How it works
From a free radar satellite to an alert on an engineer's screen. We buy the satellite data as a commodity and build the part that matters: turning raw radar into a trustworthy answer about whether the ground is moving.
Acquire free radar
We pull the Sentinel-1 radar archive over the asset from the European Copernicus programme, going back to 2015. Radar sees through cloud and works day or night, which optical satellites cannot. For faster revisit we task commercial radar (ICEYE, Capella) only where it is needed.
Bought, not builtMeasure phase change between passes
Each time the satellite passes, we compare the radar phase against earlier passes to form interferograms. A shift in phase corresponds to the ground moving toward or away from the satellite, down to millimetres. We build on pre-processed interferograms from the COMET LiCSAR service where they exist, to skip the slowest step.
Bought, not builtBuild a displacement time series
We run small-baseline InSAR inversion across the whole stack of interferograms to produce a cumulative displacement history for every coherent point in the area, referenced to stable bedrock, and corrected for atmospheric noise. The output is a clean movement curve per point over time.
Detect anomalies
This is the part that is ours. A model separates normal settlement from genuine anomalous movement, flags points that accelerate away from a consolidation trend, and cuts false alarms, using the deep-learning approach behind the landslide-detection model I built and deployed across Fugro. A moving point is only useful once you know it is the dangerous kind of moving.
Rheality's layerReport and alert, with provenance
Findings arrive as a monitoring dashboard, a compliance-ready report, and an alert to the operator's engineer when a point crosses threshold. Every figure carries an audit trail (which satellite scenes, which dates, which settings) so it stands up as independent verification. The engineer decides; we surface the signal early.
Rheality's layerSee it replayed over the 2019 Brumadinho dam collapse.
Open the demo →The optical sibling of this engine runs live over the Three Gorges reservoir today.
How the live pipeline works →Rheality provides decision-support monitoring data, not a safety guarantee. See
Terms & Disclaimer. Rheality Limited, Hong Kong.
Step photos via Wikimedia Commons: Rama (CC BY-SA 2.0 FR); Roberto Monti (CC BY-SA 4.0); Z22 (CC BY-SA 3.0); U.S. Navy (public domain); Blogtrepreneur (CC BY 2.0).