  Citation: Fox, G.C.; Rundle, J.B.; Donnellan, A.; Feng, B. Earthquake Nowcasting with Deep Learning. GeoHazards 2022, 3, 199–226. https://doi.org/10.3390/ geohazards3020011 Academic Editors: Gerassimos A. Papadopoulos and Tiago Miguel Ferreira Received: 24 December 2021 Accepted: 2 April 2022 Published: 15 April 2022 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affil- iations. Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). GeoHazards Article Earthquake Nowcasting with Deep Learning Geoffrey Charles Fox 1,2, * , John B. Rundle 3,4,5 , Andrea Donnellan 4 and Bo Feng 6 1 Biocomplexity Institute and Initiative, University of Virginia, Charlottesville, VA 22904, USA 2 Computer Science Department, University of Virginia, Charlottesville, VA 22904, USA 3 Physics and Astronomy and Geology, University of California, Davis, CA 95616, USA; john.b.rundle@gmail.com 4 Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109, USA; andrea.donnellan@jpl.nasa.gov 5 Santa Fe Institute, Santa Fe, NM 87501, USA 6 Luddy School of Informatics, Computing, and Engineering, Indiana University, Bloomington, IN 47408, USA; fengbo@iu.edu * Correspondence: vxj6mb@virginia.edu or gcfexchange@gmail.com Abstract: We review previous approaches to nowcasting earthquakes and introduce new approaches based on deep learning using three distinct models based on recurrent neural networks and trans- formers. We discuss different choices for observables and measures presenting promising initial results for a region of Southern California from 1950–2020. Earthquake activity is predicted as a function of 0.1-degree spatial bins for time periods varying from two weeks to four years. The overall quality is measured by the Nash Sutcliffe efficiency comparing the deviation of nowcast and observation with the variance over time in each spatial region. The software is available as open source together with the preprocessed data from the USGS. Keywords: deep learning; earthquake; nowcasting; Southern California; LSTM; transformer; attention; time series; Nash Sutcliffe efficiency; AI for science 1. Introduction Earthquake forecasting is an old but challenging problem with many interesting characteristics. In studying this, we not only hope to shed light on this socioscientific challenge but also lead to new methods based on deep learning that can be applied in other areas. Perhaps the most essential characteristic is the nature of its challenge. Namely, it is unlikely that building a new zettascale supercomputer will accurately simulate quakes and lead to reliable earthquake predictions [1]. As a phase transition in a system with unknown boundary conditions and phenomenological models (as for friction), it is not obvious that earthquakes are the solution of a set of differential equations or that accurate probabilities of large events can be computed. For these reasons, we have recently chosen to focus on earthquake nowcasting [2], which is the estimation of hazard in the present, the immediate past, and the near future. Earthquake nowcasting is an archetype of data-intensive problems characteristic of the fourth paradigm of scientific discovery [3] in that we suppose that the patterns of previous events hold clues to the future. This is a bit different from other studies where machine learning has been successful. For example, language translation and image recognition are very complicated problems, but one can see the natural patterns from grammar, words, segments, colors, etc., that are clearly but complexly correlated with what we want to discover. Machine learning successfully learns this complex relationship between inputs and predictions [4]. Earthquake nowcasting challenges AI to discover “hidden variables” in a case where their existence is not as clear as in other successes of machine and deep learning. These ideas are illustrated in Figure 1 where we are pursuing the right side data-intensive approach, rather than developing a theoretical model and determining unknown parameters from GeoHazards 2022, 3, 199–226. https://doi.org/10.3390/geohazards3020011 https://www.mdpi.com/journal/geohazards