Bugfixes for crop model

[bbye]

Fixes bugs introduced into crop model
Contains:
* Harvest of phosphorus
* Add grain phosphorus to allometry
* Remove retranslocation during harvest
* Fixes allocation when peak lai is reached
* Remove unnecessary print statements

This commit fixes various bugs that are in the crop model from
the phosphorus code, dynamic nutrients, and some misc others.
First, harvest of the phosphorus is added including a 1-yr
product pool. Also, the grain phosphorus demand is included
in the calculation of phosphorus allometry. Retranslocation
terms are removed from the offset litterfall since crops are
annual and completely harvested. Finally, the allocation when
crops reach peak lai is modified to prevent unrealistic
biomass partitioning in roots and stems.

[Non-BFB] - Non Bit-For-Bit

[bbye]

Pass acme developers tests on Edison. There will be one conflict with the eca bugfix (#962), please use my changes to the crop code that are in the CNPhenologyMod.

[bishtgautam]

@bbye : Thanks for the heads up on conflicts in CNPhenologyMod.F90. What do we need to add a test for the crop model to protect it? What is the the coarsest resolution that has few grids cells with non-zero crop fraction? What should be the duration of the crop test (2yrs)?

[bbye]

@bishtgautam : A 2-year simulation would be best since crops aren't planted until the second year anyway. @rljacob and I discussed the possibility of using a restart simulation to shorten the duration of the test. I generally run at a 2 degree resolution, but one could run a lower resolution than that, the US Midwest is pretty crop intensive. If it saves time, we could also do a point simulation which would speed up the test too.

[bishtgautam]

Apart from the global grids, NCAR is using the following single point grids for testing the crop model:

    1x1_smallvilleIA      (desc: 1x1 Smallville Iowa Crop Test Case -- only valid for DATM/CLM compset) (nx: 1) (ny: 1) 
    1x1_numaIA            (desc: 1x1 Numa Iowa -- only valid for DATM/CLM compset) (nx: 1) (ny: 1) 

How about using them for ALM testing?

[bbye]

Yes, either of those would work. We just need to make sure we use the correct surface map data (some of the files only have the generic crop types). Any surface map with mp24 in the name should work.

[bishtgautam]

What are the compsets that work for the crop model (ICLM45CNCROP, ICLM45BGCCROP, ICLM45BGCDVCROP)?

[bbye]

For sure, ICLM45CNCROP and ICLM45BGCCROP. In theory, ICLM45BGCDVCROP will also work (I haven't tested it), but since crops don't have a dynamic vegetation component, they should run the same as the BGC compset.

[bishtgautam]

Should components/clm/bld/namelist_files/namelist_defaults_clm4_5.xml be modified for the use_crop=".true." from clm_params.c151111.nc to clm_params_c160303.nc?

[bbye]

Yes it should.

[bishtgautam]

I made a commit to your branch regarding the parameter file. Please update your branch on Edison (git pull --ff-only) and run the following two simulations on Edison to ensure that crop model is indeed active for the domain.

./create_newcase -silent -case 1x1_numaIA.ICLM45BGCCROP \
-res 1x1_numaIA -mach edison -compiler intel -compset ICLM45BGCCROP -project acme 
cd 1x1_numaIA.ICLM45BGCCROP
./cesm_setup
./xmlchange -file env_run.xml -id STOP_N -val 40
./xmlchange -file env_run.xml -id STOP_OPTION -val nmonths
./xmlchange -file env_run.xml -id PIO_TYPENAME -val netcdf
./*.build

./create_newcase -silent -case 1x1_smallvilleIA.ICLM45CNCROP \
-res 1x1_smallvilleIA -mach edison -compiler intel -compset ICLM45CNCROP -project acme 
./cesm_setup
./xmlchange -file env_run.xml -id STOP_N -val 3
./xmlchange -file env_run.xml -id STOP_OPTION -val nyears
./xmlchange -file env_run.xml -id PIO_TYPENAME -val netcdf
./*.build

[bbye]

I'm trying to update the branch with git pull --ff-only, but I keep getting the message "There is no tracking information for the current branch."

