Every GIS layer is stored in one of two basic ways. Vector data represents features as points, lines and polygons with exact coordinates. Raster data divides the area into a grid of cells and stores a value in each one. Knowing which is which explains why some maps look crisp and others look pixelated, why some files are huge, and which analysis each one suits.

Vector data: points, lines and polygons

Vector data describes discrete features with clear edges:

  • Points: wells, bus stops, trees, addresses
  • Lines: roads, rivers, pipes, transit routes
  • Polygons: parcels, buildings, lakes, zoning districts, census tracts

Each feature has a row of attributes: a parcel's owner and zoning, a road's name and speed limit. Common formats are GeoPackage, shapefile, file geodatabase and GeoJSON.

Raster data: a grid of cells

Raster data describes continuous surfaces and images. Each cell (pixel) holds one value, such as elevation, temperature, a satellite band or a land cover class. The cell size, or resolution, sets the detail: a 1 m raster can show a driveway, a 30 m raster cannot. GeoTIFF, and its web-friendly Cloud Optimized GeoTIFF version, is the standard format.

The same features as vector and raster dataLeft, a lake polygon, a river line and a well point drawn as vector shapes. Right, the same features converted to a grid of cells; the lake edge becomes stepped and the well fills a whole cell.Vector: shapes with exact edgesRaster: a grid of 20-unit cells
The same lake (polygon), river (line) and well (point) stored as vector (left) and raster (right). In the raster, edges become steps and the well fills a whole cell. It was made by testing the centre of each cell against the shapes.

Side by side

Raster and vector compared
AspectVectorRaster
Best forFeatures with clear edgesContinuous surfaces and imagery
ExamplesParcels, roads, buildings, boundariesElevation, satellite images, land cover, temperature
PrecisionExact coordinatesLimited by cell size
AttributesMany fields per featureUsually one value per cell
Typical analysisOverlay, network routing, joins, countsMap algebra, slope, terrain, image classification
File sizeSmall for most featuresGrows fast with area and resolution

Converting between them

Data often has to switch form. Rasterizing turns vectors into a grid, for example parcels into a cost surface. Vectorizing turns a raster into shapes, for example a classified land cover image into polygons. Both lose something: rasterizing turns smooth edges into steps and forces small features to a whole cell; vectorizing creates jagged outlines and many tiny slivers that need cleaning up. The cell size you choose controls how much is lost.

Which should you use?

  • Use vector when features have legal or physical edges and you need their attributes: parcels, utilities, zoning, addresses.
  • Use raster for things that vary continuously or come from sensors: terrain, imagery, heat, rainfall, vegetation.
  • Use both for most real projects. A suitability analysis might score raster slope and land cover, then summarise the result by vector parcels.

Frequently asked questions

Is a satellite image raster or vector?

Raster. Each pixel stores the brightness measured in one or more bands of light.

Are web map tiles raster?

Traditionally yes, as pre-drawn images. Many modern web maps use vector tiles, which send the shapes and draw them in the browser.

Why does my raster look blocky when I zoom in?

You've reached its cell size. Zooming can't add detail that the data doesn't contain.

Working with mixed GIS data?

We'll get your data into shape for the job.

From geodatabase design to raster analysis, we make data that works together.

Talk to usSee our services