[Liwg-core] large impact on SMB of reducing Hmax from 10 m w.e to 1 m w.e.
M.Vizcaino at tudelft.nl
Tue Jan 24 14:09:23 MST 2017
thanks for your email.
(Raymond, please correct me if I am wrong) He started from cold start conditions. He span-up the snowpack by running some cycles with 1979-1989 ERA-Interim until snow volume equilibrated,
but I am not sure if he checked for equilibrated temperatures.
Let's discuss offline between us three on the different results of our Hmax-sensitivity runs, and we'll come back to the group with a consensus.
On Jan 24, 2017, at 7:19 PM, Kampenhout, L. van (Leo) wrote:
are the 1m runs started from cold start conditions? Beware that this probably introduces a bias because of the initial soil temperature of 250K, which is too cold for the ablation zone. It takes some time to equilibrate deep soil temperatures.
In my most recent ERA Interim tests I find a GIS melt difference of just 13% going from a 1m to 10m snowpack.
On 24 Jan 2017, at 17:51, Bill Sacks <sacks at ucar.edu<mailto:sacks at ucar.edu>> wrote:
Hi Miren and others... also including liwg-core,
I will not try to weigh in on the ideal settings here – others on this list are better qualified in this respect.
However, for now I'd just like to say that I'd prefer if we keep the same parameters applied for all glaciers, both because (a) what you're suggesting will require some extra code development, which is challenging to argue for at this point given the deadlines we are up against, and (b) unless there are physical justifications or other strong reasons, we generally try to avoid having different physics in different parts of the world.
On Jan 24, 2017, at 8:59 AM, Miren Vizcaino <M.Vizcaino at tudelft.nl<mailto:M.Vizcaino at tudelft.nl>> wrote:
please see results from Raymond with ERA-Interim forcing for the downscaled SMB (acab) and the CLM-SMB calculated from components
the last two rows in the table correspond to using the RACMO mask
The distribution of ablation areas are better modeled with Hmax = 1 m. w.e. than Hmax=10 m w.e., the integrated melt and SMB improve, and so does the refreezing fraction (50% vs. 70%; RACMO2.1.’s is 40%)
The results strongly suggest that Hmax controls the simulation of ablation areas. However, I am not sure if part of the problem is the initial condition (here #119 for the Hmax=10 m w.e. simulation) or if the model would drift anyway
to the “snow thickness clustering” situation. Checking this requires an extra long run
The Hmax-problem requires further investigation, but my quick thinking by now is: CESM, as opposed to RACMO, operates at 100 x 100 km scale, and cannot get the desired sub-scale sensitivity to e.g. inter-annual atmospheric
variations, if same elevation classes are permitted to drift to high snow thickness. We also have an issue with the albedo gap between 0.68 (snow) and 0.4 (ice), and I am not sure if the fractional snow cover fixes this in the way
My proposal is to run Greenland with Hmax= 1 m w.e. by now - and keep Antarctica with Hmax = 10 m w.e. I guess that the big downside is that in the accumulation area of the GrIS the new snow scheme will not give sensible results in
terms of snow density profiles, heat conduction, etc.,… but maybe Leo/Jan can have separate settings for runs with e.g. an static present-day GrIS?
Please let me know your thoughts here, thanks!
Begin forwarded message:
From: Raymond Sellevold - CITG <R.Sellevold-1 at tudelft.nl<mailto:R.Sellevold-1 at tudelft.nl>>
Subject: Slide from today
Date: January 24, 2017 at 1:30:49 PM GMT+1
To: Miren Vizcaino <M.Vizcaino at tudelft.nl<mailto:M.Vizcaino at tudelft.nl>>
Here is the slide from today that you wanted. Where it’s RACMO in the end reflects that I used the RACMO mask from Leo.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Liwg-core