Detectability of Melt-Rich Lenses in Magmatic Reservoirs From Teleseismic Waveform Modeling

Seismic imaging is the most commonly used technique to constrain the size, geometry, and current state of active magma reservoirs in the Earth's crust. However, unequivocal detection of eruptible magma bodies (>0.5 melt fraction) have not been reported yet. In this study, using teleseismic waveform modeling applied to synthetic data, we investigate the limitations of seismic inversions on resolving idealized melt-rich layers with thicknesses smaller than the characteristic wavelength of the teleseismic waves. We show that inverting for melt-rich layers with thicknesses of about 0.2 km, consistent with the average thickness of erupted layers associated with voluminous caldera eruptions, yields significant underestimation of the inferred melt fractions and overestimation of the thickness of that layer. We further extend our synthetic tests to study the effect of noise, variable thickness of the input melt-rich layer, and different time windows from the seismograms on the quality of inverted parameters (thickness and melt fraction of the melt-rich layer). We find that (1) thicker melt-rich layers can be better resolved, (2) longer time windows (similar to 4-6 s) from seismograms reduce the trade-off between melt fraction and layer thickness observed in the inversion results, and (3) noise emphasizes the melt fraction-thickness trade-off and leads to a bias toward thicker lower melt fraction layers retrieved from the inversions. Our synthetic study therefore supports that the lack of detection of melt-dominated horizons from seismic inversions does not preclude their existence and can be caused by a current limit in our ability to detect them.