Multi-tiling developments in ORCHIDEE

During the ORCHIDEE retreat in May 2019, it was decided to move forward with multi-tiling developments in ORCHIDEE. A working group of interested was formed, and we will keep this page designated to the discussion and implementations concerning these developments.

Brainstorming on multi-tiling developments

As agreed on during the meeting on 26/6 2019, a brainstorming phase was decided to aid us in deciding our needs in relation to multi-tiling implementation. Thus, everybody is welcome to make comments and suggestions in this section.

Meeting of 26/06/2019

Present: Anthony Bernus, Agnès Ducharne, Anne Sofie Lansø, Catherine Ottlé, Chao Yue, Frederique Cheruy, Josefine Ghattas, Jan Polcher, Martin Menegoz, Nicolas Vuichard, Philippe Peylin, Vladislav Bratiskov

After a short summary/introduction​ on the current status of the developments in ORCHIDEE, the discussions commenced. Although coupling to the atmosphere is acknowledged to be important and needs prioritizing, we agreed to start by focusing on the changes in ORCHIDEE itself in relation to multi-tiling implementations. JP did mention an interesting paper for the atmospheric coupling DeVrese paper 2016​.

Flexibility In terms of flexibility, do we wish to include cases of:

• 1 energy budget, and keep several soil water budgets (like the current trunk allowing some sort of back-ward compatible) -YES
  • 1 soil column, for several SFTs each with an energy budget? Not agreed*

*The second case was not agreed on, but arguments for keeping this case were:

• Agro-ecology systems where crops and trees would compete for the same water
• Forest systems, where species or age classes compete for water but needs different energy budgets

While arguments against it were:

• CY mentioned that there are papers showing how sparse vegetation needs individual soil columns.
• JP highlighted the physical argument of soil thermics and freezing, which for a single soil column and several energy budgets might not be feasible.

Moreover, both AD and JP stressed the importance of hydrological connectivity. It is a limitation that we in SECHIBA do not have this lateral connection. AD suggested that some of the routing scheme could be implemented in hydrol to aid in this, but she was not sure that all the complexity of the routing scheme is needed, then higher resolution might be better.

Structure of the code Two possibilities were mentioned: (i) move loops of number of energy budgets etc. inside the module and add dimensions to the variables, or (ii) completely restructuring the code such that we have no module variables and all variables are managed from a centralized place.

Limitations Will speed be a limitation when adding new dimension? Is there any part of the code (hydrology, soil thermics or energy) that would slow it down substantially, thus serving as an argument to limit the flexibility? Has any recent profiling of the code been conducted that could answer this question? If we will move forward with CN-CAN the answer is the two-way radiative transfer scheme.

What will the limitations be of computational memory, when adding new dimensions?

Nomenclature in ORCHIDEE We all agreed to redefine the usage of PFT and NOBIO in the code and include them all in an array of Surface Functional Types (SFT). Questions were raised to the nomenclature for the grouping SFTs for energy budgets and soil columns. Some suggestions were given such as Eco-Hydrology grouping/unit, EHU, EFT etc., but nothing has yet been decided.

Who and Where? Which model version should the multi-tiling be implemented in? ORCHIDEE-CN-CAN seems the most obvious choice when taking the time perspective into account. Who should do/lead the work on implementing multi-tiling in ORCHIDEE? Ideally, we should acquire funding for someone to work on the implementation supervised by JG or other permanents. But before we apply for funding, we should get a clearer picture of what we need. But the implementation of SFTs, could possibly be done by VB or others.

Next steps

• As a first step, we agree to proceed in 2 stages and to start by the separation of the NOBIO from PFTs. This part should be quite easy since NOBIO concerns only SECHIBA, and will allow to start the redefinition of PFTs in SFTs and to continue the current developments of lakes.
• Brain storming phase during the coming months. A shared document will be created for everyone to comment on (ORCHIDEE wiki could be an option).
• This should help us obtaining a better idea of what we need, before asking for funding to be used for future changes of code and structure of ORCHIDEE in relation to multi-tiling implementations.
• Next meeting will be in the fall, where the shared document should aid us in moving forward with the discussion.