Machine learning and earthquake forecasting-next steps

Deep learning forecasting of large induced earthquakes via precursory signals

Precursory phenomena to earthquakes have always attracted researchers' attention. Among the most investigated precursors, foreshocks play a key role. However, their prompt identification with respect to background seismicity still remains an issue. The task is worsened when dealing with low-magnitude earthquakes. Despite that, seismology and, in particular real-time seismology, can nowadays benefit from the use of Artificial Intelligence (AI) to face the challenge of effective precursory signals discrimination. Here, we propose a deep learning method named PreD-Net (precursor detection network) to address precursory signal identification of induced earthquakes. PreD-Net has been trained on data related to three different induced seismicity areas, namely The Geysers, located in California, USA, Cooper Basin, Australia, Hengill in Iceland. The network shows a suitable model generalization, providing considerable results on samples that were not used during the network training phase of all the sites. Tests on related samples of induced large events, with the addition of data collected from the Basel catalogue, Switzerland, assess the possibility of building a real-time warning strategy to be used to avoid adverse consequences during field operations.

Sharpen data-driven prediction rules of individual large earthquakes with aid of Fourier and Gauss

Predicting individual large earthquakes (EQs)' locations, magnitudes, and timing remains unreachable. The author's prior study shows that individual large EQs have unique signatures obtained from multi-layered data transformations. Via spatio-temporal convolutions, decades-long EQ catalog data are transformed into pseudo-physics quantities (e.g., energy, power, vorticity, and Laplacian), which turn into surface-like information via Gauss curvatures. Using these new features, a rule-learning machine learning approach unravels promising prediction rules. This paper suggests further data transformation via Fourier transformation (FT). Results show that FT-based new feature can help sharpen the prediction rules. Feasibility tests of large EQs ([Formula: see text] 6.5) over the past 40 years in the western U.S. show promise, shedding light on data-driven prediction of individual large EQs. The handshake among ML methods, Fourier, and Gauss may help answer the long-standing enigma of seismogenesis.

Deep Neural Networks for Predicting Single-Cell Responses and Probability Landscapes

Engineering biology relies on the accurate prediction of cell responses. However, making these predictions is challenging for a variety of reasons, including the stochasticity of biochemical reactions, variability between cells, and incomplete information about underlying biological processes. Machine learning methods, which can model diverse input-output relationships without requiring a priori mechanistic knowledge, are an ideal tool for this task. For example, such approaches can be used to predict gene expression dynamics given time-series data of past expression history. To explore this application, we computationally simulated single-cell responses, incorporating different sources of noise and alternative genetic circuit designs. We showed that deep neural networks trained on these simulated data were able to correctly infer the underlying dynamics of a cell response even in the presence of measurement noise and stochasticity in the biochemical reactions. The training set size and the amount of past data provided as inputs both affected prediction quality, with cascaded genetic circuits that introduce delays requiring more past data. We also tested prediction performance on a bistable auto-activation circuit, finding that our initial method for predicting a single trajectory was fundamentally ill-suited for multimodal dynamics. To address this, we updated the network architecture to predict the entire distribution of future states, showing it could accurately predict bimodal expression distributions. Overall, these methods can be readily applied to the diverse prediction tasks necessary to predict and control a variety of biological circuits, a key aspect of many synthetic biology applications.

Material Data Identification in an Induction Hardening Test Rig with Physics-Informed Neural Networks

Physics-Informed neural networks (PINNs) have demonstrated remarkable performance in solving partial differential equations (PDEs) by incorporating the governing PDEs into the network's loss function during optimization. PINNs have been successfully applied to diverse inverse and forward problems. This study investigates the feasibility of using PINNs for material data identification in an induction hardening test rig. By utilizing temperature sensor data and imposing the heat equation with initial and boundary conditions, thermo-physical material properties, such as specific heat, thermal conductivity, and the heat convection coefficient, were estimated. To validate the effectiveness of the PINNs in material data estimation, benchmark data generated by a finite element model (FEM) of an air-cooled cylindrical sample were used. The accurate identification of the material data using only a limited number of virtual temperature sensor data points was demonstrated. The influence of the sensor positions and measurement noise on the uncertainty of the estimated parameters was examined. The study confirms the robustness and accuracy of this approach in the presence of measurement noise, albeit with lower efficiency, thereby requiring more time to converge. Lastly, the applicability of the presented approach to real measurement data obtained from an air-cooled cylindrical sample heated in an induction heating test rig was discussed. This research contributes to the accurate offline estimation of material data and has implications for optimizing induction heat treatments.

