[Liwg-core] Rain-snow repartitioning

Lipscomb, William Henry lipscomb at lanl.gov
Fri Jan 13 08:08:41 MST 2017


Hi all,


I'd like to suggest that Leo launch these two runs (if he hasn't already started them):

* GLC ramp -2 to 2

* GLC ramp -2 to 0


These runs would only affect glacier columns, and would use the actual (dry-bulb) temperature.  At this point I'm hesitant to change the ramp globally, because (1) there isn't a strong physical justification and (2) there could be undesirable changes for land gridcells outside Greenland and Antarctica.


As for the wet-bulb temperature, I'd be willing to code up a calculation today (it's officially my day off, so I have some time) if people feel strongly about it, and then Leo could try some additional tests over the weekend.  But I'm also willing to leave things as is, since the wet-bulb temperature is close to the dry-bulb temperature in the conditions we care about, and since the non-glacier calculation (which we would not change) still uses the dry-bulb temperature.


I agree with Bill S. that we should make a decision as soon as we can, erring on the side of simple changes that get the job done.


Cheers,


Bill L.


________________________________
From: Bill Sacks <sacks at ucar.edu>
Sent: Friday, January 13, 2017 7:47 AM
To: Lenaerts, J.T.M. (Jan)
Cc: Lipscomb, William Henry; Sarah Bradley - CITG; liwg-core at cgd.ucar.edu
Subject: Re: [Liwg-core] Rain-snow repartitioning

Hi all,

Jean-Francois wants a science freeze by the end of January. If we want this included, I think we really need to know exactly what we want, and have it implemented and tested, by early next week – because there needs to be time after that for it to get incorporated and tested with all of the other final changes that are underway. With that in mind, I liked Leo's suggestion of just getting some test runs going with some simple changes that could be implemented immediately. If folks want a more substantial change, such as using the wet bulb temperature, I think this really needs to be coded up ASAP – like, today – so that test runs can be started over the weekend.

I hate to be pushy like this, but there is a growing sense of urgency for getting the science freeze done, which may not be apparent to those who aren't at NCAR.

Bill S

On Jan 13, 2017, at 6:34 AM, Lenaerts, J.T.M. (Jan) <j.lenaerts at uu.nl<mailto:j.lenaerts at uu.nl>> wrote:

Hi Bill,

Over melting glacier surfaces, there are frequent ‘summer inversions’ observed, where the surface is limited to zero (as it is melting) but the air temperatures are clearly positive. But the same holds for any melting surface, so this would be applicable to any place where the ground is covered with snow.

Is RH available in atm2lnd routine, where the repartitioning is performed? If so, I don’t see any reason why we wouldn’t convert to wet bulb temperature. As you mention, precipitation is usually associated with a (near-)saturated near-surface atmosphere, so the actual and wet bulb temperatures will be mostly similar. That said, I would be in favour of shifting the range, and using wet bulb temperatures.

All the best,

Jan

On 12 Jan 2017, at 17:27, Lipscomb, William Henry <lipscomb at lanl.gov<mailto:lipscomb at lanl.gov>> wrote:

Hi all,

I’d like to go back to the discussion we started yesterday about how we might justify a different ramp for glacier land units compared to other land units.

I read the Sims & Liu paper last night and found it interesting.  This is what I took away:

* The main determinants of the rain/snow temperature threshold are (1) near-surface wet-bulb temperature (not actual temperature) and (2) low-level lapse rate.  See their Fig. 8c.  As the low-level lapse rate decreases (as we’d see with a temperature inversion), the threshold temperature decreases, and the range of the transition increases.  So for example, if we look at the right side of the plot (large, positive low-level lapse rate), the transition from ~20% snow probability to 80% probability takes place over a wet-bulb temperature range of about 1.5 C to 0 C.  But at the left side of the plot (negative low-level lapse rate, this transition occurs over a range of about 0.5 C to -2.5 C.

