Abstrak

The Gaussian wavelet is widely used as a shaping wavelet for scattered wave imaging with P receiver functions due to widespread use of the iterative deconvolution method. We show the Gaussian wavelet degrades the resolution of plane wave migration by comparing results from the latest USArray data shaped with Gaussian and Ricker wavelets. We use simulations of primary conversions from the 410 and 660 km discontinuity to show this is a property of the algorithm and not the data. Simulations also show the more conventional common conversion point (CCP) method is not subject to this behaviour for flat horizons, but the CCP method penalizes dipping horizons focusing only nearly horizontal features for any choice of shaping wavelet. We explain these results using the concept of migration impulse response for an individual data sample. Applications to data from USArray show dramatic improvements in the resolution of plane wave migration images produced using Ricker wavelet in comparison to a comparable resolution a Gaussian shaping wavelet. The 410 and 660 discontinuities are resolved to higher precision, and we find the upper mantle and transition zone are full of previously unresolved dipping horizons that remain to be interpreted.