pymdurs

Python bindings for the Rust library rsmdu — a high-performance reimplementation of pymdu (Python Urban Data Model). Geospatial data processing for urban analysis, with integration with IGN (Institut Géographique National) APIs and UMEP toolchains.

• PyPI package: pypi.org/project/pymdurs
• Latest version:

Shadow	Mean radiant temperature (Tmrt)	Thermal comfort (UTCI)

📋 Table of Contents

• Installation
• Quick Start
• Core Classes
• Geometric Data Modules
• Requirements
• UMEP Integration
• Examples
• API Reference
• Versions and releases

Installation

Install uv

uv is the Python package manager used by this project (see uv.lock). Recommended installation via the standalone Astral installer:

macOS / Linux:

curl -LsSf https://astral.sh/uv/install.sh | sh

Windows (PowerShell):

powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

Other methods: Homebrew (brew install uv), pipx (pipx install uv), or PyPI.

After installation, restart your terminal. Update: uv self update (standalone installer only).

Create a virtual environment:

uv venv .venv --python 3.13

From PyPI (recommended)

The package is published on PyPI. To install the latest version:

uv pip install pymdurs

Optional morphometrics stack (DetectionUrbanTypes):

uv pip install "pymdurs[urban]"

On Windows, the PyPI wheel bundles GDAL, GEOS, and PROJ via auditwheel repair — no system GDAL installation is required.

For a specific version (see the PyPI release history):

uv pip install "pymdurs==<version>"

From source

GDAL prerequisites (build only — not required for pip install pymdurs on Windows):

Prerequisites: Rust (required to build):

Windows:

# Download and run rustup-init.exe from https://rustup.rs/
# Or use PowerShell:
Invoke-WebRequest -Uri https://win.rustup.rs/x86_64 -OutFile rustup-init.exe
.\rustup-init.exe

macOS:

curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

Linux:

curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

After installation, restart your terminal or run:

source $HOME/.cargo/env

Build (maturin develop)

With uv: uv run maturin develop --target <target>

macOS: linker can't find GDAL / "library 'gdal' not found"
If you upgraded GDAL or PROJ with Homebrew, the linker may still use old paths. Clean and point pkg-config at the current install, then rebuild:

cd pymdurs && cargo clean && cd ..
export PKG_CONFIG_PATH="/opt/homebrew/lib/pkgconfig"
maturin develop --target aarch64-apple-darwin

Build and install:

# Clone the repository
git clone https://github.com/rupeelab17/pymdurs.git
cd pymdurs

# Install maturin (Python-Rust build tool)
uv pip install maturin

# Build and install pymdurs
cd pymdurs

# For Apple Silicon (ARM64) - use native target
maturin develop --target aarch64-apple-darwin

# For Intel Mac (x86_64) - use default or specify target
maturin develop --target x86_64-apple-darwin

# Or let maturin auto-detect (may require rustup for cross-compilation)
maturin develop

Note: On Apple Silicon, if you get an error about missing x86_64-apple-darwin target, use --target aarch64-apple-darwin explicitly.

Quick Start

import pymdurs

# Create a BuildingCollection
buildings = pymdurs.geometric.Building(
    output_path="./output",
    defaultStoreyHeight=3.0
)

# Set bounding box (WGS84 coordinates)
buildings.set_bbox(-1.152704, 46.181627, -1.139893, 46.18699)

# Set CRS (optional, defaults to EPSG:2154 for France)
buildings.set_crs(2154)

# Download and process buildings from IGN API
buildings = buildings.run()

# Convert to pandas DataFrame
df = buildings.to_pandas()
print(df.head())

Core Classes

PyBoundingBox

Represents a geographic bounding box with min/max coordinates.

bbox = pymdurs.PyBoundingBox(min_x=-1.15, min_y=46.18, max_x=-1.14, max_y=46.19)

GeoCore / PyGeoCore

Base class providing common geospatial functionality (CRS, output paths, etc.).

