= Solution
A dense array does not correct an incorrect <velocity> model. The faster southwest block permits shorter <travel times> than an assumed uniform medium, with <refraction> favouring paths through fast rock. First arrivals may include refracted or <seismic head wave> paths rather than the assumed direct straight paths. Fitting those arrivals with one constant <speed> can make an <earthquake> on the fault appear closer to southwest stations and farther from northeast stations.
The sign follows even from a two-station <linearization>. Put southwest and northeast stations at $x=-L,+L$, and the true event at $x=0$ with depth $h$. In a constant-speed model,
$$
\delta(t_{\rm SW}-t_{\rm NE})
\simeq\frac{2L}{v\sqrt{L^2+h^2}}\,\delta x.
$$
A negative observed differential residual therefore produces \b[$\delta x<0$, an apparent southwest displacement]. A common origin-time adjustment cannot remove this directional residual. The stated kilometre-scale offset is a model bias, not proof that the fault dips towards the apparent <epicentres>. Its exact magnitude cannot be predicted without the actual velocities and station geometry; the question supplies that magnitude as an observation.
The relevant contrast is independently documented at Parkfield: higher velocities lie southwest of the fault, with lower velocities northeast.<Scientific Advances at Parkfield by United States Geological Survey>[Velocity-structure discussion, higher speeds southwest and lower speeds northeast]{ref} Account for the lateral structure and correct phases, then relocate and test whether the bias remains.
\Image[/past-exam-of-the-physics-of-the-earth-as-a-planet-course-of-the-university-of-cambridge/2014/paper-1-pep-location-bias.png]
{title=Illustrative velocity contrast and the direction of a constant-speed location bias}
Back to article page
