DGVM

meeting of 04/05/2018

The objective of this first meeting was to make the point on the DGVM part of ORCHIDEE. The meeting included the usual ORCHIDEE-TEAM but also people from LSCE CLIM team who are interested by DGVM development (M Kageyama, P Braconnot, P sepulchre). the main point of the discussion was around discussion on the state of art of the model, strategy for the future both in term of development and organization to improve what is done currently

state of the art

the objective was here to give an overview of recent developments and issues still to solve.

• First a presentation has been done by D Zhu who improved the representation of boreal ecosystems to summarize the changes done
• Then a presentation of P Braconnot was done to show results of the long term simulation of Holocene where DGVM was applied to reconstruct the dynamic of vegetation.

then we had a discussion on technical issues to solve:

for the current version:

• the is still problem for tropical forest where there is a confusion, in the reconstructed distribution between evergreen and raingreen forests. But it as been solve by W. Chen who observed a too low carbon use efficiency (and then too small biomass in evergreen forest. Correcting this seems to correct the confusion in PFTS
• P. Peylin point out that the new description of bare soil albedo which is prescribed spatially from satellite could be a problem for paleo simulation as albedo is not estimated independently of vegetation (especially for dense vegetation) and then could be incorrect when vegetation change. But the original formulation based on soil class is still operational and can be an alternative
• P Ciais indicate an issue on treatment of managed land. Indeed ORCHIDEE in DGVM mode can deal with prescribed crop land areas but not with pasture. It as been implemented already in MICT but this should be now implemented in the trunk
• P Braconnot indicate an instability on the long term run making the DGVM crash sometime. This is to be investigated

for the near future:

we should think about compatibility with new versions of ORCHIDEE. As point out by S. Luyssaert, the main problem will be with CAN where there is a totally different representation of the canopy (with age and diameter classes) which change the competition. So this is a very interesting potential as it will give a more process based representation of competition than is it done currently but will require works to be done

for the longer term:

• there is some studies that will begin in the near future on representation of paleo PFT that could take into account for difference in functioning (P. Sepulchre). Or interaction between vegetation and animals 5P Ciais, D Zhu)
• We should think also to the next generation of DGVM. in particular for the moment there is a separation between ecosystem functioning and vegetation dynamics which is artificial and related to the fact we don't explicitly represent all the process that drive the dynamics but this should change in the future that should allow to avoid this separation. The inclusion of DGVM in CAN will be a first step toward this. A coupling with models like PHENOFIT could also be a way to have an explicit representation of vegetation fitness and reproductive success.

strategy

We discuss then how to improve the development of vegetation dynamics of vegetation in the future as it was clearly not very good in the past compared to the rest of the model. general points:

• Have regular meeting (3 to 6 month) with CLIM team on the aspect
• Having some people from CLIM coming to Tuesday ORCHIDEE meetings

search for people dedicated to this topic and sharing Phd between CLIM and ORCHIDEE team

• a dedicated web page (this page !)

more concrete short term points:

• check if D Zhu modification are in the current trunk and include recent modification of MICT on parameters of tropical forest and a flag to include pasture in prescribed vegetation areas
• find the reason of instability pointed by Pascal
• make diagnostic on simulations with CMIP6