One of the steps in the experiments in the BCCVL is setting the constraint area. In this step you can select in which area you want to train your model. Only data from this area will be used in your model, even if your selected occurrence or true absence datasets contain records outside the selected constraint area. Note that if you are generating pseudo-absence or background records these will only be generated in the selected constraint area.
There are four different options to set the constraint area of your experiment:
1: Convex hull
The default constraint area in the BCCVL is a convex hull (= minimum polygon) around the selected occurrence records. You can add a buffer (offset) around this convex hull, which can be useful if your species is only present in a small restricted range. To do this you enter the amount of kilometres that you want to use as offset in the box, and click "Add offset". The blue outline will change from the convex hull to the convex hull + offset (see examples in image below).
In a Multi-Species Distribution Modelling experiment a convex hull will be generated for each species separately, and in a Migratory Modelling experiment a convex hull will be generated for each subset separately. The default constraint in a Climate Change experiment is the constraint area from your input SDM.
2: Pre-defined region
If you are interested in training your model in a particular area, you can use one of the pre-defined regions that we have available in the BCCVL. To do this you first select the region type. Currently we have four region types available: Local Government Areas (LGA), National Resource Management Regions (NRM), Australian States and Territories, and IBRA 7 regions. Once you have selected a region type, the list of regions will show up in the next drop down menu. At this moment you can only select 1 pre-defined region for your model. Once you have found the region you want to use, click "Add to map". You can also add a buffer around the pre-defined region.
3: Environmental Envelope
If you want to train your model across the entire geographic range of your selected climate or environmental data, you can use the Environmental Envelope option. The outer boundaries of the environmental envelope are visualised as a green box on the map. Note that if you use a climate/environmental dataset with a large extent, such as Australia or global, using this option will generate pseudo-absence or background points across that entire extent which might not be ideal if your species only occurs in a particular area. This option can however be useful if you have environmental data for a particular region.
4: Draw a polygon on the map
You can also draw your own polygon on the map to define the constraint area. To do this click "Draw Cosntraints On Map" and draw a polygon on the map.
If you clear any selection, your model will automatically use the convex hull.
Note that the results of your experiment will include two maps with the predicted probability of presence of your species:
- Constraint map: this map only shows predictions in the constraint area, where the model was trained.
- Unconstraint map: this map shows predictions across the entire geographic extent of your climate/environmental data (i.e. the environmental envelope). This is useful to see where your species could be present outside its current range (if you used the convex hull as constraint area) or outside your pre-defined region.
Please let us know if you have particular pre-defined regions that you would like to see added to the list. We are working on an option that lets you upload your own shape file to use as constraint area, so watch this space for more!