Hi @mathieumorlighem
Alrighty, I have a follow up question. The domain extends out into a fjord where there is no ice (but there will be advance in that area during a transient run). I've set everything as grounded because I assumed it wouldn't matter how a no-ice area is categorized, but this means there's no defined grounding line in md.mask.ocean_levelset. I can change that by setting a grounding line, would it help with the line search failure?
I can also run the inversion on a test domain where there is ice at the moment and set that to the final mesh later on. Would this be a good work around?
Honestly, I've been struggling with this inversion for a few months now and really appreciate the support. 🙂