Treatment and control cell sampling

[alanalutz]

Description

Implements a new sampling strategy for generating 1km2 treatment and control cells, to replace src/psm/ps_grid_creation.py. Generates a new set of treatment and control cells for all 30 test PAs.

Treatment cells are a grid of all fully interior 1km2 cells within a PA (if the PA is small), or a random sample of fully interior km2 cells (if the PA is large). Control cells are a random sample of unprotected 1km2 cells that are within a given distance of a PA, but outside of a given buffer distance from the PA. They must also fall fully outside a given buffer distance from any PA.

The PSM model is updated to use these newly generated grid cells instead of the old ones from psm_grid_creation.py

Created, updated, or replaced Files and/or tables

• src/psm/get_interior_cells.py: Iterates through a set of PAs and generates a set of valid interior (treatment) 1km2 cells for each PA. If the PA is <500 km2, a grid of all valid interior cells is generated. If the PA is >500 km2, a random sample of valid interior cells is generated.
• src/psm/get_exterior_cells.py: Iterates through a set of PAs and generates a random sample of valid exterior (control) 1km2 cells for each PA. These are unprotected cells that fall within a given distance of a PA, but outside of a given buffer distance from that PA.
• notebooks/get_exterior_cells.ipynb: Test notebook for generating exterior (control) cells; was converted to src/psm/get_exterior_cells.py.
• notebooks/ps_model.ipynb: Updated to use the newly generated treatment and control cells instead of the ones generated by ps_grid_creation.py

Notes for reviewers

I would appreciate close review on this PR, since this sampling approach is used (largely unchanged) in the current TPARA methodology.

_{Stack created with GitHub Stacks CLI • Give Feedback 💬}

[cjthomas730]

Can you please address this point, I'm not sure if I'm missing something in the CRS setting, or if this is just a nitpick