Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

How to manage crop areas in FATES-SP mode. #760

Open
rosiealice opened this issue Jul 1, 2021 · 23 comments
Open

How to manage crop areas in FATES-SP mode. #760

rosiealice opened this issue Jul 1, 2021 · 23 comments

Comments

@rosiealice
Copy link
Contributor

In FATES-SP mode, we currently do not consider crop areas in the assertion of LAI/SAI/HTOP etc. This means that in comparison with CLM5-SP runs, we under-predict the LAI where there are/should be crops.

My workaround for this so far has been to simply modify the surface dataset to add the crop area to the NAT_VEG area. I did this to facilitate direct comparison and identify biases between CLM5-SP and FATES-SP, but it's obviously the wrong approach in the long run.

I'm not sure how we should handle this as in general, though. FATES currently runs only on the NAT_VEG landunit/column. Solutions to this include

  1. Run FATES (SP) on both the NAT_VEG and CROP land units.
  2. Introduce some FATES-SP specific logic that expands the NATVEG area to include the crop area and PFTs and then just run FATES once.

I'm not really sure how exactly to implement the first, but it seems like potentially the more elegant thing to do. Does anyone have any thoughts/potential other solutions/pointer on how to active FATES on a crop column?

@dlawrenncar @danicalombardozzi @lawrencepj1 @wwieder @glemieux @ckoven @rgknox @aldivi @ekluzek @olyson ?

@rgknox
Copy link
Contributor

rgknox commented Jul 1, 2021

In the long run, don't we also want the existing crop model to run simultaneously with fates, each on their own columns? Would this facilitate direct comparisons?

@rosiealice
Copy link
Contributor Author

I guess you could have the CLM5 crop model running in native CLM5-SP mode in parallel? Right?

@rosiealice
Copy link
Contributor Author

I also just realized that irrigation pertains to SP mode (whereas most of the crop model does not), so if we had SP-crops-in-FATES, they would need irrigating?

@rgknox
Copy link
Contributor

rgknox commented Jul 1, 2021

I guess you could have the CLM5 crop model running in native CLM5-SP mode in parallel? Right?

That is what I was thinking

@wwieder
Copy link

wwieder commented Jul 1, 2021 via email

@lawrencepj1
Copy link

lawrencepj1 commented Jul 1, 2021 via email

@rosiealice
Copy link
Contributor Author

Righto. So the generic SP crop sits on its own land unit doing SP things and potentially being irrigated? Right?

@ckoven
Copy link
Contributor

ckoven commented Jul 1, 2021

seems like your option (1) is the best plan, as it would be a first step towards a full FATES w/ crops implementation.

@lawrencepj1
Copy link

lawrencepj1 commented Jul 1, 2021 via email

@danicalombardozzi
Copy link

danicalombardozzi commented Jul 3, 2021 via email

@rosiealice
Copy link
Contributor Author

Thanks all. That's interesting @lawrencepj1 that the crop land unit doesn't occur at all without irrigation. Which means either we
a) keep make the NATVEG single instance of FATED to account of generic crops PFTs.

Doing only this would make FATES-SP incompatible with irrigation. To remedy that I suppose we could do either
b) Run another instance of FATES-SP on the irrigated landunit
Or
c) Run the CLM5-SP crop model on the irrigated landunit.

I'm not sure which is harder out of b and c. I suspect b) might be more straightforward and give cleaner answers. (Noting that this is just for SP mode, wherein I don't think the crop-model specific terms have an impact, right?

Either way, making a) work is on the critical path I think...

@dlawrenncar
Copy link

dlawrenncar commented Jul 6, 2021 via email

@ekluzek
Copy link
Collaborator

ekluzek commented Jul 8, 2021

Note, the CTSM issue about using the CTSM crop landunit is here. This issue is different but related to this one here.

ESCOMP/CTSM#1046

This would use the CTSM big-leaf code to runs crops, which could also be done for FATES-SP mode.

@billsacks
Copy link
Member

Summary from ctsm-software discussion:

Can we have FATES handle crops in SP mode similarly to what's done in non-FATES – i.e., treating crops as grasses, but with a different subgrid structure – one crop per column? This way crops can use the same physics as natural veg in SP mode.

We may need to use Ryan's new column-level logical flag in more places.

