1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: % grlsample.tex: this sample file is for articles formatted with LaTeX2e,
3: % Modified June 2004
4: %
5: % This template is set up logically, with commands and instructions
6: % given in the order necessary to produce a final output that will
7: % satisfy AGU requirements.
8: %
9: % PLEASE DO NOT USE YOUR OWN MACROS
10: %
11: % All questions should be e-mailed to author.help@agu.org.
12: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
13: %
14: % ARTICLE MODE
15: %
16: % PLEASE USE THE GALLEY OPTION TO SUBMIT YOUR PAPERS
17: % IF YOU HAVE MULTI-LINE EQUATIONS
18: % The galley option produces single spaced, single column output
19: %
20: % LaTeX2e (galley):
21: %\documentclass[galley,grl]{agu2001}
22: \documentclass[grl]{agu2001}
23: % LaTeX2e (draft):
24: %\documentclass[draft,grl]{agu2001}
25: %
26: % LaTeX2.09 (galley):
27: %\documentstyle[galley,grl]{agu2001}
28: % LaTeX2.09 (draft):
29: %\documentstyle[galley,grl]{agu2001}
30: %
31:
32: %% ------------------------------------------------------------------------ %%
33: %
34: % IMAGE DISPLAY
35: %
36: %% ------------------------------------------------------------------------ %%
37: %
38: % Uncomment the following line if you need to include images
39: \usepackage{graphicx}
40: %
41: % PLEASE NOTE: WHEN YOU SUBMIT YOUR LATEX FILE TO GEMS, COMMENT OUT ANY COMMANDS
42: % THAT INCLUDE GRAPHICS.
43: % (See FIGURES section near the end of the file)
44:
45: %% ------------------------------------------------------------------------ %%
46: %
47: % ENTER PREAMBLE
48: %
49: %% ------------------------------------------------------------------------ %%
50:
51: \authorrunninghead{PIEGARI et al.}
52: % Author names in capital letters,
53:
54: \titlerunninghead{A cellular automaton for landslides}
55: % Shorter version of title entered in capital letters
56:
57: % Author address will appear at end of article, may repeat
58: % this command for each author.
59: \authoraddr{
60: %E. Piegari, V. Cataudella, R. Di Maio, L. Milano, and
61: %M. Nicodemi. Dip. di Scienze Fisiche, Universit\'a di
62: %Napoli ``Federico II'', INFN and CNR-Coherentia, I-80126 Napoli, Italy.
63: piegari@na.infn.it,
64: cataudella@na.infn.it, dimaio@na.infn.it, milano@na.infn.it,
65: mario.nicodemi@na.infn.it}
66:
67: \begin{document}
68:
69: %% ------------------------------------------------------------------------ %%
70: %
71: % ENABLE IMAGE DISPLAY WHILE USING DRAFT MODE
72: %
73: %% ------------------------------------------------------------------------ %%
74: %
75: % Uncomment the following code (as well as \usepackage{graphicx} above)
76: % if you need to include images in draft mode
77: %\setkeys{Gin}{draft=false}
78: %
79: % PLEASE NOTE: WHEN YOU SUBMIT YOUR LATEX FILE TO GEMS, COMMENT OUT ANY COMMANDS
80: % THAT INCLUDE GRAPHICS.
81: % (See FIGURES section near the end of the file)
82: %
83:
84: %% ------------------------------------------------------------------------ %%
85: %
86: % TITLE
87: %
88: %% ------------------------------------------------------------------------ %%
89:
90:
91: \title{A cellular automaton for the factor of safety field in landslides modeling}
92:
93: %
94: % e.g., \title{Terrestrial Ring Current:
95: % Origin, Formation and Decay $\alpha\beta\Gamma\Delta$}
96:
97: %% ------------------------------------------------------------------------ %%
98: %
99: % AUTHORS AND AFFILIATIONS - 3 methods
100: %
101: %% ------------------------------------------------------------------------ %%
102:
103: % Method 1 (for all journals, except Reviews of Geophysics, which
104: % should use method 3):
105: % For three or fewer author/affiliation blocks, use \author{} and \affil{}
106:
107: \author{E. Piegari, V. Cataudella, R. Di Maio, L. Milano, and M. Nicodemi}
108: \affil{Dip. di Scienze Fisiche, Universit\'a di Napoli ``Federico II'',
109: INFN, CRdC AMRA, CNR-Coherentia, Napoli, Italy}
110:
111: % ---------------
112: % Method 2 (for all journals, except Reviews of Geophysics, which
113: % should use method 3):
114: % For more than three author/affiliation blocks,
115: % use \author{\altaffilmark{}} and \altaffiltext{}
116: % \altaffilmark will produce footnote;
117: % matching altaffiltext will appear at bottom of page.
118: % May use \\ to start a new line.