# Access GeoCore from any geometric module
buildings = pymdurs.geometric.Building(output_path="./output")
geo = buildings.geo_core
print(f"CRS: EPSG:{geo.epsg}")
print(f"Output path: {geo.output_path}")

Geometric Data Modules

All geometric modules follow a similar API pattern:

1. Create an instance with output_path
2. Set bounding box with set_bbox(min_x, min_y, max_x, max_y) (WGS84)
3. Optionally set CRS with set_crs(epsg_code)
4. Run processing with run() or module-specific methods
5. Access results via get_geojson(), to_pandas(), or file paths

🏢 Building / BuildingCollection

Load and process building data from Shapefiles, GeoJSON, or IGN API.

import pymdurs

# Create BuildingCollection
buildings = pymdurs.geometric.Building(
    output_path="./output",
    defaultStoreyHeight=3.0
)

# Set bounding box (WGS84)
buildings.set_bbox(-1.152704, 46.181627, -1.139893, 46.18699)

# Set CRS (optional)
buildings.set_crs(2154)

# Download from IGN API and process
buildings = buildings.run()

# Convert to pandas DataFrame
df = buildings.to_pandas()

# Access GeoCore
geo = buildings.geo_core
print(f"CRS: EPSG:{geo.epsg}")

Features:

• Automatic height processing (storeys × default height or alternative height field)
• Mean district height calculation (weighted by area)
• Integration with pandas for tabular operations
• Support for multiple input formats (Shapefile, GeoJSON, IGN API)

🗻 DEM (Digital Elevation Model)

Download and process DEM data from IGN API via WMS-R.

import pymdurs

# Create Dem instance
dem = pymdurs.geometric.Dem(output_path="./output")

# Set bounding box (WGS84)
dem.set_bbox(-1.152704, 46.181627, -1.139893, 46.18699)

# Set CRS (optional)
dem.set_crs(2154)

# Run DEM processing (downloads from IGN WMS-R)
dem = dem.run()

# Get output paths
tiff_path = dem.get_path_save_tiff()
mask_path = dem.get_path_save_mask()

print(f"DEM saved to: {tiff_path}")
print(f"Mask saved to: {mask_path}")

Features:

• Automatic download from IGN WMS-R service
• GeoTIFF generation with proper CRS
• Mask generation for DEM boundaries
• Optional shape parameter for resampling

📋 Cadastre

Download cadastral parcel data from IGN API via WFS.

import pymdurs

# Create Cadastre instance
cadastre = pymdurs.geometric.Cadastre(output_path="./output")

# Set bounding box (WGS84)
cadastre.set_bbox(-1.152704, 46.181627, -1.139893, 46.18699)

# Set CRS (optional)
cadastre.set_crs(2154)

# Download from IGN API
cadastre = cadastre.run()

# Get GeoJSON data
geojson = cadastre.get_geojson()

# Save to GeoJSON file
cadastre.to_geojson(name="cadastre")

📊 IRIS (Statistical Units)

Download IRIS statistical units from IGN API via WFS.

import pymdurs

# Create Iris instance
iris = pymdurs.geometric.Iris(output_path="./output")

# Set bounding box (WGS84)
iris.set_bbox(-1.152704, 46.181627, -1.139893, 46.18699)

# Set CRS (optional)
iris.set_crs(2154)

# Download from IGN API
iris = iris.run()

# Get GeoJSON data
geojson = iris.get_geojson()

# Save to GeoJSON file
iris.to_geojson(name="iris")

🌳 COSIA (Land Cover)

Download COSIA land-cover data from the IGN API.

import pymdurs

# Create Cosia instance
cosia = pymdurs.geometric.Cosia(output_path="./output")

# Set bounding box (WGS84)
cosia.set_bbox(-1.152704, 46.181627, -1.139893, 46.18699)

# Set CRS (optional)
cosia.set_crs(2154)

# Download from IGN API
cosia = cosia.run_ign()

# Get output path
tiff_path = cosia.get_path_save_tiff()
print(f"COSIA raster saved to: {tiff_path}")

Note: COSIA data is downloaded as a raster TIFF. See examples/cosia_from_ign.py for a complete workflow including vectorization and conversion to UMEP format.

