Nico Schliwa, Alice-Agnes Gabriel, Jan Premus, František Gallovič

https://doi.org/10.1029/2024JB029410

Nonlinear Processes in Geophysics

The evolution of multifractal structures in various physical processes, such as climatology, seismology, or volcanology, serves as a crucial tool for detecting changes in corresponding phenomena. In this study, we explore the evolution of the multifractal structure of volcanic emissions with varying energy levels (observed at Colima, Mexico, during the years 2013–2015) to identify clear indicators of imminent high-energy emissions nearing 8.0×10⁸ J. These indicators manifest through the evolution of six multifractal parameters: the central Hölder exponent (α₀); the maximum and minimum Hölder exponents (α_max, α_min); the multifractal amplitude (); the multifractal asymmetry (); and the complexity index (CI), calculated as the sum of the normalized values of α₀, W, and γ. Additionally, the results obtained from adapting the Gutenberg–Richter seismic law to volcanic energy emissions, along with the corresponding skewness and standard deviation of the volcanic emission data, further support the findings obtained through multifractal analysis. These results, derived from multifractal structure analysis, adaptation of the Gutenberg–Richter law to volcanic emissions, and basic statistical parameters, hold significant relevance in anticipating potential volcanic episodes of high energy. Such anticipation can be further quantified using an appropriate forecasting algorithm.

Nature: Communications Earth & Environment

After large-magnitude earthquakes, a crucial task for impact assessment is to rapidly and accurately estimate the ground shaking in the affected region. To satisfy real-time constraints, intensity measures are traditionally evaluated with empirical Ground Motion Models that can drastically limit the accuracy of the estimated values. As an alternative, here we present Machine Learning strategies trained on physics-based simulations that require similar evaluation times. We trained and validated the proposed Machine Learning-based Estimator for ground shaking maps with one of the largest existing datasets (<100M simulated seismograms) from CyberShake developed by the Southern California Earthquake Center covering the Los Angeles basin. For a well-tailored synthetic database, our predictions outperform empirical Ground Motion Models provided that the events considered are compatible with the training data. Using the proposed strategy we show significant error reductions not only for synthetic, but also for five real historical earthquakes, relative to empirical Ground Motion Models.

Earth and Space Science

Tsunami simulations require powerful computational resources to be performed efficiently. Although the modern graphics processing units (GPUs) allow the acceleration of this kind of simulations, they can still last many minutes or even hours for simulations which have to deal with very high spatial resolutions or simulation times. In this paper, we propose a method to predict the alert or inundation level of a tsunami generated by an earthquake using deep learning methods. In particular, we train multilayer perceptron (MLP) neural networks for predicting the alert level due to a tsunami at given coastal locations. Ensemble methods are used to improve the predictions of the neural networks. Tsunamis caused by ruptures of several fault segments at different time instants, application to real events, probabilistic forecasting and comparison with other machine learning algorithms are also addressed. Results on realistic scenarios confirm that good accuracies are obtained. The inference times of the trained networks and ensembles are also very low, lasting less than one second to predict the results of thousands of simulations. The proposed method could be used in a tsunami early warning system along with the application of scaling laws.

Alice‐Agnes Gabriel, Thomas Ulrich, Mathilde Marchandon, James Biemiller, John Rekoske

The Seismic Record

The 2023 Turkey earthquake sequence involved unexpected ruptures across numerous fault segments. We present 3D dynamic rupture simulations to illuminate the complex dynamics of the earthquake doublet. Our models are constrained by observations available within days of the sequence and deliver timely, mechanically consistent explanations of the unforeseen rupture paths, diverse rupture speeds, multiple slip episodes, heterogeneous fault offsets, locally strong shaking, and fault system interactions. Our simulations link both earthquakes, matching geodetic and seismic observations and reconciling regional seismotectonics, rupture dynamics, and ground motions of a fault system represented by 10 curved dipping segments and embedded in a heterogeneous stress field. The 7.8 earthquake features delayed backward branching from a steeply branching splay fault, not requiring supershear speeds. The asymmetrical dynamics of the distinct, bilateral 7.7 earthquake are explained by heterogeneous fault strength, prestress orientation, fracture energy, and static stress changes from the previous earthquake. Our models explain the northward deviation of its eastern rupture and the minimal slip observed on the Sürgü fault. 3D dynamic rupture scenarios can elucidate unexpected observations shortly after major earthquakes, providing timely insights for data‐driven analysis and hazard assessment toward a comprehensive, physically consistent understanding of the mechanics of multifault systems.