Appendix: Build a model of your own system
Background Readings:
- This primer: at least chapters 1-3; chapters 5-6, 8-9 are also useful.
- Soetaert and Hermann 2009. All of it is useful, but especially the chapters used in this primer.
Steps
- Identify the scope and focus of your model, including,
- state variables (with units),
- spatial boundaries,
- time scale, or temporal boundaries,
- inputs and outputs for the system,
- fluxes between state variables,
- the state variables upon which the fluxes depend (e.g., litterfall from trees to soils depends directly on only trees but not soils); this gives you the order of the rate equations.
- Create a budget, including,
- amount in each state variable,
- annual fluxes between state variables; if you can’t find these in the literature, figure out ways to get very rough approximate estimates.
- Write out a model:
- for each flux, write an explicit mathematical function for each process in that flux (see, e.g., the NPZD model). (You will estimate rate paramters later.)
- for each state variable, write an explicit rate equation using the functions for the fluxes in the previous step.
- estimate each parameter in one of these ways:
- use algebra to calculate instantaneous rates from annual fluxes and the magnitude of the state variables upon which the fluxes depend, or
- if there is an estimate in relevant literature for your particular rate parameter, use that, or
- use both approaches and try them both out.
- Code the model in R, including,
- writing an ODE function, a vector of parameters; start with three or four years only.
- run and then troubleshoot the model.
- Consider including data from an external source (chap. 9)
- Perform sensitivity analysis (chap. 6)
- Consider letting R try to fit rate parameters through optimization (chap. 8).
- Present your findings in a simple written form and in a short oral presentation.
- Five minute (5 min) presentation should include:
- Your question.
- Introduction to the ecological system, including the items listed in 1. above related to the scope of your model.
- Clearly labelled model diagram (boxes and arrows)
- Corresponding system of model equations.
- Explanation of parameter estimates.
- Lessons from the sensitivity analysis.
- Findings and conclusions.
- One page written explanation of the model that includes all the elements in your presentation. Use elements of your slides as appropriate.