🛰️ LiDAR

Download and process LiDAR point cloud data from IGN WFS service.

import pymdurs

# Create Lidar instance
lidar = pymdurs.geometric.Lidar(output_path="./output")

# Set bounding box (WGS84)
lidar.set_bbox(-1.154894, 46.182639, -1.148361, 46.186820)

# Set CRS (optional)
lidar.set_crs(2154)

# Generate CDSM from vegetation and water classes
classification_list = [3, 4, 5, 9]  # Vegetation and water
lidar.run(file_name="CDSM.tif", classification_list=classification_list)

# Generate DSM from ground and buildings classes
classification_list = [2, 6]  # Ground and buildings
output_path = lidar.run(file_name="DSM.tif", classification_list=classification_list)

print(f"DSM saved to: {output_path}")
# Output contains 3 bands: DSM, DTM, CHM

Features:

• Downloads LAZ files from IGN WFS service
• Processes point clouds to create DSM, DTM, and CHM rasters
• Filters by LiDAR classification classes
• Outputs multi-band GeoTIFF files

LiDAR Classification Classes:

• 2 = Ground
• 3 = Low Vegetation
• 4 = Medium Vegetation
• 5 = High Vegetation
• 6 = Buildings
• 9 = Water

🏢 RNB (French National Building Registry)

Download building data from RNB API.

import pymdurs

# Create Rnb instance
rnb = pymdurs.geometric.Rnb(output_path="./output")

# Set bounding box (WGS84)
rnb.set_bbox(-1.152704, 46.181627, -1.139893, 46.18699)

# Set CRS (optional)
rnb.set_crs(2154)

# Download from RNB API
rnb = rnb.run()

# Get GeoJSON data
geojson = rnb.get_geojson()

# Save to GPKG file
rnb.to_geojson(name="rnb")

🛣️ Road

Download road segment data from IGN API.

import pymdurs

# Create Road instance
road = pymdurs.geometric.Road(output_path="./output")

# Set bounding box (WGS84)
road.set_bbox(-1.152704, 46.181627, -1.139893, 46.18699)

# Set CRS (optional)
road.set_crs(2154)

# Download from IGN API
road = road.run()

# Get GeoJSON data
geojson = road.get_geojson()

# Save to GeoJSON file
road.to_geojson(name="road")

🌳 Vegetation

Calculate vegetation from IGN IRC images using NDVI (Normalized Difference Vegetation Index).

import pymdurs

# Create Vegetation instance
vegetation = pymdurs.geometric.Vegetation(
    output_path="./output",
    write_file=False,
    min_area=0.0
)

# Set bounding box (WGS84)
vegetation.set_bbox(-1.152704, 46.181627, -1.139893, 46.18699)

# Set CRS (optional)
vegetation.set_crs(2154)

# Process vegetation (downloads IRC, calculates NDVI, filters)
vegetation = vegetation.run()

# Get GeoJSON data
geojson = vegetation.get_geojson()

# Save to GeoJSON file
vegetation.to_geojson(name="vegetation")

Features:

• Downloads IRC (Infrared Color) images from IGN API
• Calculates NDVI = (NIR - Red) / (NIR + Red)
• Filters pixels with NDVI < 0.2
• Polygonizes raster and filters by minimum area

💧 Water

Download water body data from IGN API.

import pymdurs

# Create Water instance
water = pymdurs.geometric.Water(output_path="./output")

# Set bounding box (WGS84)
water.set_bbox(-1.152704, 46.181627, -1.139893, 46.18699)

# Set CRS (optional)
water.set_crs(2154)

# Download from IGN API
water = water.run()

# Get GeoJSON data
geojson = water.get_geojson()

# Save to GeoJSON file
water.to_geojson(name="water")

🌡️ LCZ (Local Climate Zone)

Load Local Climate Zone data from external sources.

import pymdurs

# Create Lcz instance
lcz = pymdurs.geometric.Lcz(output_path="./output")

# Set bounding box (WGS84)
lcz.set_bbox(-1.152704, 46.181627, -1.139893, 46.18699)

# Load from URL (zip file containing shapefiles)
lcz = lcz.run()

# Get GeoJSON data
geojson = lcz.get_geojson()

# Get LCZ color table
table_color = lcz.get_table_color()

# Save to GeoJSON file
lcz.to_geojson(name="lcz")

Features:

