The ground has just moved. Find where.
- Choose a caseAn earthquake recorded by our stations.
- Pick the wavesOn each record click the P-wave, then the S-wave.
- LocateWadati diagram, least squares and the map.
- Get reviewedFive checks, graded from A to D.
Open a case
Loading the records…
The theory behind the game
Locating an earthquake means finding four unknowns: two coordinates of the epicentre, the depth and the origin time. All we have are the times at which the waves reached the stations.
1. Two waves, two speeds
A P-wave and an S-wave leave the source together at the origin time t₀ and travel the distance r with speeds Vp and Vs:
t_P = t₀ + r / Vp, t_S = t₀ + r / Vs
Subtracting removes the unknown origin time:
r = (t_S − t_P) × Vp·Vs / (Vp − Vs)
With Vp = 6.0 km/s and Vs = 3.5 km/s in the crust the factor is 8.4 km/s. This is the quick rule “about 8 km for every second”, and one distance from each station gives a circle on the map.
2. The Wadati diagram
In 1933 the Japanese seismologist Kiyoo Wadati noticed that the same two equations give a straight line:
t_S − t_P = (Vp / Vs − 1) × (t_P − t₀)
Plot S − P against the arrival time of P for several stations. The slope of the line gives the ratio Vp / Vs, and the point where the line crosses zero is the origin time, with no need to know where the earthquake was. For an ideal elastic solid Vp / Vs = √3 ≈ 1.73; real crust is usually between 1.7 and 1.8.
3. Least squares: Geiger’s method
In 1910 Ludwig Geiger proposed the method on which earthquake location still rests. Choose a trial source and origin time, calculate when the waves should arrive at each station, and compare with what was observed. The differences are the residuals. The best location is the one that makes the sum of squared residuals smallest:
Σ (t_observed − t_calculated)² → minimum
The game does exactly this with your six arrival times, trying thousands of points on the map. The RMS residual shows how well one source explains all your picks: a few tenths of a second is good.
4. How the reviewer checks you
Picking: each of your six picks is compared with the time read by our analysts. Fit: the RMS residual of the least-squares solution. Geometry: the azimuthal gap. Stability: the location is repeated 40 times with every pick shifted at random by up to 0.2 s, a typical reading error; the spread of the results is the uncertainty of your epicentre. Result: the distance to the epicentre of the network, which was found with more stations.
5. Why three stations are not enough
With three stations and a simple model the solution is possible but fragile. Seismologists look at the azimuthal gap, the largest angle around the epicentre with no station in it. When all stations lie on one side, as for earthquakes outside the network, the distance is well known but the direction is not, and depth can hardly be found at all. More stations, on all sides, and a layered model of the crust are what make a network location reliable.