Skip to content

Latest commit

 

History

History
165 lines (127 loc) · 5.56 KB

README.md

File metadata and controls

165 lines (127 loc) · 5.56 KB
Raw

swemaps

Maps of Sweden in GeoParquet for easy usage.

The parquets have been created from files published by Statistics Sweden and The Swedish Agency for Economic and Regional Growth. The map data includes counties, municipalities and FA regions. The original geometries have been transformed from SWEREF 99 TM (EPSG:3006) to WGS 84 (EPSG:4326) for better out-of-the-box compatibility with interactive and web-based toolkits such as Folium and Plotly. The column names have also been somewhat sanitized (e.g. KnKod -> kommun_kod).

The package gets you the file path so that you can load it with your prefered tool, for example PyArrow or GeoPandas. An extra convenience function is included to quickly convert a PyArrow Table object to GeoJSON.

Made for Python with inspiration from swemaps2.

Municipalities and counties

Municipalities Counties
municipalities counties

PyArrow example with Plotly

>>> import plotly.express as px
>>> import pyarrow.parquet as pq
>>> import swemaps

# Load the map for the specified type
>>> kommuner = pq.read_table(swemaps.get_path("kommun"))

>>> kommuner.column_names
['kommun_kod', 'kommun', 'geometry']

# The convenience function returns GeoJSON from a PyArrow table object
>>> geojson = swemaps.table_to_geojson(kommuner)

# Here's a dataframe with municipalities and some random values that we can plot
>>> df.head()
shape: (5, 2)
┌──────────┬───────┐
│ Kommun   ┆ Value │
│ ---      ┆ ---   │
│ str      ┆ i64   │
╞══════════╪═══════╡
│ Ale      ┆ 544   │
│ Alingsås ┆ 749   │
│ Alvesta  ┆ 771   │
│ Aneby    ┆ 241   │
│ Arboga   ┆ 763   │
└──────────┴───────┘

# Use Plotly to create a choropleth using the dataframe and GeoJSON
>>> fig = px.choropleth(
        df,
        geojson=geojson,
        color="Value",
        locations="Kommun",
        featureidkey="properties.kommun",
        projection="mercator",
        color_continuous_scale="Viridis",
        fitbounds="locations",
        basemap_visible=False,
    )

You might want to subset the map of municipalities for a specific county or a group of counties. Since the geometry is loaded as a PyArrow table the filter operation is straightforward.

>>> import pyarrow.compute as pc

>>> kommuner.schema 

kommun_kod: string
kommun: string
geometry: binary
  -- field metadata --
  ARROW:extension:metadata: '{"crs":{"$schema":"https://proj.org/schemas/' + 1296
  ARROW:extension:name: 'geoarrow.wkb'
-- schema metadata --
geo: '{"version":"1.1.0","primary_column":"geometry","columns":{"geometry' + 1621

# County code for Skåne is 12
>>> kommuner = kommuner.filter(pc.starts_with(pc.field("kommun_kod"), "12"))

>>> geojson = swemaps.table_to_geojson(kommuner)

You could also use list comprehension on the GeoJSON to filter it.

>>> geojson["features"] = [
        feature
        for feature in geojson["features"]
        if feature["properties"]["kommun_kod"].startswith("12")
        ]

Anyway, now we can plot Skåne.

>>> skane = px.choropleth(
        df,
        geojson=geojson,
        color="Value",
        locations="Kommun",
        featureidkey="properties.kommun",
        projection="mercator",
        color_continuous_scale="Viridis",
        fitbounds="locations",
        basemap_visible=False,
        title="Skåne municipalities"
    )

skane.show()

skåne

GeoPandas and plotnine

Another possibility is to load the GeoParquet into a GeoDataFrame.

>>> import geopandas as gpd

>>> gdf = gpd.GeoDataFrame.read_parquet(swemaps.get_path("lan"))

>>> gdf.head()

lan_kod            lan                                           geometry
0      01     Stockholms  MULTIPOLYGON (((17.24034 59.24219, 17.28475 59...
1      03        Uppsala  MULTIPOLYGON (((17.36606 59.61224, 17.35475 59...
2      04  Södermanlands  MULTIPOLYGON (((15.95815 58.96497, 15.8613 58....
3      05  Östergötlands  MULTIPOLYGON (((14.93369 58.13112, 14.89472 58...
4      06     Jönköpings  MULTIPOLYGON (((14.98311 57.9345, 15.00458 57....

# And with matplotlib installed as well we can have quick look
>>> gdf.plot()

län

For best results with plotnine you can either reproject to SWEREF 99 TM or set the aspect ratio in coord_fixed(). A ratio of 1.96 should be near optimal.

>>> gdf = gpd.read_parquet(swemaps.get_path("kommun"))

# Insert some random values
>>> gdf["value"] = np.random.randint(1, 600, size=len(gdf["kommun"]))

# Reproject back to SWEREF 99 TM
>>> gdf = gdf.to_crs(epsg=3006)

>>> (
    ggplot(gdf, aes(fill="value"))
    + geom_map(show_legend=False)
    + coord_fixed() # Or skip the reprojection above and use ratio=1.96 here
    + scale_fill_cmap("YlGnBu")
    + theme(
        axis_ticks=element_blank(),
        panel_background=element_rect(fill="white"),
        axis_text_x=element_blank(),
        axis_text_y=element_blank(),
    )
  )
SWEREF 99 TM WGS 84
sweref99tm wgs84