River sequence problems

[MartinDix]

Suites are currently using ancillaries from /scratch/tm70/kr4383/cylc-run/ancil-gen-22-04-2026/share/data/n96e_mom025_20260316 and
/scratch/tm70/kr4383/cylc-run/ancil-gen-22-04-2026/share/data/n96e_mom025_20260316/qrparm.landfrac.

The qrparm.rivseq files here have some blocked outflow points which is likely affecting water balance. They're also missing the Black Sea - Mediterranean connection

Unfortunately at the moment it's a manual correction for these. See also #48.

[anton-seaice]

Might be able to use remapping weights like

https://github.com/ACCESS-NRI/om3-scripts/blob/main/mesh_generation/generate_rof_weights.py

to ensure all runoff gets moved into the ocean (as long as the runoff reaching the mediator isn't lost to a land mask)

[kieranricardo]

@MartinDix the suites are still using your fixed river sequence from a while back: /g/data/vk83/prerelease/configurations/inputs/access-cm3/ancil/n96e_momO1_20201102/qrparm.rivseq.20241001, but all the other ancillaries from /scratch/tm70/kr4383/cylc-run/ancil-gen-22-04-2026 are being used.

I haven't checked whether this matches the new land mask (I'm not sure how the UM calculates the landfraction on the river grid), but we're not getting the massive water conservation errors we used to have.

@anton-seaice yeah I think your mapping will automatically fix this! I think the UM still outputs the river runoff if the outflow points are land, it just gets lost in cap/mediator.

[anton-seaice]

Yea I think so