Monitoring and predicting fault slip behaviors in subduction zones are essential for understanding earthquake cycles and assessing future earthquake potential. We developed a data assimilation (DA) method for fault slip monitoring and short-term prediction of slow slip events (SSEs), which was applied to the 2010 Bungo Channel SSE in southwest Japan. The observed geodetic data were quantitatively explained using a physics-based model with DA. We investigated short-term predictability by assimilating observation data with limited periods. Without prior constraint on fault slip style, observations solely during slip acceleration predicted the occurrence of a fast slip; however, the inclusion of slip deceleration data successfully predicted a slow transient slip. With prior constraint to exclude unstable slip, the assimilation of data after the SSE occurrence predicted a slow transient slip. This study provided a tool using DA for fault slip monitoring and prediction based on real observation data.

Geodetic fault slip inversions have been generally performed by employing a least squares method with a spatial smoothing constraint. However, this conventional method has various problems: difficulty in strictly estimating non-negative solutions, assumption that unknowns follow the Gaussian distributions, unsuitability for expressing spatially non-uniform slip distributions, and high calculation cost for optimizing many hyper-parameters. Here, we have developed a trans-dimensional geodetic slip inversion method using the reversible-jump Markov chain Monte Carlo (rj-MCMC) technique to overcome the problems. Because sub-fault locations were parameterized by the Voronoi partition and were optimized in our approach, we can estimate a slip distribution without the spatial smoothing constraint. Moreover, we introduced scaling factors for observational errors. We applied the method to the synthetic data and the actual geodetic observational data associated with the 2011 Tohoku-oki earthquake and found that the method successfully reproduced the target slip distributions including a spatially non-uniform slip distribution. The method provided posterior probability distributions with the unknowns, which can express a non-Gaussian distribution such as large slip with low probability. The estimated scaling factors properly adjusted the initial observational errors and provided a reasonable slip distribution. Additionally, we found that checkerboard resolution tests were useful to consider sensitivity of the observational data for performing the rj-MCMC method. It is concluded that the developed method is a powerful technique to solve the problems of the conventional inversion method and to flexibly express fault-slip distributions considering the complicated uncertainties.

We consider a Bayesian multi-model fault slip estimation (BMMFSE), in which many candidates of the underground-structure model characterized by a prior probability density function (PDF) are retained for a fully Bayesian estimation of fault slip distribution to manage model uncertainty properly. We performed geodetic data inversions to estimate slip distribution in long-term slow slip events (L-SSEs) that occurred beneath the Bungo Channel, southwest Japan, in around 2010 and 2018, focusing on the two advantages of BMMFSE: First, it allows for estimating slip distribution without introducing strong prior information such as smoothing constraints, handling an ill-posed inverse problem by combining a full Bayesian inference and accurate consideration of model uncertainty to avoid overfitting; second, the posterior PDF for the underground structure is also obtained through a fault slip estimation, which enables the estimation of sequential events by reducing the model uncertainty. The estimated slip distribution obtained using BMMFSE agreed better with the distribution of deep tectonic tremors at the down-dip side of the main rupture area than those obtained based on strong prior constraints in terms of the spatial distribution of the Coulomb failure stress change. This finding suggests a mechanical relationship between the L-SSE and the synchronized tremors. The use of the posterior PDF for the underground structure updated by the estimation for the 2010 L-SSE as an input of the analysis for the one in 2018 resulted in a more preferable Bayesian inference, indicated by a smaller value of an information criterion.

Recently, Uchida et al. [2016, Science] revealed the periodic changes in the interplate locking in the northeast Japan subduction zone based on the activity of small repeating earthquakes and terrestrial crustal deformation data. They found that slow slip on the plate interface has occurred repeatedly at intervals of from 2 to 6 years, depending on the location. In the northern part of the Japan Trench, a tsunami earthquake occurred with rupturing the shallow plate boundary in 1896, but it is not well understood whether the coseismic slip fully released the interseismic slip deficit, and whether periodic slow slip events occur near the trench or not. In order to investigate the interplate locking state in this region, we have just started a research project titled “Head and tail of massive earthquakes: Mechanism arresting growth of interplate earthquakes” (JSPS KAKENHI Grant Number JP19H05596), under which GNSS-Acoustic observation to detect the seafloor crustal deformation will be performed more frequently than ever before. To accomplish frequent observations, we have been developing automatic GNSS-A data acquisition system using an unmanned surface vehicle, the Wave Glider. As a first observation, we have performed GNSS-A observation at a seafloor station off Aomori Prefecture in this July. The Wave Glider (SV3-240) was equipped with 2 GNSS antennas, acoustic transducer, MEMS gyro, and their control and logging units. The data acquisition from these sensors and the autonomous activation of the seafloor transponders were successfully executed only with turning the power supply to the payload on/off from land via a satellite communication. The Wave Glider rarely strayed off the configured course, and the solar panels generated enough power to perform the observations although the weather was mostly cloudy during the operation. Now, we are processing the obtained data, and the results will be presented at the meeting.