• Loads LCZ data from zip URLs
• Built-in LCZ color table (17 LCZ types)
• Spatial filtering by bounding box
• Shapefile support (requires GDAL)

📐 DXF → Cosia (Python helper)

Convert ENON-style DXF layers to shapefiles and Cosia-weighted overlays (pure Python on pymdurs.geometric).

from pymdurs.geometric import dxf_to_polygon_shp, dxf_to_cosia_and_weighted_layers

gdf = dxf_to_polygon_shp("paysage.dxf", "output/dxf_polygons.shp")
weighted = dxf_to_cosia_and_weighted_layers(
    "output/dxf_polygons.shp",
    "output/dxf_cosia_weighted.shp",
)

Features:

• Tree layers (ENON-arbres*) → circumscribed circles
• Layer → Cosia class + overlay weight (LAYER_PROPERTIES)
• Shared color table: pymdurs.geometric.TABLE_COLOR_COSIA / pymdurs.geometric_helpers

See examples/dxf2shp_example.py.

🏙️ DetectionUrbanTypes (Python helper)

Cluster buildings into urban morphotypes (momepy morphometrics + Clustergram). Requires optional extra urban.

from pymdurs.geometric import DetectionUrbanTypes

detection = DetectionUrbanTypes(output_path="./output")
detection.set_bbox(-1.152704, 46.181627, -1.139893, 46.18699)
detection.set_crs(2154)
detection = detection.run(nbr_cluster=4)
gdf = detection.to_gdf()
detection.to_gpkg("detection")

uv pip install "pymdurs[urban]"

See examples/detection_urban_types_example.py.

Requirements

Python

• Python >= 3.8
• pandas >= 1.0.0
• numpy < 2.0.0 (for compatibility with numexpr and other dependencies)

Note: If you encounter NumPy 2.x compatibility issues, install NumPy 1.x:

uv pip install 'numpy<2.0.0'

Optional Dependencies

For advanced workflows and examples:

# Geospatial operations
uv pip install geopandas rasterio pyproj shapely

# For UMEP integration
uv pip install "solweig @ git+https://github.com/UMEP-dev/solweig.git"
uv pip install umep  # Optional

UMEP Integration

The full UMEP workflow runs in two steps under examples/, using the same output directory (./output/umep_workflow) and the same study area (La Rochelle bbox by default).

Step 1 — Land cover (cosia_from_ign.py)

Downloads the COSIA orthophoto from the IGN API, vectorizes polygons by RGB color, reclassifies them to UMEP format (buildings, vegetation, water, etc.), and produces:

• landcover.tif — land-cover raster compatible with SOLWEIG
• terrain.shp, terrain.geojson — vector subset (bare soil and impervious surfaces)

python examples/cosia_from_ign.py

Step 2 — Thermal analysis (umep_workflow_new.py)

Builds on step 1 outputs and collects the remaining urban data via pymdurs:

1. DEM from the IGN API
2. DSM / CDSM from IGN LiDAR (WFS)
3. Clip rasters to the mask (DEM_clip.tif, DSM_clip.tif, CDSM_clip.tif, landcover_clip.tif)
4. SOLWEIG (solweig): sky view factor (SVF), Tmrt, thermal comfort (UTCI)

umep_workflow_new.py expects landcover.tif in ./output/umep_workflow/. Without that file, the SOLWEIG step is skipped (a warning is printed).

# On Apple Silicon (ARM64), add the x86_64 Rust target first:
rustup target add x86_64-apple-darwin

# Install solweig:
uv pip install "solweig @ git+https://github.com/UMEP-dev/solweig.git"

# Run the workflow (after cosia_from_ign.py):
python examples/umep_workflow_new.py

Note: solweig currently requires the x86_64-apple-darwin Rust target even on Apple Silicon — a limitation of the solweig package itself. Place a weather EPW file (e.g. la_rochelle_2025.epw) in examples/ for the SOLWEIG step.

Main outputs: clipped rasters, PNG/GIF previews, Tmrt/UTCI time series in ./output/umep_workflow/.

