GNSS Live monitors electromagnetic anomalies in the world's most strategically important maritime chokepoints using NASA's CYGNSS satellite constellation. When military radar, electronic warfare systems, or unusual vessel activity disturbs the RF environment over the ocean surface, CYGNSS detects it.
The platform provides passive civilian satellite intelligence derived entirely from publicly available NASA data — no classified sources, no proprietary sensors.
Anomaly spikes have been independently validated against confirmed Houthi attack events across the October 2023 — October 2025 campaign period.
| Date | Event | Match |
|---|---|---|
| 27 Nov 2023 | Missiles at USS Mason | Confirmed |
| 13 Dec 2023 | Ardmore Encounter attack | Confirmed |
| 15 Dec 2023 | Maersk suspends ops | Confirmed |
| 01 Jan 2024 | Iranian warship Alborz | Confirmed |
| 16 Jan 2024 | Post-Poseidon Archer | Confirmed |
| 19 Feb 2026 | Houthi mobilisation | Confirmed |
| Tier | Lift vs baseline | Near-attack rate |
|---|---|---|
| All events | 2.6x | 36.6% |
| Dual-signal only | 3.6x | 50.0% |
Baseline rate: 13.8% (random chance)
CYGNSS Level 2 Science Data Record V3.2, sourced from NASA PO.DAAC under Creative Commons Zero licence. Each daily file contains all eight satellite observations merged into a single NetCDF file (~150MB). The Red Sea bounding box (lat 11-20°N, lon 41-45°E) yields approximately 5.3 valid observation hours per day.
The Normalised Bistatic Radar Cross Section (NBRCS) measures GPS signal reflectivity from the ocean surface. Elevated NBRCS relative to a 24-hour rolling baseline indicates unusual electromagnetic conditions — consistent with military radar, electronic warfare systems, or abnormal vessel traffic.
A rolling z-score is computed per hour against a 24-hour window. Events exceeding z > 1.5 are flagged as anomalies.
Inspired by Gorman (GPS World, May 2026), a second detection channel uses cross-channel NBRCS variance — the standard deviation of NBRCS readings across the five DDM channels within each observation. When a jammer is active, channels disagree significantly, elevating this variance independently of the mean signal level.
Events where both channels fire simultaneously are flagged as dual-signal — the highest confidence detection tier with 3.6x lift over baseline.
For significant events, a parametric 1/r² inverse-distance model is fitted to the spatial distribution of anomalous observations. The source position is optimised using SciPy Nelder-Mead. Confidence is reported as CEP — circular error probable — estimated via 200-iteration bootstrap resampling.
L1 reactive download is triggered for events above z > 2.5, enabling four-method detection (noise floor, spatial grid, SNR hole, NBRCS drop) and sub-20km localisation accuracy.
CYGNSS covers latitudes 38°N to 38°S. Coverage above 38°N (Ukraine, Baltic, Nordic) is not available from this data source. Localisation accuracy varies from <15km on clean single-source days to >100km on multi-source or low-observation days. All classifications are probabilistic estimates, not confirmed attributions.
All endpoints return JSON. Authenticate with your API key in the request header.
For API access, enterprise data feeds, or custom zone coverage — contact harry@gnsslive.io
All derived intelligence is produced from NASA CYGNSS data (CC0 licence). Source data attribution required. No warranty on accuracy — all classifications are probabilistic estimates for situational awareness only.