Skip to main content

Compare imagery over time

A two-image swipe is useful, but a defensible change study records why the images can be compared and what the difference actually shows.

1. State the feature and period​

Examples:

  • wetland extent between 1950 and 2025
  • river channel position before and after a major flood
  • buildings at a marae through several decades
  • plantation harvest and regrowth
  • shoreline/dune movement

2. Record each image​

Capture:

source
survey_or_scene_id
acquisition_date
resolution
orthorectified_yes_no
crs
licence
cloud_or_quality_notes

3. Align the sources​

Prefer already orthorectified imagery. If a historic frame needs georeferencing, choose stable control away from the feature whose movement you are investigating.

4. Control visual differences​

Season, tide, river flow, sun angle, drought, crop cycle and image processing can all mimic change. A darker wetland in one image does not automatically mean more water.

5. Digitise only what the imagery supports​

Create a separate feature layer for each date or use a valid_from/image_date field. Do not draw a sub-metre boundary from a coarse satellite pixel.

6. Quantify carefully​

Area, distance and rate-of-change calculations inherit the positional uncertainty of both images. Report a sensible precision.

7. Add documentary evidence​

Aerial imagery can show that a building, track or channel is visible. Historical records can help explain what it was. Use Archives to map and Place names for that next step.

QGIS tools​

Useful techniques include raster transparency, swipe/comparison plugins, temporal layers, vector digitising and raster calculator for multispectral indices. For simple visual comparison, two map themes or side-by-side layouts can be more understandable than an automated classification.

Last reviewed: 26 August 2026