GPP biases

[rosiealice]

Adding a thread to discuss the high GPP bias in FATES-SP mode.

Initially we thought that increasing the absorbed radiation would increase ET such that a lower ball berry slope could be used, decreasing GPP accordingly. On creating a parameter set that matches the absorbed radiation of the CLM 5 code, FATES still has high GPP unless vcmax is significantly reduced.

Potential differences with CLM5 (which has much lower GPP biases that are not so mono directional) include:

1. Absence of daylength adjustment in vcmax in FATES ( @JAHolm working on this)
2. LUNA causing low vcmax in CLM5 and this dampening GPP
3. Photosynthetic acclimation (I tested this and the impact was minimal)
4. Absence of TPU limitation in FATES
5. Absence of smoothing in FATES (changing this has an effect on the inferred optimal vcmax but it was relatively minor)
6. Differences in the calculation of jmax.

The first priority is thus I think to assess a simulation where vcmax is identical between models (I.e. set to 50) to assess whether the CLM5 vcmax model is effectively running with very low photosynthetic capacity. I will try to do this next week...