compatibility with PyLandStats

[mohdalattar]

I would be a really good idea to come up with a function that bridge between OSMnx and PyLandStats. Image I extract urban green spaces from OSMnx say

gdf = ox.geometries.geometries_from_place('Manhattan', tags={'leisure': 'park', 'landuse': 'grass'})

Meaning basic landscape metrics will be extremely beneficial to calculate patch indices for green spaces which is a common practice using imagery via Fragstats. I believe this would be a great extension for "Measuring Diversity" part [http://docs.momepy.org/en/stable/user_guide/weights/diversity.html]

[martinfleis]

As far as I remember, pylandstats is a raster-oriented library. momepy is a vector one. And the bridge between a GeoDataFrame coming from the OSMnx and a raster required by pylandstats is practically just a use of geocube's make_cube, not? So momepy is not involved at any point. I would say that raster is out of scope for our project and if we had some usage for that, I would try to send a PR elsewhere, like to pylandstats directly.

cc @martibosch

[martibosch]

Hello @mohdalattar,

yes, pylandstats works with categorical land use/land cover rasters. Therefore, while it is perfectly possible to use it with OSM data, the latter needs to be rasterized first (like @martinfleis mentioned, this can be done with make_cube or also with rasterio.

Otherwise, there is a tool named geograph (I have not used it myself) which can compute landscape metrics from vector data.

I hope this helps. Best,
Martí