Spatial Data

How we represent the world in a computer

• Vector
• Raster

Vector

Represent each entity with a shape or geometry. Simple features:

• (Multi-)Points
• (Multi-)Lines
• (Multi-)Polygons

Divide space into a finite set of entities

[Source]

[Source]

Raster

Use an image and control pixel colors to encode value

The value assigned for each cell represents the attribute of that cell

• Continuous variables, surfaces (temperature, density, elevation…)
• Satellite images (land cover, land use…)

[Source]

[Source]

Spatial Data Formats

Spatial Data Formats

In principle…

• Points, lines, polygons –> Vector formats
• Images, surfaces –> Raster formats

But these boundaries are blur…

Traditionally

Vector

• Single files: shapefiles, etc.
• Client-server (geo-)DBs: PostGIS, etc.

Raster

• Mostly (single) image formats: GeoTIFF, etc.

However…

Many of these formats were designed for an offline world, so display some of the following:

• Binary (“non-streamable”)
• “Unqueriable”
• Complex format structures

Modern formats

New formats have appeared in part “fixing” those issues, but also responding to web needs:

• Streamable (e.g. GeoJSON)
• Queriable (e.g. PostGIS/Geopackage)
• Single file (e.g. .mbtiles)

Let’s explore a bit more on a couple of them…

GeoJSON

Advantages

• Plain text, (human) readable
• Streamable
• Well integrated with web standards (JSON)

Excellent exchange format

Disadvantages

• Plain text, inefficient
• Non-queriable
• Vector only

Tilesets (.mb/vtiles)

Map Tiles

Advantages

• Queriable (SQLite)
• Fast access to large maps with limited resources (client/server model + queriable format)
• Some (vector tiles) are stylable

Disadvantages

• Designed for serving not analysing
• A dataset needs to be stored at several zoom levels
• Once created, hard to modify (e.g. reproject)

Selecting the right format

Selecting the right format

No silver bullet…

Quiz

1. Large dataset of tweets you want to analyse PostGIS/Geopackage
2. Drone imagery to make available for workshop participants MBTiles
3. Street basemap to provide context to a small dataset you want to make available on the web Vector tiles
4. The small dataset from 3. GeoJSON