urbanflow.RasterGrid

class urbanflow.RasterGrid(coordinate_reference_system='EPSG:32633', cell_size=1)

Bases: object

RasterGrid class. Stores the following information: 1. grid: an np.ndarray[np.uint8] with unsigned integer values. 2. transform an Affine transform object to transform points onto the grid 3. legend: a python dict describing the different grid values and their meaning.

Parameters:

  • coordinate_reference_system (str)

  • cell_size (float)

cell_size

Square cell width/height in CRS units

Type:

float

Interface
__init__
Type:

constructor

.save(filepath)
Type:

saving the grid, as small as possible

.laod(filepath)
Type:

load the grid, as quickly as possible

__init__(coordinate_reference_system='EPSG:32633', cell_size=1)

Constructor given a grid, transform, and legend

Parameters:

  • cell_size (float) – Square cell width/height in CRS units. Defaults to 1

  • coordinate_reference_system (str) – Name of the CRS of this grid. Defaults to EPSG:32633.

Methods

__init__([coordinate_reference_system, ...])

Constructor given a grid, transform, and legend

from_geojson_dataframes(buildings, streets, ...)

Rasterize buildings, streets, and canals of Venice into a RasterGrid.

load(filepath)

Create a RasterGrid object from a .npz compressed file.

save(filepath)

Save a RasterGrid to a .npz file.

to_image([scale, palette])

Convert a uint8 grid of states into a PIL Image.
classmethod from_geojson_dataframes(buildings, streets, canals, *, courtyards=None, auto_courtyards=True, coordinate_reference_system='EPSG:32633', cell_size=1, legend={'building': 2, 'canal': 3, 'courtyard': 4, 'ocean': 0, 'street': 1})

Rasterize buildings, streets, and canals of Venice into a RasterGrid.

Parameters:

  • buildings (GeoDataFrames) – All must share the same CRS.

  • streets (GeoDataFrames) – All must share the same CRS.

  • canals (GeoDataFrames) – All must share the same CRS.

  • cell_size (float) – Square cell width/height in CRS units.

  • legend (Dict[str, int]) – A python dict that contains what cell values contian what. The dict must define: “ocean” : (default 0) “street” : (default 1) “building” : default 2) “canal” : (default 3)

  • courtyards (GeoDataFrame)

  • auto_courtyards (bool)

  • coordinate_reference_system (str)

Return type:

RasterGrid object with grid, transform, and legend.

classmethod load(filepath)

Create a RasterGrid object from a .npz compressed file.

save(filepath)

Save a RasterGrid to a .npz file.

Parameters:

filepath (str)

to_image(scale=1, palette=None)

Convert a uint8 grid of states into a PIL Image.

Parameters:

  • grid (ndarray[int] shape (rows, cols)) – Cell codes: 0=ocean, 1=street, 2=building (extend as you wish).

  • scale (int, default 1) – How many output pixels per cell (must be ≥1). 2 doubles width/height.

  • palette ({state: (R, G, B)}, optional) – Custom colors. Unspecified states default to white.

Return type:

PIL.Image.Image (mode “RGB”)

Notes

  • Uses pure NumPy → very fast, even for multi-million-cell grids.

  • Nearest-neighbour up-scaling keeps the crisp blocky CA look.