Map of Didi in Zarr format
A map built from OpenStreetMap data, ready to download
Customize the map
Open this place's area in the constructor and adjust everything to your needs
Open in constructor →Map passport
What this map contains by default
This is how the map is built as-is — every setting can be changed in the constructor.
.zarr
About Zarr
Map of Didi in Zarr — an export of real OpenStreetMap data for the selected area.
Zarr is a multi-dimensional array storage format for scientific data. It supports chunked storage, compression, and parallel access. Ideal for climate data, satellite time series, and geospatial data.
Which layers are in the file
This file for the selected area contains:
- Roads
- Buildings
- Water
Coordinate systems
WGS 84 (EPSG:4326), selectable in the wizard
Units
Meters (projected in the chosen coordinate system)
| Category | Cloud |
| File extension | .zarr |
| Tier | Open Beta |
How to use Zarr for Didi
- Python — zarr.open(zarr.ZipStore('map.zarr.zip'), mode='r') reads the archive without unpacking; the feature_count and land_use arrays then behave like ordinary NumPy arrays.
- xarray and Dask — unpack the directory and open it through open_zarr: the arrays are computed chunk by chunk and do not require holding everything in memory.
- Attributes — the root group attributes hold the coordinate system, the bounding box of the area, the grid step and the list of layers; from them the geographic referencing of the cells is reconstructed.
Programs that open Zarr
FAQ: Zarr for Didi
Why is Zarr better than GeoTIFF?
Zarr supports multi-dimensionality (time, bands), parallel access, and flexible compression.
Can Zarr be read over the network?
Yes, Zarr supports reading from S3, GCS, HTTP via fsspec.
Is it vectors or raster inside?
Raster: the area is rasterized into a grid where feature_count holds the number of objects in a cell and land_use holds the code of the dominant layer. Vector geometry is not carried over into Zarr; for that there are GeoParquet and FlatGeobuf.
What is the cell size?
0.001 degrees by default, roughly a hundred meters. Over small areas the step is refined automatically so the grid does not degenerate; the ceiling is 10,000 × 10,000 cells.
How do I open the archive without unpacking it?
Through ZipStore: zarr.open(zarr.ZipStore('map.zarr.zip'), mode='r'). For xarray with its open_zarr the directory is still better unpacked.
