1. The demo modules in the SpaDES package

   • caribouMovement
     
   • fireSpread
     
   • randomLandscapes

2. Simple forest dynamics model (with caribou) (LCC2005)

3. SpaDES-Landis Biomass Succession

4. Wolf IBM (wolfAlps)

5. A shiny app on shinyapps.io (e.g., Proof of concept)

• The demo modules are simple, proof of concept modules.

• There are 3 different types of modules:

  • Event (fireSpread)
  • Agent (caribouMovement)
  • Data generation (randomLandscapes)

• How to find them:

# Locate SpaDES on your computer; they are in the sampleModule folder
whereSpaDES <- find.package(package  = "SpaDES")
whereSamples <- system.file('sampleModules', package = 'SpaDES')
file.path(whereSamples, 'caribouMovement', 'caribouMovement.R') %>%
  file.edit()

• More modules than SpaDES demo
• Vegetation component is simple, but data-driven (LCC 2005)
• Neither caribou, nor fire are data-driven

• During simulation, it shows:

  • maps
  • data summary plots
  • text legends
  • colour matching between map and histogram
  • modular plotting (each module is unaware of other plots)

• Freely available from the SpaDES-Modules repository
• This repository is intended to be like CRAN, but for SpaDES modules; makes them easy to distribute to others
• How to find it:

## specify where to save files
moduleDir <- file.path(tempdir(), 'SpaDES_modules')

## download all necessary modules
downloadModule('LCC2005', path = moduleDir, data = TRUE)

## look at module or helper .Rmd file
openModules('LCC2005', path = moduleDir) 

• This is a rewrite in R of a core LANDIS vegetation dynamics module
• Slightly different implementation, making it faster under many situations
• Cross-platform (LANDIS is Windows-only)

• Allows scientist easier access to code, e.g.,

  • making growth climate sensitive
  • changing mortality functions
  • dividing up large problems into small parallel problems

• Not yet published

Example results

• Features:

  • Shows a random location in western boreal forest
  • Uses shiny to make it an interactive web page
  • Shows maps
  • Data summaries
  • Pre-run results

  • Can be easily published online, once public:

    (top right)

Running the app

shiny::runApp("~/GitHub/LandwebApp/")

• Agent based model of demography and dispersal of wolves in the Italian Alps
• R port of the original SELES model that was used in (Marucco and McIntire 2010)
• Can run this on your computer, in R

• Features:

  • Incremental visuals
  • One module
  • Progress bar
  • Caching (automatic saving)
  • Parallel simulations
  • Summarizing across parallel runs

• Go to model

• Freely available from the SpaDES-Modules repository

downloadModule("wolfAlps", path = moduleDir, data = TRUE)

• The .Rmd file shows how to run it:

openModules("wolfAlps", path = moduleDir)

https://spades.shinyapps.io/ForestChange_ProofOfConcept/

• Features:

  • Uses the shiny package to create dynamic web page

  • hosted on shinyapps.io because it's very easy to do so (click on "Publish" button in RStudio):

    (top right)

  • Allows user to change parameters

  • Shows some ideas about modularity (click modules on and off), only partially implemented

During this workshop we will learn how to:

• break down ecological processes into discrete events;
• organize ecological processes into SpaDES modules;
• rapidly prototype and develop complex ecological models in SpaDES;
• parameterize and evaluate simulation models;
• perform scenario analyses, e.g., for decision support;
• build portable, reproducible models;
• share models, analyses, and code.

• package help docs (e.g., overview of SpaDES package):

?SpaDES

• package vignettes (link)
• SpaDES wiki (link)
• SpaDES user group (link)
• bug reports (link)

Thinking the SpaDES way (slides)