LAUSR

We demonstrate double-difference (DD) tomography, a method involving differential measurements between stations, for 2-D and 3-D adjoint inversions based on realistic source–receiver distributions, with a focus on the global scale. We first present 2-D synthetic inversion results using cross-correlation traveltime and L2 waveform difference objective functions. Introducing a weighting term to DD objective functions based on the number of measurement pairs per station speeds up convergence and reduces bias in the final inverted model due to uneven data coverage. We next demonstrate frequency-dependent multitaper DD measurements in a 3-D experiment with real earthquake data by computing global-scale gradients. At the global scale, careful selection of station pairs is required for differential measurements in terms of geographical distance or geological context. In our technique, if no suitable pairs are found for a particular station, the DD measurement reduces to a classical misfit measurement. Furthermore, we compare 2-D and 3-D DD results with those from corresponding conventional misfits. By exploiting previously unused information in the recorded wavefield, DD tomography shows promise for balancing the gradient and speeding up convergence, especially around dense regional seismic networks.