ENT LAI too high

[rosiealice]

The LAI of ENTs is high in recent simulations because of a shift to using an l2fr of 0.75 for ENTs.

This is evident in run with the most recent parameter tag
https://ns9560k.web.sigma2.no/datalake/diagnostics/noresm/rosief/i2000.ne16pg3_tn14.fatesnocomp.noresm3_0_beta09.CPLHIST.BT3_LSP_MS3.25_BT1EBT.2026-01-24/OBS_comparison/ANN/i2000.ne16pg3_tn14.fatesnocomp.noresm3_0_beta09.CPLHIST.BT3_LSP_MS3.25_BT1EBT.2026-01-24_compare_ELAI_MODIS_ANN_0025-0030.png

and with CRUJRA

@kjetilaas has point plots from the boreal forest, e.g.
https://ns9560k.web.sigma2.no/datalake/diagnostics/noresm/kjetisaa/NorESM_Key_Simulations/n1850.ne16pg3_tn14.noresm3_0_beta10.Run6.2026-01-23/Point_Plots/Point_Timeseries_n1850.ne16pg3_tn14.noresm3_0_beta10.Run6.2026-01-23_Boreal_ilamb_406-435_Seasonal.png

which are from 1850, but show that in summer time we are not all that far off.

My simulations are from y2000, and these plots compared the maximum monthyl LAI in the data product and the model

.

[rosiealice]

At the same time, we have had persistant issues with the boreal forest trees being too small (references for this?) but this is not related to a deficit in GPP,

Nor is it an issue with the CUE: which seems reasonable (if on the low side with higher LAI configurations here).

These parameters we chose to look at in the using a previous ensemble which suggested that the ratio between LAI and NPP was controlled by d2bl1 in particular. Which make sense as this is the primary lever on how much carbon goes to leaves for a given amount of NPP:
https://ns9560k.web.sigma2.no/datalake/diagnostics/noresm/rosief/ppe_diags/ens_v1_soilc/npp1850_lai2000/npp_lai_ratio_vs_parameters_ENTs.png

In the ensemble I did yesterday, this is confirmed:
https://ns9560k.web.sigma2.no/datalake/diagnostics/noresm/rosief/ppe_diags/ppe_2000_v10_entlai/npp_lai_ratio/npp_lai_ratio_vs_params_years28-28_ENT.png

[rosiealice]

This plot shows quite cleanly what is the relationship with NPP and LAI and these parameters
https://ns9560k.web.sigma2.no/datalake/diagnostics/noresm/rosief/ppe_diags/ppe_2000_v10_entlai/npp_lai_ratio/npp_lai_ratio_vs_params_years28-28_ENT.png

NPP is controlled quite tightly by l2fr (presumably via its impact on root MR), but LAI is controlled mostly by d2bl1.

Interestingly, in this system, LAI does not seem to have a big impact on GPP, meaning that this feedback is not very strong, and so we can alter LAI down without actually having a big impact on NPP, I think.
https://ns9560k.web.sigma2.no/datalake/diagnostics/noresm/rosief/ppe_diags/ppe_2000_v10_entlai/gpp/gpp_vs_parameters_ENT_linear.png

[rosiealice]

Here is another way of looking at that relationship:

Going back to the LAI plot then:

It seems to me that the conclusion of this is that we should reduce the ENT d2bl further, to maybe We could probably get away with reducing the l2fr to increase CUE/NPP and VEGC, but I am not sure that we want to do that? (@JessicaNeedham ?)

We haven't actually explored the space that gets us to the actual data range. If we were to do any further experiments, that might be helpful. The minimum we looked at here, 0.0275, is OK, but pushing it a bit lower would likely look better.

[rosiealice]

I have set off a run to test this and a few of the other biases which alters d2bl1 for ENT
NorESMhub/NorESM3_land_only_dev_simulations#96

[rosiealice]

Update on present state of ENT biases

[rosiealice]

So, I turned down the d2bl again. The changes I made look like this:

<  fates_alloc_store_priority_frac = 0.6, 0.8, 0.8, 0.8, 0.8, 0.8, 0.8, 0.8, 
---
>  fates_alloc_store_priority_frac = 0.4, 0.8, 0.8, 0.8, 0.8, 0.8, 0.8, 0.8, 
1060,1061c1060,1061
<  fates_allom_d2bl1 = 0.04, 0.025, 0.05, 0.01, 0.04, 0.07, 0.07, 0.07, 0.05, 
<     0.018, 0.018, 0.018, 0.04, 0.04 ;
---
>  fates_allom_d2bl1 = 0.04, 0.03, 0.05, 0.01, 0.04, 0.07, 0.07, 0.07, 0.05, 
>     0.022, 0.022, 0.022, 0.04, 0.04 

and the parameter file is here:
/cluster/work/users/rosief/paramscratch/fates_params_sci.1.88.6_api.42.0.0_14pft_nor_sci3_api1_c260219.nc

Diagnostics for this new parameterisation are here:
https://ns9560k.web.sigma2.no/datalake/diagnostics/noresm/rosief/i2000.ne16pg3_tn14.fatesnocomp.noresm3_0_beta11.CRUJRA.14hlmpft_LUCy2k_2000FSx16pft_14fpft_d2bl1.2026-02-20/

and should be compared to
https://ns9560k.web.sigma2.no/datalake/diagnostics/noresm/rosief/i2000.ne16pg3_tn14.fatesnocomp.noresm3_0_beta11.CRUJRA.14hlmpft_LUCy2k_2000FSx16pft_14fpft.2026-02-19/

(I will write these up into runs shortly, and my comparison diagnostics crashed but I will double check my command), BUT after 20 years:
the LAI looks better:
BEFORE:

AFTER

GPP, on the other hand, is not very affected by this:
BEFORE

AFTER

[rosiealice]

fwiw @JessicaNeedham this doesn't seem to affect NPP either, so shouldn't imply anything for biomass. Which is almost wierd tbh, but...