Herald of the
Russian Academy of Sciences |
On a National Scientific
Monitoring System G. G.
Malinetskii, A. V. Podlazov, and I. V. Kuznetsov* For
almost a decade, the interaction between researchers who forecast, analyze,
and evaluate the consequences of emergencies, disasters, and catastrophes
and those who should, in fact, prevent and parry them has been developing
according to the same scenario. First, scientists bring forward what is, in
their opinion, a reasonable idea, discuss it at conferences, write papers,
and, finally, convince the corresponding ministries that the idea is really
worth studying. The words of scientists are then repeated many times over,
and the corresponding topic is placed at the command of industrial science or
is included into an academic program. Research into the topic is not properly
coordinated, and in the general case, it never reaches the point of
generating serious forecasts, the organization of monitoring, or the establishment
of a center in which experts would assume responsibility for the problem in
case of a crisis. Industrial institutes report the fulfillment of all their
plans, but the idea is discredited and ends up in a deadlock. One of the main
reasons for this state of affairs is the absence of judgment criteria as to
whether the work has been finished or not and the peculiar
"dropout" of the gleaned results from the feedback circuit, which
should have provided for the coordination of research. The idea
of controlling the risk of natural and technological disasters has,
unfortunately, followed this path [I]. The concept of monitoring and
forecasting social instabilities seems to be headed in the same direction
[2]. The implementation of a program for the evaluation of Russia's strategic
risks—threats that the country is now facing or will be facing in the near
future and that may change its historical path [3]—continues to face serious
challenges. The purpose of the notes that follow is to call attention to the
extreme topicality of a scientific program that forecasts disasters and
calamities, as well as crisis phenomena in modem Rus- ' Georgii
Gennadievich Malinetskii. Dr. Sci. (Phys.-Math.), is a deputy director of the
Keldysh Institute of Applied Mathematics. RAS. Andrei Viktorovich Podlazov.
Cand. Sci. (Phys.-Math.), is a senior researcher at the Keldysh Institute of
Applied Mathematics. RAS. Igor' Vasil'evich Kuznetsov, Cand. Sci. (Tech.),
is a deputy director of the RAS International Institute of the Theory of
Earthquake Forecasts and Mathematical Geophysics. sia, and a project of the national
scientific monitoring system [4]. WHY DO WE HAVE NO MODERN
FORECASTING AND PREVENTION OF DISASTERS? At a meeting with the leadership
of the Russian Academy of Sciences on December 3, 2001, the Russian
president pinpointed problems associated with the independent examination of
government decisions and forecasting and the prevention of calamities and
disasters in the natural and technological spheres, as well as of social
instabilities, as priorities that are presently facing Russia's scientific
community. That is, political decisions were adopted that are necessary in
order to establish a national system of forecasting and preventing dangerous
phenomena and processes. At the same time, several research institutes
are ready to cooperatively engage the potential solutions of this problem. The
concepts of strategic risks and crisis analysis have been
elaborated, and it has been agreed that it is necessary to establish a national
scientific monitoring system, which would bring together the necessary
informational flows and researchers, describe dangerous phenomena and
processes [4], and generate databases, models, and algorithms, as well as
coordinate the efforts of bodies that have the authority and resources to
prevent calamities and disasters. This area of research has received backing
from the RAS Presidium [5]. Despite the fact that several years have elapsed since the political decision, its serious implementation has not yet begun. On November 13, 2003, the Russian Security Council and the Presidium of the State Council of the Russian Federation discussed the issue "On Measures Protecting Critically Important Security Infrastructure Facilities and the Country's Population from Natural and Anthropogenic Threats and Terrorist Manifestations." The responsibility for the scientific support of these tasks rests with the Russian Academy of Sciences and the Russian Ministry of Education and Science, who are to draw up a state interdepartmental complex
program and start research. Yes neither the Academy nor the government has
established a procedure for examining and adopting state interdepartmental
programs. This means that the procedure should be adopted first. The attempts
that have been undertaken have yielded no results. An important political
decision has not found its practical implementation. Meanwhile, forest fires in Siberia, peat
