Mapbox Vector Tile Specification

Mapbox Vector Tile Specification
Name	Mapbox Vector Tile Specification
Developer	Mapbox
Initial release	2013
Stable release	2.1
File extension	.mvt
Genre	Geospatial vector tile format

Mapbox Vector Tile Specification is a compact binary format for encoding tiled vector map data designed to deliver geometry and attribute information efficiently to web and mobile clients. It was developed to support high-performance rendering in environments that use OpenStreetMap, Mapbox GL JS, Mapbox Mobile, Leaflet, and other mapping libraries, enabling client-side styling and interaction with vector features. The specification unifies concepts from Protocol Buffers, Web Mercator, and tile pyramids used by services like Google Maps, Esri ArcGIS, and Bing Maps.

Overview

The specification defines a standardized tiling scheme and binary encoding aimed at reducing bandwidth and CPU overhead for applications similar to Mapbox Studio, Carto, Mapnik, QGIS, and ArcGIS Pro. It addresses requirements arising in projects integrating data from OpenStreetMap, Natural Earth, USGS National Map, Ordnance Survey, and national mapping agencies, facilitating use in contexts such as Strava Metro, Uber Movement, CityMobil2, and disaster-response platforms like Humanitarian OpenStreetMap Team. The design balances interoperability with clients including Tiler, Tippecanoe, TileStache, and server stacks based on Node.js, Go (programming language), Python (programming language), and C++.

File Format and Encoding

Tiles are encoded using Protocol Buffers binary messages to form a compact ".mvt" payload; this approach parallels other binary encodings used by Google Protocol Buffers adopters such as Apache Kafka consumers in geospatial pipelines. Encoded files typically include layer blocks, feature messages, and key/value attribute tables; tools like tilelive, tegola, and mbutil produce compatible outputs. The format emphasizes deterministic ordering and integer encoding to keep files small for transport over protocols used by HTTP/2, CDN providers, and mobile networks in projects by Akamai Technologies, Fastly, and Cloudflare.

Tile Geometry and Features

Geometry types supported include points, lines, and polygons with topologies analogous to feature models in GeoJSON, Shapefile, and GML datasets. Each feature carries attribute indices that reference shared key and value tables, a model used in systems such as PostGIS, SpatiaLite, and GeoServer to minimize duplication. Encoding of geometry commands (moveTo, lineTo, closePath) follows a compact integer command stream similar to vector graphics primitives used by SVG renderers and graphics engines like Skia and Cairo.

Coordinate Systems and Extent

Tiles are defined within a localized integer grid (extent) mapped to projected coordinate spaces like Web Mercator (EPSG:3857) commonly used by OpenStreetMap, Google Maps, Bing Maps, and Mapbox Streets. The integer extent parameter (often 4096) provides a quantized coordinate system comparable to tiling schemes in XYZ tile pyramids and quadkey systems used by Bing Maps Tile System. The specification permits aligning vector coordinates with raster tile grids used in XYZ tiles, TMS, and slippy map conventions employed by CartoDB and Mapbox Classic.

Protobuf Schema and Versioning

The protobuf schema defines Layer, Feature, and Value messages and reserves fields for future extension; versioning follows principles similar to those in Protocol Buffers and software versioning practices used by projects like Semantic Versioning. Backward-compatible evolution of the schema is essential for long-lived platforms including Mapbox Studio Classic, Mapbox GL Native, and third-party renderers such as Tangram and Deck.gl. The spec documents how to handle unknown fields, optional values, and feature IDs to interoperate with vector tile generators such as Tippecanoe, mbtiles, and server exporters in GeoServer.

Performance and Use Cases

Vector tiles are optimized for use cases where client-side styling, dynamic labeling, and progressive loading matter, as in applications like Mapbox GL JS, OpenLayers, Cesium, Kepler.gl, and mobile SDKs for iOS and Android. They reduce over-fetching compared to whole-dataset transfers used in WFS services and enable hardware-accelerated rendering pipelines similar to those in WebGL and Metal. Performance considerations include tile size limits, geometry simplification drawn from algorithms in Douglas–Peucker algorithm implementations, and attribute pruning techniques used by cartographic production systems such as Natural Earth and national mapping agencies.

Implementations and Tools

A broad ecosystem implements the specification: vector tile creators like Tippecanoe, ogr2ogr, and tile-join; servers and tile caches such as tileserver-gl, TileServer GL, Tegola, TileStache, TileServer PHP; client libraries including Mapbox GL JS, Mapbox GL Native, OpenLayers, and Leaflet.VectorGrid. Conversion, inspection, and debugging tools include vtzero, mapbox-vector-tile-cpp, mapbox/vector-tile-js, mapbox/tile-reduce, and integrations with databases like PostGIS and BigQuery GIS. Cloud and enterprise platforms supporting vector tiles include services by Mapbox, Carto, Esri, Google Cloud, and Amazon Web Services.

Category:Geospatial data formats