How to Detect Water in the Mantle Wedge of a Subduction Zone Using Seismic Anisotropy

In approximately one fourth of worldwide subduction zones, seismic observations indicate a rotation from trench‐normal to trench‐parallel fast axis orientations in the mantle wedge. To interpret this observation we predict the evolution of crystal lattice preferred orientation in mantle wedge material as a function of the amount of water by using a model of polycrystal deformation (D‐Rex) coupled with an analytical corner flow. The resulting seismic signature is obtained from synthetic seismic wave propagation through this mantle wedge. We identify that progressive hydration produces the rotation of fast axis orientations and can generate between the two zones of trench‐parallel and trench‐normal fast axis orientations a morph zone with very small anisotropy and a related decrease in P and S wave velocities. Such a morph zone is not produced by trench‐parallel flow, hence this signature can be used to detect water in the mantle wedge. Plain Language Summary A subduction zone's mantle wedge can have a complex pattern of seismic anisotropy where the fast direction often rotates from trench‐parallel close to the trench to trench‐normal in the backarc. This pattern can be interpreted as induced by either 3‐D trench‐parallel flow or by the presence of water close to the trench. Almost all models so far favored the trench‐parallel flow hypothesis, usually based on indirect or complementary indicators such as the evolution of geochemical signatures of volcanoes along the arc. Here we examine a seismic anisotropy observational signature that can be used to discriminate between the two explanations. The concept is defined using an interdisciplinary approach linking a direct modeling of the flow in the subduction wedge and a computation of seismic wave propagation in anisotropic media. We define a unique water‐induced signature that is the presence of a “morph zone” characterized by a weak anisotropy and a decrease of seismic velocities. We apply the model to the Lau Basin where we find this predicted signature, demonstrating for the first time that water rather than trench‐parallel flow is responsible for the observed anisotropy pattern there. Key Points Development of trench‐parallel anisotropy in a mantle wedge is modeled using a model of crystals deformation and dynamic recrystallization Progressive hydration toward the trench produces a “morph zone” with a weak anisotropy and a decrease of VP and VS Applied to the Lau Basin the model shows that water rather tha