Locate the Quake

A seismology lab: find a hidden earthquake from noisy seismograms, size it, map its shaking and see how much early-warning time a network can buy.

Click a seismogram to place a pick on that station. Use P on the first onset (vertical trace) and S on the big horizontal arrival. Wheel to zoom time, drag to pan. Alt-click or right-click removes a pick.

Display gain ×1.0

Each station with both picks gives a distance: d = (t_S − t_P) / (1/V_s − 1/V_p) ≈ 8.4 km/s × (S−P). The source lies on a circle around it. With noisy picks the circles do not meet at a point.

Assumed depth for the circles 10 km

The locator searches every point (x, y, depth) for the one whose predicted P and S times best match your picks. Origin time is solved at each node. The misfit map shows how well-determined the position is.

Local magnitude M_L uses the peak amplitude on a simulated Wood-Anderson seismometer with a distance correction (−log A₀). Moment magnitude M_w comes from the low-frequency plateau of the S-wave spectrum, corrected for spreading and attenuation.

Spectrum window 20 s

A ground-motion prediction equation gives PGA and PGV from magnitude and distance; the soil layer amplifies them; the result is converted to Modified Mercalli intensity (MMI) and a rough damage estimate per town.

Include soil amplification

Stations trigger on the first P wave (STA/LTA). When K stations agree, an alert is issued after the processing latency. Towns whose S-wave arrives later than the alert get a warning; the rest are in the blind zone.

Processing + telemetry latency 4.0 s Stations needed to alert (K) 4 STA/LTA threshold 4.0

Drag stations to move them, or switch tool to add or remove stations on the map.

Time since origin 0.0 s

Model: V_p = 6.0, V_s = 3.5 km/s crust over a 25 km Moho (V_n = 8.0 km/s). The locator assumes a uniform half-space, so distant picks are slightly biased, as in real networks. All data are synthetic.