Adaptively Smoothed Seismicity Earthquake Forecasts for Italy

Adaptively Smoothed Seismicity Earthquake Forecasts for Italy 1 Maximilian J. Werner1*, Agn`es Helmstetter2, David D. Jackson3, Yan Y. Kagan3, and Stefan Wiemer1 2 June 23, 2021 3 first version submitted 23 March 2010 to the CSEP-Italy special issue of the Annals of Geophysics 4 revised version submitted 28 June 2010 5 1 Swiss Seismological Service, Institute of Geophysics, ETH Zurich, Switzerland. 6 2 Laboratoire de G´eophysique Interne et Tectonophysique, Universit´e Joseph Fourier and Centre National 7 de la Recherche Scientifique, Grenoble, France. 8 3 Department of Earth and Space Sciences, University of California, Los Angeles, USA. 9 * Corresponding author: 10 Maximilian J. Werner 11 Swiss Seismological Service 12 Institute of Geophysics 13 ETH Zurich 14 Sonneggstr. 5 15 8092 Zurich, Switzerland 16 mwerner@sed.ethz.ch 17 Abstract 18 We present a model for estimating the probabilities of future earthquakes of magnitudes m ≥4.95 19 in Italy. The model, a slightly modified version of the one proposed for California by Helmstetter et al. 20 [2007] and Werner et al. [2010a], approximates seismicity by a spatially heterogeneous, temporally homo- 21 geneous Poisson point process. The temporal, spatial and magnitude dimensions are entirely decoupled. 22 Magnitudes are independently and identically distributed according to a tapered Gutenberg-Richter 23 magnitude distribution. We estimated the spatial distribution of future seismicity by smoothing the 24 locations of past earthquakes listed in two Italian catalogs: a short instrumental catalog and a longer 25 instrumental and historical catalog. The bandwidth of the adaptive spatial kernel is estimated by op- 26 timizing the predictive power of the kernel estimate of the spatial earthquake density in retrospective 27 forecasts. When available and trustworthy, we used small earthquakes m ≥2.95 to illuminate active fault 28 1 arXiv:1003.4374v2 [physics.geo-ph] 28 Jun 2010 structures and likely future epicenters. By calibrating the model on two catalogs of different duration 29 to create two forecasts, we intend to quantify the loss (or gain) of predictability incurred when only a 30 short but recent data record is available. Both forecasts, scaled to five and ten years, were submitted to 31 the Italian prospective forecasting experiment of the global Collaboratory for the Study of Earthquake 32 Predictability (CSEP). An earlier forecast from the model was submitted by Helmstetter et al. [2007] to 33 the Regional Earthquake Likelihood Model (RELM) experiment in California, and, with over half of the 34 five-year experiment over, the forecast performs better than its competitors. 35 1 Introduction 36 In this article, we document the calibration of a previously published, time-independent model of earth- 37 quake occurrences to the region of Italy. We extracted probabilities of future m ≥4.95 shocks for a 38 five- and ten-year period in a format suitable for prospective testing within the Italian earthquake pre- 39 dictability experiment [Schorlemmer et al., 2010a]. Previously, Helmstetter et al. [2007] calculated a 40 probabilistic earthquake forecast for m ≥4.95 for the region of California over a five year duration. The 41 forecast is currently being tested within the Regional Earthquake Likelihood Model (RELM) experiment 42 [Field, 2007]. After more than half of the five years over, the forecast cannot be rejected by a suite of 43 tests and performs better than competing forecasts [Schorlemmer et al., 2010b]. Werner et al. [2010a] 44 made small modifications to the model by Helmstetter et al. [2007] and re-calibrated it on updated data 45 to generate a new earthquake forecast for California. This forecast is under test within the California 46 branch of the global Collaboratory for the Study of Earthquake Predictability (CSEP) [Jordan, 2006; 47 Zechar et al., 2009]. To calculate future earthquake potential in Italy, we used the same model with some 48 minor modifications. One modification concerns the estimation of the completeness threshold, which was 49 difficult to estimate at the small spatial scales that were possible with the high quality data set available 50 in California [Werner et al., 2010a; Helmstetter et al., 2007]. Instead, we set a single magnitude threshold 51 for the entire region. 52 Smoothed seismicity models, such as the present one, usually do not incorporate geological or tectonic 53 observations. Rather, the models are calibrated on the seismicity data available from earthquake catalogs. 54 One may justifiably question the hypothesized validity that the short (decadal) periods covered by high- 55 quality instrumental catalogs suffice to forecast the locations of future large earthquakes that have very 56 2 low occurrence probabilities. Even if the spatial distribution of seismicity is reasonably stable up to 57 geological timescales, estimating this distribution from a short time window of observations is difficult. 58 A partial solution to this problem

…(Full text truncated)…

This content is AI-processed based on ArXiv data.