Extended fault radiation, rupture propagation & directivity — graduate level
The Haskell model represents a finite rectangular fault of length L and width W. Slip D propagates unilaterally at constant rupture velocity Vr, producing a moving dislocation on the fault plane.
Each point on the fault slips at rate ṡ(t) = D/τ · [H(t) − H(t−τ)] — a boxcar of duration τ (the rise time). The far-field displacement is proportional to the moment rate M̊₀(t), which is the spatial integral of ṡ over the fault as the rupture front sweeps at Vr. This integral produces a trapezoid — the Haskell STF — whose flat top has duration Tr(θ) = (L/Vr)(1 − Vr/c · cosθ) and total duration Tr + τ. No further differentiation is needed: M̊₀(t) is already the far-field waveform shape.
Observers in the rupture propagation direction (θ = 0°) see a compressed, high-amplitude pulse. Observers in the anti-rupture direction (θ = 180°) see a long, low-amplitude pulse. The apparent duration is:
The Fourier transform of the trapezoidal STF gives the ω² (Brune) spectral shape. Two corner frequencies emerge from the finite fault: fc1 from rise time τ and fc2 from the rupture duration L/Vr. Directivity shifts fc2 with azimuth.
The Haskell model links observable seismogram properties to physical fault parameters through well-defined scaling relations. Fault dimensions follow the empirical regressions of Wells & Coppersmith (1994, BSSA): log L = −2.44 + 0.59·Mw and log W = −1.01 + 0.32·Mw (all faults). Drag the moment magnitude to see how all quantities scale.