Examples

Comprehensive examples are available in the examples/ directory:

• Basic usage: building_basic.py
• IGN API integration: building_from_ign.py, dem_from_ign.py, cadastre_from_ign.py, etc.
• LiDAR processing: lidar_from_wfs.py
• COSIA workflow: cosia_from_ign.py
• DXF landscape → Cosia: dxf2shp_example.py
• Urban morphotype clustering: detection_urban_types_example.py (requires pymdurs[urban])
• UMEP workflow: umep_workflow_new.py (complete urban analysis workflow)

See examples/README.md for detailed documentation of all examples.

Typage (IDE)

Le package est annoté (PEP 561) via py.typed et des stubs .pyi générés depuis les bindings PyO3.

Après modification d’un binding Rust :

./scripts/generate-stubs.sh

Vérification locale :

uv run basedpyright pymdurs examples

Les stubs commités (pymdurs/pymdurs.pyi, pymdurs/geometric/, pymdurs/thermal/) sont contrôlés en CI (workflow type-stubs.yml).

API Reference

Common Methods

All geometric modules share these common methods:

set_bbox(min_x: float, min_y: float, max_x: float, max_y: float)

Set the bounding box in WGS84 (EPSG:4326) coordinates.

module.set_bbox(-1.152704, 46.181627, -1.139893, 46.18699)

set_crs(epsg: int)

Set the coordinate reference system (CRS) using EPSG code.

module.set_crs(2154)  # Lambert 93 (France)

geo_core: GeoCore

Access the GeoCore instance for CRS and path information.

geo = module.geo_core
print(f"CRS: EPSG:{geo.epsg}")
print(f"Output path: {geo.output_path}")

Module-Specific Methods

Building

• run() -> Building - Download and process buildings
• to_pandas() -> pandas.DataFrame - Convert to pandas DataFrame

Dem

• run(shape: Optional[Tuple[int, int]] = None) -> Dem - Download and process DEM
• get_path_save_tiff() -> str - Get DEM GeoTIFF path
• get_path_save_mask() -> str - Get mask shapefile path

Cadastre, Iris, Road, Rnb, Water, Vegetation

• run() -> Self - Download and process data
• get_geojson() -> dict - Get GeoJSON data
• to_geojson(name: str) -> None - Save to GeoJSON file

Cosia

• run_ign() -> Cosia - Download COSIA from IGN API
• get_path_save_tiff() -> str - Get COSIA raster path

Lidar

• run(file_name: str, classification_list: List[int]) -> str - Process LiDAR data
• Returns path to output GeoTIFF file

Lcz

• run() -> Lcz - Load LCZ data
• get_table_color() -> dict - Get LCZ color table

DXF helpers (Python)

• dxf_to_polygon_shp(...) - DXF → shapefile (trees as circles)
• dxf_to_cosia_and_weighted_layers(...) - Cosia-weighted overlay by layer

DetectionUrbanTypes (Python, optional urban extra)

• set_bbox / set_crs / run(nbr_cluster=4) / to_gdf() / to_gpkg(name)

Notes

API Aliases

Both Pythonic aliases and original class names are available:

• Building / PyBuilding
• Dem / PyDem
• Cadastre / PyCadastre
• Iris / PyIris
• Lcz / PyLcz
• PyBoundingBox
• GeoCore / PyGeoCore
• etc.

Coordinate Systems

• Input coordinates: Must be in WGS84 (EPSG:4326) for IGN API
• Default CRS: EPSG:2154 (Lambert 93) for French data
• Output CRS: Can be customized with set_crs()

IGN API Limitations

• Rate limiting: The IGN API may have rate limits
• Data availability: Some data may not be available for all areas
• Internet connection: Required for all IGN API operations

Versions and releases

The version is synchronized across pyproject.toml, pymdurs/Cargo.toml, and rsmdu/Cargo.toml.

• Set a version: ./scripts/set-version.sh 0.1.2
• Bump (patch/minor/major): ./scripts/bump-version.sh patch
• Release (bump + commit + tag): ./scripts/bump-version.sh patch --tag then git push && git push origin py-X.Y.Z

See docs/VERSIONING.md for details.

Support

• PyPI – pymdurs — project page and release history
• GitHub – pymdurs — source repository (Rust crate rsmdu + Python bindings)
• Examples README — detailed examples
• Versioning
• IGN Documentation