sigau-tileserver

English | Español

Pipeline for generating and serving vector tiles (MBTiles) of Bogotá's Urban Tree Census (SIGAU) and Localities datasets.

👉 Live demo: https://tileserver.juanliz.com

Features

Dataset	Description	Zoom range
sigau.mbtiles	Individual tree points	12–18
sigau-clustered.mbtiles	Pre-clustered tree data	8–15
localities.mbtiles	Bogotá locality polygons	0–18

Important

The tileserver exposes properties and attributes exactly as provided by the original sources. No additional modifications are made to the data's attributes; for full details on attributes, schemas, and licenses, consult the data sources listed in the Data Sources section.

Quick Start

Choose the option that best fits your needs:

Option	Use case	Requirements
A. Pull image	Fastest way to run	Docker
B. Build Docker locally	For custom builds or development	Docker
C. Pre-built tiles	Serve the included MBTiles directly	Node.js or Tileserver-GL
D. Manual generation	Customize parameters or use fresh data	GDAL, Tippecanoe, Tileserver-GL

Option A – Pull image

The fastest way to get started. Pull the pre-built image from GitHub Container Registry:

docker pull ghcr.io/juanliz/sigau-tileserver:latest
docker run -p 8080:8080 ghcr.io/juanliz/sigau-tileserver:latest

Then open http://localhost:8080.

Option B – Build Docker Locally

The Dockerfile will:

1. Downloads the datasets
2. Converts them to EPSG:4326
3. Generates MBTiles via Tippecanoe
4. Serves them with Tileserver-GL

Build the image yourself locally.

docker build -t sigau-tileserver .
docker run -p 8080:8080 sigau-tileserver

Then open http://localhost:8080.

Option C – Serve Pre-built Tiles

If you don't need to regenerate the tiles, you can serve the included MBTiles directly.

With npx (no installation required):

npx tileserver-gl data/

With a global installation:

tileserver-gl data/

This is a lightweight option if you already have the MBTiles files.

Option D – Manual Generation

For full control over the pipeline, follow these steps:

1. Download GeoJSON files

• Urban Tree Census (SIGAU):
  https://datosabiertos.bogota.gov.co/dataset/censo-arbolado-urbano

• Bogotá Localities (IDECA):
  https://www.ideca.gov.co/recursos/mapas/localidad-bogota-dc

2. Convert to EPSG:4326

ogr2ogr -f GeoJSON -t_srs EPSG:4326 \
  arboladourbano_wgs84.geojson arboladourbano.geojson

ogr2ogr -f GeoJSON -t_srs EPSG:4326 \
  localities_wgs84.geojson localities_raw.geojson

3. Generate MBTiles with Tippecanoe

Here are the commands to create each MBTiles file. You can adjust parameters as needed.

Individual tree points (sigau.mbtiles):

tippecanoe \
  -o sigau.mbtiles \
  -l arboladourbano \
  --drop-densest-as-needed \
  --extend-zooms-if-still-dropping \
  --no-feature-limit \
  --no-tile-size-limit \
  --minimum-zoom=12 \
  --maximum-zoom=18 \
  --force \
  arboladourbano_wgs84.geojson

Clustered tree points (sigau-clustered.mbtiles):

tippecanoe \
  -o sigau-clustered.mbtiles \
  --layer=clusters \
  --no-feature-limit \
  --no-tile-size-limit \
  --minimum-zoom=8 \
  --maximum-zoom=15 \
  --cluster-distance=40 \
  --cluster-maxzoom=15 \
  --cluster-densest-as-needed \
  arboladourbano_wgs84.geojson

Localities polygons (localities.mbtiles):

tippecanoe \
  -o localities.mbtiles \
  --layer=localities \
  --no-feature-limit \
  --no-tile-size-limit \
  --minimum-zoom=0 \
  --maximum-zoom=18 \
  --coalesce-densest-as-needed \
  --extend-zooms-if-still-dropping \
  --force \
  localities_wgs84.geojson

4. Serve with Tileserver-GL

Copy the generated .mbtiles files to your data/ directory and run:

npx tileserver-gl data/

Data Sources

Dataset	Source	License
Bogotá Localities	IDECA	CC BY 4.0
Urban Tree Census	Datos Abiertos Bogotá	CC BY 4.0

Why not use the SIGAU REST service directly?

The REST service is provided by SIGAU at the following endpoint: https://sigau.jbb.gov.co/arcgis/rest/services/IDECA/CensoArbol/MapServer/0. It operates as an ArcGIS Feature Layer (version 10.81) rather than a tileserver. This setup means that data must be retrieved through direct queries to the endpoint, which poses performance constraints given the large number of tree-point geometries for Bogotá. The service enforces a maximum record limit (2000 features per request), requiring either multiple paginated requests or spatial subdivisions to retrieve the full dataset. It returns point geometries (WKID 4686) with a broad attribute set, but lacks mechanisms for efficient spatial tiling or caching of vector tiles. As a result, any application relying on this service must implement complex strategies for spatial partitioning, local caching, and optimized querying to avoid excessive load times and numerous requests, since the service is not designed to support high-density real-time data delivery directly.

License

• Datasets: Licensed by IDECA and Datos Abiertos Bogotá under CC BY 4.0.
  Attribution is required when using or redistributing the data.

• Code: This repository's scripts and configuration are licensed under the MIT License.