How ecological/environmental/intervention models will be organized?

[Zapiano]

Context/common ground for the discussion:

Currently within ADRIA.jl there is an ecological model (ADRIAmod) that can be used to run many scenarios, combining the effect of distinct Environmental, Intervention and Ecological factors. That allows us to do all sorts of analysis - such as sensitivity and rule extraction, to cite just two.

Recently we developed a new model for the coral growth/survival (which lives in a separate repository and will have a distinct name from ADRIAmod). The new model doesn't account for mortality due to disturbances (such as heat stress or cyclones), nor does it model interventions (such as seeding or fogging). The new model also doesn't implement natural adaptation due to heat stress mortality. All those things are implemented within ADRIA.

The way things are set now, ADRIA has a function called run_model that triggers several calls to run_scenario. Inside the run_scenario function we have an initial setup and then we run a sequence of timesteps. As part of each timestep, a call is made to the new model.

Discussion

Since ADRIA is supposed to be an analysis tool, that in theory could be used with any coral/ecological model, it would be interesting to decouple ADRIAmod from ADRIA. That includes all Environmental, Ecological (including Reproduction), Intervention and Heritability related mechanics. Pretty much the entire run_model function, plus some related functions. That excludes the decision module. Some possible solutions for that are

1. Move everything that we today call ADRIAmod to the same repo as the new model and give it a single name;
2. Move everything that we call ADRIAmod to a separate repo called.. ADRIAmod (or something else) and this model will interact with the new model the same way as today;
3. Keep this inside ADRIA, maybe move this to a separate module or so;
4. Something else;

[Zapiano]

@ConnectedSystems @DanTanAtAims and me had an initial discussion about this in an online meeting./

Conclusions

• We don't want to bring "everything that we call ADRIAmod" to the new model.
  • The new model only represents growth/survival dynamics, and we don't want to mix that with the other dynamics represented by ADRIAmod
• We agree that it would probably be a good idea to move ADRIAmod outside ADRIA.jl repo
• To make this decision, we need a diagram (or some diagrams) of the distinct mechanics implemented by ADRIAmod so we can have a clear vision

Actions

• In the next semester we won't change this (not our focus right now)
• We need to start drawing these diagrams as a first step in the next few months so we can plan the next steps