ESCOMP
Use bilinear mapper for states?
From a conversation earlier with @mvertens , it seemed like CMEPS is currently using area conservative mapping for states as well as fluxes, at least for some components (e.g., lnd -> atm). This differs from the CESM behavior with CPL7, which used bilinear mapping for states. It seems like this should either be changed to match the CESM behavior with CPL7 or be presented for broader scientific review and approval.
@mvertens I am not trying to suggest any particular urgency by opening this issue, but I thought it would be helpful to have issues to track outstanding discrepancies between CMEPS and CPL7.
@mvertens points out that some states need to be mapped conservatively - e.g., from ocean to atmosphere. I'm not sure, then, what (if anything) used bilinear mapping in CPL7. From a quick look at the code, it looks like lnd -> atm did, but generally this wasn't invoked because lnd & atm were typically run on the same grid. If it turns out that all of the typically-remapped fields used conservative remapping, then maybe this issue can be closed without the need for broader discussion.
Discussion with Mariana today: this is still an issue, but currently we aren't running / testing any trigrid configurations with nuopc, and until that becomes something that anyone wants to use, this is low priority.