Skip to content

Extracting from WFS Services

This guide walks through extracting building footprints from a WFS (Web Feature Service), enriching metadata, generating PMTiles for web visualization, and publishing to Source Cooperative.

Live Example

The result of this tutorial is published at:

Prerequisites

  • Portolan CLI installed (pipx install portolan-cli)
  • tippecanoe for PMTiles generation
  • AWS credentials configured for your target bucket

Step 1: Explore the WFS Service

First, discover available layers without extracting:

portolan extract wfs \
  "https://geoservices.wallonie.be/geoserver/inspire_bu/ows" \
  --dry-run

Dry run - would extract 2 layers:
  • inspire_bu:BU.Building_building_emprise (ID: 0)
  • inspire_bu:BU.Building_building_lod1 (ID: 1)

Two layers available: building footprints (building_emprise) and 3D buildings (building_lod1).

Step 2: Extract Building Footprints

Extract 100,000 features from the footprints layer:

portolan extract wfs \
  "https://geoservices.wallonie.be/geoserver/inspire_bu/ows" \
  buildings \
  --layers "inspire_bu:BU.Building_building_emprise" \
  --limit 100000 \
  --auto

Portolan automatically:

  1. Downloads features via WFS with pagination
  2. Converts to GeoParquet with Hilbert spatial ordering
  3. Generates STAC catalog with proper metadata
  4. Seeds metadata.yaml from ISO 19139 records (if available)

Metadata Auto-Extraction

Since v0.6.0, portolan extract wfs automatically fetches ISO 19139 metadata records from the WFS service, populating license, description, keywords, and contact info.

Step 3: Enrich Metadata

The extraction creates two metadata files:

  • buildings/.portolan/metadata.yaml — Catalog level
  • buildings/inspire_bu_.../.portolan/metadata.yaml — Collection level

Review and complete these files:

buildings/.portolan/metadata.yaml
contact:
  name: "Your Name"
  email: "your@email.com"

license: "CC-BY-4.0"
license_url: "https://creativecommons.org/licenses/by/4.0/"

title: "Belgium Building Footprints - Wallonia"
description: |
  INSPIRE-compliant building footprints for the Wallonia region.

keywords:
  - buildings
  - footprints
  - belgium
  - inspire
  - geoparquet

source_url: "https://geoservices.wallonie.be/geoserver/inspire_bu/ows"
attribution: "Service Public de Wallonie (SPW)"

License Verification

Always verify the license from the source. Check _license_info_from_source in the auto-generated metadata.yaml for hints from the WFS service.

Step 4: Generate PMTiles

PMTiles enables efficient web-based map rendering. Generate from GeoParquet:

cd buildings/inspire_bu_bu_building_building_emprise_865a73

ogr2ogr -f GeoJSONSeq \
  -s_srs EPSG:3035 \
  -t_srs EPSG:4326 \
  /vsistdout/ \
  inspire_bu_bu_building_building_emprise_865a73.parquet | \
tippecanoe \
  -o inspire_bu_bu_building_building_emprise_865a73.pmtiles \
  -z14 -Z4 \
  --layer=buildings \
  --force \
  --read-parallel \
  --drop-densest-as-needed

CRS Reprojection

Some WFS services return coordinates in projected CRS (like EPSG:3035) even when the GeoParquet metadata says WGS84. The -s_srs flag overrides the source CRS for correct reprojection.

Then add the PMTiles to your catalog:

cd buildings
portolan add inspire_bu_.../*.pmtiles

Step 5: Generate READMEs

Create human-readable documentation from STAC metadata:

portolan readme --recursive

This generates README.md files at catalog and collection levels.

Step 6: Push to Source Cooperative

Configure your remote in .env:

buildings/.env
PORTOLAN_REMOTE=s3://us-west-2.opendata.source.coop/username/dataset/
PORTOLAN_PROFILE=source-coop

Push to the bucket:

portolan push --verbose

Summary

Step Command
Explore portolan extract wfs URL --dry-run
Extract portolan extract wfs URL OUTPUT --layers PATTERN --limit N
Metadata Edit .portolan/metadata.yaml files
PMTiles ogr2ogr \| tippecanoe pipeline
Add tiles portolan add *.pmtiles --stac-geoparquet
READMEs portolan readme --recursive
Push portolan push --verbose

See Also