119:
120: %\authors{E. Piegari, \altaffilmark{1}
121: %V. Cataudella,\altaffilmark{1}
122: % R. Di Maio,\altaffilmark{1}
123: % L. Milano,\altaffilmark{1}
124: % and M. Nicodemi \altaffilmark{1}}
125:
126: %\altaffiltext{1}
127: %{Dipartimento di Scienze Fisiche, Universit\'a di Napoli, Napoli, Italy.}
128:
129: %% ------------------------------------------------------------------------ %%
130: %
131: % ABSTRACT
132: %
133: %% ------------------------------------------------------------------------ %%
134:
135: % Do NOT include any \begin...\end commands within
136: % the body of the abstract.
137:
138: \begin{abstract}
139: Landslide inventories show that the statistical distribution of the area of
140: recorded events is well described by a power law over a range of decades.
141: To understand these distributions, we consider a cellular automaton to model
142: %based on a dissipative dynamical variable associated to
143: a time and position dependent factor of safety. The model is able to
144: reproduce the complex structure of landslide distribution, as
145: experimentally reported. In particular, we investigate the role of
146: the rate of change of the system dynamical variables, induced by an
147: external drive, on landslide modeling and its implications on hazard
148: assessment. As the rate is increased, the model has a crossover from
149: a critical regime with power-laws to non power-law behaviors. We
150: suggest that the detection of patterns of correlated domains in
151: monitored regions can be crucial to identify the response of the
152: system to perturbations, i.e., for hazard assessment.
153: \end{abstract}
154:
155: %% ------------------------------------------------------------------------ %%
156: %
157: % BEGIN ARTICLE
158: %
159: %% ------------------------------------------------------------------------ %%
160:
161: % The body of the article must start with a \begin{article} command,
162: % and an \end{article} command must be placed at the end of the file,
163: % before \end{document}.
164: %
165: % If using draft mode \end{article} must follow the references section.
166:
167: \begin{article}
168:
169: %% ------------------------------------------------------------------------ %%
170: %
171: % TEXT
172: %
173: %% ------------------------------------------------------------------------ %%
174:
175: %\section
176: {\bf Introduction } As for earthquakes and forest fires, there is a
177: compelling evidence that the landslide frequency-size distributions
178: are power-law functions of the area \citep{Turcotte_02}. The
179: presence of these broad distributions has crucial consequences on
180: both the basic understanding of these phenomena and the practical
181: and relevant purposes, such as the evaluation of natural hazards.
182: Here, we introduce a cellular automaton that is aimed at modeling
183: the general features of landslides, and is focused on the dynamical
184: evolution of a space and time dependent factor of safety field. This
185: model is very simple, but it is able to give a comprehensive picture
186: of the avalanching phenomena and to reproduce some well-known
187: properties of landslide distributions.
188:
189: Several authors invoked the paradigm of self-organized criticality (SOC)
190: \citep{Bak_87,Jensen_98,Turcotte_99}
191: to explain landslide distributions
192: \citep{Turcotte_02,Pelletier_97,Hertgarten_00}.
193: %The idea of SOC \citep{Bak_87}, applied to many media exhibiting avalanche
194: %dynamics \citep{Jensen_98,Turcotte_99}, refers to the tendency of
195: %natural systems to self-organize into a critical state where even
196: %an infinitesimal perturbation can cause huge responses (i.e., events
197: %with sizes of the order of the system) with a probability of
198: %occurrence that follows a power-law.
199: Although the ``critical'' nature of the present phenomenon is not
200: yet assessed and many authors believe that deviations from power-law
201: appear to be systematic for small landslides data
202: \citep{Stark_01,Pelletier_97,Brardinoni_04,Malamud_04}, several
203: regional landslide inventories records show robust power-law
204: distributions of large events with an exponent around $\alpha \sim
205: 2.5$ \citep{Turcotte_02}, ranging approximately from $\alpha \sim
206: 1.75$ for rockfalls to $\alpha \sim 2.8$ for mixed landslides (see
207: \citep{Dussauge_03,Faillettaz_04} and references therein). The
208: universality of such an exponent is still debated (see
209: \citep{Turcotte_02,Dussauge_03,Faillettaz_04,Malamud_04} and
210: references above), and its reported values are far from the one in
211: the original ``sandpile model'' \citep{Bak_87}, where $\alpha \sim
212: 1.0$. Recently, the reported values of $\alpha$ have been obtained
213: by introducing two-thresholds mechanisms in models that relate
214: landslide dynamics both to SOC \citep{Pelletier_97,Hertgarten_00}
215: and to non SOC cellular automata \citep{Faillettaz_04}. Furthermore,
216: models based on $\Gamma$ \citep{Malamud_04} or Pareto
217: \citep{Stark_01} distributions have been proposed.
