Subduction zones, read in new detail: fossil melt fronts, two-way hydration, and rupture traced to the sky

Fossil fronts in old mantle

In a preprint, subduction zones are being read through rock that preserves an early stage of the process rather than through the active plate boundary alone. Basal lherzolite tectonites in supra-subduction zone ophiolites act as fossil reaction fronts recording slab-melt addition during subduction infancy, so the base of an ophiolite can hold a record of melt moving into mantle rock when subduction was young. m:…7309

The Mont Albert ophiolite shows a systematic compositional, thermal, and strain gradient through its basal mantle, with fertility increasing toward the contact with underthrust lithologies. That gradient gives readers a direction to follow: change is ordered through the rock, not scattered, and fertility rises as the contact is approached. Across a longer view, basal lherzolites in ophiolites show a trend of thinning over the Phanerozoic, potentially reflecting a waning role of slab-melts in nascent subduction processes. m:…7309

Water, measured two ways

A second preprint turns to water carried into subduction zones and to how that water is measured. Combined Vp and Vs observations provide more robust constraints on mantle hydration than single-velocity measurements, because two velocities together narrow what the rock can be holding. The distinction matters for interpretation: water bound in minerals and water free in fractures do not imply the same state of the incoming plate. m:…8462

The preprint reports different patterns at three margins. Cascadia propagator wakes exhibit higher hydration levels than the surrounding mantle. Shumagin mantle low-velocity zones are characterized by trace water in thin cracks rather than significant serpentinization. Nankai subduction zone data are consistent with free water in thicker fractures and moderate serpentinization. This suggests that a low-velocity zone alone does not settle the form of water present, and that the same kind of signal can arise from different combinations of cracks, free water, and altered rock. m:…8462

Two earthquakes, slip in detail

In a preprint on the 2025 Mw 6.6 Sanriku-Oki earthquake in Japan, the main slip region exhibits a maximum slip of 1.2 m and an average stress drop within that region of 1.4 MPa. Those two numbers describe different parts of the event: how far the fault moved at its peak, and how much stress was released on average across the main slip region. The preprint also reports that slow and fast fault-slip phenomena coexist in the Sanriku-Oki region. m:…7747

That coexistence is driven by heterogeneous frictional properties under low effective normal stress and elevated pore-fluid pressure, according to the preprint. In plain terms, the fault does not behave as one uniform surface: variation in friction, combined with low effective normal stress and high pore-fluid pressure, allows slow and fast slip to occur in the same region. A separate preprint reconstructs the 2026 M7.1 Kumamoto earthquake on the Hinagu Fault zone south of the 2016 rupture termination, using a two-segment fault model for coseismic slip. m:…7747

In the Kumamoto model, peak right-lateral slip occurred at shallow depths near the hypocenter, while normal-slip was concentrated at greater depths near a fault bend. The contrast between shallow lateral motion near the hypocenter and deeper normal motion near the bend shows why a single uniform slip picture would miss structure that the two-segment model is built to capture. m:…3019

Rupture traced to the sky

A further preprint tests how earthquake rupture should be represented when modeling coseismic ionospheric disturbances, using the 2025 Mw 7.7 Myanmar earthquake. Finite-fault modeling reduces median timing error for disturbance arrivals to 202 seconds, compared with 383 seconds for point-source modeling. The improvement is large enough to matter for any use of ionospheric signals as a check on rupture, since timing is the quantity being compared. m:…2913

Arrival times and amplitudes are ordered by distance from the fault rupture rather than the epicenter, according to the preprint. That ordering supports the finite-fault picture: the signal follows the extended rupture, not a single point. Misfits in far-field modeling suggest limitations in current propagation models and potential Rayleigh-wave-induced arrivals, so the far field remains the part of the model with the clearest need for revision. m:…2913

What the set leaves open

Taken together, these preprints share a method: read the structure, not only the event. Fossil fronts order melt addition through rock, paired velocities separate forms of water, slip models separate shallow and deep motion, and finite-fault models follow an extended rupture instead of a point. Each step replaces a single summary value with a distribution, a gradient, or a spatial pattern. This suggests that subduction research is gaining precision less from new events than from finer descriptions of events and rocks already in view. m:…7309 m:…8462 m:…2913

The limits are stated inside the work itself. The thinning trend for basal lherzolites is described as potentially reflecting a waning role of slab-melts, not as a settled cause. The far-field ionospheric misfits point to propagation models that still need work. For professional readers, the useful question is therefore narrow: which of these finer descriptions hold when tested against a second margin, a second earthquake, or a second modeling approach. m:…7309 m:…2913