Skip to content

Raster Data

Goal: Understand raster data — pixel-based grids used for imagery, elevation, and continuous phenomena.

What you'll learn

  • What a raster is and how it stores values
  • Bands, cells, resolution, and bit depth
  • Common raster types and formats
  • Continuous vs discrete rasters

What is raster data?

A raster is a grid of cells (pixels), where each cell holds a value. Like a photo — but the value can mean anything: elevation, temperature, land cover class, NDVI, population density.

[ 102 | 105 | 108 | 110 ]   ← row 1
[ 103 | 107 | 109 | 111 ]   ← row 2
[ 105 | 108 | 110 | 112 ]   ← row 3
   ↑     ↑     ↑     ↑
  col1  col2  col3  col4

Each cell has:

  • A row, column position
  • A value (e.g., elevation in meters)
  • An implicit size (e.g., 30 × 30 meters)
  • A CRS (so the grid sits in the right spot on Earth)

Resolution

Resolution = the size of one pixel on the ground.

Resolution Example data Use case
30 cm Drone imagery, Maxar Building inspection
1 m NAIP aerial Detailed urban mapping
10 m Sentinel-2 Regional land cover
30 m Landsat 8/9 Long-term change detection
250 m MODIS Global vegetation monitoring
1 km Climate models Global / national thematic

Resolution vs accuracy

Higher resolution = smaller pixels = more detail. But accuracy depends on the sensor, not just pixel size.

Bands

A raster can have one or many bands.

flowchart LR
    R[RGB image] --> R1[Red band]
    R --> G[Green band]
    R --> B[Blue band]
    L[Landsat scene] --> L1[Band 1: Coastal]
    L --> L2[Band 2: Blue]
    L --> L3[...]
    L --> L11[Band 11: Thermal]

    classDef root fill:#4338ca,stroke:#312e81,color:#fff
    class R,L root
    classDef band fill:#eef2ff,stroke:#4338ca,color:#312e81
    class R1,G,B,L1,L2,L3,L11 band
  • 1 band: a single value per cell (DEM, NDVI, single satellite band)
  • 3 bands: RGB photo
  • 4 bands: RGB + near-infrared (NIR)
  • N bands: Landsat (11), hyperspectral (hundreds)

Continuous vs discrete (categorical)

Type Example Cell value
Continuous Elevation, temperature, NDVI A floating-point number
Discrete (categorical) Land cover class, soil type An integer that maps to a class (1=forest, 2=water…)

The same color ramp does not work for both. Use a sequential ramp for continuous and a categorical palette for discrete.

Common raster types

  • DEM (Digital Elevation Model)

    Elevation per cell. Used for slope, aspect, hillshade, watershed analysis.

  • Satellite imagery

    Landsat, Sentinel, MODIS, Maxar. Multi-band, used for land cover and change detection.

  • Aerial imagery

    NAIP (US), Bing aerial, drone orthos. Higher resolution, fewer bands.

  • NDVI / vegetation indices

    Computed from NIR + Red. Healthy vegetation = high NDVI.

  • Climate / weather

    Temperature, precipitation, wind grids from PRISM, ERA5, etc.

  • Population grids

    WorldPop, GHSL — raster estimates of population density.

Common raster formats

Format Extension Notes
GeoTIFF .tif Most common. Self-contained TIFF + georeferencing.
Cloud Optimized GeoTIFF (COG) .tif Modern, streamable. Used by ArcGIS Online, Living Atlas.
Esri Grid folder Esri legacy format.
MrSID, JPEG2000 .sid, .jp2 Compressed imagery.
NetCDF, HDF5 .nc, .h5 Multi-dimensional (climate, oceanography).

Bit depth

The number of bits per cell determines the range of values:

  • 1-bit: 0 or 1 (binary masks)
  • 8-bit unsigned: 0–255 (most photos, land cover classes)
  • 16-bit signed: -32,768 to 32,767 (Landsat surface reflectance)
  • 32-bit float: decimal numbers (NDVI, DEM, scientific data)

A bigger bit depth = bigger files. Match the bit depth to your data.

NoData

Cells outside the area of interest, or where the sensor failed, get a NoData value (often -9999 or 0). Always check your raster's NoData value before doing math.

Practice

Try this

  1. Download a Landsat scene from https://earthexplorer.usgs.gov or use ArcGIS Pro's Add Data → Living Atlas → Sentinel-2.
  2. Open the Symbology pane for the raster — try different stretch types (Standard Deviation, Min/Max, Percent Clip).
  3. Use Raster Functions → NDVI to compute vegetation health.
  4. Compare two NDVI rasters from different seasons.

Next up

→ Attribute Tables — the data behind every vector layer.