Excitation approximation strategy for reflection traveltime inversion

Full waveform inversion (FWI) plays an important role in the realm of exploration geophysics since its ability to restore high-resolution subsurface structures. Reflection traveltime inversion (RTI) can provide FWI with a good initial model containing accurate low-wavenumber components, which is essential for the success of FWI. However, due to the necessary migration/demigration process, every iteration of RTI requires six times as many forward calculations, and the amount of calculation and storage is several times that of FWI. High-performance computing can accelerate RTI calculations, but they are almost helpless to the consumption of massive storage. Thus, in this letter, the background and perturbed wavefields with excitation approximation are used in RTI to overcome the storage trouble. In our RTI, the strategies of source-wavelet convolution and wavefield direction decomposition are introduced to solve the problems of source signature missing and multipathing in the excitation approximation wavefields. Numerical examples have demonstrated that the excitation approximation RTI can provide good initial models and reduce the storage burden.