← Home

Forensic Waterfall

ENF chronolocation waterfall — cross-correlating a recording against months of power grid reference data

6
Correlation:
0.00.51.0
Match Detail
Grid ENF
Extracted ENF
Months Scanned
-
Peak PMCC
-
Chronolocation Match
-
Temporal Resolution
-
Peak Prominence
-
ENF Grid Region
-

Multi-period search strategy: When the recording date is unknown, the ENF fingerprint must be correlated against every position in potentially months or years of reference data. This waterfall visualizes that search — each row scans one month of TSO reference data at hourly resolution.

Computational cost: A 30-minute recording correlated against 6 months of reference data requires ~4,320 PMCC evaluations (720 hours × 6 months). Each evaluation computes Pearson correlation over 1,800 data points (30 min × 60 samples/min). Real-world searches over years of data use FFT-based cross-correlation for efficiency.

Peak prominence: The confidence metric measures how much the best match stands out from the background. A true match typically has PMCC > 0.7, while background noise correlations hover around 0.05–0.25. High prominence (z > 5) indicates a reliable chronolocation.

Reference databases are available from power-grid-frequency.org (CC-By 4.0), UK National Grid ESO, and GridRadar.net (REST API).

Note: All data shown is synthetic. Grid frequency patterns are procedurally generated using a mean-reverting stochastic process. No actual grid recordings or reference databases are used. Built in collaboration with enf_fingerprinting_tool (Bellingcat).