Slab temperature beneath the Tohoku region, Northeast Japan, constrained from surface heat flow and the location of the blueschist-out boundary

SUMMARY This study investigates the following three issues in numerical models of the thermal structure of subduction zones, using the Tohoku region in Northeast Japan as an example: (1) A steady state is often assumed in models, (2) quantitative assessment of the uncertainty in the predicted temperatures is lacking and (3) surface heat flow has been used to constrain many of the models. I found that, at least under the model setting of this study, a steady state may be safely assumed as long as only surface heat flow within 150 km of the trench is used to constrain the model. I used Bayesian inference to predict the thermal structure, with surface heat flow near the trench and the location of the blueschist-out boundary in the oceanic crust as observational constraints. The depth of slab–mantle kinematic decoupling, effective friction coefficient and rate of radiogenic heat production in the upper island arc crust were constrained simultaneously to be $\sim$80–100 km, 0.03–0.08 and 1.5–2.16 $\mu$W m$^{-3}$, respectively, although the decoupling depth is sensitive to the assumed location and temperature of the blueschist-out boundary. The uncertainties in slab temperature reach $\sim$450 K at depths of $<$100 km and 100 K for greater depths, which are substantial. To reduce these uncertainties, it is necessary to reduce the uncertainty in the input parameters and obtain additional observational constraints.