218:
219: Here, we consider a model inspired to an {\em anisotropic} version
220: of the Olami-Feder-Christensen (OFC) \citep{Olami_92} cellular
221: automaton and subject to a {\em finite driving rate}
222: \citep{Hamon_02}. The model describes the evolution of a space and
223: time dependent factor of safety, which is investigated for the first
224: time in the present framework. In particular, we outline the
225: essential role played by the rate of change of the system dynamical
226: variables (variation of pore water pressure, lithostatic stress,
227: cohesion coefficients, etc., \citep{Helley_04,Iverson_00}) induced
228: by external triggers. We find the model to be at the edge of the SOC
229: limit. Actually, such a limit, which is achieved only in the
230: asymptotic condition of vanishing driving rate, is hardly attainable
231: in a real landslide process. The model is able to reproduce
232: power-law distributions with exponents very close to the observed
233: values. Power-laws are robust even though their exponent smoothly
234: depends on system parameters (e.g., time derivative of the factor of
235: safety and its dissipation level, see below). In this sense,
236: although the SOC paradigm to some extent may be applied to
237: landslides \citep{Turcotte_02}, the idea of universality, within
238: this model, must be restricted to the shape of the frequency-size
239: distribution rather than to its exponent, as can be deduced from
240: some catalogues \citep{Dussauge_03,Faillettaz_04}. Finally, in
241: presence of strong driving rates we find that the model has Gaussian
242: behaviors. We examine below the implications of our results on
243: hazard assessment.
244:
245:
246: %\section
247: {\bf The Model } The empirical Mohr-Coulomb failure criterion
248: establishes that landslides occur when the shear stress exceeds a
249: maximum value, which is given by $\tau_{max}=c + (\sigma
250: -u)\tan\phi$, with $\sigma$ the total normal stress, $u$ the
251: pore-fluid pressure, $\phi$ the angle of internal friction of the
252: soil and $c$ the cohesional (non-frictional) component of the soil
253: strength \citep{Terzaghi_62}. In literature, the factor of safety,
254: $FS$, against slip is defined by the ratio of the maximum shear
255: strength $\tau_{max}$ to the disturbing shear stress $\tau$
256: \begin{equation}
257: FS = \frac{\tau_{max}}{\tau}.
258: \end{equation}
259: If $FS>1$ resisting forces exceed driving forces and the slope
260: remains stable. Slope failure starts when $FS = 1$. Although the
261: practical determination of $FS$ is difficult, simple one-dimensional
262: infinite-slope models can quantify how $c,u$ and $\phi$ influence
263: the Coulomb failure and show that the ground-water term has the most
264: widely ranging influence \citep{Iverson_97}. Traditional models
265: generally treat soils and rocks as continuous porous media that obey
266: to the Darcy's law. Actually, field evidence indicates that the
267: hydrology of some natural slopes is strongly influenced by
268: discontinuities such as fractures and macropores. In practice,
269: observations of large spatial and temporal fluctuations of water
270: flow, within slopes at different sites, support the assertion that
271: water-flow paths and permeability continually change within the
272: slopes and also during the failure, providing different local values
273: of the pore pressures and of the cohesion
274: \citep{Iverson_97,Helley_04,Iverson_00}.
275:
276: In order to take into account the complex non-homogeneous structure
277: of a slope in the above failure condition, we consider a site and
278: time dependent factor of safety, $FS$. In particular, we approximate
279: a natural slope by a two-dimensional (square) grid and define on
280: each cell, $i$, of the lattice a local variable $e_i = 1/FS_i$. Such
281: a local inverse factor of safety is the fundamental dynamical
282: variable of our model. The presence of diffusion, dissipative and
283: driving mechanisms acting in the soil, such as those on the water
284: content, inspires the dynamics of our model, which is defined by the
285: following operations. Starting from a random and ``stable'' initial
286: configuration ($e_i<1$ $\forall i$), the system is subject to
287: changes caused by some external trigger, as for instance a uniform
288: rainfall, and the values of $e_i$ on each cell of our grid change at
289: a given rate $\nu$, $e_i \rightarrow e_i + \nu$. For the sake of
290: simplicity, we consider here only a uniform driving rate, $\nu$, but
291: different choices can be made to simulate the effect of different
292: hydrologic and external triggering mechanisms. The model is driven
293: as long as $e_i < 1$ on all sites $i$. Then, when the generic site
294: $i$ becomes unstable (i.e., overpasses the threshold, $e_i\geq 1$),
295: it relaxes with its neighbors according to the rule:
296: \begin{itemize}
297: \item[] $e_i \rightarrow 0$; \quad \quad $e_{nn} \rightarrow e_{nn}+f_{nn} e_i$,
298: \end{itemize}
299: \noindent where the index $nn$ denotes the nearest neighbors of site
300: $i$ and $f_{nn}$ is the fraction of $e_i$ toppling on $nn$ (after
301: failure we set $1/FS=0$ for simplicity, as any other finite level
302: would work \citep{Jensen_98}). This kind of chain relaxations
303: (``avalanches'') is considered to be instantaneous compared to the
304: time scale of the overall drive and it lasts until all sites are
305: below threshold. The model is said to be conservative if
306: $C=\sum_{nn} f_{nn} =1$. Since many complex dissipative phenomena
307: (such as evaporation mechanism, volume contractions, etc.
