Reduced-order modelling of Cascadia’s slow slip cycles

SUMMARY Slow slip events (SSEs) modulate the earthquake cycle in subduction zones, yet understanding their physics remains challenging due to sparse observations and high computational cost of physics-based simulations. We present a scientific machine-learning approach using a data-driven reduced-order modelling (ROM) framework to efficiently simulate the SSE cycle governed by rate-and-state friction in a Cascadia-like 2-D subduction setting. Our approach projects fault slip, slip rate and state-variable trajectories onto a spline-based latent space, which is subsequently emulated using proper orthogonal decomposition and radial basis function interpolation. Achieving a speedup of $\sim 360\,000 \times$ compared to volumetric simulations, the ROMs enable comprehensive parameter exploration and Bayesian Markov chain Monte Carlo (MCMC) inversion. By smoothly interpolating between the physics-based simulations, the ROMs reveal complex dependencies that might be overlooked with coarser parameter space sampling. Our analysis reveals complex, nonlinear dependencies of SSE characteristics on the width and magnitude of the deep, low-effective-normal-stress region while holding frictional parameters constant. We show that, to first order, the recurrence time of SSEs is linearly dependent on the normalized fault width, defined as the SSE zone width divided by the characteristic nucleation length, in agreement with previous work. However, at second order, the recurrence interval can change more rapidly with small variations in the SSE zone width. We identify a region of steep, nonlinear dependence of the recurrence interval on the normalized fault width, which we attribute to the extent of the velocity-weakening portion of the subduction interface that produces SSEs. Our MCMC inversion constrained by Northern Cascadia SSEs observations indicates near-lithostatic pore fluid pressure ($99.6 \pm 0.17 \, \rm per\, cent$ lithostatic) and positions the upper frictional transition zone at $30.4 \pm 2.8$ km depth, consistent with geophysical observations. The inversion resolves the deep SSE-portion of the slab spanning $45 \pm 16$ km with low effective normal stress of $3.8 \pm 1.4$ MPa. We discuss how varying the fault frictional parameters affects the MCMC-inverted parameter values. This framework provides a new tool for advancing the physics-based understanding of SSEs and subduction zone faulting mechanics. By systematically linking megathrust properties such as fluid pressure and fault strength to rate-and-state friction governed slow slip cycle characteristics, such as recurrence interval, our approach helps to constrain the first and second-order physics-based controls and the uncertainties of how subduction zones slip.