Raster processing using Python Tools

Geospatial Concepts: Raster Data


Teaching: 5 min
Exercises: 0 min
  • What is a raster?

  • What sorts of information does a raster typically model?

  • What are the major characteristics of a raster dataset?

  • What assumptions does the format imply?

  • Understand the raster data model

What is a raster?

Unlike vectors, where features have discreet boundaries (which is useful for storing data like country borders, land parcels and streets), rasters are useful for storing data that varies continuously. At its heart, a raster is:

In the 1950’s raster graphics were noted as a faster and cheaper (but lower-resolution) alternative to vector graphics.

Bands in Landsat 7 (bottom row of rectangles) and Landsat 8 (top row)
Graphic created by L.Rocchio & J.Barsi.
Landsat 8 Band 1 (“Ultra Blue”) Landsat 8 Band 3 (“Green”) Landsat 8 Band 9 (“Cirrus”)
Landsat 8: Ultra Blue Landsat 8: Green Band Landsat 8: Blue Band

What makes a raster geospatial?

A raster is just an image in local pixel coordinates until we specify what part of the earth the image covers. This is done through two pieces of metadata that accompany the pixel values of the image:

Spatially-aware applications are careful to interpret this metadata appropriately. If we aren’t careful (or are using a raster-editing application that ignores spatial information), we can accidentally strip this spatial metadata. Photoshop, for example, can edit GeoTiffs, but we’ll lose the embedded CRS and geotransform!

Common Types of Raster Datasets

Examples of common raster datasets include:

Limitations of the Raster Format

Key Points