308: \citep{Fredlund_93}) contribute to a dissipative stress transfer, we
309: consider the non-conservative case $C<1$, which is different from
310: previous landslide models \citep{Pelletier_97,Hertgarten_00}. Since
311: gravity individuates a privileged direction, we consider an
312: anisotropic model where the fraction of $e_i$ moving from the site
313: $i$ to its ``downward'' (resp. ``upward'') neighbor on the square
314: lattice is $f_d$ (resp. $f_u$), as $f_l=f_r$ is the fraction to each
315: of its ``left'' and ``right'' neighbors. In particular, we assume
316: $f_u < f_d$ and $f_l =f_r < f_d$. This choice of parameters is made
317: in the attempt to sketch the complex relaxation processes occurring
318: in a slope failure. The conservation level, $C$, and the anisotropy
319: factors, $f$'s, which we assume to be uniform, are actually related
320: to the local soil properties (e.g., lithostatic, frictional and
321: cohesional properties), as well as to the local geometry of the
322: slope (e.g., its morphology). The rate of change of the inverse
323: factor of safety, $\nu$, which is induced by the external drive
324: (e.g., rainfall) and is related to soil and slope properties,
325: quantifies how the triggering mechanisms affect the time derivative
326: of the FS field.
327: %Laboratory experiments
328: %%about slope failures could be crucial for
329: %might help testing the model and relating its
330: %parameters to physical properties.
331:
332:
333: %\section
334: {\bf Numerical Results }
335: We consider a $64 \times 64$ square
336: lattice, implementing both cylindrical (open along the vertical axis
337: and periodical along the horizontal axis) and open boundary
338: conditions, which do not give appreciable differences.
339: %we checked differ in the slopes of the distribution curves for less
340: %than $1\%$.
341: Once the system has attained a stationary state in its dynamics, we
342: study the probability distribution, $P(s)$, of avalanches of size
343: $s$. During a run (we treat statistics of over $10^9$ events per
344: run) the conservation level $C$ and the rate, $\nu$, are kept fixed.
345: Examples of the frequency-size distribution of avalanches, $P(s)$,
346: %in the anisotropic non conservative case $C=0.4$, with $f_u
347: %/f_d=2/3$, $f_l/f_d =5/6$,
348: are shown in figures \ref{scaling} and \ref{scaling_C}. In figure
349: \ref{scaling}, the different curves correspond to different values
350: of the rate $\nu$, for $C=0.4$, and in figure \ref{scaling_C} to
351: different values of $C$, for $\nu=5\cdot 10^{-3}$.
352:
353: In the limit of very small driving rate, i.e., $\nu\rightarrow 0$,
354: the distribution of events, $P(s)$, exhibits the typical SOC
355: structure (see figure \ref{scaling}): a power law characterized by a
356: critical exponent $\alpha$, $P(s)\sim s^{-\alpha}$, in agreement
357: with the experimental evidence on medium and large landslides
358: \citep{Turcotte_02,Pelletier_97,Hertgarten_00,Faillettaz_04,Brardinoni_04,Malamud_04,Dussauge_03},
359: followed by a size dependent exponential cutoff \citep{Jensen_98}.
360: By increasing the rate $\nu$, the power-law regime shifts towards
361: larger sizes and at some point the probability distribution
362: apparently shows a maximum for a value $s^*$. There are two regimes:
363: for large landslides ($s>s^*$) the above structure $P(s)\sim
364: s^{-\alpha}$ is found, while for small events ($s<s^*$) an
365: increasing function of $s$ is observed. Such a complex structure
366: %, with a peak separating two regions,
367: is absent in SOC models, instead a maximum is found in landslide
368: inventory maps for small landslide data, although there is no
369: consensus about the nature of such a feature
370: \citep{Stark_01,Brardinoni_04,Malamud_04}. The values of the
371: power-law exponent, $\alpha$, by varying the rate, $\nu$, and $C$
372: are very close to those experimentally found
373: \citep{Turcotte_02,Dussauge_03}. As in the original isotropic OFC
374: model \citep{Olami_92}, the critical exponent decreases with the
375: level of conservation, $C$ (see inset in figure \ref{scaling_C}).
376: The value of $\alpha$ slightly changes with the anisotropic ratios
377: $f_d/f_d$ and $f_u/f_d$, except when they get too small
378: \citep{Piegari_05} where, as found also in other models
379: \citep{Amitrano_99, Amitrano_03, Faillettaz_04}, the event size
380: distribution is considerably modified.
381: %A detailed analysis of $\alpha$ as function of
382: %$\nu$, $C$ and the anisotropy will be given elsewhere \citep{Piegari_05}.
383: The power-law regime is crucially robust to changes in system
384: parameters. For instance, in the case of figure \ref{scaling} it can
385: be found for $\nu$ up to approximately $10^{-2}$, all over the range
386: $C\in[0.4,0.8]$. It is worth noting that the $\alpha$ values here
387: obtained are comparable to those found in models of failure in fiber
388: bundle \citep{Hansen_Hammer92,Hansen_Hammer94,Hidalgo_02}.
389:
390: As it can be seen in figure \ref{scaling}, a further increase of the
391: driving rate (above $\nu\sim 10^{-2}$) causes a crossover to a
392: markedly different regime where power-laws are no longer apparent
393: and a bell shaped (Gaussian) distribution emerges, whose peak shifts
394: towards larger sizes and shrinks up. Such a behavior is to be
395: expected since for strong driving rates all internal correlations
396: are washed out.
397:
398: Summarizing, the conservation level, $C$, and the time derivative of
399: $1/FS$, $\nu$, turn out to be important to determine landslide
400: probability distributions: in the limit $\nu\rightarrow 0$, the
401: model is indeed in the SOC class; for small but finite $\nu$ the
402: system is at the edge of SOC and the critical behaviors are still
403: largely observed; finally, as $\nu$ gets large enough, Gaussian
404: properties are found. Thus, in the small $\nu$ regime, our model
405: reproduces the general properties of existing catalogs and can help
406: interpreting them, while in the large $\nu$ regime it foresees a
407: different class of behavior. Nevertheless, for the sake of clarity,
408: we have considered the simple case where the rates $\nu$ and $C$ are
409: uniform. Thus, the distributions of figure \ref{scaling} may be not
410: directly comparable to landslide inventories, which gather events
411: with non-uniform driving rates.
412:
413: Pictures of a typical ``avalanche'' in the different regimes
414: discussed above are plotted in figure \ref{map} (upper panels) with
415: the corresponding values of the factor of safety, $FS_i$, on the
416: model grid after the avalanche (lower panels). The snapshots in the
417: upper panels, taken in two systems driven at different rates (left
418: $\nu=5\cdot 10^{-3}$, right $\nu=5\cdot 10^{-2}$), show a typical
419: event with size $s=230$ (such a value is chosen because it has
420: approximately the same probability in the two cases, see figure
421: \ref{scaling}):
422: the system on the left is in the power-law regime; %, $\nu=5\cdot 10^{-3}$;
423: the one on the right is in the non power-law regime. %, $\nu=5\cdot 10^{-2}$.
424: The difference of the avalanche geometry in the two cases is
425: impressive. Domino effects are crucial to determine a
426: ``catastrophic'' event when the system is governed by a power-law
427: statistics, where a huge compact landslide is present
428: \citep{Pietronero_91} with a typical size of the order of the system
429: size (left-upper panel). Conversely, large events are expected at
430: higher $\nu$ (where indeed the average size $\langle s\rangle$ is
431: much larger than $230$), but in such a regime a typical event with
432: $s=230$ is made of many tiny unconnected avalanches (summing up to
433: $s=230$).
434:
435: Interestingly, even though the $P(s)$ is very different in the two
436: cases, the probability distribution, $P(FS)$,
437: of the spatial values of $FS$ on the grid has a
438: similar Gaussian shape, with comparable averages
439: $\langle FS\rangle$ and fluctuations $\langle \Delta FS^2\rangle$
440: ($\langle FS \rangle=2.20$ and $\langle \Delta FS^2\rangle=0.16$
441: for $\nu=5\cdot 10^{-3}$; $\langle FS \rangle=2.53$ and
442: $\langle \Delta FS^2\rangle=0.06$ for $\nu=5\cdot 10^{-2}$)
443: laying far above the instability threshold $FS=1$.
444: Thus, a measure of just an average safety factor on the
445: investigated area could provide only very partial information about
446: the statistics governing landslide events.
447:
448: The origin of the striking difference of the $P(s)$ in the two
449: considered cases traces back to the relative extension of spatial
450: correlations of the factor of safety, $FS$, which is derived from the
451: correlation function:
452: \begin{equation}
453: C(\vec z)={\langle FS(\vec r)FS(\vec r+\vec z)\rangle-\langle FS(\vec
454: r)\rangle^2 \over \langle FS(\vec r)^2\rangle-\langle FS(\vec r)\rangle^2}
455: \end{equation}
456: where $FS(\vec x)$ is $FS$ at position $\vec x$ (here we take $\vec
457: z$ along the direction of the slope) and the average is over the
458: system sites. As it is well-known \citep{Jensen_98}, we find that
459: $C(\vec z)\propto \exp(-z/\xi)$, where $\xi$ is the spatial
460: correlation length of $FS$. The value of $FS_i$ on site $i$ is shown
461: in the lower panels of figure \ref{map}, in gray scale, for the same
462: cases pictured in the upper panels:
463: %(left $\nu =5\cdot 10^{-3}$, right $\nu =5\cdot 10^{-2}$):
464: patterns of large correlated areas (regions with similar values of
465: $FS_i$, i.e., the same color) are apparent in the left bottom panel
466: and, in practice, absent in the right one. In the power-law regime
467: (e.g., $\nu=5\cdot 10^{-3}$), the correlation length, $\xi$, is of
468: the order of the system size; thus, even a very small perturbation
469: (say, a drop of water) at one single point can trigger huge system
470: responses. Instead, in the non power law regime (e.g., $\nu=5\cdot
471: 10^{-2}$) large-scale correlations are absent; here, large events
472: trivially occur just because the strong external driving rate makes
473: likely that many cells simultaneously approach the instability
474: threshold. The detection of patterns of correlated domains (i.e.,
475: the size of $\xi$) in investigated areas results, thus, to be a
476: crucial tool to identify the response of the system to
477: perturbations, i.e., for hazard assessment.
