A potential bug regarding nitrogen uptake by crops and trees

[yunan-l]

Hey, I noticed that the crop and tree nitrogen-uptake routines appear to compare the remaining plant N demand against n_uptake, which is still initialized to zero at that point, rather than against the calculated potential uptake nupsum.

In src/crop/nuptake_crop.c, the n_uptake is initialized to zero at the beginning of the function. After accumulating the potential NO3 and NH4 uptake in nupsum, the code uses:

if(n_uptake>*n_plant_demand-pft->bm_inc.nitrogen)
  n_uptake=*n_plant_demand-pft->bm_inc.nitrogen;
else
  n_uptake=nupsum;

The tree routine contains the same pattern in src/tree/nuptake_tree.c.

By contrast, the grass implementation compares nupsum with the remaining demand in src/grass/nuptake_grass.c:

if(nupsum>*n_plant_demand-(vegn_sum_grass(pft)+pft->bm_inc.nitrogen))
  n_uptake=*n_plant_demand-(vegn_sum_grass(pft)+pft->bm_inc.nitrogen);
else
  n_uptake=nupsum;

Is this a bug :)

[cmueller-pik]

Dear @yunan-l thanks for raising this issue!
Indeed, that looks wrong.
Essentially, this means that crops and grasses exudate N to soils if they have more N than currently requested (i.e. if *n_plant_demand-pft->bm_inc.nitrogen is smaller zero (n_uptake) , i.e. negative), making buffering N impossible.
I will run a test on fixing this and report back.

[yunan-l]

Hi Christoph, I think there is another bug in npp_crop.c: root respiration is not subtracted from crop NPP.

The root respiration (roresp) is included when growth respiration is calculated and also added to the autotrophic respiration (RA) output, but not subtracted from crop NPP. The code in src/crop/npp_crop.c:

gresp=(assim-roresp-soresp-presp)*param.r_growth;
if(gresp<0.0)
  gresp=0.0;
npp=assim-soresp-presp-gresp;

/* ... */

getoutput(&pft->stand->cell->output,RA,config)+=
  (soresp+roresp+presp+gresp)*pft->stand->frac;

The calculation is correct in previous versions. The root and storage respiration are combined in rosoresp, which is subtracted from NPP.

[cmueller-pik]

Thanks Yunan!
Initial testing shows that the nuptake fix makes broadleaved trees grow better, which is good. More scrutiny is needed to confirm all aspects, but I think that's a general improvement in model performance.
The crop_npp fix eliminates more or less the productivity gain from the nuptake fix at the global aggregation, but I haven't looked into spatial patterns or temporal dynamics yet.

[yunan-l]

That's great to hear :)