ASIS:Minutes 03 06 2008

Understanding contemporary change in the West Antarctic ice sheet Project meeting, 3rd June 2008, SPRI Notes

Present: 	Tony Payne, Anne Le Brocq, Andreas Vieli, Poul Christoffersen, Vicky Lee, Marion Bougamont. Copied to: 	Gwendolyn Leysinger-Vieli, Steve Price, Rupert Gladstone.

Agreed actions for next meeting

1.	Bristol.
a.	AJP to ask JLB whether his datasets for ice surface, bedrock and ice-shelf base will be available to the project. ** NOTE ** JLB has agreed this – the data should save our project a lot of effort.

b.	RG to set up a wiki for the project with access restricted to project members for the time being (plus Bill Lipscomb, Jeff Ridley and Jonathan Gregory).

c.	VL to continue work on thickness solver: to extend to 2d and to use velocity from higher-order stress solver. Initial work on zero order model. Need to liaise with SFP about Los Alamos thickness solver, as well as Bueler method. ''SP comments: Bueler method (fully explicit) is “working”, but for stream/shelf cases requires very small time steps. An improvement may come from splitting the solution into two parts, an implicit solution for the diffusive part (via subroutines already in GLAM) and an explicit solution for the advective part (simple upwinding). Currently, the diffusive part of the solution in fast flowing / flattish regions appears to be what is limiting the calculation w.r.t. a stable time step. I am planning to experiment w/ this in the next week or two. If there is another “working” dH/dt scheme that Vicki has come up with, I can help her to patch it into the current version of the code. Bill still has plans to put incremental remapping scheme in, but that probably won’t happen until end of summer / early fall. ''

d.	AJP to provide 2-d ice shelf model as separate, standalone code, including thickness solver, for use at SPRI/Durham and to test nested grid and automatic differentiation packages.

e.	AJP to ask SFP for information on numerical details of coupling between higher order model and basal model (i.e., specifying a yield stress) for the wiki. ''SP comments: I’ve rejigged the code a bit so that the “betasquared” sliding parameter is now called from a subroutine that contains numerous options for specification (specified via a namelist variable and a case construct). My plan is to patch Marion’s code in here as one of the allowable cases. I’m hoping to try and get that working (compiling at least) in the next few weeks as well. Note that there may be some issues w/ how well the iteration on Tau0 is working, but perhaps AJP and SFP can discuss that offline at some point. ''

f.	RG and VL to complete work on 1-d. higher order model with nested grid for presentation at meetings (e.g., WAIS) in September. AJP to provide 1-d. stress code; VL to provide thickness solver.

g.	AJP to convene a meeting on the future of GLIMMER: what should the base code be; changes to physics and numerics; coding standards etc? Initial thoughts attached below. Most likely this will have to be done remotely either as video/audio meetings or via a jointly edited document. ''SP comments: Note that Bill has taken parts from a slightly older version of Glam and altered them so that the HO solver in Glam can be called from either Glimmer OR Glam. We’ve confirmed that (1) these additional subroutines will compile in Glimmer and (2) the altered subroutines in Glam give identical results to that from previous versions of Glam. My preference would be to migrate more recent updates (e.g. debugging by SFP, RG, and Gethin, other improvements by SFP) into these altered subroutines so that further code development takes place along a branch that (at least initially) may be used in either Glimmer or Glam. This would allow us to take advantage of the better parts of Glimmer (coupler, temperature balance, basal hydrology, etc.) and Glam but still run Glam in it’s simple stand-alone mode as well. We assume that this branch would be kept separate from the publicly available parts of Glimmer for the time being. ''

2.	Durham.
a.	ALeB to work on combining data sets in Arc and adding boundary conditions (accumulation, air temperature, geothermal heat flux and geology), as well as available velocity information.

b.	ALeB to add GLIMMER tutorial to wiki.

c.	GLV to obtain accumulation data from BAS.

d.	AV/GLV to work on adding thickness to inversion code for application to ice shelves.

e.	Initial application for PIG; perhaps both basal drag (beta-sq) parameter and ice rheology?

f.	Longer-term work on parameterization of shear-margin weakness.

3.	Cambridge.
a.	PC to send datasets for geology and geothermal heat flux to ALeB.

Next meeting in Durham during week beginning 27th October 2008 (28th?)

Other notes from SFP:

''I’ve implemented a simple 2d sheet/stream/shelf coupled model with a grounding line that is free to migrate. Initial experiments indicate that the model will achieve a steady state. That is to say, nothing too crazy happens when the domain, consisting of a coupled stream and shelf, is allowed to evolve over time. Currently, I’m doing experiments to see if the g.l. behaves “sensibly” and/or what type of grid spacing is required to make that the case (by sensible I’m thinking something like neutral stability for simple bedrock topographies). Note that I’m using a Pattyn+ (JGR, 2006) type interpolation of the g.l. position and, if this works well, it is fairly straightforward to extend it from a flowline to plan view. ''

Appendix 1: Note on basal hydrology
Anne’s estimate of basal water flux: 		20×103 m2 yr-1

Width of grid cell: 				5 km

Total flux that must pass through cell	:	100×106 m3 yr-1

Change units:					100×106 m3 yr-1/ 3×107 s yr-1 = 3.33 m3 s-1

Measured flow velocity:			7.5×10-3 m s-1 (is this right Anne?)

Cross-sectional area need to discharge flow: 	 3.33 m3 s-1/7.5×10-3 m s-1 = 444 m2

Poul’s estimate of aspect ratio:		0.1

Area of channel:				0.1 width2

Total width of channel in 5-km cell:		66 m (1.32%)

Poul – what is your estimate of the distance over which the channel would have an impact on till water?

In any event, initial conclusion is that if water flows in channels (and the observed velocity is accurate) then very few channels are required to drain the water, which (to me) implies that there is little interaction between water in channel and porewater in till.