How large is the subducted water flux? New constraints on mantle regassing rates

Estimates of the subducted water (H2O) flux have been used to discuss the regassing of the mantle over Earth history. However, these estimates vary widely, and some are large enough to have reduced the volume of water in the global ocean by a factor of two over the Phanerozoic. In light of uncertainties in the hydration state of subducting slabs, magma production rates and mantle source water contents, we use a Monte Carlo simulation to set limits on long-term global water cycling and the return flux of water to the deep Earth. Estimates of magma production rates and water contents in primary magmas generated at ocean islands, mid-ocean ridges, arcs and back-arcs are paired with estimates of water entering trenches via subducting oceanic slab in order to construct a model of the deep Earth water cycle. The simulation is constrained by reconstructions of Phanerozoic sea level change, which suggest that ocean volume is near steady-state, though a sea level decrease of up to 360 m may be supported. We provide limits on the return flux of water to the deep Earth over the Phanerozoic corresponding to a near steady-state exosphere (0–100 meter sea level decrease) and a maximum sea level decrease of 360 m. For the near steady-state exosphere, the return flux is 1.4 − 2.0− 0.3+ 0.4 × 1013 mol/yr, corresponding to 2–3% serpentinization in 10 km of lithospheric mantle. The return flux that generates the maximum sea level decrease over the Phanerozoic is 3.5− 0.3+ 0.4 × 1013 mol/yr, corresponding to 5% serpentinization in 10 km of lithospheric mantle. Our estimates of the return flux of water to the mantle are up to 7 times lower than previously suggested. The imbalance between our estimates of the return flux and mantle output flux leads to a low rate of increase in bulk mantle water content of up to 24 ppm/Ga.

► We use a Monte Carlo simulation to set new limits on the mantle H2O return flux.
► Model constrained by estimates of mantle H2O output and sea level reconstructions.
► Old return flux estimates too large to reflect long-term mantle–exosphere exchange.
► Our new estimate suggests a low net regassing rate of 24 ppm H2O per Ga.
► Low bulk slab H2O content suggests juvenile H2O is required for ocean island source.