Calculating area of rasterized square

[trygvels]

Hi!
First of all, I very much appreciate your work, it is great!

My question is, is it possible to calculate the area of the polygon that one is gridding using the make_geocube method and save it to the data-column?

Once I convert my geometry-column I lose all information about the area of the layer, but I would like to know what percentage of a particular square was filled with my geometry layer.

Cheers,
Trygve

[snowman2]

It is likely not easy to get. One method would be to polygonize the raster cells and then intersect them with the input geometry. The output of that could be used to calculate the overlap percentage.

[snowman2]

What you have makes sense to me. The only thoughts I have are to try shapely 2.0 or pygeos with geopandas and/or try https://github.com/geopandas/dask-geopandas.

[trygvels]

I managed to get a few more speedups with the following code (Posting here for anyone who might have a similar problem):

def dissolve_shuffle(ddf, by=None, **kwargs):
    """Shuffle and map partition"""
    meta = ddf._meta.dissolve(by=by, as_index=False, **kwargs)
    shuffled = ddf.shuffle(
        by, npartitions=ddf.npartitions, shuffle="tasks", ignore_index=True
    )
    return shuffled.map_partitions(
        gpd.GeoDataFrame.dissolve, by=by, as_index=False, meta=meta, **kwargs
    )

def calculate_area_overlap_dask(area_name, df_grid, layer, nthreads=8,):
    # Reduce dataframe
    layer = layer[["geometry"]]
    df_grid = df_grid[["geometry"]]

    # Overlay layer on grid
    _overlay = gpd.overlay(layer, df_grid, how="intersection")
    _overlay = dg.from_geopandas(_overlay, npartitions=nthreads)

    # calculate area of split layer
    _overlay["area"] = _overlay.area
    _overlay = _overlay.compute()
    df_grid = dg.from_geopandas(df_grid, npartitions=nthreads)

    # assign smaller chunks to respective squares
    joined = dg.sjoin(df_grid, _overlay, how="inner").reset_index()

    # sum polygons in grid using smart suffle
    scored_grid = dissolve_shuffle(joined, "index",)
    scored_grid = scored_grid.compute()
    scored_grid = scored_grid.set_crs(25833)
    return scored_grid

This sped it up quite a bit. I am currently looking into your suggestions, but does shapely 2.0 support GeoPandas? I get a bunch of errors if a upgrade it.

[snowman2]

Possible that geopandas hasn't made the updates to be compatible yet. I see this PR: geopandas/geopandas#2570

[trygvels]

Hmm, it seems different to the issue I'm encountering.
I get AttributeError: module 'shapely.geos' has no attribute 'lgeos', so there seems to be some in PYGEOS_SHAPELY_COMPAT, which is called when I read a file with Fiona as the engine.

[snowman2]

This may be the PR: geopandas/geopandas#2275