Source effects on surface wave group travel times and group velocity maps

In most seismic surface wave studies observed group travel times are interp reted as time delays due entirely to the wave propagation along the wave pa th, and source effects are considered as negligibly small. This is in contr ast with observed phase times where correction for the source phase is gene rally acknowledged to be mandatory. An important, yet unanswered, question is how neglecting source group time (SGT) in broadband surface wave studies will affect the accuracy of the measured group velocity curves and the tom ographic maps constructed from these measurements. We consider here the eff ect of SGT on group velocity measurements for fundamental Rayleigh waves an d report on its dependence on period (10-200 s), source mechanism, and sour ce depth. Varying these parameters strongly affects the magnitude and azimu thal pattern of SGT shifts and we present statistics of certain salient fun ctionals that characterize this dependence. SGT is negligible for periods l ess than about 75 s and for earthquake shallower than about 25 km. At longe r periods and for deeper events, average SGT corrections are greater than 1 0 s in magnitude, which for continental scale studies translates into group velocity perturbations of 1-2%. We estimate the bias caused by uncorrected SGT in inversions for Rayleigh wave group velocity maps across the Eurasia n continent. The largest perturbations to these maps (up to 1-2% for the 50 -s period and up to 5% for the 100-s period) are found near the periphery o f the continent where ray coverage is poor. From these results, some statis tical estimates for adjacent wave paths (clusters), and the fact that SGT c orrections display considerable sensitivities to earthquake depths, we conc lude that the effects of SGT on group velocity tomographic images may safel y be ignored at periods less than about 75 s and for shallow sources. Altho ugh such corrections are appreciable at longer periods for events deeper th an about 25 km and should in principle be applied, the inherent inaccuracy of present day CMT solutions and group velocity measurements make these cor rections practically non-essential for current group velocity tomographic s tudies. (C) 1999 Elsevier Science B.V. All rights reserved.