Why do subduction plates dive downwards, but in the results of the case run, do they move outwards?
It looks like the fault normal direction for the bottom of the slab has flipped from what it used to be. The fault normal is picked arbitrarily when the fault cohesive cells are created. In 3D, when the fault normal has a positive Z component, reverse slip is positive. As a result of the flipped fault normal, the sense of slip also flipped sign, so left-lateral slip is negative (instead of positive) and reverse slip is negative (instead of positive). You can check this by examining the normal_dir vector in the fault info files output.
In this current case (v4.1.3) with the current mesh posted on GitHub, the top of the slab has a fault normal with a positive Z component and the bottom of the slab has a fault normal with a negative Z component. This means the provided input files are correct for the top of the slab but the input files need to be corrected for the bottom of the slab (flipping the sign for the slip components).
I will create a GitHub issue so that we can develop a method for resolving this issue. We do not want to introduce inconsistencies for cases with fault surfaces with larger variations in strike and dip.
1 Like