Rikuto Fukushima | Searcharxiv

arXiv2026

Physics-informed deep learning links geodetic data and fault friction

Fault slip modeling, based on laboratory-derived friction laws, has significantly enhanced our understanding of fault mechanics. Agreement between model predictions and observations supports the hypothesis that observed slip diversity, including fast earthquakes and slow transient slips (Slow Slip Events; SSEs), originates from frictional heterogeneity. However, quantitative assessments of frictional heterogeneity from geodetic observations while fully incorporating fault mechanics are lacking due to the difficulties of high-dimensional optimization. In this study, we aim to address this gap using Physics-Informed Neural Networks (PINNs) to link frictional heterogeneity with geodetic observations. PINNs employ a neural network to represent the spatially variable frictional properties, making their estimation feasible. Targeting the 2010 Bungo SSE in southwest Japan, our estimation reveals heterogeneous friction coinciding with localized SSE nucleation in southwest Shikoku, and subsequent westward propagation. The calculated fault slip of SSE successfully reproduces the spatio-temporal pattern of observed surface displacements. This PINN-based inversion provides a mechanically consistent fault slip model validated through quantitative comparison with observations. Furthermore, we forecast the future fault slip evolution, demonstrating the importance of assimilating observations spanning multiple SSE cycles. Our results demonstrate the potential of PINN for advancing understanding of fault mechanics and enabling physics-based fault slip forecasting.

physics.geo-ph↗

arXiv2026

PINN-based short-term forecasting of fault slip evolution during the 2010 slow slip event in the Bungo Channel, Japan

Monitoring and forecasting fault slip evolution are fundamental for understanding earthquake cycles and assessing future seismic hazards. This study proposes a physics-based data assimilation framework that integrates geodetic observations with fault mechanics introducing spatial heterogeneity in frictional properties, with a particular focus on short-term fault slip forecasting. The proposed method employs physics-informed neural networks (PINNs) to calculate fault slip evolutions and to optimize the spatial distribution of frictional properties and is applied to the 2010 slow slip event beneath the Bungo Channel, southwest Japan, by changing the data period to be assimilated. When only the initial phase of slip acceleration is assimilated, a velocity-weakening frictional region is inferred beneath southwest Shikoku, corresponding to the initial nucleation are of the slow slip event. Out results demonstrate that the PINN-based data assimilation framework successfully forecasts slow transient slip even when only slip acceleration data are assimilated, whereas forecasts based on frictionally homogeneous models result in unstable fast slip. This difference can be interpreted as a consequence of introducing frictional heterogeneity, which allows both the characteristic size of the slipping region and the critical nucleation size to be variable, leading to stable slip evolution consistent with observations. When longer observation periods are assimilated, a velocity-strengthening region emerges around the slip-weakening patch, progressively restricting the direction of slip propagation. This velocity-strengthening region is interpreted as a mechanical constraint imposed by fault physics, linking the slip regions required to reproduce the observed geodetic time series. The results highlight the capability of PINN-based data assimilation incorporating geodetic observations and fault mechanics.

physics.geo-ph↗