Need to check some other subgrid setup (e.g., initSubgridMod, subgridMod) and set up of filters, as well as any new idiosyncracies that are coming in with the FATES-SP PR.

Need to double check that there are no assumptions in FATES that there is only one column that it's operating on, particularly in terms of subgrid weights.

@rosiealice
Copy link
Contributor Author

Thanks @billsacks

On your first point, I think it broadly makes most sense to use the FATES physiology for crops in FATES-SP mode. Most of the crop model is downstream of GPP, as far as I remember, but is there anything special about what it does with Vcmax? (that is an input parameter in the default carbon-only FATES).

How does the sub grid structure work in CLM5-SP for crops? Is it -not- one column per crop type?

I agree it would be super useful to check on the sub-grid setup, filters etc. This is the part I am foggiest on wrt making this work. Thanks :)

@danicalombardozzi
Copy link

@rosiealice: I can answer part of this. In CLM-SP, crops are treated the same as grasses but with higher Vcmax values and also the option to irrigate for the irrigated fraction.

I believe the subgrid structure is still one column per crop type, but the SP configuration uses only 16 PFTs total, with PFTs 15 & 16 as crops: generic C3 rainfed and generic C3 irrigated. On the land use time series (or surface dataset if running a time slice), the generic crops are on the crop land unit, similar to the structure for BGC, while the rest of the PFTs are on the natural vegetation land unit.

@rosiealice
Copy link
Contributor Author

Here is where we decide whether a column is handled by FATES, as I see it
https://github.com/ESCOMP/CTSM/blob/01d4cf03eeac00b7d8b99c92cdb41064aeaf1a77/src/utils/clmfates_interfaceMod.F90#L539\

So we are choosing the columns with an 'istsoil' label. Do we think that adding the 'istcrop' labeled columns to this would get us to the point where FATES would at least try and run on the crop columns (@rgknox @billsacks?)

Notwithstanding that then I'd also have to sort out how to make the crop instances use the correct PFTs and areal fractions.

I'd test it to find out but Cheyenne is down this week...

@rgknox
Copy link
Contributor

rgknox commented Jul 27, 2021

@rosiealice , I believe that gets us half way there
The other part of it is to make sure that those crop columns are added to the ctsm-side filters and lists (but I'm guessing they already are added to the correct filters for the non-fates sp case?).
There must be more to the ctsm side of things, my sense of the needs on that side are murkier. For instance, are there crop history diagnostics that are important to maintain?

Alternatively, we could push the crop fraction of the grid-cell onto the naturally vegetated portion of the gridcell, thereby increasing the weight of the natveg LU to be the sum of natveg+crop, then set the crop LU weight to 0, and make sure that the crop fates pft gets that fraction of the now shared column. This would seem to be the pretty simple to implement and would seem to just work without having to manipulate anything on the ctsm side, aside from those two weights. But we would lose out on any history diagnostics that we want maintained on the ctsm side.

@dlawrenncar
Copy link

dlawrenncar commented Jul 27, 2021 via email

@billsacks
Copy link
Member

@rosiealice - I agree with @rgknox that, to first order, the line you identified seems to be the key thing here. As we discussed a couple of weeks ago in the ctsm-software meeting, though, this is one of these changes where the hard part is thinking through / identifying where there might be implicit assumptions that are violated by this change. (See #760 (comment) for some initial thoughts on that.) This could be done via looking through relevant parts of the code (though off-hand I'm not sure I can identify what to check... figuring that out can be the hardest part), and/or through careful examination of many aspects of the model results to verify that things are working as they should. (I would normally be inclined towards the former, but I was impressed by your recent careful analysis of the causes of differences between FATES-SP and Big Leaf CTSM; a similar careful analysis of FATES's handling of crops could give us confidence that things are all working as intended.)

@rosiealice
Copy link
Contributor Author

see #817

@adrifoster
Copy link
Contributor

@rosiealice @ckoven for my SP calibration I am treating the crops as c3 grasses. Is this a sufficient solution?

@wwieder
Copy link

wwieder commented Nov 7, 2024

Suggestion is to run these as generic grasses in FATES-SP. This can be done on the parameter file. Would be resolved with https://github.com/NGEET/fates/pull/817/files

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
Status: No status
Development

No branches or pull requests

10 participants