Crop implementation in FATES: ideas and brainstorming

[rosiealice]

@samsrabin and I chatted a bit yesterday (IRL, in Potsdam!) about the necessary steps we would need to make for FATES to have a crop model. Background is that the existing agriculate impact models used in ISIMIP seems to a) have pretty underwhelming performance in general and b) systematically greatly underestimate the impact of extreme events on crops. So I think it would be useful if we collectively, in the medium term, try and leverage the mortality architecture we have in FATES for this purpose.

We discussed that to implement something like the CLM5 crop model (which is based on IBIS) in FATES, the following steps are needed:

1. Specific crop PFTs
2. Passing of existing CLM planting date windows into FATES
3. Implementation of temperature limits on planting date (https://escomp.github.io/ctsm-docs/versions/release-clm5.0/html/tech_note/Crop_Irrigation/CLM50_Tech_Note_Crop_Irrigation.html)
4. Logic for planting (where does seed carbon come from?)
5. Implementation of 'phased' phenological allocation scheme, where allocation to tissues varies between emergence, growing, grain filling phases, passage between which is a function of GDD (I suspect this is the most work).
6. For simulatons with nutrients, retranslocation logic also applies.
7. Harvest, (based in CLM5 on a growing degree day logic https://escomp.github.io/ctsm-docs/versions/release-clm5.0/html/tech_note/Crop_Irrigation/CLM50_Tech_Note_Crop_Irrigation.html)

Of course, applying irrigation and fertilizer logic would require crops to be on their own soil columns, which is a wider topic (though I couldn't find an actual issue? )

Have I missed anything here?

@samsrabin thinks that the more advanced crop models in the ISIMIP Agriculture sector (for example) have allocation phases that a) vary by PFT and b) can have a larger number of growth stages. I feel like like adding PFT specific parameters that can control the number of stages and the logic of allocation as a PFT specific vector might be a thing we could potentially pull off in FATES using the existing parameter architecture, albeit with a bit of work. It might pave the way for us being able to merge calibrations from the more advanced crop models (many of which, if I understand correctly, do not have the same degree of coupled hydrology, physiology and functionality that FATES does). @samsrabin can you remember the names of the specific crop models that do this?

Further potential work might involve adding waterlogging and heat-stress induced damage and/or mortality functions as these are major causs of crop failure.

Anyone interested in implementing any of this can use this discussion as a starting point. We (at CICERO) have resources to work on BECCS, which might include some elements of this logic, later this year/in 2026.

[huitang-earth]

Hi @rosiealice @samsrabin, thanks for the initiative.

Some features that may be added to this line of work, for instance, (1) allowing crop LAI to be prescribed using higher spatio-temporal remote sensing dataset (such as Sentinel-2). This may make the now-casting or hinder-casting at individual sites more accurate. (2) A systematic separation of the annual and perennial crop in the light of this paper may be interesting.

Hope these ideas are useful for planning the work.