478:
479:
480: %\section
481: {\bf Conclusions } To summarize, we have investigated a continuously
482: driven anisotropic cellular automaton model for the characterization
483: of landslide size distributions. The model may help in interpreting
484: the general behaviors observed in real systems. In particular, we
485: have found that different values of the driving rate give rise to
486: different statistical distributions of events. The determination of
487: correlated domains in the factor of safety becomes crucial for
488: landslide classification and, consequently, for hazard assessment.
489:
490: %% ------------------------------------------------------------------------ %%
491: %
492: % ACKNOWLEDGMENTS
493: %
494: %% ------------------------------------------------------------------------ %%
495:
496: \begin{acknowledgments}
497: Work supported by MIUR-PRIN '04, CRdC-AMRA.
498: \end{acknowledgments}
499:
500: %% ------------------------------------------------------------------------ %%
501: %
502: % REFERENCE LIST AND TEXT CITATIONS
503: %
504: %% ------------------------------------------------------------------------ %%
505: %
506: % If you use BiBTeX for your References, please do not send
507: % your bibliography database. Copy the reference list
508: % from your .bbl file into your article file before submission:
509: %
510: %1. Run LaTeX on your LaTeX file.
511: %
512: %2. Run BiBTeX on your LaTeX file.
513: %
514: %3. Open the new .bbl file containing the reference list and
515: %copy all the contents into your LaTeX file after the
516: %acknowledgments section;
517: %
518: %4. Comment out the old \bibliographystyle and \bibliography commands.
519: %
520: %5. Run LaTeX on your new file before submitting.
521:
522: %Failure to follow these instructions will require manual
523: %intervention through hard keying of information,
524: %which can introduce errors.
525:
526: %\bibliographystyle{agufull04}
527: %\bibliography{SOC}
528:
529: \begin{thebibliography}{19}
530: \providecommand{\natexlab}[1]{#1} \expandafter\ifx\csname
531: urlstyle\endcsname\relax
532: \providecommand{\doi}[1]{doi:\discretionary{}{}{}#1}\else
533: \providecommand{\doi}{doi:\discretionary{}{}{}\begingroup
534: \urlstyle{rm}\Url}\fi
535:
536: \bibitem[{\textit{Amitrano et~al.}(1999)\textit{Amitrano et al.}}]{Amitrano_99}
537: Amitrano, D., J.~R.~Grasso, D.~Hantz (1999), From diffuse to
538: localised damage through elastic interaction, \textit{Geophys. Res.
539: Lett.}, \textit{26}, 2109.
540:
541: \bibitem[{\textit{Amitrano}(2003)\textit{Amitrano}}]{Amitrano_03}
542: Amitrano, D. (2003), Brittle-ductile transition and associated
543: seismicity: Experimental and numerical studies and relationship with
544: the b value, \textit{J. Geophys. Res.}, \textit{108B1}, 2044.
545:
546: \bibitem[{\textit{Bak et~al.}(1987)\textit{Bak, Tang, and Wiesenfeld}}]{Bak_87}
547: Bak, P., C.~Tang, and K.~Wiesenfeld (1987), Self-organized
548: criticality: An explanation of 1/f noise, \textit{Phys. Rev. Lett.},
549: \textit{59}, 381.
550:
551: \bibitem[{\textit{Brardinoni and Church}(2004)}]{Brardinoni_04}
552: Brardinoni, F., and M.~Church (2004), Representing the landslide
553: magnitude-frequency relation: Capilano river basin, British Columbia,
554: \textit{Earth Surf. Process. Landforms}, \textit{29}, 115.
555:
556: \bibitem[{\textit{Dussauge et~al.}(2003)\textit{Dussauge et al.}}]{Dussauge_03}
557: Dussauge, C., J.~R.~Grasso, and A.~Helmstetter (2003), Statistical
558: analysis of rockfall volume distribution: implication for rockfall
559: dynamics, \textit{J. Geophys. Res.}, \textit{108(B6)}, 2286.
560:
561: \bibitem[{\textit{Faillettaz et~al.}(2004)\textit{Faillettaz, Louchet, and
562: Grasso}}]{Faillettaz_04}
563: Faillettaz, J., F.~Louchet, and J.~R.~Grasso (2004), Two-threshold
564: model for scaling laws of noninteracting snow avalanches,
565: \textit{Phys. Rev. Lett.}, \textit{93}, 208,001.
566:
567: \bibitem[{\textit{Fredlund and Rahardjo}(1993)}]{Fredlund_93}
568: Fredlund, D.~G., and H.~Rahardjo (1993), \textit{Soil Mechanics for
569: Unsatured Soils}, Wiley-Interscience, New York.