I don’t have a good sense of how much the wet-bulb temperature in our simulations differs from the actual (dry-bulb) temperature.  Has anyone ever looked at this?  If I pull out my handy psychrometric chart, I see that at 90% relative humidity, T_wet = 0 C corresponds to T_dry ~1 C.  So in cold, moist conditions the values aren’t too different, but a ramp based on dry-bulb temperatures would use values that are a bit higher than those in Fig. 8c.

Given our deadlines, I don’t think we’re in a position to diagnose lapse rates in the lower troposphere and pass that information to CLM.  (Maybe for CESM3!)  But if we’re confident that low-level lapse rates are generally lower over a melting glacier surface than over non-glacier surfaces, I think we could defend a different (lower and broader) ramp for glacier land units.

Also, a comment on lapse rates.  For glacier land units, we use a globally uniform lapse rate of 6 deg/km.  This is used to convert surface temperature at the mean grid cell elevation to surface temperature at different elevations in different parts of the grid cell.  This value is not the same as the low-level lapse rate.  It might be that there’s a temperature inversion at all grid-cell locations, but it would still be appropriate to assume higher temperature at lower elevations.

Coming back to Leo’s simulations:
* I doubt we can justify a ramp from -5 to 0.  But a range of (-2,x) where x is between 0 and 2 seems to be defensible.
* It may be easier to justify a change in the GLC-only ramp than a change in the global ramp.
* There are empirical formulas for computing wet-bulb temperature from T and RH.  If we coded one of these formulas, we could use web-bulb temperature for our ramp.

I’m OK with spending the hours on BG runs, since these experiments are high priority for the release.  But if we have a consensus on which physical settings would be easiest to justify, we could start with those.

Cheers,

Bill L.

On Jan 12, 2017, at 5:02 AM, Kampenhout, L. van (Leo) <l.vankampenhout at uu.nl<mailto:l.vankampenhout at uu.nl>> wrote:

Hi all,

Coming back to my previous email, I realised that these are no cheap runs so I would like to ask your input on whether these runs are sensible enough to justify the cost.
Currently, there are about 400k core hours on our Yellowstone budget remaining (project P93300601, LIWG production ), and about 4M on Cheyenne.
The estimated cost of 5 x 10-year BG runs is ~110k core hours, i.e. around 25% of the remaining budget.
Doing BG instead of FG runs remove the need to do a baseline run because we can use 126 for that. Also they are simpler to setup.
Please let me know what you think!

Cheers,
Leo



On 12 Jan 2017, at 11:06, Kampenhout, L. van (Leo) <L.vanKampenhout at uu.nl<mailto:L.vanKampenhout at uu.nl>> wrote:

Hi all,

I will do the following FG runs, 10 years each.


  *   global default ramp
  *   global ramp A -2 to 2
  *   global ramp B -5 to 0
  *   GLC ramp A -2 to 2 (only affects glacier columns)
  *   GLC ramp B -5 to 0
  *   global default ramp, with modified restart file where Hmax has been capped at 1.0 m  (the run itself again allows Hmax = 10)

This last run builds on the idea that melt in Northern Greenland is hindered by an unrealistically thick snowpack , resulting from the offline spinup with GSWP3 forcing.
All runs continue off 126 @ year 0043

I will let you know when they are done.

Leo


On 12 Jan 2017, at 08:52, Lenaerts, J.T.M. (Jan) <j.lenaerts at uu.nl<mailto:j.lenaerts at uu.nl>> wrote:

Hi all,

I think we’ll have to appreciate that we lose some physical justification, and this is a way to reduce the melt bias. I agree that we can partly justify shifting the ramp by the poorly constrained lapse rate, and frequent occurrence of inversions over the ice sheets.

I asked Leo to set up some FG-runs (ramp -5 to 0, -2 to 2, and 0 to 2), and if he can do that, we’ll analyse the output together. We will then ask Dave if he can look at the impact on seasonal snow. My remaining concern is that we won’t see the impact of higher runoff on the ocean state in these runs, but I think these changes are pretty small compared to the overall Arctic basin freshwater input.

All the best,

Jan











On 12 Jan 2017, at 02:44, Lipscomb, William Henry <lipscomb at lanl.gov<mailto:lipscomb at lanl.gov>> wrote:


My sense was that 0 to -5 was more of a fishing expedition, with numbers first suggested by me.  If it’s feasible, I think it would be great to run some short simulations with a warmer lower bound and/or different range.

