Best practices for developing and coupling BMI-compliant models

[PeishiJiang]

Hello,

I have a general question regarding how to develop and couple BMI-compliant models? While it is a big question, any detailed suggestions and best practices would be much appreciated.

Suppose I have two models: one land surface model written in Fortran and one river routing model written in C. To couple both models and standardize the coupling process, my current immature plan is as follows:

1. Convert the land model with a Fortran-based BMI wrapper
2. Convert the routing model with a C-based BMI wrapper
3. Use Babelizer to bring the two BMI-compliant models into Python
4. Develop my own Python-based workflow to couple the two BMI-compliant models

I got a few questions regarding this plan:

• Do you know if there is any caveat that I need to pay attention to for this workflow?
• For step 4, can I use pymt or landlab rather than developing my own workflow? If so, how to put models into either system?
• If the land model uses parallel computing to run multiple cells simultaneously, will the BMI be compatible with this setup?

Thanks in advance!
Peishi