Low GPP for MTC10,11 and 12

[luyssaert]

The current parameter values result in a too low GPP for MTC12 in North America and Europe. The GPP is especially low in the core cropping areas (400 gC m-2 y-1) and increases to 1050 gC m-2 y-1 towards the more northern edges of the cropping areas. This could hint at excessive water stress. Increasing K_LATOSA did not result in a higher GPP for MTC12 but it did help to get an acceptable GPP for MTC13.

[luyssaert]

r6432 was run without water stress (FM1, unstressed) and with the water stress switched on for FM1 and FM2.

Name	#PFT	GPP FM1, unstressed	GPP FM1, stressed	GPP FM2, stressed	NPP/GPP	July LAI	Comment
BroadLeavedEvergreenTropical	2	-	-	-	-	-	-
BroadLeavedRaingreenTropical	3	-	-	-	-	-	-
NeedleleafEvergreenTemperate	4	1200	1000	800	0.5	-	Check RDI
BroadLeavedEvergreenTemperate	5	1200	1200	1000	0.5	-	OK
BroadLeavedSummergreenTemperate	6	1200	1200	1200	0.5	-	OK
NeedleleafEvergreenBoreal	7	600	600	300	0.5	-	Check RDI
BroadLeavedSummergreenBoreal	8	1000	1000	700	0.5	-	OK
LarixSpBoreal	9	1100	1100	1100	0.5	-	In the Alps
!C3GrassTemperate	10	800	300	300	0.6	~3.5	Check GPP & water stress
!C4GrassTemperate	11	-	-	-	-	-	-
!C3AgricultureTemperate	12	800	300	300	0.6	~2.0	Check GPP & water stress
!C4AgricultureTemperate	13	1000	300	300	0.6	~2.2	Check GPP & water stress
!C3GrassTropical	14	-	-	-	-	-	-
!C3GrassBoreal	15	700	300	300	0.6	~2.0	Check GPP & water stress

Overall the GPP is on the low side but it could be for different reasons for different PFTs.

[luyssaert]

In r6432 PFT 12 and 13 are still growing. In r6444 PFT12 and 13 are no longer growing.

In r6444 the short vegetation (PFT10-15) hardly grows after 12 years and even the first years the vegetation dies over the course of three years. The first year the LAI, its phenology and plant_status all look OK. The second year, plant_status seems to work properly but the initial LAI is very low. There is still some labile and reserve N left but less than the first year. The same pattern was already present in r6432. This may have been the cause of the low overall GPP and LAI.

For PFT10 the LABILE_M_c+RESERVE_M_c is low at the end of the first year, the LABILE_M_n+RESERVE_M_n is higher at the end than at the beginning of the year, hence the N dynamics are unlikley to be the first order cause of the problem.

[luyssaert]

Compared PFT10 against PFT6. The following observations were made:

• In PFT6 the LAI develops quickly towards max LAI during spring. In PFT10 the max LAI is reached in september. This issue has been fixed for the trees in r6016. This contributes to differences in the reserves and labile pools.

-> Similar to the tree PFTs, grass and crops PFTs now pre-calculate the allocatable carbon based on the plant available nitrogen. This improves the allocation considerable. See r6462.

• Root conductivity has been increased for trees (incl. PFT6). This resulted in a low root mass (~10 g m-2) and thus makes growing a full canopy easier. In PFT10 roots and canopy are of the same magnitude. This is not the problem the first year and does not explain the difficulties to grow a canopy the second year.

-> See ticket #654 which needs to be fixed first!

• At the start of phenology almost all reserves and labile are invested in the roots in PFT10. Almost no carbon goes into the canopy (10e-4). PFT6 and PFT10 use a different approach of calculating the initial leaf mass. This could contribute to the problem.

-> fixed in r6460

• Plant n uptake depends on iroot + isapbelow but for grasses isapbelow is zero. So plant uptake is zero in winter because iroot is zero in winter. This does not explain the difficulties to grow a canopy the second year because there seems to be sufficient N in the labile and reserve pools.

-> fixed in r6459

• Lab_fac of trees and grasses was very different in the sense that lab_fac is above 0.3 for large part of the growing season in grasses (affecting some other calculations). For trees lab_fac is rarely above 0.3 and thus hardly has any impact.

-> in r6462 lab_fac was fixed to 0.1 for grasses and crops. A cleaner solution would be to remove lab_fac. See ticket #655.

[luyssaert]

Running r6470 with the FG2 over Europe and Africa does no longer contains any hints of too low GPP. GPPs seems to be very reasonable given their LAI (which are also reasonable).