Bill L.


On Jan 11, 2017, at 6:31 PM, Marcus Löfverström <marcusl at ucar.edu<mailto:marcusl at ucar.edu>> wrote:

Hi Bill,

I am intrigued by this idea. It will of course require some carefully crafted sentences when formulating this in the model documentation but I think it could be a way to defend having a different partitioning over glaciated surfaces.

Question primarily for Jan and Leo: what was the rational for choosing 0C and -5C as end values in the ramp you guys tested the other day? Perhaps you mentioned it in an earlier email but I seem to have missed out on most of the discussion about these experiments -- how they were set up and what results you got. Did you have a reason for selecting precisely these values or was it more of "a fishing expedition" (as I reviewer once wrote when I didn't motivate a parameter choice in a paper). Is it possible to run a few shorter simulations (perhaps 10 yr F cases) where you test the impact of different choices along the line of Bill S's email above? Sorry for nagging if you have already done that.

Best,
Marcus

On Wed, Jan 11, 2017 at 5:35 PM, Lipscomb, William Henry <lipscomb at lanl.gov<mailto:lipscomb at lanl.gov>> wrote:
Hi Bill,

Thanks for the papers.  I’ll look at them this evening.

One other thought to add to the mix:  For glacier landunits we’re doing a fairly crude temperature adjustment (globally uniform lapse rate).  Using a somewhat unphysical ramp—or a ramp different from that used in other landunits—could perhaps be rationalized as a fix for a crude lapse rate.

But I feel like I’m handwaving.  I’d be happy to have some backup from the European folks tomorrow :-)

Bill L.


On Jan 11, 2017, at 3:31 PM, Bill Sacks <sacks at ucar.edu<mailto:sacks at ucar.edu>> wrote:

Thanks for your thoughts Bill and Jan.

I did a quick literature search, which turned up the attached papers. I have given these only the briefest of skims, mostly looking at the figures, but they suggest to me that – for the land surface as a whole – the current ramp from 0 - 2 C is more physically justifiable than a -2 to 2 C ramp. In fact, if anything, the ramp should be shifted more in the positive direction – say from 0 - 3 C. Someone else should look at these more closely to make sure I'm interpreting them correctly. There's still the possibility that the relationship could be different over the ice sheet, I suppose, as well as the possibility that we can acknowledge that we're making this change in order to get good answers despite lack of physical justification....

Bill S



<Dai-2008-Geophysical_Research_Letters.pdf>
<Sims and Liu 2015.pdf>
<Stewart et al 2015.pdf>

On Jan 11, 2017, at 3:09 PM, Lipscomb, William Henry <lipscomb at lanl.gov<mailto:lipscomb at lanl.gov>> wrote:

Hi Bill S.,

You raise some good questions.  I think the answer to (1) is yes.  We would be exchanging an excess of snow for an excess of rain.  My reasoning would be that it may be better to get the right answer (good SMB) for the wrong reason (excessive rain) than the wrong answer (poor SMB) for a different wrong reason (excessive snow).  Either way, we have too much precip, and we’re tuning to limit the damage.

A subtlety, though, is whether the recently improved precip is causally related to colder temperatures.  If so, then the knobs we’re using to reduce precip are not very effective knobs for reducing SMB.  Do others have thoughts on this?

As for (2), I’d expect that in warmer climates we’ll see more rain.  With modified repartitioning, we'd be starting from a different preindustrial baseline, with the usual caveats about possible differences in climate sensitivity when you have a different baseline.

I haven’t looked at Antarctica.  Maybe Jan and Leo have?

Thanks,

Bill L.





On Jan 11, 2017, at 2:39 PM, Bill Sacks <sacks at ucar.edu<mailto:sacks at ucar.edu>> wrote:

Hi all (just including LIWG folks here),

I was just giving Bette a recap of our discussion on rain-snow repartitioning, and this raised a couple of questions for us:

(1) For Greenland, is it correct that a change to the rain-snow partitioning would lead to more rain, which is a degradation relative to observations – but that Jan and Bill L feel we can (and should) live with this degradation in order to boost the melt?

