I have been tasked with subsampling a data set of cameras to determine whether we can get away with fewer cameras in our camera grid. The dataset already has detection rates for each species at each subplot. I have never subsampled before in R and was wondering how best to go about this? Hoping to do this in R. Thanks for the help in advance!
Havent tried anything quite yet. Don't know where to start.
This is more of a process problem than an R problem. You need to decide a whole bunch of things before you make decisions and having meaningful information to do it is critical.
The easiest way to approach this, in my mind, is geospatial analysis.
Create coverages for each camera, angles and distances and turn those into some kind of a geospatial entity, a polygon or buffer.
Create a set of locations (approximate is good) of each event, with a species designator. Map them out with different colored markers.
Spend some time comparing the known events by species with current camera locations.
Then you can consider where to place the least amount of cameras to get the greatest coverage.
I would likely do a viewshed analysis that is tuned to the angle of view and resolution and such of your camera, using the topography, any objects with elevations that might obscure your view as well as the points you have entered to assess how to best configure the cameras.
There is a package called windfarmGA in R for the viewshed itself. See the method here:
This is how I would proceed because it allows you to know for sure if your current sightings would have been detected in a different configuration.
You could also do some kind of simulation, but I feel like this path properly executed would be the most reliable. And there are many Geospatial packages in R to help you build the maps and layers needed.
Or you could use QGIS and do the work in there.
