Functions to Clip Point, Line and Polygon Data Like Those that exist in other spatial GIS tools

Hi geopandas dev team:

The Issue / Question That I have

I am trying to come up with a consistent way to “clip” data that is in line with a typical GIS (think arcgis or qgis) workflow as follows:

• Clip takes a point, line or polygon feature to be clipped and a boundary polygon as the clipping extent
• Clip outputs a point, line or polygon feature that has ONLY features contained in the boundary polygon. Each feature if it’s a line or poly is thus “CLIPPED” to that extent if that is required.
• Clip maintains the attributes of the clipped point, line or polygon features

On the back end - clipping points, lines or polygons is actually quite different. On the front end however - to a user - it is the “same” input and output. input: some (shapefile) i want to clip to an extent output: some clipped shapefile

What I tried initially - overlay()

Initially i tried to use overlay() with the how="intersection" argument. however that approach only seems to support polygon features. Did I do something wrong? I spent some time on this.

What I did / Tried

I am trying to create a simpler way for my students to “clip” a point, line or polygon layer using a boundary polygon object that exposes less of the back end differences between these object types when processed.

Below i’ve created a set of functions to make up a parent function that clips points, lines or polygons. I wondered if you could have a look at my code to help me determine if my approach is the most efficient one. I haven’t built any tests – yet but will.

Question 2 – If applicable would this fit in geopandas? Or is there a better approach?

My next question if you don’t mind and if i am able to find an approach that fits the project – is if the geopandas project is open to such a function housed in geopandas? It is a task I and my colleagues do often!

Any feedback is appreciated. I do still owe you another PR for more documentation surrounding dissolve as well.

The Functions I Wrote

# Create function to clip point data using geopandas


def clip_points(shp, clip_obj):
    '''
    Docs Here
    '''

    poly = clip_obj.geometry.unary_union
    return(shp[shp.geometry.intersects(poly)])

# Create function to clip line and polygon data using geopandas


def clip_line_poly(shp, clip_obj):
    '''
    docs
    '''

    # Create a single polygon object for clipping
    poly = clip_obj.geometry.unary_union
    spatial_index = shp.sindex

    # Create a box for the initial intersection
    bbox = poly.bounds
    # Get a list of id's for each road line that overlaps the bounding box and subset the data to just those lines
    sidx = list(spatial_index.intersection(bbox))
    shp_sub = shp[shp.index.isin(sidx)]

    # Clip the data - with these data
    ne_roads_sub['geometry'] = ne_roads_sub.intersection(poly)

    # Return the clipped layer with no null geometry values
    return(ne_roads_sub.geometry.notnull())

## I just updated the above as I realized I could do this with one less line by updating the geometry col

# Final clip function that handles points, lines and polygons


def clip_shp(shp, clip_obj):
    '''
    '''
    if shp["geometry"][0].type == "Point":
        return(clip_points(shp, clip_obj))
    else:
        return(clip_line_poly(shp, clip_obj))

[jorisvandenbossche]

I think there are two separate issues here:

• overlay can definitely be improved to work with other geometries than polygons. See also the last bullet point in https://github.com/geopandas/geopandas/issues/706, and I seem to remember we were recently talking about this (maybe on gitter?).

• Apart from overlay, I think it is still valuable to have a separate clip function. It is conceptually simpler as a full overlay (you “just” want to clip the left geodataframe by the right polygon, and not “combine” it with the right geodataframe). See also https://github.com/geopandas/geopandas/issues/956 and https://github.com/geopandas/geopandas/issues/921. @lwasser I think the code (and example) you show above from earthpy looks really good (that is also more or less how I would implement a clip function). Would you like to contribute this clip module to geopandas?

[martinfleis]

I will reopen this issue as it is something I feel geopandas should support. Preferably as an enhancement of overlay to work on all types of geometries.