Comparison of ‘whole-pulse’ cross-correlation and Scripps interactive traveltime measurement procedures. Solid and dashed lines display perturbed and unperturbed waveforms, respectively. The centre of the anomaly is situated at a distance S=10λ from the source. (a) Raw seismograms p_het(r, t) and p_hom(r, t) at an on-axis (φ=0°) receiver, situated at a distance R=30λ behind an a=2λ, ε=+3 per cent anomaly (left) and an a=2λ, ε=−3 per cent (right) anomaly. (b) Best-fit alignment of p_het(r, t) and p_hom(r, t−δT_ccm^φ=0), as determined by automated cross-correlation (19). (c) Interactive alignment of the initial upswings of p_het(r, t) and p_hom(r, t−δT_sio^φ=0). The Scripps ‘initial-swing’ traveltime shifts δT_sio^φ=0 are significantly less healed than the automated ‘whole-pulse’ shifts δT_ccm^φ=0.