The spatiotemporal heterogeneity of

The Gutenberg–Richter

Spatial and temporal heterogeneity is an important topic in

A model named the Asperity Likelihood Model (ALM) based on the above assumptions
has been developed and used to forecast future earthquakes (Wiemer and
Schorlemmer, 2007; Gulia et al., 2010). Research on the temporal
heterogeneity of

However, some research results show that the apparent variability in

Based on the above viewpoints, the calculation reliability for research on
the spatiotemporal heterogeneity of

In the traditional calculation of the Guttenberg–Richter magnitude–frequency

Among those data-driven approaches, Si and Jiang (2019) developed a method
using the continuous distribution function (hereafter referred to as the OK1993
model) given by Ogata and Katsura (1993), which has the advantage of
simultaneously determining the minimum magnitude of completeness and
obtaining

The OK1993 model uses the seismic detection rate function

The construction of the data-driven approach can be achieved by Voronoi
tessellation with limited boundaries. Voronoi tessellation refers to a unique
set of continuous polygon partitioning schemes

The maximum likelihood calculation of the OK1993 model parameter is not
performed when the number of earthquakes contained in a Voronoi cell is

In the above calculation steps, the setting of the maximum number of nodes, the number of random throws, etc. has obvious subjectivity. However, due to the fact that the data-driven approach actually obtains a very stable final result when the number of divisions and the number of grid nodes are sufficient (Si and Jiang, 2019) – for example, when the maximum number of nodes is 100, each type of node is randomly thrown 1000 times – the final result obtained when 1000 optimal models are selected is almost the same as the result of this paper.

The 2019 Changning

Distribution of seismicity in the Changning area. The red
dots show the aftershocks of the Changning

We used earthquake catalogs and bulletins provided by the Sichuan Regional
Seismic Network from 1 January 2009 to 17 July 2019. To obtain relatively reliable parameters such as the epicenter location and focal depth, the
double-difference algorithm hypoDD (Waldhauser and Ellsworth, 2000) was used to relocate the earthquakes. Among the data we used are a total of 21 246 seismic events that meet the requirements of the hypoDD method with not less than four arrivals, including 516 649 P-wave arrivals and 506 809 S-wave arrivals, and 59 permanent seismic stations and temporary seismic stations are used which are located in Sichuan and the surrounding provinces. We used a 12-layer one-dimensional crustal velocity model (Xie et al., 2012) during the relocation. The ratio of

A total of 18 371 earthquake events were relocated (Fig. 1), of which the
smallest event had a magnitude of

From the spatial distribution of the relocated earthquakes shown in Fig. 1, the aftershocks are mainly distributed in the northwest direction of the
mainshock epicenter and extend along the Changning anticline with a length of
about 27 km, which is much longer than the rupture scale of about 10 km for
a

In the aftershock sequence of the Changning

To facilitate the calculation of

Distribution of seismicity for

According to the technical process of the data-driven approach described
above, after Voronoi tessellation, calculation of the BIC values, and
selection of the optimal models, the ensemble median (

An example of calculating the parameters of the OK1993 model in terms of the frequency–magnitude distribution based on a data-driven approach.

We calculated the distribution of the ensemble median

The spatial distribution of the ensemble median

Figure 4c and d show the distribution of the ensemble median

Figure 5 shows the spatial distribution of the median absolute deviation (MAD) of

The spatial distribution of the median absolute deviation (MAD) of the

Considering that the

Spatiotemporal distribution of the ensemble median

Spatiotemporal distribution of the median absolute deviation (MAD) of the

It can be seen in Fig. 6a that before the Changning

From the results before and after the Changning

Compared with Fig. 6a, the results in Fig. 6b show that before the Changning

In the pattern of

There is still much controversy over the temporal variation pattern of

For the spatiotemporal heterogeneity of the

In addition,

To reveal whether there is spatiotemporal heterogeneity of

The

The

The

Although the distribution characteristics of

The MATLAB code for the data-driven method used to compute the

The relocated earthquake catalog can be obtained by contacting Changsheng Jiang (jiangcs@cea-igp.ac.cn).

The supplement related to this article is available online at:

CJ, LH and FL contributed to the conceptualization. CJ prepared the manuscript with contributions from all authors. CS and LH wrote software for data processing. GL, FY, JB, and ZS contributed to the revised manuscript version.

The authors declare that they have no conflict of interest.

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

The earthquake catalog used in this paper was provided by the Sichuan Earthquake Agency. The Multi-Parametric Toolbox 3.0 (

This research has been supported by the program of the China Seismic Experimental Site (grant no. 2019CSES0106); the program of Basic Resources Investigation of Science and Technology (grant no. 2018FY100504); the National Natural Science Foundation of China (grant no. U2039204); and the Special Fund of the Institute of Geophysics, China Earthquake Administration (grant no. DQJB20X11).

This paper was edited by Filippos Vallianatos and reviewed by two anonymous referees.