Detection and modelling of strong topography of mid-mantle structures beneath the North Atlantic

Abstract

SUMMARY There is mounting evidence for the presence of seismic reflectors in the mantle at a depth of around 1000 km, however, the cause for these reflectors is not yet well established and published observations show a range of depth estimates for these structures. Discussion is also still ongoing whether it is one global or several different reflectors. Here, we investigate the presence of seismic reflectors beneath the North Atlantic using a large number of PP and SS underside reflections. We analyse over 2600 earthquakes with Mw ≥ 5.7 and use array seismic methods to improve the visibility of the small-amplitude reflected signals. The measured time lag between PP/SS arrivals and their corresponding precursors on robust stacks are used to estimate the depth of the reflector. Our results reveal the presence of mid-mantle structure beneath the North Atlantic in a depth range of ∼700–1300 km, consistent for both P- and S-wave observations. The reflector depth is shallower than 1000 km beneath the southern part of the investigation area and deepens seemingly abruptly towards the northern part of the North Atlantic. We find polarity variations in a region of strong depth change which we assume to be due to wave interference. Using 3-D waveform modelling, we implement models with strong topography as well as models with two overlapping reflectors and two separated reflectors and show that a large step in reflector depth over a short lateral distance, as imaged with our data set, can be resolved, while the discrimination between strong topography and two overlapping and separated reflectors, respectively, is also possible. The variations in precursor polarity can also be observed in our synthetic data in the region of strong topography confirming our assumption of wave interference for generating apparent precursor polarity and waveform changes. While the opposite polarity can be produced by the two overlapping reflectors as well, the strong topography model is preferred in this region, since we do detect only one reflected signal in our observations.