Computational meaningfulness as the source of beneficial cognitive biases

The human brain has evolved to solve the problems it encounters in multiple environments. In solving these challenges, it forms mental simulations about multidimensional information about the world. These processes produce context-dependent behaviors. The brain as overparameterized modeling organ is an evolutionary solution for producing behavior in a complex world. One of the most essential characteristics of living creatures is that they compute the values of information they receive from external and internal contexts. As a result of this computation, the creature can behave in optimal ways in each environment. Whereas most other living creatures compute almost exclusively biological values (e.g., how to get food), the human as a cultural creature computes meaningfulness from the perspective of one's activity. The computational meaningfulness means the process of the human brain, with the help of which an individual tries to make the respective situation comprehensible to herself to know how to behave optimally. This paper challenges the bias-centric approach of behavioral economics by exploring different possibilities opened up by computational meaningfulness with insight into wider perspectives. We concentrate on confirmation bias and framing effect as behavioral economics examples of cognitive biases. We conclude that from the computational meaningfulness perspective of the brain, the use of these biases are indispensable property of an optimally designed computational system of what the human brain is like. From this perspective, cognitive biases can be rational under some conditions. Whereas the bias-centric approach relies on small-scale interpretable models which include only a few explanatory variables, the computational meaningfulness perspective emphasizes the behavioral models, which allow multiple variables in these models. People are used to working in multidimensional and varying environments. The human brain is at its best in such an environment and scientific study should increasingly take place in such situations simulating the real environment. By using naturalistic stimuli (e.g., videos and VR) we can create more realistic, life-like contexts for research purposes and analyze resulting data using machine learning algorithms. In this manner, we can better explain, understand and predict human behavior and choice in different contexts.

A comprehensive suite of earthquake catalogues for the 2016-2017 Central Italy seismic sequence

The protracted nature of the 2016-2017 central Italy seismic sequence, with multiple damaging earthquakes spaced over months, presented serious challenges for the duty seismologists and emergency managers as they assimilated the growing sequence to advise the local population. Uncertainty concerning where and when it was safe to occupy vulnerable structures highlighted the need for timely delivery of scientifically based understanding of the evolving hazard and risk. Seismic hazard assessment during complex sequences depends critically on up-to-date earthquake catalogues—i.e., data on locations, magnitudes, and activity of earthquakes—to characterize the ongoing seismicity and fuel earthquake forecasting models. Here we document six earthquake catalogues of this sequence that were developed using a variety of methods. The catalogues possess different levels of resolution and completeness resulting from progressive enhancements in the data availability, detection sensitivity, and hypocentral location accuracy. The catalogues range from real-time to advanced machine-learning procedures and highlight both the promises as well as the challenges of implementing advanced workflows in an operational environment.

On the Use of High-Resolution and Deep-Learning Seismic Catalogs for Short-Term Earthquake Forecasts: Potential Benefits and Current Limitations

Enhanced earthquake catalogs provide detailed images of evolving seismic sequences. Currently, these data sets take some time to be released but will soon become available in real time. Here, we explore whether and how enhanced seismic catalogs feeding into established short-term earthquake forecasting protocols may result in higher predictive skill. We consider three enhanced catalogs for the 2016-2017 Central Italy sequence, featuring a bulk completeness lower by at least two magnitude units compared to the real-time catalog and an improved hypocentral resolution. We use them to inform a set of physical Coulomb Rate-and-State (CRS) and statistical Epidemic-Type Aftershock Sequence (ETAS) models to forecast the space-time occurrence of M3+ events during the first 6 months of the sequence. We track model performance using standard likelihood-based metrics and compare their skill against the best-performing CRS and ETAS models among those developed with the real-time catalog. We find that while the incorporation of the triggering contributions from new small magnitude detections of the enhanced catalogs is beneficial for both types of forecasts, these models do not significantly outperform their respective near real-time benchmarks. To explore the reasons behind this result, we perform targeted sensitivity tests that show how (a) the typical spatial discretizations of forecast experiments ( ≥ 2 km) hamper the ability of models to capture highly localized secondary triggering patterns and (b) differences in earthquake parameters (i.e., magnitude and hypocenters) reported in different catalogs can affect forecast evaluation. These findings will contribute toward improving forecast model design and evaluation strategies for next-generation seismic catalogs.

Revealing the spatiotemporal complexity of the magnitude distribution and b-value during an earthquake sequence

The Magnitude–Frequency-Distribution (MFD) of earthquakes is typically modeled with the (tapered) Gutenberg–Richter relation. The main parameter of this relation, the b-value, controls the relative rate of small and large earthquakes. Resolving spatiotemporal variations of the b-value is critical to understanding the earthquake occurrence process and improving earthquake forecasting. However, this variation is not well understood. Here we present remarkable MFD variability during the complex 2016/17 central Italy sequence using a high-resolution earthquake catalog. Isolating seismically active volumes (‘clusters’) reveals that the MFD differed in nearby clusters, varied or remained constant in time depending on the cluster, and increased in b-value in the cluster where the largest earthquake eventually occurred. These findings suggest that the fault system’s heterogeneity and complexity influence the MFD. Our findings raise the question “b-value of what?”: interpreting and using MFD variability needs a spatiotemporal scale that is physically meaningful, like the one proposed here.

The authors introduce a new perspective to study the spatiotemporal behavior of the magnitude–frequency distribution: spatially isolating seismogenic zones to provide an appropriate scale to resolve the b-value. Among those zones, the b-value behaved remarkably throughout the 2016 central Italy sequence.

Deep-learning seismology

Seismic waves from earthquakes and other sources are used to infer the structure and properties of Earth's interior. The availability of large-scale seismic datasets and the suitability of deep-learning techniques for seismic data processing have pushed deep learning to the forefront of fundamental, long-standing research investigations in seismology. However, some aspects of applying deep learning to seismology are likely to prove instructive for the geosciences, and perhaps other research areas more broadly. Deep learning is a powerful approach, but there are subtleties and nuances in its application. We present a systematic overview of trends, challenges, and opportunities in applications of deep-learning methods in seismology.