Teleseismic P wave spectra from USArray and implications for upper mantle attenuation and scattering

Abstract

Teleseismic P wave amplitude spectra from deep earthquakes recorded by USArray are inverted for maps of upper mantle Δt* for multiple frequency bands within 0.08–2 Hz. All frequency bands show high Δt* regions in the southwestern U.S., southern Rocky Mountains, and Appalachian margin. Low Δt* is more common across the cratonic interior. Inversions with narrower frequency bands yield similar patterns, but greater Δt* magnitudes. Even the two standard deviation Δt* magnitude for the widest band is ∼2–7 times greater than predicted by global QS tomography or an anelastic olivine thermal model, suggesting that much of the Δt* signal is nonthermal in origin. Nonthermal contributions are further indicated by only a moderate correlation between Δt* and P travel times. Some geographic variations, such as high Δt* in parts of the cratonic interior with high mantle velocities and low heat flow, demonstrate that the influence of temperature is regionally overwhelmed. Transverse spectra are used to investigate the importance of scattering because they would receive no P energy in the absence of 3‐D heterogeneity or anisotropy. Transverse to vertical (T/Z) spectral ratios for stations with high Δt* are higher and exhibit steeper increases with frequency compared to T/Z spectra for low Δt* stations. The large magnitude of Δt* estimates and the T/Z spectra are consistent with major contributions to Δt* from scattering. A weak positive correlation between intrinsic attenuation and apparent attenuation due to scattering may contribute to Δt* magnitude and the moderate correlation of Δt* with travel times.

Seismic High Attenuation Region Observed Beneath Southern New England From Teleseismic Body Wave Spectra: Evidence for High Asthenospheric Temperature Without Melt

“…Hwang et al () measures differential attenuation by analyzing teleseismic P wave amplitude and indicates highly attenuating uppermost mantle mainly beneath the border of New York and Vermont. Cafferky and Schmandt () found dispersed high attenuation in the upper mantle beneath Vermont, Western New York, New Jersey, and Southern New England using teleseismic P wave data recorded by USArray. The geographical differences between these studies indicate complexity in the attenuation structure at this region.…”

“…Hwang et al (2009) measured differential attenuation Δt à P range from À0.24 s and À1.09 s within North America, and the mean difference in Δt à P between TNA (Tectonic North America) and SNA (Stable North America) is 0.23 s. In addition to the large-scale variation between TNA and SNA, their Δt à P map also indicates highly attenuating uppermost mantle mainly beneath the border of New York and Vermont ( Δt à P reaches up to 0.4 s comparing to the neighboring stations located at the craton). Cafferky and Schmandt (2015) North America, they measured much dispersed high attenuation in upper mantle beneath Vermont, Western New York, New Jersey, and Southern New England. An approximately linear, north-south striking band of high attenuation region in this area is observed in the vicinity of the Appalachian Mountain.…”

“…Recent seismic attenuation studies across North America show that the upper mantle beneath northeastern North America is highly attenuating compared to the shield (Bao et al, ; Cafferky & Schmandt, ; Hwang et al, ). Hwang et al () measures differential attenuation by analyzing teleseismic P wave amplitude and indicates highly attenuating uppermost mantle mainly beneath the border of New York and Vermont.…”

“…The geographical differences between these studies indicate complexity in the attenuation structure at this region. The relatively large measurement error in both Hwang et al () and Cafferky and Schmandt () (ranges from 15% to 25%) may also contribute to the geographical differences. Furthermore, both teleseismic P wave studies focus on the upper mantle beneath whole North America but not the high attenuated northeastern North America.…”

Global rupture parameters for deep and intermediate‐depth earthquakes

“…We find that a constant t * = 0.3 s provides a good fit to our spectra. This t * value agrees with previous deep earthquake studies [ Ye et al ., ] and with the estimated one from Cafferky and Schmandt []. An alternative approach will be the use of empirical Green's function [e.g., Abercrombie , ], but at teleseismic distances SNR for the candidate EGF would be quite low, especially at long periods.…”