uMORPH: Global Urban Canopy Parameters Dataset

uMORPH is a global dataset of 7 urban canopy parameters (UCPs) comprising 24 gridded layers at 100 m resolution, derived from ~2.38 billion building footprints with heights. It covers 98.4% of cities above 50,000 population and 94.5% of global population.

Jain, M., Ortiz, L. E., & McPhearson, T. (2026). uMORPH: Global LOD1 buildings and 100-meter urban canopy parameters dataset. (forthcoming)

Global distribution of 2°×2° tiles in uMORPH v1.0, color-coded by continent. Red points denote city centres with population > 50,000.

Area-weighted mean building height (H_aw) at 100 m resolution for 30 representative cities spanning all continents.

---

Dataset Products

uMORPH is distributed as three complementary products:

Product	Format	Coverage	Description
LOD1 Building Morphology	GeoParquet	Per 2°×2° tile, organised by continent	Building footprints + heights (2.5D); ~2.38B polygons
UCP Rasters	Cloud-Optimized GeoTIFF (COG)	Global, 1 file per UCP layer	24 layers at 100 m, WGS84
WRF Geogrid Binaries	Binary + index	Continental ZIP archives	WRF-ready, compatible with SLUCM, BEP, BEP+BEM

Full dataset available at: [data portal link — pending]

---

UCPs Provided

#	Variable	Symbol	Layers	Description
1	Mean Building Height	H_mean	1	Arithmetic mean height per grid cell
2	Area-Weighted Mean Height	H_aw	1	Height weighted by footprint area
3	Std Dev of Building Height	H_std	1	Population std dev (ddof=0)
4	Plan Area Fraction	λ_p	1	Union of building footprints / cell area
5	Building Surface-to-Plan Area Ratio	λ_B	1	(Roof area + wall area) / cell area
6	Frontal Area Index	FAI	4	At 0°, 45°, 90°, 135° wind directions
7	Building Height Distribution	p(z)	15	Area-weighted fractions, 0–75 m in 5 m bins

See docs/ucp_definitions.md for full formulae and units.

---

Pipeline Overview

01_tile_system_umorph.py          # Grid generation + city-tile mapping (local)
02_tiled_3d_morphology.py         # Overture Maps download + GHSL height infilling (HPC array job)
03_tiled_compute_ucp.py           # UCP computation → UTM + WGS84 rasters (HPC)
04_continent_ucp_merge.py         # Per-continent tile mosaic (HPC)
05_global_cog_ucp.py              # Global VRT → Cloud-Optimized GeoTIFF (HPC)
06_shp_to_geoparquet.py           # LOD1 SHP tiles → GeoParquet for distribution (HPC)

Scripts 02–06 are designed for SLURM batch execution. Each has a paired .sh job script in hpc/job_scripts/.

Geogrid binary conversion (product 3) was performed using the GIS4WRF QGIS plugin for uMORPH v1.0. See docs/geogrid_ingestion.md.

---

Data Sources

Source	Version	Use
Overture Maps	v1.9.0–v1.10.0 (May–Jun 2025)	Building footprints + native heights
GHSL ANBH	R2023A (E2018)	Height infilling for footprints lacking native heights
SimpleMaps World Cities	v1.90	City locations + population filter

---

Data

Tile Index (data/tile_index/)

Two files are distributed:

File	Description
worldcities_over_50k_tile_bounds.csv	Master tile index: tile_id, min_lon, min_lat, max_lon, max_lat, continent (2,624 tiles)
worldcities_over_50k_tiles.shp (+ sidecars)	Tile polygons, EPSG:4326

City centre points are not distributed as they derive directly from the SimpleMaps World Cities database (CC-BY 4.0). To regenerate locally:

# Download free SimpleMaps CSV from https://simplemaps.com/data/world-cities
python pipeline/01_tile_system_umorph.py \
    --input_cities /path/to/worldcities.csv \
    --output_dir data/tile_index/ \
    --pop_threshold 50000

Sample Tiles

Sample GeoParquet tiles are available from the uMORPH data portal [link pending]. Recommended tiles for testing:

Tile ID	City	Size
R030_C034	Phoenix, AZ	~115 MB
R026_C047	Chicago, IL	~358 MB
R029_C031	Los Angeles, CA	~407 MB

import geopandas as gpd
gdf = gpd.read_parquet("R030_C034.parquet")
print(gdf.crs)      # EPSG:4326
print(len(gdf))     # number of buildings
print(gdf.columns)  # id, version, sources, height, height_fil, was_filled, ghsl_pixel, geometry

---

Requirements

Data prerequisites (download separately)

1. Overture Maps CLI — pip install overturemaps==0.14.0
2. GHSL ANBH raster — download GHS_BUILT_H_ANBH_E2018_GLOBE_R2023A_4326_3ss_V1_0.tif from the GHSL portal. Update ghsl_raster path in 02.
3. SimpleMaps World Cities CSV — free version from simplemaps.com/data/world-cities. Update input_worldcities path in 01.

