Category: Time series and analysis

Change Detection

Change detection identifies meaningful differences between dates, such as harvest, flooding, deforestation, or construction.

Also known as: change analysis, alerting

Expanded definition

Change detection compares observations across time to identify where something changed on the ground. Methods range from simple thresholds on indices to sophisticated time-series models.

The hard part is separating real change from noise and artifacts. Clouds, shadows, haze, BRDF effects, and inconsistent preprocessing can all look like change.

Robust change detection usually includes quality masks, normalization, and a definition of what counts as change for the application. For monitoring, the goal is often stable alerts with known false positive behavior, not just high sensitivity in one-off tests.

Related terms