it seem unlikely to be relevant for the ACCESS-N48. I do not interpolate vegetation types and I also do not have any restart files, as I start from a cold start.
One thing Iād like to know about the cold start in rose-cylc:
When does the interpolation of cm2.astart happen? Does the UM model do it internally? Or does cylc do an interpolation to a temporary restart that gets fed into the UM?
If cylc creates a restart file for the N48 grid, where do we find it?
Interpolation happens in the reconfiguration stage. The output should be getting put under share/data somewhere I think
1 Like
Thanks Scott. In that case, I think we need to look at the interpolated restart file in the problem case vs the original (N48) case. My guess is that something is going wrong in the interpolated restart file @atteggiani @ars599 @Dietmar_Dommenget
1 Like
my guess is in the restart dump file the land-related fields have not changed when you introduce a new āislandā. Such as the leave area index, or the land tiles.
Lynn from BoM has touched on that and itās on Githut.
If possible I am happy to have another online meeting with Abhik.
David, if there is any chance could you please document how you get through the runs and debud)? Your documentation is weel-arm
Regards,
Arnold
I will make a wiki on the forum of how I configured ACCESS-ESM1.5 with land-sea mask changes.
If we figure out whatās going on in the N48 case, weāll certainly report back here!
2 Likes
This seems not likely to me. Note, I can remove land or added land and it does run for 5yrs without crash. But in some cases it does not and crash right at start. I do have to edit the UM land fractions to make it work. To me this suggests something is wrong in how the interpolation works.
1 Like
If I could have the job id, I would give a check. I spend more time on rose/cylc system. I will post the potential issue from the error message and see if I can fix or if @Scott , @dkhutch @atteggiani could give some advices later after I post the error message.
2 Likes
I think there is an issue with the remapping files used in the coupling (the rmp*.nc files). They need to be updated with the new masks. We can try this tomorrow.
I canāt see any problems with the atmosphere restart files, BTW.
Surely we can write some test routines to check for compatibility amongst these various fields/remappings etc?
Would be great to streamline this process! Currently just aiming to make a working prototype.
1 Like
For sure. I wasnāt implying you should do this, rather that such tests should be available to make this process much much much easier.
1 Like
It looks like we have made some progress on the N48 case. Dietmar ran with new coupler remapping (rmp*nc) files and he is no longer getting crazy temperatures in the Drake Passage closed case. @ars599 @Dietmar_Dommenget @Scott @atteggiani
3 Likes
Can I suggest updating this web or copying it somewhere under NRI with these two latest new sets from David and Dietmar?
Hi Arnold,
I am definitely keen to make available my scripts and some instructions on what Iāve done so far. I will have to defer that to next week, as this week is fully booked!
Cheers, David
1 Like