SUMMARY
Ambient noise autocorrelations can be used to reconstruct the seismic reflection response of the Earth structure beneath single stations using continuous recordings without the need for either active sources or earthquakes. In the last decade, this technique has emerged as an inexpensive approach with the potential to provide similar information to that from the classical receiver function (RF) analysis. Previous studies have located and mapped discontinuities at different crustal depths with ambient noise autocorrelation functions (ACFs) by applying different processing techniques. An ambient noise ACF provides the body-wave reflectivity of the local structure, assuming a homogeneous distribution of noise sources. An effective method design is required in order to determine a reliable reflection response. Here, we review the theory behind the ambient noise ACF method and design a workflow to obtain the P-wave reflectivity with a special focus on the Moho depth. In particular, we calculate a smooth function to fit and subtract the zero-lag component in the time domain, i.e., the large-amplitude signal near 0 lag time in the ACF. The zero-lag component can interfere with the reflection component, so its removal allows us to increase the frequency band to use. We band-pass filter the ACFs between 1 and 6 seconds. We also derive and apply a phase shift correction in the ACFs due to the integration of a homogeneously distributed noise field dominated by distant sources from deep below, such as teleseismic sources. Both linear and a non-linear, phase-weighted stacks are used. Linear stacking is used to identify the main interfaces since it ensures the linearity of the processing steps; nevertheless, non-linear, phase-weighted stacking help validate the coherent signals. We test and apply our method to continuous vertical recordings from three stations in Ireland and five stations in different cratons and obtain clear P-wave reflection from the Moho and other crustal and upper-mantle discontinuities in most cases. However, noise coming from local heterogeneities, non-homogeneous distribution of the ambient noise sources or instrumental noise is also expected. Therefore additional, a priori information is desirable to help identify key phases in single ACFs. We compute synthetic ACFs using P-wave velocity (VP) models from controlled-source profiles in Ireland. The relatively complex ACF traces obtained at the stations in Ireland show a close data-synthetic match for the Moho and mid-crustal discontinuities. The ACF traces from the stations in different cratons are directly compared with receiver functions showing overall agreement and offering complementary information on the origin of the signal.