ASIS:Minutes 26 03 2009

Understanding contemporary change in the West Antarctic ice sheet
Project meeting, 26th March 2009, Durham Notes

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

Summary
The project has had a slight hiatus since the last meeting (Peak District, September 2008). This is because Marion has been on maternity leave and Rupert has moved projects to work on coupling with the Hadley Centre models under NCEO/JCRP. We therefore mostly discussed the work that Anne has been doing. Steve continues to offer great support on the higher-order modelling from Los Alamos. Gwendolyn is now funded by a Dorothy Hodgkin Fellowship and continues to do work that is highly relevant to the project.

Funding
With Marion’s return from leave, the postdoc funding is as follows:

•	Bristol: The position will be advertised shortly and we hope to fill it with someone who has C++ skills and can work with the chombo adaptive-mesh software. We hope to have someone start in the summer for a period of 2 ½ years.

•	Cambridge: Marion is funded until the end of 2010.

•	Durham: Anne is funded until February 2010.

Datasets
Anne’s work on collating data sets. We looked at the data that had been gathered and regridded; the overall package looks very impressive, however several issues remain.

•	It would be very helpful to have observed thinning rates (just trends from last 10-20 years). Tony has since spoken to David Vaughan about this and it seems that Hamish Pritchard’s data set would be ideal and available to us – Anne please contact Hamish.

•	It would be helpful to have the data available as netcdf files.

•	Some work is still required on the bathymetry under ice shelves (in particular in the Amundsen Sea). No data exist so whatever is done will be artificial but should, nonetheless, be sensible. Interpolation using ice-free ocean and grounded ice points was discussed; as was forcing the presence of troughs by inserting artificial data points under the shelves and interpolating. Anne to look at papers that discuss possible bathymetry (eg Andersen etc) and suggest a scheme.

•	Nunataks need to appear in the mask file.

•	The velocity validation set is very useful but has only partial coverage (although all major areas of interest are covered). It would be helpful to have a second dataset that fills in the blanks (ice divide mostly). We discussed ways of doing this: combining with balance velocity or interpolating. Anne should speak to Laura Edwards about this; such a map was an output of Laura’s PhD work.

•	The geology data needs to be completed. We agreed the following scheme: calculate driving stress using geometry data and Kamb-Echelmeyer smoothing; extend velocity data (either balances or by interpolation, see above); obtain rebounded bedrock (nothing very sophisticated required); based on these data apply 3-4 classes (low stress, high velocity and marine: soft; low velocity and terrestrial: hard; etc etc). This will need some tuning to get a sensible map.

•	Andreas to ask Ian Joughin about his calculated basal traction values (or beta-sq) for Siple, Amundsen and Ronne sectors.

We agreed that there would be a good paper on the creation of the database and the various issues resolved during this process.

Automatic differentiation
Gwendolyn briefly discussed her work on the use of automatic differentiation software, in particular the TAPANADE package. This shows great potential and, in a 1-d. case study, produces results that agreed with those found by Andreas when doing the work ‘by hand’. The next step will be to differentiate with respect to other variables, such as thickness etc. The technique could also be applied to the 2-d. model (Tony to supply simplified code).

Development paths
We discussed the complexities of developing GLIMMER within the framework of the Hadley Centre models, as well as issues on the future linkage between GLIMMER and CISM. We decided that Anne and Marion should use GLAM as their main model for the time being and, in particular, for work on WAIS. This should avoid delay to the project.

In the meantime, Bristol and Los Alamos will continue work on adaptive mesh refinement and making GLIMMER more modular and incorporating the higher-order solvers.

Project plan for 2009
We then discussed work for the rest of 2009 and agreed the following outline.

Next meeting in Bristol during June?