I intend to model InVEST pollination for one country. The idea is to differentiate the model for five subregions. These subregions have all different relative bee abundance values.
Question: Is it possible to run the pollination model using five different input guild tables varying in the relative abundance values?
Hi @sibylles -
While the model will not take 5 different tables by default, I can think of two options:
1/ Run the model 5 times, with inputs that are specific to each subregion, then combine the results in post-processing.
2/ If you want to only use one model run: You could make new categories for the bees, like “Bombus - subregion 1”, “Bombus - subregion 2”, which would allow you to assign different parameters to them, based on subregion. Depending on which model parameters you’re changing for each subregion, you might also need to edit the land use raster, creating new categories that correspond to, for example “forest - subregion 1”, “forest - subregion 2” etc. That way you would be able to assign different values for nesting and floral resources etc.
Thank you very much vor the options. Yes indeed, post-processing is necessary.
Option 2 also needs post-prcessing, because InVEST will model Bombus-subregion 1 for the entire country etc. Considering the biophysical table I suppose it is ok, because I have the right values for each grid cell over the entire country.I higher degree of detail is not possible. This in contrary to the species table. I expect a strong improvement by a higher degree of detail (e.g. subregions in comparison to country relative species abundance as input).
Yes it would be nice to have one model run, because I have many model scenarios (climate scenarios, time frame, different crops), so building additional 5 models per model run…
Option 2 also needs post-prcessing, because InVEST will model Bombus-subregion 1 for the entire country etc.
That’s why I also mentioned altering the land cover map. So instead of having one “forest” class that covers the entire country, you could have 5 forest classes, where “forest - subregion 1” is only located in subregion 1, “forest - subregion 2” is only located in subregion 2 etc. The forest class itself wouldn’t change, but you would have finer granularity for the purpose of assigning pollination parameters, and could use a single land cover map and single set of parameter tables, and do a single model run.
Of course, it’s probable that you’ll need to do post-processing regardless, since that’s a typical part of modeling, so we produce results that are tailored to the project’s needs.