![Pairs (tstart, Mc) for six sequences. a Sequences of 1995, 2003 and 2016 M7.1 and 2010 M7.2 earthquakes; b sequences of 2009 and 2016 M7.8 earthquakes. Solid lines correspond to equation Mc=M-4.5-0.76log10(tstart)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$M_{\text{c}} = M - 4.5 - 0.76{ \log }_{10} (t_{\text{start}} )$$\end{document} (Helmstetter et al. 2006)](https://www.researchgate.net/publication/318221743/figure/fig3/AS:959996951027736@1605892517421/Pairs-tstart-Mc-for-six-sequences-a-Sequences-of-1995-2003-and-2016-M71-and-2010_Q320.jpg)
- A preview of this full-text is provided by Springer Nature.
- Learn more
Preview content only
Content available from Pure and Applied Geophysics
This content is subject to copyright. Terms and conditions apply.
Long-Delayed Aftershocks in New Zealand and the 2016 M7.8 Kaikoura Earthquake
P. SHEBALIN
1
and S. BARANOV
2
Abstract—We study aftershock sequences of six major earth-
quakes in New Zealand, including the 2016 M7.8 Kaikaoura and
2016 M7.1 North Island earthquakes. For Kaikaoura earthquake,
we assess the expected number of long-delayed large aftershocks of
M5?and M5.5?in two periods, 0.5 and 3 years after the main
shocks, using 75 days of available data. We compare results with
obtained for other sequences using same 75-days period. We esti-
mate the errors by considering a set of magnitude thresholds and
corresponding periods of data completeness and consistency. To
avoid overestimation of the expected rates of large aftershocks, we
presume a break of slope of the magnitude–frequency relation in
the aftershock sequences, and compare two models, with and
without the break of slope. Comparing estimations to the actual
number of long-delayed large aftershocks, we observe, in general, a
significant underestimation of their expected number. We can
suppose that the long-delayed aftershocks may reflect larger-scale
processes, including interaction of faults, that complement an iso-
lated relaxation process. In the spirit of this hypothesis, we search
for symptoms of the capacity of the aftershock zone to generate
large events months after the major earthquake. We adapt an
algorithm EAST, studying statistics of early aftershocks, to the case
of secondary aftershocks within aftershock sequences of major
earthquakes. In retrospective application to the considered cases,
the algorithm demonstrates an ability to detect in advance long-
delayed aftershocks both in time and space domains. Application of
the EAST algorithm to the 2016 M7.8 Kaikoura earthquake zone
indicates that the most likely area for a delayed aftershock of
M5.5?or M6?is at the northern end of the zone in Cook Strait.
1. Introduction
A major earthquake of M7.8 occurred near the
coast of New Zealand on 13 November 2016. The
earthquake has initiated a significant tsunami with
amplitude more than 4 m. The earthquake fault was
located in about 100 km from the fault zone of the
Canterbury earthquake sequence. The Canterbury
sequence started on 3 September 2010 (Mw7.2) near
Darfield. An earthquake of Mw6.2 occurred in the
Eastern part of its fault zone about 6 months later, on
21 February 2011. The epicenter was located in
Christchurch, the second most populous city in New
Zealand, and the earthquake caused huge damage
including about 200 casualties. Further large events
with magnitude 5.5 and higher occurred in the area on
13 June 2011, on 22 December 2011, and on 14
February 2016. The proximity of the 2016 M7.8 and
2010 M7.1 fault zones raises a question: should we
expect a similar scenario, with a set of successive
large long-delayed aftershocks?
Various methods were developed recently aimed
to an operational forecasting of aftershocks (Ger-
stenberger et al. 2005; Omi et al. 2013,2016; Steacy
et al. 2014; Cattania et al. 2014). Some aftershock
forecasting models are being tested in real time in the
New Zealand earthquake forecast testing center
(Gerstenberger and Rhoades 2010). Most of those
models are based on the idea of independent com-
bining of well-known Gutenberg–Richter and
Omori–Utsu relations, the idea first proposed by
Reasenberg and Jones (1989). All those models are
designed to assess the expected rates of seismic
events in specified space–time–magnitude volumes.
An important direction in development of the earth-
quake rate models is the ETAS model (Ogata 1983)
and its modifications, including applications in New
Zealand (Harte 2014). Another important new ten-
dency is combining different models, including
hybrid statistical and physics-based models (Rhoades
2013; Rhoades et al. 2014,2016; Shebalin et al. 2014;
Cattania et al. 2014).
The Canterbury earthquake sequence has strongly
reactivated the interest to the problem of forecasting
aftershocks. Recently, a retrospective analysis of
1
Institute of Earthquake Prediction Theory and Mathemati-
cal Geophysics, Russian Academy of Sciences, Moscow, Russia.
E-mail: p.n.shebalin@gmail.com
2
Kola Branch of the Geophysical Survey of Russian Acad-
emy of Sciences, Apatity, Russia.
Pure Appl. Geophys. 174 (2017), 3751–3764
Ó2017 Springer International Publishing AG
DOI 10.1007/s00024-017-1608-9 Pure and Applied Geophysics
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
