1: %\documentclass[twocolumn,prl]{revtex4}
2: \documentclass[twocolumn,showpacs,preprintnumbers,superscriptaddress,amsmath,amssymb,prl]{revtex4}
3:
4: \usepackage{graphicx}% Include figure files
5: \usepackage{dcolumn}% Align table columns on decimal point
6: \usepackage{bm}% bold math
7: \usepackage{amsmath}
8: \setlength{\topmargin}{0.5cm}
9:
10: \newcommand{\be}{\begin{equation}}
11: \newcommand{\ee}{\end{equation}}
12: \newcommand{\ba}{\begin{eqnarray}}
13: \newcommand{\ea}{\end{eqnarray}}
14: %\nofiles
15:
16: \begin{document}
17:
18: \preprint{APS preprint}
19:
20: {\bf Comment on ``Analysis of the Spatial Distribution Between
21: Successive Earthquakes'' by Davidsen and Paczuski}
22:
23: \noindent
24:
25: \noindent Davidsen and Paczuski [1] claim to have found evidence
26: contradicting the
27: theory of aftershock zone scaling in favor of scale-free
28: statistics. We present four elements showing that
29: Davidsen and Paczuski's results may be insensitive to the existence of
30: physical length scales associated with aftershock zones or mainshock
31: rupture lengths, so that their claim is unsubstantiated.
32:
33: Firstly, the power law exponent $\delta$ of their probability
34: density distribution (pdf) for distances between pairs of
35: successive earthquakes in southern
36: California is less than $1$. Therefore the
37: exponent cannot hold for the tail because the pdf cannot be
38: normalized and hence can at most describe an intermediate asymptotic (if
39: any). The real tail of the pdf must behave differently, independent of
40: any finite size scaling.
41:
42: Secondly, we performed tests that show the extreme sensitivity of
43: the suggested power law to the duration of the catalog. We took the
44: same catalog from the Southern California Earthquake Data Center as in
45: [1] with the same parameters but contrasted in Figure 1 the highly active
46: 6-month period from June 1, 1992 until December 31, 1992 including
47: the June 28 M7.3 Landers earthquake (crosses) with the remainder of the
48: catalog both before and after Landers (circles). Firstly, we see that
49: removing 6 months of data from a 17 year period causes the power law
50: to disappear. Secondly, the
51: Landers aftershocks show clear signs of scales, such as the bump
52: marked by an arrow, which may be connected with the
53: simultaneous aftershocks of the June 28 M$6.4$ Big Bear event and the
54: July 11 M$5.7$ Mojave earthquake and the rupture length of Landers.
55:
56: Thirdly, we repeated the same analysis for Japan and northern
57: California and found no evidence of robust power laws. For Japan
58: (Figure 2, circles), we used the JMA catalog
59: from January 1984 to
60: December 2001 within $(120.0^o E, 150.0^o E)$ by
61: $(25.0^o N, 45.0^o N)$
62: above magnitude thresholds $m_d=3.0, 3.4 $ and $4.0$
63: resulting in $34163, 13275$ and $6757$ pairs with distances larger
64: than $2$km, respectively. For northern California (Figure 2, crosses), we
65: used data in the period from January 1984 until December 2004 (28274
66: events above magnitude $2.4$). Our results suggest that the power law
67: found by Davidsen and Pazcuski [1] may not be robust with
68: respect to location, catalog and window size.
69:
70: Finally, we show in the inset of figure 2 that a model [2] that
71: explicitely obeys
72: aftershock zone scaling can
73: reproduce the observed histogram, demonstrating that the statistic may
74: not be sensitive to the scales. We simulated a seismic
75: catalog using a 3D version of the ETAS model [2] that explicitely includes
76: the scale of each mainshock rupture length $l_r(m)=0.02 \times 10^{0.5m}$
77: in the spatial aftershock decay distribution with distance $d$
78: according to $P(d) \sim (l_r(m)+d)^{-(1+\mu)}$ and
79: initiated the catalog with a Landers-like $M7.3$ mainshock. We used
80: the parameters ($b=\alpha=1, p=1.1, c=0.0001, k=0.0022, \mu=2$,
81: background=5.0 per day in a 700km by 700km window). We show
82: that the same analysis of data generated with aftershock zone scaling
83: leads to an apparent power law that shows no sign of the aftershock
84: zone scale $l_r \sim 90km$.
85:
86: \begin{figure}[h!]
87: \includegraphics[width=7cm]{fig1.eps}
88: \caption{\label{Fig1} Distribution of the epicentral distances between
89: successive earthquakes in southern California for magnitude
90: thresholds $2.4, 3.0, 3.4$ (light to dark markers): 6 month period
91: from June through December 1992 (crosses) consisting mainly of
92: Landers and its aftershocks contrasted with the remainder of the 17
93: year catalog (circles).
94: }
95: \end{figure}
96:
97:
98: \begin{figure}[h!]
99: \includegraphics[width=7cm]{fig2.eps}
100: \caption{\label{Fig2} Distribution of the epicentral distances between
101: successive earthquakes in the JMA catalog (circles) from January 1984 until
102: December 2000 above magnitude thresholds $3.4$, $4.0$ and $4.4$ and
103: in northern California (crosses) from January 1984 until December 2004
104: above magnitude thresholds $2.4$, $3.0$ and $3.4$ (data from the
105: Northern California Earthquake Data Center). Inset: ETAS model
106: simulation.
107: }
108: \end{figure}
109:
110: \vskip -1cm
111:
112: \begin{thebibliography}{99}
113:
114: \bibitem{DP} J. Davidsen and M. Paczuski, Phys. Rev. Lett. 94,
115: 048501 (2005).
116:
117: \bibitem{Ogata88} Ogata, Y., J. Am. Stat. Assoc. 83, 9 (1988).
118:
119: \end{thebibliography}
120:
121:
122: M.J. Werner $^1$ and D. Sornette$^{2,3}$,\\
123: $^1$ IGPP, UCLA, and ESS, UCLA, California 90095-1567, USA
124: $^2$ LPMC, CNRS UMR 6622 and Univ. Nice, 06108 Nice, France
125: $^3$ MTEC, ETH Zurich, Switzerland
126:
127:
128:
129: \end{document}
130:
131:
132: