Projection issue

andreasplach

Hello everybody,

I am trying to do simulations for Greenland. So far I always used the SeaRISE development data set (Greenland_5km_dev1.2.nc) and DomainOutline.exp also used in the ISSM Greenland example. But now I would like to use more up to date data namely surface velocity from Rignot and Mouginot 2012 and bed + ice thickness from Morlighem et al. 2014.

I think the Rignot and Mouginot 2012 and Morlighem et al. 2014 data sets use the same projection - Polar Stereographic North espg with origins 90N and 45W and secant 70N.

I thought the SeaRISE data and the DomainOutline.exp uses the same projection, but looking more closely I noticed that the reference lat/lon are different: Polar Stereographic North but with different origins of 90N and 39W and secant 71N.

I somehow need to project the two data sets together, I guess, but I can not figure out how...

It would be great if somebody could post a few code lines for reading the two newer data set into ISSM. I guess I am not the first one to use this data sets with ISSM. Sorry, I am not sure if I did a good job explaining, but I am totally confused at this point after trying for way too long to get this right... I tried a lot around including xy2ll()/ll2xy() and a python package to project data (pyproj), so far no success...

Any tips are highly appreciated!

Cheers

Andreas

PS:

I attach a plot of the imported surface velocities to illustrate my problem. I just can not make the projection of the domain and the velocities fit.

nicoleschlegel

Hi Andreas,

Please see step 7 of the Greenland example. Here, we translate a smb forcing from a 45 70 projection into the 39 71 projection of the model.

We save the smb to a mat file (x1,y1,smbmean) in step 6.

%convert mesh x,y into the Box projection

= xy2ll(md.mesh.x,md.mesh.y,+1,39,71);

= ll2xy(md.mesh.lat,md.mesh.long,+1,45,70);

%Interpolate and set surface mass balance

index = BamgTriangulate(x1(:),y1(:));

smb_mo = InterpFromMeshToMesh2d(index,x1(:),y1(:),smbmean(:),xi,yi);

The trick is to translate the model mesh coordinates into the coordinate system of your input data, and then pass the

transformed x and y into the interpolation routine. The resulting vector should be projected correctly for your model.

I hope this helps.

Thank you,

Nicole

andreasplach

Hei Nicole,

thank you so much! The data is projected correctly now, but did you every experience any memory issues when using BamgTriangluate(). The velocity data has a pretty high resolution which results in a matrix 17946 x 10018 and if I try to use the full matrix it seems like I am running out of memory. So far I just got it working when using only every 3rd value of the velocity array, which is probably not good for further use.

Cheers

Andreas

nicoleschlegel

Hi Andreas,

Your matrix is large, and will take a while for matlab to deal with. Unfortunately, matlab is a terrible memory hog, so this is a typical problem with input data that has such high resolution. Please note that you do not have to use InterpFromMeshToMesh2d, you can use any interpolation routine that is available in matlab, or for matlab. Other routines (e.g. griddata) may not require that you pass in an index value, and you may be able to choose more efficient routines/options for your purposes.

Thank you,

Nicole

andreasplach

Hei Nicole,

I am using the python interface for ISSM, and unfortunately python does not seem to be more efficient than matlab concerning memory :-). Thanks for the tip about using another interpolation which does not require an index. I will try that.

Thanks for your help!

Cheers

Andreas

wudipao88

Hi Andreas,

Do you still have the surface velocity data from Rignot and Mouginot 2012? If so, could you provide me with a copy? Thank you.