570:
571: \bibitem[{\textit{Hamon et~al.}(2002)\textit{Hamon, Nicodemi, and
572: Jensen}}]{Hamon_02}
573: Hamon, D., M.~Nicodemi, and H.~J. Jensen (2002), Continuously driven
574: OFC: A simple model of solar flare statistics,
575: \textit{Astronomy\&Astrophysics}, \textit{387}, 326.
576:
577: \bibitem[{\textit{Hansen et~al.}(1994)\textit{Hansen and Hemmer}}]{Hansen_Hammer94}
578: Hansen, A., P.~C. Hemmer (1994), Burst avalanches in bundles of
579: fiber - local v.s. global load-sharing, \textit{Phys. Lett. A, 184},
580: 394.
581:
582: \bibitem[{\textit{Helley et~al.}(2004)\textit{Helley, Burgmann, Ferretti,
583: Novali, and Rocca}}]{Helley_04}
584: Helley, G.~E., R.~Burgmann, A.~Ferretti, F.~Novali, and F.~Rocca,
585: Dynamics of slow-moving landslides from permanent scatterer analysis,
586: \textit{Science}, \textit{304}, 1952.
587:
588: \bibitem[{\textit{Hemmer et~al.}(1992)\textit{Hemmer and Hansen}}]{Hansen_Hammer92}
589: Hemmer, P.C., A.~Hansen (1992), The distribution of simultaneous
590: fiber failures in fiber-bundels, \textit{J. Appl. Mech., 59}, 909.
591:
592: \bibitem[{\textit{Hertgarten and Neugebauer}(2000)}]{Hertgarten_00}
593: Hertgarten, S., and H.~Neugebauer (2000), Self-organized criticality
594: in two-variable models, \textit{Phys. Rev. E}, \textit{61}, 2382.
595:
596: \bibitem[{\textit{Hidalgo et~al.}(2002)\textit{Hidalgo et al.}}]{Hidalgo_02}
597: Hidalgo, R.C., Y.~Moreno, F.~Kun, and H.~J.~Herrmann (2002),
598: Fracture model with variable range of interaction \textit{Phys. Rev.
599: E 65}, 046148.
600:
601: \bibitem[{\textit{Iverson et~al.}(2000)\textit{Iverson, Reid, Iverson, Lahusen,
602: an~J.~E.~Mann, and Brien}}]{Iverson_00}
603: Iverson, M., M.~E. Reid, N.~R. Iverson, R.~G. Lahusen, M.~L.
604: an~J.~E.~Mann, and D.~L. Brien (2000), Acute sensitivity of
605: landslide rates to initial soil porosity, \textit{Science},
606: \textit{290}, 513.
607:
608: \bibitem[{\textit{Iverson et~al.}(1997)\textit{Iverson, Reid, and
609: Lahusen}}]{Iverson_97}
610: Iverson, R.~M., M.~E. Reid, and R.~G. Lahusen (1997), Debris-low
611: mobilization
612: from landslide, \textit{Annu. Rev. Earth Planet. Sci.}, \textit{25}, 85.
613:
614: \bibitem[{\textit{Jensen}(1998)}]{Jensen_98}
615: Jensen, H.~J. (1998), \textit{Self-Organized Criticality: emergent
616: complex behavior in physical and biological systems}, Cambridge
617: University Press, Cambridge.
618:
619: \bibitem[{\textit{Malamud et~al.}(2004)\textit{Malamud, Turcotte, Guzzetti, and
620: Reichenbach}}]{Malamud_04}
621: Malamud, B.~D., D.~L. Turcotte, F.~Guzzetti, and P.~Reichenbach
622: (2004), Landslide inventories and their statistical properties,
623: \textit{Earth Surf. Process. Landforms}, \textit{29}, 687.
624:
625: \bibitem[{\textit{Olami et~al.}(1992)\textit{Olami, Feder, and
626: Christensen}}]{Olami_92}
627: Olami, Z., H.~J.~S. Feder, and K.~Christensen (1992),
628: Self-organized
629: criticality in a continuous, nonconservative cellular automaton modeling
630: earthquakes, \textit{Phys. Rev. Lett.}, \textit{68}, 1244.
631:
632: \bibitem[{\textit{Pelletier et~al.}(1997)\textit{Pelletier, Malamud, Blodgett,
633: and Turcotte}}]{Pelletier_97}
634: Pelletier, J.~D., B.~D. Malamud, T.~Blodgett, and D.~L. Turcotte
635: (1997),
636: Scale-invariance of soil moisture variability and its implications for the
637: frequency-size distribution of landslide, \textit{Eng. Geol.}, \textit{48},
638: 255.
