The initial running of recon-all autorecon-1 autorecon-2 for a sample scan resulted in a bad skull strip. I am attempting to troubleshoot the issue through running just the skull strip portion of the recon-all pipeline again, followed by adjusting the watershed parameters. 

During re-running -skullstrip, I am given a mri_watershed Error: read failed (the em_register portion of the -skullstrip flag ran without interruption). I have attached the corresponding log.

Is there a way to fix this watershed error (when I have tried to run the watershed command by itself, the command is not recognized)?