Radial profiles of seismic attenuation in the upper mantle based on physical models

S U M M A R Y Thermally activated, viscoelastic relaxation of the Earth’s materials is responsible for intrinsic attenuation of seismic waves. Seismic observations have been used to define layered radially symmetric attenuation models, independent of any constraints on temperature and composition. Here, we interpret free-oscillation and surface wave attenuation measurements in terms of physical structures, by using the available knowledge on the physical mechanisms that govern attenuation at upper-mantle (<400 km) conditions. We find that observations can be explained by relatively simple thermal and grain-size structures. The 1-D attenuation models obtained do not have any sharp gradients below 100 km, but fit the data equally well as the seismic models. The sharp gradients which characterize these models are therefore not required by the data. In spite of the large sensitivity of seismic observations to temperature, a definitive interpretation is limited by the unknown effects of pressure on anelasticity. Frequency dependence of anelasticity, as well as trade-offs with deeper attenuation structure and dependence on the elastic background model, are less important. Effects of water and dislocations can play an important role as well and further complicate the interpretation. Independent constraints on temperature and grain size expected around 100 km depth, help to constrain better the thermal and grain-size profiles at greater depth. For example, starting from a temperature of 1550 K at 100 km and assuming that the seismic attenuation is governed by the Faul & Jackson’s (2005) mechanism, we found that negative thermal gradients associated with several cm grain sizes (assuming low activation volume) or an adiabatic gradient associated with ∼1 cm grain size, can explain the data. A full waveform analysis, combining the effects on phase and amplitude of, respectively, elasticity and anelasticity, holds promise for further improving our knowledge on the average composition and thermal structure of the upper mantle.