Code we could consider deleting

• rifts
• tetras
• flowband models
• ice model coupling
• ADOLC support
• remove external packages we are not using (and old versions of MPI/PETSc etc)
• LA theta XTaylor Hood Uzawa Pressure etc
• we should have a discussion about long term support for multiple interfaces
• tao
• KML (MATLAB can do most of it natively ​https://www.mathworks.com/help/map/exporting-vector-data-to-kml.html)

Things to improve/address

• [Mathieu] too many memory allocations
• [Surendra] get rid of Esa solver, adapt newly developed solid Earth capabilities instead
• [Mathieu] make model a dynamic class (do not set all fields by default). This would help integrate new solutions that may not be ice oriented into ISSM (e.g. GIA, Kriging, etc)
• [Mathieu] treat material like friction: create/call class on the fly instead of carrying it in a dataset
• [Mathieu] Fast SSA (with static array allocation P1/triangle etc), would require to delete coupling....
• [Surendra] Code Ivins GIA module in C++....
• [Nicole] Thermal solution in GEMB improvement in speed (possible use of solvers instead of explicit loop)
• [Nicole] Deal with heavy size of model forcing arrays / implement downscaling of GEMB input (this was also a request from AWI!)
• [Mathieu] MEX modules should not link to the entire ISSM library, it creates all sorts of issues with global variables, MPI/PETSc, etc. May need to compile smaller libraries (e.g. bamg, ContourToNodes) so that both issm.exe and the MEX modules can link to it.

New capabilities

• [Mathieu] Julia interface? may not be mature enough for what we want to do
• [Nicole] Coupling of GEMB to the thermal model
• [Mathieu / Nicole] Age tracking / advection of the snow layer