Persistent Change Monitoring

Persistent Change Monitoring is EarthDaily's AI-driven, broad-area change detection service for defence and intelligence organisations, also marketed to civil government agencies. Manual image analysis and periodic surveillance leave gaps in coverage, and imagery collected under inconsistent conditions produces false positives that overwhelm analysts.

The service runs machine-learning models on data from the EarthDaily Constellation, which collects daily at 5 m resolution with a consistent local overpass time and viewing geometry across 22 calibrated spectral bands. The models detect artificial structures and construction, vehicle concentrations, ground disturbance, infrastructure change and maritime activity such as vessel movements, while suppressing benign seasonal and weather variation; EarthDaily states precision and recall exceeding 85% over operationally challenging environments. Changes are detected and surfaced within 24 hours through a secure web platform with an alerting dashboard, role-based access, LLM-assisted pattern-of-life analysis and optional human-in-the-loop validation. Integration is through C4ISR-compatible APIs and ESRI-compatible exports, with support for modular cloud environments including NATO's Maven Smart System, and detections can cue tasking of other ISR assets. Monitoring scales from individual installations to continental theatres, and the models are retrained and audited for drift quarterly.

For civil government, the same capability is applied to deforestation and land-use change, natural hazards and disaster progression, urban expansion and infrastructure monitoring at national scale. EarthDaily is applying the approach in SMATAK, a multi-year border-monitoring project with Defence Research and Development Canada and the RCMP, due to conclude with an end-to-end demonstration in 2027.