Additional Model Validation
Additional Model Validation
If resources permit, additional steps may be taken to do a more rigorous validation of the model. These options are listed below, in order from least resource-intensive to most resource-intensive.
- Re-run the model using another modeling tool having a similar underlying approach and compare the two sets of results.
- Example: If the modeling tool you initially used is built on a least-cost path approach (e.g., ArcGIS’ least-cost path functions), choose another least-cost path tool (e.g., one of the components in the Connectivity Assessment Toolkit) and run your inputs through the second tool to get another set of results for comparison.
- Re-run the model with different settings and a smaller spatial extent and compare the two sets of results for consistency. Note that it will be important to really understand how the modeling tool works in order to determine whether differences between the two sets of results are significant or not.
- Conduct a detailed validation against field data or new field work to confirm that some subset of the areas of connectivity shown in the model results are actually being used as predicted by the model.
- Example: If a corridor through gas fields has been identified for pronghorn and land management entities are taking management action around this, it will be important to confirm pronghorn are actually using the area as hoped (e.g., through the use camera traps).
If the model validation shows inconsistent or unexpected results, those results should be examined carefully to understand whether and how to adjust the model inputs or parameters to re-run it and obtain a meaningful set of results. To re-iterate, it is important to be careful to not “over-calibrate” the model to match expectations, and to consider whether unusual results are in fact a reasonable possibility for the species being assessed before making adjustments to the model.