[erichlf]

You need to do something like git pull --ff-only origin

[bbye]

Good news, bad news. The numaIA run failed with permission denied - looks like @bishtgautam already found that out too. The smallvilleIA run completed with crops, however several phosphorus variables (but not all of them) are nan in the output file.

[bishtgautam]

On Edison, I downloaded fsurfdat for 1x1_numaIA. Can you try again?

[bbye]

I tried again and the model runs for those resolutions with crops. I managed to fix the nan's, but now the pbalance error is back. I must have introduced a new bug that masked the old bug. Hold off until I can get to the bottom of this.

[bbye]

OK, I managed to get all the bugs out of the crop model. The phosphorus bug is still there (related to #974), but at least now I can avoid the pbalance error when the use_nitrif_denitrif flag is false ( I cannot avoid the error by only setting suplphos='ALL'). Also, I can confirm @bishtgautam that both crop simulations work (1x1_smallvilleIA and 1x1_numaIA).

[bishtgautam]

@bbye : To get 1x1_numaIA.ICLM45BGCCROP and 1x1_smallvilleIA.ICLM45CNCROP running successfully, does one need to add use_nitrif_denitrif = .false. in usr_nl_lnd? Also, is this PR now ready to be merged? Or, should I wait for a fix for #974?

[bbye]

Yes, one needs to set use_nitrif_denitrif = .false. for the compsets to be run. By default, the 1x1_smallvilleIA is already set to false, only the 1x1_numaIA needs to be adjusted. This PR is ready to be merged, I don't think it needs to wait for #974. It also passed all the acme_dev tests.

[bishtgautam]

@qzhu-lbl : While merging @bbye's PR, there are few conflicts in CNPhenologyMod. The conflicts can be seen at https://gist.github.com/bishtgautam/dea9084a1722e9c75ad7f7abbb695dd4. The major conflicts are:

1. In the if (ivt(p) >= npcropmin) loop.
2. Else condition of if (offset_counter(p) == dt).

Are @bbye's changes compatible with the ECA model?

[bishtgautam]

@bbye : I have added a test for CROP model, which ran successfully on Edison. If @qzhu-lbl has no objection to your version of CNPhenologyMod, I will merge your branch to next today.

[bbye]

Thanks. I can chime in to help resolve the conflicts. The code in the "if (ivt(p) >= npcropmin)" loop should reflect my modifications, but the code under the else condition of the offset_counter should be that provided by @qzhu-lbl . @qzhu-lbl, can you confirm?

[qzhu-lbl]

@bbye could you explain more about: why crop litter fluxes only occur in the end of offset period (if (offset_counter(p) == dt) then)? What about during the offset period (offset_counter(p) > dt)? Does crop model also need grainp_to_food and grainn_to_food fluxes?

[bbye]

@qzhu-lbl : The crop offset period is only one timestep so it never enters the loop where offset_counter(p) > dt. As soon as the crop reaches maturity, it is harvested. Since these are annual crops, they don't have an offset transition, nor do they transfer nutrients back to storage pools. New crops are planted each year from a seed stock, not from the previous years growth. The grainx_to_food fluxes have been replaced with hrv_grainx_to_prod1x (where x is c, n, or p), which then places all of the grain into the harvest product pool.

[qzhu-lbl]

@bbye I think your code between [if (ivt(p) >= npcropmin) then ------- else] is good for ECA module, thanks.. @bishtgautam You can also clean up the commented codes if you want.

[bishtgautam]

@bbye, @qzhu-lbl : I have merged this PR into next. Can you both verify that CNPhenologyMod.F90 in next preserves both of your changes?

[bbye]

Thanks @bishtgautam , looks OK to me.