Statistical power of spatial earthquake forecast tests

Abstract

Summary The Collaboratory for the Study of Earthquake Predictability (CSEP) is an international effort to evaluate earthquake forecast models prospectively. In CSEP, one way to express earthquake forecasts is through a grid-based format: the expected number of earthquake occurrences within 0.1○ × 0.1○ spatial cells. The spatial distribution of seismicity is thereby evaluated using the Spatial test (S-test). The high-resolution grid combined with sparse and inhomogeneous earthquake distributions leads to a huge number of cells causing disparity in the number of cells and the number of earthquakes to evaluate the forecasts, thereby affecting the statistical power of the S-test. In order to explore this issue, we conducted a global earthquake forecast experiment, in which we computed the power of the S-test to reject a spatially non-informative uniform forecast model. The S-test loses its power to reject the non-informative model when the spatial resolution is so high that every earthquake of the observed catalog tends to get a separate cell. Upon analyzing the statistical power of the S-test, we found, as expected, that the statistical power of the S-test depends upon the number of earthquakes available for testing, e. g. with the conventional high-resolution grid for the global region, we would need more than 32000 earthquakes in the observed catalog for powerful testing, which would require approximately 300 years to record M ≥ 5.95. The other factor affecting the power is more interesting and new; it is related to the spatial grid representation of the forecast model. Aggregating forecasts on multi-resolution grids can significantly increase the statistical power of the S-test. Using the recently introduced Quadtree to generate data-based multi-resolution grids, we show that the S-test reaches its maximum power in this case already for as few as 8 earthquakes in the test period. Thus we recommend for future CSEP experiments the use of Quadtree-based multi-resolution grids, where available data determine the resolution.

A pre-seismic anomaly detection approach based on graph attention isomorphism network

“…However, when an earthquake occurs, the stability of the geomagnetic signal is disrupted, resulting in a change in the fractal dimension. Therefore, in this paper, the fractal dimension of the geomagnetic data is extracted as a nonlinear feature, which can be calculated by equation (1).…”

“…The number of seismic events and corresponding magnitudes in the research area are shown in table 3. The value of k is fixed between [1,5] for experimentation to find the best sparsity level. Figure 9 shows F1 in the classification results under different values of k, and it can be seen that the best classification performance of the detection model is achieved with k = 3.…”

“…Earthquakes are among the most devastating natural disasters worldwide, leading to significant casualties and economic losses for human society. Furthermore, they can trigger secondary disasters such as tsunamis, landslides, and fires, resulting in severe environmental damage [1]. Pre-seismic anomaly detection is a vital component of earthquake monitoring [2] and seismic forecasting systems (SFSs) [3,4].…”

“…Anomaly detection methods can be broadly categorized into two main branches [12]: (1) unsupervised detection methods and (2) supervised detection methods. Unsupervised anomaly detection methods typically identify anomalous data through distance-based similarity [13], density-based similarity [14], decision trees [15], and so on.…”

The need for open, transdisciplinary, and ethical science in seismology

“…The EPOS Thematic Core Service for Seismology (Haslinger et al, 2022) enables homogenized monitoring efforts and collaboration based on seismic waveform data (ORFEUS), rapid earthquake information (EMSC), and expertise in seismic hazard and risk assessments (EFEHR); thus connecting the different assets along the disaster cycle. Also in RISE, some open science assets have been created, such as the pyCSEP toolkit, an open source software for developing and testing probabilistic earthquake forecasts (Savran et al, 2022a,b), so-called reproducibility packages that contain code, data, and other resources to reproduce research outcomes without additional effort (e.g., Bayona et al, 2022Bayona et al, , 2023Khawaja et al, 2023), an open sensor firmware platform that supports creating real-time monitoring networks (quakesaver.net), and a dynamic exposure model based on crowd-sourced/citizen-science building data (Schorlemmer et al, 2020). These developments set an example for making the fundamental assets of dynamic (seismic) risk assessment available.…”