CFTs of input and output not matching

[Andrie]

Dear Team,

I was running a CLM5.0-only simulation with the I2000Clm50BgcCrop compset where I planted additional trees. I was reading in a different surfdat file where the tree coverage (natpft 1-8) is increased while the crop area is decreased accordingly (PCT_CROP). As a check that my simulation ran correctly, I was writing out the PCT_NAT_PFT and PCT_CFT variables where I encountered a surprising problem: If I compared the cfts one-by-one they were not matching up.

After some more checking, I found out that my simulation did run correctly but that for some reason some of the cfts are mapped to inactive types in the output (according to the 3rd column of the Table 2.26.1 2.26. Crops and Irrigation — ctsm release-clm5.0 documentation). When I now add together all inactive crop types to the active ones such that I get 16 active cfts (or 8 active cfts with one irrigated and one rainfed version), then the changes in cfts are matching up between the input and outputted PCT_CFT variables.

I hope that the above is understandable. I was just wondering if there is a reason why the cfts would get mapped to inactive types when outputting the PCT_CFT variable.

Thanks a lot!

 

[slevis]

I'm glad that you got your simulation to work.

From reading the same section (2.26. Crops and Irrigation — ctsm release-clm5.0 documentation)
I understand that the model should merge inactive types with the active types listed in column 4 of the same table.

Did you instead find that the model did not do that, and you had to do that manually? That's what I understood from your description.

 

[Andrie]

Hi,

sorry a bit difficult to explain: Basically if I take the input file (which I based on the existing I2000Clm50BgcCrop surfdata file) and I plot each of the 64 cfts then only the active cfts show some distribution across the globe. However, when I make the same plot using the PCT_CFT variable that I outputted from the runs then some of the cfts which are inactive according to the table have non-zero/non-nan values over land. Only when I add together all inactive cfts to each active cft (i.e. ending up with the 16 active cfts) then my input and output data agrees.

I attached two plots (the files were to big to include them as a file so I am copying a subsection of the plot here)
1. cft_baseline_inputoutput: The first one shows the 64 cfts separately with the input on the left and the output on the right hand side (the cfts are labelled 0 to 63 as is the case in the output where 10 corresponds to 25, 11 to 26 etc in the table mentioned).
2. cft_summarised: This plot shows that the total distribution for each cft type that is active (+ added are the distributions from inactive cfts that are mapped to one of the active cfts). The first column is the input, the second the outputted PCT_CFT and the 3 the difference of column of 1 and 2 which is always 0.

A friend also said that he had the same strange cft output in his simulation so it does not seem to be a one time occurrence.

Is that better to understand now?

cft_baseinelin_intputput:


cft_summarised

 

[oleson]

For that compset and a resolution of f09 using the supported development release of the model (release-cesm2.2.0) the surface dataset is:

/glade/campaign/cesm/cesmdata/inputdata/lnd/clm2/surfdata_map/release-clm5.0.18/surfdata_0.9x1.25_hist_78pfts_CMIP6_simyr2000_c190214.nc

Is that the version you've modified?
I see non-zero PCT_CFT for inactive and active crop types in that dataset.
However, when I do a short run with that dataset, I only see non-zero PCT_CFT for active crop types in the history files. Using the 0-63 index on the file, these types are 2,3,4,5,8,9,26,27,46,47,52,53,60,61,62,63. So the inactive types are being collapsed to the active types properly as far as I can tell.
I'm not sure why the inactive types would be non-zero in your simulation. Have you tried looking at output from a default simulation and confirmed you get the right results?

 

[Andrie]

Hi Keith,

you are completely right. I found my mistake and I am now also getting that the input has inactive+active and the output only has active cfts.
Thanks a lot for the help!