Errors in inverting ice rigidity

wudipao88

Now I am trying to invert the rigidity of grounded ice. The ice flow equation is approximated by SIA, the cost function and its coefficients are already set, but the stress balance solution is always blocked by the following error:

[2] ??? Error using ==> ./modules/ModelProcessorx/Control/UpdateElementsAndMaterialsControl.cpp:153
[2] UpdateElementsAndMaterialsControl error message: Control size not supported


[1] ??? Error using ==> ./modules/ModelProcessorx/Control/UpdateElementsAndMaterialsControl.cpp:153
[1] UpdateElementsAndMaterialsControl error message: Control size not supported


[3] ??? Error using ==> ./modules/ModelProcessorx/Control/UpdateElementsAndMaterialsControl.cpp:153
[3] UpdateElementsAndMaterialsControl error message: Control size not supported


[0] ??? Error using ==> ./modules/ModelProcessorx/Control/UpdateElementsAndMaterialsControl.cpp:153
[0] UpdateElementsAndMaterialsControl error message: Control size not supported

 
job aborted: 
[ranks] message 
 
[0-3] process exited without calling finalize 
 
---- error analysis ----- 
 
[0-3] on DESKTOP-GCF3UT7 
E:\BaiduSyncdisk\ISSM\ISSM-binaries\bin/issm.exe ended prematurely and may have crashed. exit code 1 
 
---- error analysis ----- 
loading results from cluster

I don't know where the problem is, thank you!

mathieumorlighem

In short, inversions are not implemented with SIA because you can determine the rheology analytically, although the error message you are getting indicates that the size of md.materials.rheology_B is wrong (it is neither equal to numberofvertices, nor numberofelements). In any case, the surface speed (with SIA) is simply

v = -2*(rho*g)^n*surf.^(n-1).*slope.*A/(n+1).*H.^(n+1)

where

A = md.materials.rheology_B.^-3;

(see contrib/morlighem/sia.m). If you know the surface speed, you can find B easily.

I hope this makes sense
Cheers
Mathieu

wudipao88

mathieumorlighem Hi mathieu,
Thanks for the help! According to the analytical equation you gave, the md.materials.rheology_B value I calculated based on the surface velocity is on the magnitude of 10¹², which seems to be wrong.

If md.materials.rheology_B is not inverted by surface velocity, but a steady-state temperature field is calculated first, and then the md.materials.rheology_B value is calculated using Paterson() based on the temperature field, is this reasonable?

mathieumorlighem

you could try that but you are putting a lot of faith in the paterson (or cuffey) table and SIA, which is extremely sensitive to surface slopes (need to be smoothed over several ice thicknesses).
Either you made a mistake in your calculation or the slope is too high locally, so SIA needs very stiff ice. I recommend smoothing the surface slopes first.