The CSV has columns: t_ms, ax, ay, az, mag

• t_ms: milliseconds since recording start
• ax, ay, az: linear acceleration on each axis (gravity removed when available)
• mag: √(ax² + ay² + az²) — magnitude

To look for earthquake-like signals: filter for sustained elevated mag (not single spikes — those are bumps). Real quakes show a P-wave then a larger S-wave over several seconds. Plot magnitude over time in Python/pandas, Excel, or feed into a spectrogram (FFT) to spot the 1–10 Hz seismic band.

You can also cross-check timestamps against earthquake.usgs.gov.

A phone accelerometer is roughly 1000× less sensitive than a real seismometer. Realistically you'll detect:

• ✓ Trucks, footsteps, doors slamming
• ✓ M4+ quakes within ~50 km
• ✓ M5+ within ~200 km
• ✗ Distant teleseismic waves
• ✗ Anything below the phone's noise floor (~0.01 m/s²)

iOS Safari will likely throttle motion events when screen locks. Keep screen on for best results.