datacarpentry
lesson 3 -- include points of discussion
Lesson 3 includes a couple of questions for which it would be good to provide solutions in the lesson are to include in the instructor guide. e.g.,
- does it make sense to have 30m trees in Harvard?
- what's the difference in the output generated by overlay and manual raster math?
Related to https://github.com/datacarpentry/r-raster-vector-geospatial/issues/86
For the question about tree height, see #86 For the difference in output between overlay and manual raster math - I believe that there isn't a difference. The only difference should be in the speed of operations. Could I get confirmation from one of the @datacarpentry/geospatial-maintainers? @jsta @obrl-soil @r4space
My understanding is that manual raster math requires that the entire layer be loaded into memory. If you try manual raster math with very large layers you will run likely run out of memory. overlay solves this issue by performing calculations in chunks so the entire layer doesn't need to be loaded. I would guess for small rasters that manual raster math is faster but I'd be surprised if speed of operations was a priority for most people.