(2) Particularly if (1) is true: Since it feels like this achieves the right answer (net SMB) for the wrong reasons (too much rain): Has any thought been given to whether this makes sense in a climate change (future or paleo) world?

(3) Has anyone looked at what this change to rain-snow partitioning does over Antarctica? We want to be sure that we're not substantially degrading the Antarctica SMB from this change....

Thanks,
Bill S
_______________________________________________
Liwg-core mailing list
Liwg-core at cgd.ucar.edu<mailto:Liwg-core at cgd.ucar.edu>
http://mailman.cgd.ucar.edu/mailman/listinfo/liwg-core

---
William Lipscomb
Los Alamos National Laboratory
Group T-3, MS B216
Los Alamos, NM 87545
505-667-0395<tel:(505)%20667-0395>






---
William Lipscomb
Los Alamos National Laboratory
Group T-3, MS B216
Los Alamos, NM 87545
505-667-0395<tel:(505)%20667-0395>





_______________________________________________
Liwg-core mailing list
Liwg-core at cgd.ucar.edu<mailto:Liwg-core at cgd.ucar.edu>
http://mailman.cgd.ucar.edu/mailman/listinfo/liwg-core




--
Marcus Löfverström (PhD)
Post-doctoral researcher
National Center for Atmospheric Research
1850 Table Mesa Dr.
80305 Boulder, CO, USA

https://sites.google.com/site/lofverstrom/

---
William Lipscomb
Los Alamos National Laboratory
Group T-3, MS B216
Los Alamos, NM 87545
505-667-0395




_______________________________________________
Liwg-core mailing list
Liwg-core at cgd.ucar.edu<mailto:Liwg-core at cgd.ucar.edu>
http://mailman.cgd.ucar.edu/mailman/listinfo/liwg-core

- - - - - - - - - - - - - - - - - - - - - - - - - - -
Jan Lenaerts
IMAU, Utrecht University
@lenaertsjan<https://twitter.com/lenaertsjan> || IMAU website<http://www.staff.science.uu.nl/~lenae101/> || CU<http://www.colorado.edu/lab/icesheetclimate/> website<http://www.colorado.edu/lab/icesheetclimate/>





_______________________________________________
Liwg-core mailing list
Liwg-core at cgd.ucar.edu<mailto:Liwg-core at cgd.ucar.edu>
http://mailman.cgd.ucar.edu/mailman/listinfo/liwg-core

_______________________________________________
Liwg-core mailing list
Liwg-core at cgd.ucar.edu<mailto:Liwg-core at cgd.ucar.edu>
http://mailman.cgd.ucar.edu/mailman/listinfo/liwg-core

_______________________________________________
Liwg-core mailing list
Liwg-core at cgd.ucar.edu<mailto:Liwg-core at cgd.ucar.edu>
http://mailman.cgd.ucar.edu/mailman/listinfo/liwg-core

---
William Lipscomb
Los Alamos National Laboratory
Group T-3, MS B216
Los Alamos, NM 87545
505-667-0395




_______________________________________________
Liwg-core mailing list
Liwg-core at cgd.ucar.edu<mailto:Liwg-core at cgd.ucar.edu>
http://mailman.cgd.ucar.edu/mailman/listinfo/liwg-core

- - - - - - - - - - - - - - - - - - - - - - - - - - -
Jan Lenaerts
IMAU, Utrecht University
@lenaertsjan<https://twitter.com/lenaertsjan> || IMAU website<http://www.staff.science.uu.nl/~lenae101/> || CU<http://www.colorado.edu/lab/icesheetclimate/> website<http://www.colorado.edu/lab/icesheetclimate/>





_______________________________________________
Liwg-core mailing list
Liwg-core at cgd.ucar.edu<mailto:Liwg-core at cgd.ucar.edu>
http://mailman.cgd.ucar.edu/mailman/listinfo/liwg-core

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cgd.ucar.edu/pipermail/liwg-core/attachments/20170113/b0995842/attachment-0001.html>


More information about the Liwg-core mailing list