Update study area to account for exclusion zones

[brynpickering]

Reopening PR from old repo: timtroendle/possibility-for-electricity-autarky#7

Fix for #1

For completeness, original PR text is below.

--

Now an arbitrary number of exclusion zones can be defined in the model scope. These can only be rectangular.

I can't test this on the whole workflow, since I have eternal package conflicts, but the defined 'atlantic islands' exclusion zone leads to the following study area:

If defining a polygon that fits within the exclusion zone (e.g. small_hole = shapely.geometry.Polygon(((-25, 35), (-24, 35), (-24, 40), (-25, 40)))), then study_area.intersects(small_hole) and study_area.contains(small_hole) are both False.

If defining a polygon that partially fits within the exclusion zone (e.g. big_hole = shapely.geometry.Polygon(((-25, 35), (-10, 35), (-10, 45), (-25, 45)))), then study_area.intersects(big_hole) == True and study_area.contains(big_hole) == False.

MWE

import shapely.geometry
import shapely.errors
import geopandas as gpd

def _study_area(config):
    """
    Create a bounding box for the study area, and cut out holes for all defined
    exclusion zones. For plotting purposes, exclusion zones and the bounding box are
    defined in opposite orientations, see https://github.com/geopandas/geopandas/issues/951
    """
    if config["scope"].get("exclusion_zones", {}) and isinstance(config["scope"]["exclusion_zones"], dict):
        holes = [
            (
                (exclusion_zone["x_max"], exclusion_zone["y_min"]),
                (exclusion_zone["x_max"], exclusion_zone["y_max"]),
                (exclusion_zone["x_min"], exclusion_zone["y_max"]),
                (exclusion_zone["x_min"], exclusion_zone["y_min"])
            )
            for exclusion_zone in config["scope"]["exclusion_zones"].values()
        ]
    else:
        holes = []

    study_area = shapely.geometry.Polygon(
        ((config["scope"]["bounds"]["x_min"], config["scope"]["bounds"]["y_min"]),
         (config["scope"]["bounds"]["x_min"], config["scope"]["bounds"]["y_max"]),
         (config["scope"]["bounds"]["x_max"], config["scope"]["bounds"]["y_max"]),
         (config["scope"]["bounds"]["x_max"], config["scope"]["bounds"]["y_min"])),
        holes=holes
    )

    if study_area.is_valid is False:
        raise shapely.errors.TopologicalError(
            "Invalid study area geometry. "
            "Ensure that exclusion zones do not share a border with the study bounds."
        )
    else:
        return study_area

config = {'scope':{
    'bounds': {
        'x_min': -30, # in degrees east
        'x_max': 37,  # in degrees east
        'y_min': 30,  # in degrees north
        'y_max': 75,  # in degrees north
    },
    'exclusion_zones':{
        'atlantic_islands':{
            'x_min': -29.5, # in degrees east
            'x_max': -15,  # in degrees east
            'y_min': 31,  # in degrees north
            'y_max': 41  # in degrees north
        }
    }}
}

study_area = _study_area(config)

small_hole = shapely.geometry.Polygon(((-25, 35), (-24, 35), (-24, 40), (-25, 40)))
big_hole = shapely.geometry.Polygon(((-25, 35), (-10, 35), (-10, 45), (-25, 45)))

print(
    "contains small hole:", foo.contains(small_hole)
    "intersects small hole:", foo.intersects(small_hole)
    "contains big hole:", foo.contains(big_hole)
    "intersects big hole:", foo.intersects(big_hole)
)

[timtroendle]

Looks good, but some more tests would be good.

[timtroendle]

I am not convinced by this configuration mechanism (which I introduced and) which we are extending here. But I guess it may not the right moment to fix that.

[brynpickering]

Yeah, I didn't edit the config mechanism. Best to fix it universally in a different PR

[timtroendle]

Could this maybe move to utils.py and have a few unit tests (especially including the error that you are catching)?

In that case I'd not hand the config over to it, but rather a bounds object (probably a dict) and a exclusion zones object (probably list of dicts).

[timtroendle]

If you change the default.yaml, I think you should test whether the entire workflow still works. Alternatively, create another config file. Even then, it would be good to test the workflow.

The package conflicts should be fixed at the moment.

[brynpickering]

sure thing. I haven't checked the functioning of the workflow in this repo on any of these PRs, I have to admit...

[brynpickering]

@timtroendle I'm closing this in preference of #9, which also includes exclusion zone schema and tests