How to convert LUH2's landclasses to the landclasses needed by the CLM

[CGL]

Hi,all!

Recently I've been trying to use the new GCAM data to convert it into land use data that can be driven by CLM (related post). I still have some questions about how LUH2 is converted to CLM standard format for land use driver data. My CESM version is 2.1.3.

1. When converting to LUH2 format with GCAM, this article (Harmonization of global land use change and management for the period 850–2100 (LUH2) for CMIP6) demonstrates that only 5 land classes were converted.But in CLM5-Land-Data-Technote it converts out to 15+32 classes. I understand that it was obtained with the current satellite data and FAO etc.

2. I guess that LUH2 is just an intermediate pft dataset and the number of land classes can be arbitrary. The conversion of the final number of landclasses and the standard landclasses is done through CLM5's mksrfdata conversion.I'm not sure if I'm understanding this correctly?


3.If I have a pft's landtime series file, the land classes in it don't match the CLM land classes. How does the classic method at such times assign less landclasses to more landclasses?

 

[CGL]

For the third point, I would like to explain it in detail.
I have now obtained the pft and cft landuse_timeseries files, but the land use type does not match the LUH2. According to workflow of CLM5 Land Use Data Tool，does a conversion with the mksrfdata tool allow me to get the standard landuse_timeseries types that match the CLM model?

My understanding is that the existing remotely sensed data and the IAM output of the future simulation data to get a global pft grid data. This pft grid is then fed into the mksrfdata tool . During this process, the landuse type of the pft grid is reconciled by the mksrfdata tool. Maybe it can also be used to make adjustments to my landuse_timeseries land use types?

 

[slevis]

@lawrencepj1 do you have input for @CGL?

On the 3rd point, I will try responding, though I may have misunderstood the questions:
- The mksurfdata tool takes data from our so called raw datasets and puts it into an fsurdat file for use later by the CLM.
- Users are welcome to generate their own raw datasets, and have the mksurfdata tool point to them instead of to the default ones.

I hope this helps at least a little. I'm afraid I likely didn't understand some of @CGL's points.

 

[CGL]

Actually,my question is related to this post.

For clarity, I try to construct a GCAM-Demeter-CESM framework. I generate a 2020-2100 landuse_timeseries file, but the landclass not match the CLM inputdata/surfdata. So I want to use mksurfdata tool to generate the standard landtime_series data.

mayebe I can directly match up the land classes at downscaling, but this would involve land class matching and standard formatting. I refer to this papaer to produce the landuse_timeseries file.

 

[slevis]

Are you hoping to achieve synchronous or asynchronous coupling between the iam and esm? Synchronous means the two models run simultaneously and communicate while running; asynchronous means that the models run separately and do not communicate directly while they run.

 

[CGL]

It is the latter that I want. I want the IAM land use output to drive CESM.

 

[slevis]

Got it. And, if I understand correctly, you wish to try @lawrencepj1's tools to generate clm "raw" datasets, to then run the mksurfdata tool to generate fsurdat/landuse files. I assume that you need to modify @lawrencepj1's tools to read the IAM output. All that sounds fine. Please direct questions about @lawrencepj1's tools to him. If he does not respond in this thread, you may wish to try email.

 

[CGL]

Yes. you understand correctly. I will email lawrence :) Thank you @slevis ! have a good day!