Environment setup

conda env create -f environment.yml
conda activate umorph_env

See environment.yml for full dependency list (Python 3.9, rasterio 1.4.3, geopandas 1.0.1, gdal 3.10.2).

---

Quickstart

Step 1 — Generate tile grid (local)

python pipeline/01_tile_system_umorph.py \
    --input_cities data/simplemaps/worldcities.csv \
    --output_dir data/tile_index/ \
    --pop_threshold 50000

Step 2 — Download buildings + infill heights (HPC)

Edit hpc/config/config.sh with your paths, then:

sbatch hpc/job_scripts/02_tiled_3d_morphology.sh

Step 3 — Compute UCPs (HPC)

sbatch hpc/job_scripts/03_tiled_compute_ucp.sh

Steps 4–5 — Mosaic + COG export (HPC)

sbatch hpc/job_scripts/04_continent_ucp_merge.sh
sbatch hpc/job_scripts/05_global_cog_ucp.sh

Step 6 — Convert LOD1 SHPs to GeoParquet (HPC)

sbatch hpc/job_scripts/06_shp_to_geoparquet.sh

---

Repository Structure

umorph/
├── README.md
├── CITATION.cff
├── LICENSE
├── environment.yml
│
├── pipeline/
│   ├── 01_tile_system_umorph.py
│   ├── 02_tiled_3d_morphology.py
│   ├── 03_tiled_compute_ucp.py
│   ├── 04_continent_ucp_merge.py
│   ├── 05_global_cog_ucp.py
│   └── 06_shp_to_geoparquet.py
│
├── hpc/
│   ├── job_scripts/
│   │   ├── 02_tiled_3d_morphology.sh
│   │   ├── 03_tiled_compute_ucp.sh
│   │   ├── 04_continent_ucp_merge.sh
│   │   ├── 05_global_cog_ucp.sh
│   │   └── 06_shp_to_geoparquet.sh
│   └── config/
│       └── config.sh
│
├── data/
│   └── tile_index/           # tile CSVs + SHPs
│
├── docs/
│   ├── ucp_definitions.md
│   ├── tile_naming.md
│   ├── geogrid_ingestion.md
│   ├── validation.md
│   └── figures/
│       ├── umorph_global_coverage.jpg
│       └── umorph_haw_30cities.jpg
│
├── wrf/
│   └── GEOGRID.TBL.ARW_LCZ_UMORPH_north_america
│
└── tests/
    └── test_ucp_formulae.py

---

Citation

@article{jain2026umorph,
  title   = {uMORPH: Global LOD1 buildings and 100-meter urban canopy parameters dataset},
  author  = {Jain, Madhavi and Ortiz, Luis E. and McPhearson, Timon},
  year    = {2026},
  note    = {Forthcoming}
}

---

License

Dataset (uMORPH v1.0 data products): Open Database License (ODbL) 1.0 — https://opendatacommons.org/licenses/odbl/

uMORPH v1.0 derives from OpenStreetMap and Microsoft Buildings data, both licensed under ODbL. The ODbL share-alike clause requires that derivative databases be distributed under the same license. You are free to use, share, and adapt uMORPH data with attribution and under the same ODbL terms.

Pipeline code: Apache License 2.0 — https://www.apache.org/licenses/LICENSE-2.0

Source data licenses:

Source	License
OpenStreetMap (via Overture Maps)	ODbL 1.0
Microsoft Buildings (via Overture Maps)	ODbL 1.0
Google Open Buildings (via Overture Maps)	CC BY 4.0
GHSL ANBH	CC BY 4.0
SimpleMaps World Cities	CC BY 4.0

---

Contact

Madhavi Jain · Urban Systems Lab, New York University · mj3848@nyu.edu