2099
B06.45 2m Air Temperature in Kelvin Feb 1995 thru Dec 2048
B06.44 Total Precipitation in cm (per month) Feb 1995 thru Nov 2099
B06.45 Total Precipitation in cm (per month) Feb 1995 thru Dec 2048
B06.44 Surface Zonal Wind in m/s Feb 1995 thru Nov 2099
B06.45 Surface Zonal Wind in m/s Feb 1995 thru Dec 2048
B06.44 Surface Meridional Wind in m/s Feb 1995 thru Nov 2099
B06.45 Surface Meridional Wind in m/s Feb 1995 thru Dec 2048
Added data for 2 additional simulations (B06.46 and B06.47) on 18 July 2001:
B06.46 2m Air Temperature in Kelvin Feb 1995 thru Nov 2099
B06.47 2m Air Temperature in Kelvin Feb 1995 thru Nov 2099
B06.46 Total Precipitation in cm (per month) Feb 1995 thru Nov 2099
B06.47 Total Precipitation in cm (per month) Feb 1995 thru Nov 2099
B06.46 Surface Zonal Wind in m/s Feb 1995 thru Nov 2099
B06.47 Surface Zonal Wind in m/s Feb 1995 thru Nov 2099
B06.46 Surface Meridional Wind in m/s Feb 1995 thru Nov 2099
B06.47 Surface Meridional Wind in m/s Feb 1995 thru Nov 2099
Tom - 2 March:
I am not surprised that after 5 years B06.44 is not getting "closer" to
B06.40, but I anticipate that you will see that it does after 10-15 years.
Of course, "closer" is relative to measures of mean+variability.
Mike - 5 March:
Okay, I'll bite. As Dan notes, we haven't yet seen anything that it
clearly not "just another ensemble member". Mary just handed me
a stack of figures from Tom Bettge's website, and they also
look a lot like B0644 is "just another member" by and large.
But this same stack confuses me a bit. Why are the "historical
(B0628)" traces sooo different from any of the B0640's? Is this
a difference between the "free" historical runs and an assimilated
initial state? Is there any chance of seeing what the assimilated
values also look like?
I'm a bit concerned, looking at these, that comparisons between
the historical (downscaled) conditions and the b0640 series is
likely to be a joke. The good news is that this seems to explain
why I have such a large "tear" (towards warmer floodier responses)
between the historical runs and the climate-change runs, when
I ran b0622 (?) and b0640 through my sierran models and then tried
to compare the two. There are a clear and immediate (from first year
on) difference between the flood (most notably) statistics from the
two runs.
Ruby - 5 March:
I think we should use the first 10-20 years of the B0640 series as the control
rather than the historical runs (B0630 series). Even after the initialization
problem is fixed for the B0640 series, I suspect they could still be different
from the historical runs. The present conditions of each series should be used
separately as the control for each set. That's my understanding.
Mike - 5 March:
Yes, I rather gather that. It'll be a major major limitation
on my grand objectives. It means that we can't know what
the long-term (say, 50-yr recurrence interval extremes) might
be in the "historical" simulation, for comparison to the ensemble
of climate-change futures (from which such statistics
could be extracted). "We're" turning out these lovely multiple
long-term futures, with only a (hmmm, multiple) snip at the
beginning to compare to. Oh well! More thought is indicated
(on my part).
My question remains, however: WHY are they so
different? I obviously have missed something here.
Are the b0640 series runs indeed initiated from
data-assimilated versions, different from the
"free" historical coupled runs? If so, where is that
"assimilated version of history"?
Tim - 5 March:
I also noticed the difference between historical and scenario runs on the
PCM page. I bet Dave can shed some light on this. If the SST in the
assimilated runs were appreciably different from those of the historical
runs in 1995 then that might explain it. I seem to remember this is the
case, but Dave will know for sure.
Dave - 5 March:
Hm, I think I am missing the gestalt of this conversation. When I look at
the plots I see the following:
1. In volume average temperature, B06.28 is maybe 0.06C cooler than the
B06.4x runs. But we changed the T,S field in the B06.4x runs to reflect
Detlef's assimilated values. I would expect the volume averaged
temperature to be systematically different in the B06.4x cases because of
this. (If *something* wasn't different in the B06.4x cases with respect
to the historical cases, there would have been no point in doing
them!) For a yardstick of what a "big" difference is, note that the
difference between the various B06.4x cases spans about 0.05C just due to
natural variability. Against that backdrop, the systematic difference due
to the assimilated field seems to me to be modest at best.
2. The salinities are different in the B06.4x cases. This is because I
changed the salinity field to compensate for the temperature difference --
i.e., it was a density-compensated temperature change. This was done to
keep dynamical shocks out of the process, which I believe is important.
3. In surface quantaties, I see no difference between the B06.28 run and
the B06.44 (fixed land initialization) run. We know that the B06.40 run
(it is specifically mentioned in some of the attached e-mail) had the land
initialization problem, so I would advise against using it for anything
that is sensitive to land values (especially surface temperature).
4. There are systematic differences in the mean kinetic energy even in the
B06.44 run. I'm not sure why, but it seems plausible that holding the T,S
values close to the assimilation targets suppresses eddy activity. But
if so, there is really no way around this, as it is part of the
assimilation process. However, I don't see that this would be a problem
for any of the applications or downscaling I have heard about. Maybe if
you were looking at eddy fields in the Southern California Bight, but why
should the mean K.E. of the ocean affect Sierra precip?
Overall, then, it seems to me that the differences I see between the
B06.28 run and the B06.44 run are within the range of what I expected to
see, are confined to the oceanic interior, and do not yet have a
discernable influence on any surface quantity. If the assimilated ocean
temperatures never end up having any effect on a surface quantity, then
we will conclude that it doesn't matter whether you assimilate 3-D oceanic
initial conditions or not, but that remains to be seen.
Mike - 5 March:
Perhaps I should feel chastised, but not quite yet. Yeah, these
are small differences in terms of degrees C, but that is not
what they represent...surely as a volume average heat
content, these must be rather large differences in the
scheme of things. (I do like the change that you made to
the land initial conditions in b0644 as suggested by the
land surface temperatures plot). Yes, the instantaneous
influence of these heat (and ice) differences is probably
small (since they are deeply buried), but I return to my
original question (recast conveniently): Is the historical run
really comparable (plus/minus the radiative
forcing) to the b0640 series...in the long haul? These are the
statistics that caused me to speak up in the first place. I'll
take yes for an answer, but not if its only an "in the short
term, yes". Sorry for my denseness on this aspect, but
I want to know how far to bother with my end of things.
If there are (what seem to me to be) large heat content
differences in the b0640s initial conditions vs historical
run conditions, how do I know what is radiatively forced
difference and what is initial conditions?