639:
640: \bibitem[{\textit{Piegari et~al.}(2005)\textit{Piegari, Cataudella, Di Maio,
641: Milano, and Nicodemi}}]{Piegari_05}
642: Piegari, E., V.~Cataudella, R.~Di Maio, L.~Milano, and M.~Nicodemi
643: (2005), submitted to Phys. Rev. E
644:
645: \bibitem[{\textit{Pietronero and Schneider}(1991)}]{Pietronero_91}
646: Pietronero, L., and W.~R. Schneider (1991), Fixed scale
647: transformation approach
648: to the nature of relaxation clusters in self-organized criticality,
649: \textit{Phys. Rev. Lett.}, \textit{66}, 2336.
650:
651: \bibitem[{\textit{Stark and Hovius}(2001)}]{Stark_01}
652: Stark, C.~P., and N.~Hovius (2001), The characterization of
653: landslide size
654: distributions, \textit{Geophys. Res. Lett.}, \textit{28}, 1091.
655:
656: \bibitem[{\textit{Terzaghi}(1962)}]{Terzaghi_62}
657: Terzaghi, K. (1962), Stability of steep slopes on hard unweathered
658: rock,
659: \textit{Geothecnique}, \textit{12}, 251.
660:
661: \bibitem[{\textit{Turcotte}(1999)}]{Turcotte_99}
662: Turcotte, D.~L. (1999), Self-organized criticality, \textit{Rep.
663: Prog. Phys.},
664: \textit{62}, 1377.
665:
666: \bibitem[{\textit{Turcotte et~al.}(2002)\textit{Turcotte, Malamud, Guzzetti,
667: and Reichenbach}}]{Turcotte_02}
668: Turcotte, D.~L., B.~D. Malamud, F.~Guzzetti, and P.~Reichenbach
669: (2002),
670: Self-organization, the cascade model, and natural hazards, \textit{Proc.
671: Natl. Acad. Sci. U.S.A.}, \textit{99}, 2530.
672:
673: \end{thebibliography}
674:
675: %% ------------------------------------------------------------------------ %%
676: %
677: % FIGURES
678: %
679: % PLEASE NOTE: WHEN YOU SUBMIT YOUR LATEX FILE TO GEMS, COMMENT OUT ANY COMMANDS
680: % THAT INCLUDE GRAPHICS (see example below).
681: %
682: % ---------------
683: % ONE-COLUMN figure example
684: % (For further instructions see FIGURE, PLATE, AND TABLES section at end of file)
685:
686: \begin{figure}
687: %\noindent\includegraphics[width=20pc]{scaling_wi.eps}
688: \caption{The probability density distribution, $P(s)$, of avalanches
689: of size $s$ is plotted for the shown values of the time derivative of
690: the inverse factor of safety, $\nu$
691: (model size $L^2=64\times64$, conservation level
692: $C=0.4$, anisotropy coefficients $f_u/f_d=2/3$ and $f_l/f_d=5/6$).
693: The power law $P(s)\sim s^{-\alpha}$ found in the limit
694: $\nu\rightarrow 0$ is partially preserved by increasing $\nu$ up to a
695: point where a bell shaped behavior is clearly observed.
696: }
697: \label{scaling}
698: \end{figure}
699:
700: \begin{figure}
701: %\noindent\includegraphics[width=20pc]{cons_mod.eps}
702: \caption{The probability distribution, $P(s)$, of avalanches of size
703: $s$ is plotted for the shown values of level of conservation, $C$
704: ($\nu=5\cdot 10^{-3}$, other param.s as in figure \ref{scaling}).
705: The {\bf inset} shows the exponent $\alpha$ (confidence interval
706: $95\%$) of the power-law fit as a function of $C$.
707: }\label{scaling_C}
708: \end{figure}
709:
710: \begin{figure}
711: %\noindent\includegraphics[width=20pc]{unica_bw.eps}
712: \caption{{\bf Top panels}: Two snapshots of a typical landslide event of size
713: $s=230$ on our $64\times 64$ grid, in two cases with a driving rate $\nu
714: =5\cdot 10^{-3}$ ({\em left figure}) and $\nu =5\cdot 10^{-2}$
715: ({\em right figure}). The cells marked in black are those which reached
716: the instability threshold.
717: %The value $s=230$ is such that the event has approximately the same
718: %probability to occur in both cases (see Fig.1).
719: {\bf Bottom panels}: The pictures plot the local value of the
720: factor of safety, FS, corresponding to the stable configurations
721: reached after the avalanches shown in the upper panels. The FS values
722: have been associated to ten
723: levels of color from white to black, in order to measure the
724: distance of a cell from its instability condition: the darker the
725: color, the farther is the cell from the instability threshold. In
726: the panels it is possible to recognize as dark areas the
727: avalanches shown in the corresponding upper grids. In particular,
728: dark areas are related to previous landslide events, as
729: the lighter areas indicate regions of future events.
730: In the {\em left figure} large correlated regions (compact areas with
731: same color) are observed, whereas their size is small in the
732: {\em right figure}.
733: }\label{map}
734: \end{figure}
735:
736:
737: %% ------------------------------------------------------------------------ %%
738: %
739: % END ARTICLE
740: %
741: %% ------------------------------------------------------------------------ %%
742:
743: \end{article}
744:
745: \end{document}
746: