uMORPH_frontend
uMORPH: Global LOD1 Buildings and 100-meter Urban Canopy Parameters Dataset
Microsite Coming Soon!
GENERAL INFORMATION
Title of Dataset: uMORPH: Global LOD1 Buildings and 100-meter Urban Canopy Parameters Dataset
Version: 1.0
Date of data collection: May - June 2025
Geographic location: Global - 11,103 cities/locales with population above 50,000 used as tile selection anchors, covering all continents. Due to the 2°×2° tile extent and 1° buffer strategy, tiles also capture smaller cities, towns, and peri-urban areas within each tile boundary.
Funding sources: This dataset was partially supported with funding from Google.org Grant No. TF2307-114864.
Author
- Name: Madhavi Jain
- ORCID: https://orcid.org/0000-0001-6660-1181
- Institution: Urban Systems Lab, Department of Environmental Studies, New York University
- Email: madhavi.jain@nyu.edu
Co-Author
- Name: Luis E. Ortiz
- ORCID: https://orcid.org/0000-0002-4248-6374
- Institution: Atmospheric, Oceanic & Earth Sciences Department, George Mason University
- Email: lortizur@gmu.edu
Co-Author
- Name: Timon McPhearson
- ORCID: https://orcid.org/0000-0002-9499-0791
- Institution: Urban Systems Lab, Department of Environmental Studies, New York University
- Email: timon.mcphearson@nyu.edu
SHARING / ACCESS INFORMATION
License (data): Open Database License (ODbL) 1.0
License (code): Apache License 2.0
Recommended dataset citation: Jain, M., Ortiz, L. E., & McPhearson, T. (2026). uMORPH: Global LOD1 buildings and 100-meter urban canopy parameters dataset. https://doi.org/10.58153/mrt3g-htt3
Related publication: Jain, M., Ortiz, L. E., & McPhearson, T. (2026). uMORPH: A Global buildings and 100-meter urban canopy parameters dataset. (Forthcoming)
Was data derived from another source? Yes. See Table below for source data licenses.
| Source Dataset | Version | License | Use in uMORPH |
|---|---|---|---|
| Overture Maps Foundation (Buildings theme) | v1.9.0–v1.10.0 (May–Jun 2025) | ODbL 1.0 | Building footprints + native heights. Includes building data from OSM, Google, Microsoft. |
| GHSL ANBH (Global Human Settlement Layer - Average Number of Building Heights) | R2023A (Epoch 2018) | CC BY 4.0 | Height infilling for footprints with missing heights |
| SimpleMaps World Cities | v1.90 | CC BY 4.0 | City locations and population filter |
DATASET COVERAGE SUMMARY
| Metric | Value |
|---|---|
| Total tiles selected | 2,624 |
| Tiles with building footprints | 2,042 |
| Tiles without footprints (mostly ocean-dominated or no data on land areas) | 582 |
| Cities/locales covered (> 50,000 population) | 11,103 / 11,283 (98.4%) |
| Global population coverage | 94.5% |
| Total building footprints with heights | ~2.38 billion |
| Spatial resolution (UCPs) | 100 meter |
| Coordinate reference system | WGS84 (EPSG:4326) |
| Overture Maps source version | v1.9.0–v1.10.0 (May-Jun 2025) |
| GHSL ANBH source version | R2023A (Epoch 2018) |
DATA & FILE OVERVIEW
The dataset is organized into four top-level directories:
1. LOD1_building_footprints_heights/
LOD1 building morphology (footprint geometry + height) in GeoParquet format. One file per 2°×2° tile, organized by continent. Tiles are named using the uMORPH R###_C### tile naming convention (see tile_naming.md).
| Path | Description | Size |
|---|---|---|
| LOD1_building_footprints_heights/africa/ | 315 GeoParquet tiles | > 45 GB |
| LOD1_building_footprints_heights/asia/ | 628 GeoParquet tiles | > 111 GB |
| LOD1_building_footprints_heights/europe/ | 430 GeoParquet tiles | > 46 GB |
| LOD1_building_footprints_heights/north_america/ | 348 GeoParquet tiles | > 33 GB |
| LOD1_building_footprints_heights/south_america/ | 267 GeoParquet tiles | > 21 GB |
| LOD1_building_footprints_heights/oceania/ | 54 GeoParquet tiles | > 1.7 GB |
Figure. Global distribution of 2°×2° uMORPH tiles, color-coded by continent. Red dots indicate city centres with population > 50,000 used as tile selection anchors. Grey filled tiles were processed but contained no Overture Maps building footprints.
2. ucp_global_cloud_optimized_geotiffs/
Global Cloud-Optimized GeoTIFF (COG) rasters for all 7 UCPs (24 layers total) at 100 m resolution in WGS84 (EPSG:4326). Each UCP is distributed as a compressed tar archive.
| Archive | Layer(s) | Description | Size |
|---|---|---|---|
| mean_building_height.tar | global_mean_building_height_100m_wgs_COG.tif | Arithmetic mean building height per 100m cell (m) | > 4.6 GB |
| area_weighted_mean_height.tar | global_area_weighted_mean_height_100m_wgs_COG.tif | Footprint area-weighted mean building height (m) | > 4.6 GB |
| std_dev_building_height.tar | global_std_dev_height_100m_wgs_COG.tif | Population std dev of building heights per cell (m) | > 4.5 GB |
| plan_area_fraction.tar | global_plan_area_fraction_100m_wgs_COG.tif | Ratio of building footprint area to cell area [0,1] | > 5.1 GB |
| surface_to_plan_area_ratio.tar | global_bldg_surface_to_plan_area_ratio_100m_wgs_COG.tif | Ratio of total building surface (roof + walls) to cell area | > 5.1 GB |
| frontal_area_index.tar | global_frontal_area_index_{0,45,90,135}deg_100m_wgs_COG.tif (4 layers) | Building frontal area index at 4 wind directions | > 21 GB |
| building_height_distribution.tar | global_building_height_distribution_{5m–75m}_100m_wgs_COG.tif (15 layers) | Area-weighted fraction of buildings per 5m height bin | > 23 GB |
Figure. Area-weighted mean building height (H_aw; in meters) at 100 m resolution for 30 representative cities spanning all continents.
3. ucp_continental_geogrid_binaries/
Weather Research and Forecasting (WRF) model ingestion ready geogrid binary archives for all 7 UCPs (24 layers), organized by continent. Compatible with SLUCM, BEP, and BEP+BEM urban canopy schemes in the WRF model.
Format: WRF geogrid binary (index + data files), one directory per UCP layer. Each continental ZIP archive contains 24 UCP directories following the NUDAPT URB_PARAM index naming convention (e.g., 91_Plan_Area_Fraction/, 92_Mean_Building_Hgt/, …, 132_Build_Hgt_Dist_70_75m/). Unzipped files may require several TBs of storage space.
| File | Continent | Size |
|---|---|---|
| africa.zip | Africa | > 2.4 GB |
| asia.zip | Asia | > 4.8 GB |
| europe.zip | Europe | > 2.4 GB |
| north_america.zip | North America | > 2.4 GB |
| south_america.zip | South America | > 1.2 GB |
| oceania.zip | Oceania | > 2.4 GB |
4. additional_data_tiles_logs/
Supporting files including the tile index, continent chunk CSVs used for HPC batch processing, and per-chunk processing statistics.
| Path | Description |
|---|---|
| tiles/worldcities_over_50k_tile_bounds.csv | Master tile index: tile_id, min_lon, min_lat, max_lon, max_lat, continent (2,624 tiles) |
| tiles/worldcities_over_50k_tiles.shp (.shx, .dbf, .prj, .cpg) | Tile polygons in WGS84 (EPSG:4326) |
| tiles/continent_chunks/ |
Per-continent HPC batch chunk files (50 tiles each) |
| logs/ |
Per-chunk processing statistics: total/missing/filled/skipped/failed polygons |
| logs/all_continents_missing_hgts_tile_ids.csv | Tiles with no Overture Maps building footprints (labeled “Processed — no footprints”) |
METHODOLOGICAL INFORMATION
Description of methods used for collection/generation of data:
uMORPH integrates building footprint and height data from Overture Maps Foundation (v1.9.0 - v1.10.0, May - Jun 2025) with height infilling from the Global Human Settlement Layer Average Number of Building Heights (GHSL ANBH, R2023A, Epoch 2018) for footprints lacking native height attributes. A scalable tile-based pipeline processes 2°×2° tiles selected on cities/locales (with 1° buffer strategy) having population above 50,000 (sourced from SimpleMaps World Cities v1.90). For each tile, building footprints are downloaded, height-infilled where necessary, and used to compute 7 UCPs across 100m grid cells after reprojection to the tile’s UTM zone for metric accuracy.
Methods for processing the data:
The full processing pipeline (scripts 01–06) is available at https://github.com/UrbanSystemsLab/uMORPH. Steps include: (1) tile grid generation; (2) Overture Maps download and GHSL height infilling; (3) UCP computation at 100m in UTM and reprojection to WGS84; (4) per-continent tile mosaic; (5) global Cloud-Optimized GeoTIFF generation; (6) LOD1 SHP to GeoParquet conversion. All processing was performed on the NYU High Performance Computing (HPC) cluster using SLURM.
Software/libraries specific information:
| Component | Version |
|---|---|
| Python | 3.9.22 |
| GDAL | 3.10.2 |
| rasterio | 1.4.3 |
| geopandas | 1.0.1 |
| pyarrow | 20.0.0 |
| overturemaps CLI | 0.14.0 |
| fiona | 1.10.1 |
| pyogrio | 0.11.1 |
| GIS4WRF (QGIS plugin) | 0.16.1 - used for WRF geogrid binary conversion |
Quality assurance: UCPs were validated against NUDAPT, UT-GLOBUS, and GloUCP across New York City and Phoenix. Full validation methodology and results are reported in Jain et al. (2026, forthcoming).
DATA-SPECIFIC INFORMATION
LOD1 GeoParquet tiles (LOD1_building_footprints_heights/)
Format: GeoParquet (Apache Parquet + GeoArrow geometry encoding), Snappy compression, EPSG:4326
| Variable | Type | Description |
|---|---|---|
| id | string | Overture Maps building feature ID |
| version | integer | Overture Maps version tag |
| sources | string | Source attribution (OSM, Microsoft Buildings, Google Open Buildings, etc.) |
| height | float | Native building height from Overture Maps (m); NaN if not available |
| height_fil | float | Final height after GHSL ANBH infilling (m); equal to height if native height available |
| was_filled | boolean | True if height was infilled from GHSL ANBH; False if native height used |
| ghsl_pixel | integer | Number of valid GHSL pixels used for infilling (0 if not filled) |
| geometry | geometry | Building footprint polygon (WGS84, EPSG:4326) |
Missing data codes: NaN for height where no native height is available and GHSL infilling returned no valid pixels.
UCP Cloud-Optimized GeoTIFFs (ucp_global_cloud_optimized_geotiffs/)
Format: Cloud-Optimized GeoTIFF (COG), LZW compression, 512×512 tile layout, EPSG:4326
Spatial resolution: 100 meter
NoData value: -9999
| UCP | Symbol | Units | NUDAPT Index | UCM Schemes |
|---|---|---|---|---|
| Mean Building Height | H_mean | meter | 92 | SLUCM |
| Area-Weighted Mean Building Height | H_aw | meter | 94 | SLUCM, BEP, BEP+BEM |
| Standard Deviation of Building Height | H_std | meter | 93 | SLUCM |
| Plan Area Fraction | λ_p | dimensionless [0, 1] | 91 | SLUCM, BEP, BEP+BEM |
| Building Surface-to-Plan Area Ratio | λ_B | dimensionless | 95 | SLUCM, BEP, BEP+BEM |
| Frontal Area Index (4 directions; 0,45,90,135 deg) | λ_f(θ) | dimensionless | 96–99 | SLUCM |
| Building Height Distribution (15 bins) | p(z) | dimensionless fraction [0, 1] | 118–132 | BEP, BEP+BEM |
Height bin labels (p(z)): 5m, 10m, 15m, 20m, 25m, 30m, 35m, 40m, 45m, 50m, 55m, 60m, 65m, 70m, 75m. Buildings with height ≥ 70m are assigned to the 75m bin (NUDAPT convention).
WRF Geogrid Binaries (ucp_continental_geogrid_binaries/)
See the geogrid ingestion guide for WRF/WPS integration instructions.