fires that caused smog over Moscow, the flood in Krasnodar krai, mudflows
near Novorossiisk, the catastrophic avalanche in the Karmadon Canyon, and
terrorism in Russia. which has taken on new proportions (the Nord-Ost, the
explosion in the Moscow metro, Beslan), show that the presidential decision
is still topical. In our opinion, the following circumstances hinder its
practical implementation: • Departmentalism and the absence of the
organizational institutions necessary' to solve complex problems. The
task posed by the Russian president requires the coordination of efforts of several
departments, which is, according to current legislation, the prerogative of
the premier or vice premier. The leaders at this level are not engaged in the
solution of the problem posed. • The inadequacy of the management
structure to the problems of organizing the forecasting and preventing
system. The necessity of attracting political leaders for the solution of
specific organizational, technical, and scientific problems means that there
is no corresponding managerial structure in the country. Assistance,
support, and funding must constantly be requested from official channels that
are, generally speaking, unable to deal with the current problems— and that,
moreover, should not be dealing with them. • Underestimation of resources necessary
to solve problems. At a conference in Yokohama in 1994, the scientific
community that addresses the analysis of crisis events called for a
transition from the stage of the mere mitigation and liquidation of the
consequences of calamities and disasters to that of their forecasting and
prevention. At the national level, developed countries have driven this
initiative home to the state machinery and risk managers. In particular,
after US President B. Clinton had defined as one of the major problems that
of forecasting, preventing, and controlling instability in the natural,
social, and anthropogenic spheres in the United States and elsewhere, the
Russian Ministry for Emergencies some time later announced the same course. In
1997, the Federal Target Program "Reducing the Risks and Mitigating the
Consequences of Natural and Anthropogenic Emergencies in the Russian Federation
until 2005" was formed, which remains active today. Unfortunately, the
program is not very efficient, a situation that is due primarily to the shortage
of allocated funds. For example, the 2000 state budget allocated
nearly 47 million rubles to research in the Russian Ministry of Emergencies,
but in reality only slightly more than 16
million were paid out. The results of the research conducted and the
efficiency of their implementation did not, in fact, become the common
property of the scientific community that deals with risk problems, to say
nothing of administrators and persons who shape government strategy. In later
years, the situation remained the same. Funding, even when it is spent on
forecast and prevention research, is very limited. Adequate target programs
at the Russian Academy of Sciences are also absent. In the present
conditions of systemic crisis, when it is often impossible to divide factors
into natural, anthropogenic, and social, and many calamities and crises must
be considered in their totality, such a state of affairs is unacceptable. • The absence of an
organizational design of the problem and the role of scientific monitoring,
if it is established, in the country's administrative system. Assume that
the necessary research has been conducted and the monitoring system has been
developed and deployed. Will it ensure that people live safer lives and that
the economy develops with more stability in relation to natural and
anthropogenic disasters? Clearly, it will not. It is also necessary to define who
it is that will use the results of forecasts and risk assessments in public
administration, in addition to how the monitoring system and experts
involved in it will communicate with decision makers. Naturally, the system
of scientific monitoring should directly relate to either the Security
Council of the Russian Federation or the president's administration, or else
it should represent an independent body under the president. The latter is
fundamentally important, because, if the system's work is successful, the
main result will be the analysis of strategic risks, threats, and dangers
that arise as a consequence of inadequate decision making or of failures that
occur in their implementation, as well as the definition of the country's
corridor of opportunities. There is also another approach, which the United
States adopted after the terrorist acts of September 11, 2001—the integration
of various monitoring systems and analytical structures in order to
coordinate security efforts. • "Information
privatization " problem unsolved. At present, the corpus of
information necessary for forecasting and preventing calamities and disasters
is dispersed among many organizations that belong to different departments.
The majority of them prefer to sell information in a situation when the
administrative levers of influence have been forfeited in this sphere. Naturally,
this situation would damage the potential of monitoring, forecasting, and
preventing systems that are based on the complex analysis of all available
information. It would be logical to change the statement of the problem
itself. Information owners should be liable for the absence of efficient
forecasts in their own sphere of competence. Until they are really interested
in the creation of a forecasting system, no amount of effort on the behalf of
researchers will be able to build such a system. • The
absence of an adequate legal basis for prediction and prevention. A wide
range of problems in connection with forecast development, monitoring organization,
computer modeling, and systems analysis of dangerous phenomena and processes
has been left outside the reach of the law. In particular, the procedure for
the use of predicted results remains practically unregulated. The calamities
of recent years have helped to identify many gaps in current Russian
legislation, as well as derelictions in its execution. Thus,
there are many necessary conditions for the construction of a system of
scientific monitoring and structures that provide strategic risk analysis
that have not been fulfilled. There are two ways out of this situation. The
first is to press for the fulfillment of these conditions and only then
commence with serious scientific research, the development of hardware and
software complexes, and the organization of information flows. The second way
is to deal simultaneously with both the implementation of an extremely important
project for the country and the adaptation of the system environment for it.
The acuteness of the problems facing the country, the existing potential and
cooperation of researchers, and the domestic experience of implementing such
large projects make the second approach preferable. THE AGENT PROBLEM Let us pose the main question concerning
forecasting and monitoring organization: Who really needs it? Which agent is
interested in ensuring that dangerous processes and crises in modem Russia
will be observed, and is there one at all? An agent
exists so long as it has clear-cut goals. It has been observed that at one
time the firefighters of New York City were paid according to the number of
extinguished fires—and the city was constantly bum-ing. Then they were paid
to maintain peace and quiet, and the number of fires dropped significantly. As
soon as the city and not fire security per se became the focus, the
efficiency of fire security increased noticeably. In other words, the fact
that those who are responsible for parrying dangers desire to be in demand—to
be bothered, that is—as little as possible is of fundamental importance. The
monitoring system would inevitably be a slave to the aspirations of these
individuals. It would either provide information about how to avert dangers
or else report on the successes of liquidating the consequences of
emergencies. In the latter case, scientific developments are completely
unnecessary. Along with the definition of objectives, the agent of management
should obtain solutions to the problems posed. Unfortunately,
there are problems with both hypostases of power in Russia. Neither a
national strat egy nor
strategies in individual fields of activity have been formulated. We should
admit that an agent interested in monitoring crisis phenomena and processes
in the country is also absent. At present, a small portion of the decisions
adopted by administrative bodies is being carried out. Under such conditions,
it becomes difficult to speak of controllability, political will, or the
very possibility of foreseeing and preventing calamities and disasters. The
same situation repeats itself at other hierarchical levels. In particular,
the Russian Academy of Sciences is neither the agent that organizes research
into the topics under consideration nor the executor of forecasts and
recommendations. Today, the Academy is not able to organize work as a single
institution or to produce an expert statement on any project or forecast of
the development of processes. It is not part of the government machinery
either. Strategy building requires a
certain organization of society, as well as opportunities for
self-organization based on the awareness of its interests and readiness to
protect them. According to the data of sociological research, such processes
occur very slowly in modern Russia. The majority of citizens do not trust the
existing social institutions, the Russian president alone having become an
exception to this rule in recent years. An analog strategy could very likely
play an organizing role during the transitional period. Obviously, this
should be organized so that, by the time society is ready to formulate and
adopt a national strategy, it will have the necessary instruments to solve
key problems. One of these instruments is a national forecasting and monitoring
system. However, the situation today is undergoing
fundamental changes. On the one hand, the centralization of state power has
sharply increased. The majority of power resources are now in the hands of
the president, and decision making and implementation is his responsibility.
On the other hand, the indicative disasters of the past years, the
inevitability of which was predicted by scientists, as well as the wave of
terrorist acts and the infrastructure crisis, have led to the awareness of
national interests in the sphere of risks, calamities, and disasters [1, 4]. We
should not exclude the possibility that an agent will appear in the nearest
future that is aware of the necessity for a serious scientific approach to
forecasting calamities and disasters and that has enough power and resources
to prevent and liquidate them. Naturally,
the place of this agent in the power structure should have clear-cut
boundaries. In essence, we are talking about an institution that is in many
respects functionally similar to the general staff in the military command
system. First, such a structure, in the ideal case, does not seek its own
interests but supports the government machinery at higher levels. Second, we
are referring to the availability of large, authentic, and continuously
retrieved and updated information flows and their generalization and
interpretation. Third, we ascribe fundamental importance not to the forecast
itself, but to the recommendations and specific action plans based on it.
which are to be perfected long before emergencies occur. Fourth and finally,
the destruction wrought on the modem world by natural and anthropo-genic
disasters and social instabilities is already comparable with that arising
from combat actions. WHAT AN INTERDEPARTMENTAL PROGRAM
OF SCIENTIFIC RESEARCH SHOULD RESEMBLE Assume that the organizational,
financial, and other problems related to an interdepartmental program, which
has been proposed by the president, have been solved. Let us discuss what
this program should resemble. Its main goal is to provide for the
"observability" of a number of dangerous phenomena and processes,
as well as of crises, in modem Russia. The monitoring, forecasting, and
decision-support systems will make it possible to prevent and efficiently
parry a wide range of dangers and threats, which will improve the efficiency
of state management, provide an opportunity to save many thousands of lives,
and prevent huge material losses. World experience has shown that investing
means in prediction and prevention is 10-100 times less expensive overall
than the costs of mitigating or liquidating the consequences of calamities
and disasters after they have occurred. A special interdepartmental
scientific program would considerably differ from federal target programs and
scientific programs of the RAS Presidium or departments of the Russian
Academy of Sciences. It would also differ from the State Scientific and
Technical Program "Security," which had been carried out since
1985 (after the Chemobyl disaster), as well as from the Federal Target Risk
Reduction Program that is currently underway. Let us stress its fundamental
differences. Since the program considers
threats, dangers, and risks that may significantly or cardinally change the
path of Russia's development, it should be based on the concept of
strategic risks [3, 4]. It envisages the formulation of several
objectives concerning the feasibility of strategic decisions made at
the state level, as well as the assessment of possible damages incurred by
the failure to carry out these decisions or the consequences that they may
lead to. The Russian Academy of Sciences has experience in dealing with
problems of this scale. Systems analysis and computer modeling of large-scale
state military and space programs made it possible in their own day to solve
problems of paramount importance. And the "price of issues" that
this program addresses is immensely high. This dictates the involvement of
many leading specialists in its development, high levels of responsibility,
and the need for close interaction between program
developers and decision makers, The assessment of strategic risks, which is
the main target of the program, is integral to strategic forecast, ñ long-term reference points, the evaluation of
the corridor of the country's potentialities, and the key tasks facing
modem Russian society. As is known, the amount 1 of acceptable and unacceptable risks and
damages, as I well as the volume of
resources that can be assigned to parry the dangers, to prevent threats, or
to liquidate the ( consequences
of previous disasters, fundamentally
' depend on the scenario of the country's development : and on
its chosen long-term course. In contrast to the ' United States, Japan, and other
developed countries, Russia presently lacks large research programs dealing
with these problems, which was demonstrated by the preparation for the
national report at the summit on sus-tainable development (Johannesburg,
2002). The national concept of sustainable development is itself inseparable
from the strategic forecast and the assessment of a number of strategic
risks [2, 6]. The research programs of the majority of departments engage the
problems associated with the implementation of chosen policies, while the RAS
scientists should naturally analyze strategic problems. There are many dangers and threats
that cannot be assessed within a single spatiotemporal interval, according to
a single scientific discipline, or by a single ministry or scientific
approach. To improve the management of the state, complex system analysis of
present and potential threats to Russia is necessary. A list of priorities
according to the different dangers and a comparison of the scope of different
threats should be included in the program's results. This is why departmental
and narrowly discipline-specific analysis is unacceptable in the
implementation of the research program under review: an interdisciplinary
approach is necessary. This predetermines a much closer interaction
between participants in the program and makes it necessary to carry out
regular updates of analytical methods and forecasts, as well as adjustments
of forecasting systems. The work
of scholars should result not in a pile of reports and scientific papers but
in an effective system of scientific monitoring, which, first and
foremost, makes it possible to assess the most important crisis phenomena in
the country's development and the dynamics of many dangerous processes in
different spheres. This determines the principal significance of system
integration and the wide usage that must be made of new information
technologies, as well as general databases, knowledge, and information
flows, which has become possible owing to the RAS telecommunications
infrastructure. More than 20 years ago, Academician Yu.A. Izrael' proposed to
organize such a system with regard to a narrower range of dangers, but at
that time the situation was not so acute, there were no necessary
infrastructural preconditions, and they failed to overcome the organizational
obstacles. At present, the creation of the scientific monitoring system is
vitally important, and this is why the scientific and organizational
potential of the Russian Academy of Sciences should be utilized to the full. Under the
present conditions of crisis and transition. it is desirable to carry out
forecasting as soon as possible. It is only natural that the program should
be open. The possibility of flexible changes in structure and priorities
and the opportunity to involve new functionaries should be built into the
program. The traditional one-year cycle of developing research programs is
too long to adequately respond to many newly arising problems, although some
tasks do require extended system work that is not to be restricted to either
one year or five. According to practices adopted by many
international organizations and a number of countries, as well as to the
instructions by the Russian president, it is desirable to reflect how the
results of decisions made, as well as threats and dangers, affect the life of
an individual, separate social groups, nationalities, and regions. For this
reason, the human dimension of the problem under review—its humanitarian
component—acquires fundamental importance. The main focus of this component
should not be on certain social, production, or other technologies (such as
armed struggle and terrorism), but rather on the deep system factors that
predetermine the use of such technologies, i.e., on the interaction between
the following: civilization, the irregularity of the socioeconomic dynamics
of different areas, and the crises of the world, of Russia, and of other civilizations. One of
the elements influencing the success of any program is the use of the
obtained scientific results in managing the country and close cooperation
between scholars and decision makers, as well as interested ministries and
departments. The objects of analysis should be the mechanisms by which
results are used to ensure the sustainability of Russia's development and
strengthen the safety of the Russian population, the technosphere, and the
biosphere. Moreover, there should be a special focus on specific measures
that make it possible to parry arising dangers and threats at different
levels. If the
program is implemented successfully, a wide range of state bodies will be
able to use its results. One may expect that the assessment of risks will
become an essential element of management and expert organizations for
the government and the largest Russian enterprises will appear. In
particular, similar work that began in the United States more than 70 years
ago has led to the creation of the RAND Corporation, whose forecasts and
developments a number of the US statespersons have connected with the key
achievements of American strategy in recent decades (there are more than 5000
high-proficiency experts among RAND Corporation personnel). In addition to
the RAND Corporation, the Institute for Complexity in Santa Fe and the Center
for Nonlinear
Studies in Los Alamos were organized, the activities of which help analyze
the changes that are taking place in the world, estimate their possible
results, and find methods of influencing various processes in the desired
way. At the first stage, the objectives of our
specialized research program are more modest: to evaluate important threats,
risks, and crises, to find ways of parrying them, and to develop a body whose
constituents will be able to carry out permanent monitoring and forecasting
based on modem scientific achievements and who are included in the managing
profile of the country. The
assessment of risks and threats, the prospects for development, choice of
priorities, and search for new possibilities, all of which is envisaged by
the program under review, are among the most effective tools to involve
the existing potential of the Russian Academy of Sciences in the nearest
future. In this respect, the statement of Academician Zh.I. Alferov to
the effect that the lack of financial support for science is a necessary but
not sufficient condition to kill it seems quite relevant. The sufficient
condition is the nonuse of the results of science. For this reason, the task
of forecasting and analyzing risks could now become a supertask for the
Russian Academy of Sciences. The work conducted towards the performance of
this task will promote the revival of both the scientific medium and the
system of higher education, which is designed for preparing research
brainpower. In other words, we are dealing with a systemic problem. The
scenario is comparable to that of a state order for production, which can
stimulate a considerable part of the economy provided it is properly
organized. THE BASE FOR THE INTERDISCIPLINARY
STUDY OF DISASTERS, CATASTROPHES, AND CRISES Traditionally, a number of scientific disciplines, each of which has its own approach to and interpretation of the problem, studies disasters, catastrophes, and crises of different origins, in addition to their causes and consequences. Such "feudal division" has a baneful influence on the general situation. It is clear that developing and making general decisions are possible only on the basis of common scientific views about risks and uniform methods of describing information. Synergetics, which has proven to be an effective interdisciplinary approach [2, 5, 7], provides such an opportunity. It is based on the idea of the existence of universal regularities in the behavior of complex systems. Let us consider the universal properties of catastrophism. The language describing disasters and catastrophes. The statistical image of catastrophic behavior is the exponential laws of distribution. Their probability density is of the following type: Fig. l.The typical form of the probability density of
values distributed according to the degree (/). exponential (2). and normal {3)
laws. (a) The speed of decrease of the probability density for the
distribution tails: ib) the density of the degree, exponential, and normal
distributions in the log-log scale: and the linear form (,./) testifies to
the scale invariance of the systems described by degree distributions. where ~ 1. Such distributions are also called
distributions with heavy tails. The distribution tail accounts for the
probability of gigantic—extraordinary—events. It is possible not to take them
into account if the probability density decreases quickly enough, as is the
case with exponential and normal distribution (Fig. 1). However, in systems
described by degree distributions, catastrophic events are not rare enough
to ignore their possibility. In other words, in the case of degree
statistics, the idea of a hypothetical, that is, only theoretically possible,
accident is inadmissible. Any accident, even the largest one, which is not
taken into account by designers, is in fact admissible. It is noteworthy that in the case
of nonstationary processes, which are considered, in particular, by the The dependence of the damage
on the number of the event in the catalogue [10] in chronological order is
presented starting from 1975: the angular coefficients of degree
approximations (thin straight lines) are 1.10 for the maximum damage and
1.47 for the accumulated one. theory of
examining operations while analyzing war conflicts, the probabilistic images
of catastrophes can be different [8]. In these problems, the principal element
is decision making, which is analyzed as a rule within the framework of the
theory of artificial intelligence [9]. In practice, when analyzing catastrophic
events, it is more convenient to consider not the probability density ii(x)
but the dependence between the size (value) of the event x and its
rank r— the number of the event on the list normalized according to
the diminution x. For degree distributions, the "rank-size"
dependence also has a degree form of Rank-size
degree dependences are characteristic of anthropogenic and natural disasters,
AIDS statistics, and the expansion of computer viruses. The danger of a phenomenon subordinate to a
degree distribution is determined by its index: the less is a, the more
dangerous it is. One can draw a conditional differentiation between
"accidents" characterized by the > 1 value and "catastrophes"
for which < 1. Even the impact of the largest
"accidents" on the total damage resulting from them is negligible
because it is formed by a great number of moderate events. However, in the case
of "catastrophes," the total damage from a number of events is
comparable to that of the largest one among them. Thus, the law of large
numbers does not work for the statistics of catastrophes ( <
1). In other words, the sample mean increases without limit as the volume of
sampling becomes greater, with no definite finite bound. Respectively, the
accumulated damage increases quicker with the increase of the sampling
volume, that is, it accelerates (Fig. 2). The
maximum damage also increases without limit in the course of time, which can
be erroneously interpreted as proof of the nonstationary nature of the pro cess.
It produces the illusion that the situation is constantly getting worse,
whereas it is in fact stably bad. In addition, the criteria that could help
to single out situations that are constantly getting worse, as the case is
now in our native country, become ambiguous. The study of phenomena for which
degree laws of distributing probabilities are characteristic requires
special non-traditional methods of statistical analysis [I]. Mechanisms
of the rise and development of catastrophic events. Systems inclined to catastrophes
are complex in the sense that they cannot be reduced to the simple sum
of their components. Otherwise, events would have arisen as a sum of a great
number of independent addends, which, according to the central limit
theorem, is normally distributed. Another consequence of degree statistics is
the scaling invariance of the systems under review, which means that
the processes taking place in them are organized in the same way on different
scales. It thus becomes impossible to factorize their behavior into a number
of independent subprocesses; an integrated description is necessary. Integrated
scale-invariant properties are observed in systems in the so-called critical
state, such as phase transitions of the second order. This can arise either
owing to a fine artificial tuning or during the self-organization of some
nonlinear systems that are far from equilibrium. The mechanism of
self-organization into the critical state is universal, which determines the
extraordinarily high diversity of self-organised critical phenomena in
nature [II]. The basic
model of the theory of self-organized crit-icality is a heap of sand. If the z;
mean slope of the sand surface is not large, the sand is static (Fig. 3). If
the slope is larger than the critical value , a
spontaneous flow of sand J occurs along the surface (see the insert on
Fig. 3). The two states correspond to non-catastrophic behavior. In the
subcritical state , nothing
happens, while, in the supercritical state z >, nothing
unexpected takes place. Large unexpected events, such as catastrophes, are
possible only in the critical point; z =, where
there is no spontaneous flow yet, but in which any fluctuation can produce
arbitrary large avalanching. A sand
heap can be put into a critical state either through manual tuning of the
governing parameter to mean z = or as a
result of self-organization upon establishing the parameter of the order to
mean J = +0. To make such self-organization possible, let us consider
the dynamics step by step, add grains of sand on the top of the heap one by
one, and wait for the end of the relaxation process. The sand flow obviously
has the minimal possible value: on average, one grain for one step of the
examination. If the
slope of the surface is small, the avalanche produced by adding a grain will
most likely not reach the edge of the heap, and the slope will begin to
increase. With a very large slope, the state of the heap becomes metastable,
that is, it will respond with a glo- Fig. 3. A sand comer. The
state of the sand is determined by the angle of slope of the surface z.
the change in which leads to a permanent phase transition (the dependence of
the order parameter on the control parameter is presented on the insert) from
the static state to the slate of sand's continuous current. Fig. 4. The cellular
automation for a sand heap. (a) The state of the system
before the crumbling avalanche: (b) the state of the system
after the crumbling avalanche; the avalanche is initiated by adding one grain
to an arbitrarily chosen cell of the top layer: the black shows the areas of
the avalanche and cells at its border that preserved stability after having
received a grain of sand. bal event to any perturbation, which will
result in the abandonment of the system by a great amount of sand, and the
slope will become smaller. Thus, there is a negative feedback that forces
the slope to assume the meaning of z = which
allows the perturbation to spread arbitrarily far. This means that,
notwithstanding the local nature of the interaction between the grains of
sand, the sand heap behaves like a single whole. The self-organized critical behavior of the system under review may be described by means of a simple cellular automation [12]. Let us compare the heap to a two-dimensional hexagonal lattice, the cells of which occupy whole numbers, characterizing the local slope of the surface (Fig. 4). If a number is more than unity, Fig. 5. A fragment of a
hierarchical system. Each element of the ith level consists of three
elements of (i – 1)th level: the system elements can be correct or
faulty (grey color), the state of each of them being determined by the state
of the constituent elements of the previous level, as well as by its own
receptivity to flaws. the cell is declared unstable and crumbles,
which manifests itself by a decrease in the number in it by 2 and a
simultaneous increase in the values of the two cells adjacent to it from
below by 1 (see Fig. 4). The horizontal layers of the lattice conditionally
correspond to the lines of the surface level, which is why the crumbling of
the cell may be regarded as if two grains are sliding down the slope. A modeling step consists of perturbation and
relaxation. The perturbation of the stable state is carried out by
increasing by unity the value of an arbitrarily chosen cell of the upper
layer, which corresponds to adding one grain to the top of the heap. If the
grain loses stability as a result of the perturbation, it crumbles and the
process of relaxation begins. The crumbling of the cell leads to an increase
in the slope of subjacent cells, which, in its turn, can disturb their
stability and so on, by the chain reaction principle. Thus, the loss of
stability by one cell can result in a crumbling avalanche, which will go on
until all the cells once again acquire stability. After that, the period of
relaxation is over, and the next modeling step begins. The
bottom edge of the lattice of the cellular automation is open; therefore,
the crumbling of a cell from the bottom layer forces two grains to leave the
system. This ensures the existence of the stationary state and the
possibility of self-organization. The
characteristic of the crumbling avalanche is its size, that is, the number of
cells in which crumbling has taken place. Avalanches are distributed by size
exponentially with an index of 1/3 [12], which is confirmed by computer
calculations. Scale-invariant distribution means that a system is inclined to
catastrophes. Its response to an elementary action does not have a specific
size of its own; therefore, gigantic events are possible in it in the
absence of clear reasons. Although one is able to indicate the grain that has
provoked this or that avalanche, the origin of catastrophes is evidently
connected not with grains but with critical properties of the system, in
which small causes can lead to large effects. The organization of systems
inclined to catastrophes. Many scale-invariant systems have hierarchical structures.
For example, the lithosphere of the earth may be represented as a system of
blocks divided by breaks. Each of these blocks is divided into smaller ones.
the latter, into smaller ones still, and so on. Geo-physicists have singled
out more than 30 hierarchical levels in the earth's crust—from tectonic
plates of thousands of kilometers in length to grains of rock of a millimeter
in size. Large earthquakes are usually accompanied by many recurrent
shocks—aftershocks that in a cascade redistribute the stress down by the
hierarchy of the breaks. The preparation of an earthquake occurs through a
reverse cascade in which stress is transmitted, moving upward from the bottom
layers of the hierarchy to the upper ones. A natural example of a
hierarchical anthropogenic system is the system of administrative or military
management. The success of solving problems at any given level of management
is determined by the effectiveness of the subjacent levels. The electorate is
also a hierarchical system. It too is divided into several groups with their
own interests. Each of them consists of smaller subgroups and so on—down to
the individual voter. We are able to observe the behavior of hierarchical
systems only at the upper levels of the hierarchy (earthquakes, performing
orders, and voting results). However, the reasons for these events lie at the
bottom levels, and it is important to realize how the levels interact. Let us consider the hierarchical
system a fragment of which is presented in Fig. 5 [13, 14]. The system is
divided into levels that may be interpreted as the degrees of description
detail (the lower the level, the more details are accessible). Each element
of the level i' > 0 consists of three elements of the previous
level i - 1. The elements of the system can be correct or faulty. Assume
the state of elements at the bottom level i = 0 to be fully arbitrary
and the concentration of faulty elements to be p0. The elements of subjacent levels combined in
threes in an element of the next level transmit their state to it according
to the degree of its receptivity to flaws k. The degree of receptivity
of an element means the minimal number of faulty components sufficient to
make it faulty too. Let us consider a spontaneous rise and correction of
flaws to be impossible. In this case, the system may consist of elements of
all three types with k = 1, 2, and 3; that is, elements become faulty,
respectively, if only one component is faulty, if at least two components
are faulty, and if all the three components are faulty. Let us denote the
share of elements of the k type through sk, with . The main question concerning the system described is the following: what is its state at the upper levels (up to the last one containing only one element, which is the system itself) under the specified concentrations of flaws at the lowest level and shares of elements of different receptivity? A change
in the concentration of faulty elements p with the ascent by one level
is presented through the following mapping:
It always
has two trivial fixed points p = 0 and p = 1, which correspond
to flawless and fully faulty states. and a critical point which
must meet the condition 0 < pc < 1 to have physical
meaning. The
relative positions and stability of the fixed points depend on the -^ values.
On the constitution diagram of a hierarchical system, the parametric space
is divided into four areas (phases): in two of them, one trivial fixed point
is stable and the other one is unstable, while the critical point is absent altogether;
in the other two, the critical point does exist (Fig. 6). Let us consider in
more detail some examples of the system's behavior that correspond to each
area of the parametric space. 1. With s1
= 1 and s3 = s2 = 0 a single faulty part
of it is sufficient for the appearance of a faulty element. Respectively, the
only fixed stable point of the mapping p = 1 and any nonzero
concentration of flaws at the bottom level lead to the faultiness of the
entire system. 2. With s3
= 1 and .s2 = s1 = 0, all its parts
must be faulty for a faulty element to appear. Respectively, the only fixed
stable point of the mapping p = 0 and any nonzero concentration of
correct elements at the bottom level guarantee the correctness of the entire
system. 3. With s2
= 1 and s1 = s3 = 0, a faulty element
appears if more than half of its parts are faulty. Both extreme values p =
0 and p = 1 are stable, and the state of the system as a whole is
determined by the concentration of flaws at the bottom level. If p0
< pc. = 1/2, the system will be correct (as variant 2),
and if p0>pc faulty (as variant 1). Only in
the case of p0 = pc;. will the critical
concentration of flaws be preserved from level to level. 4. If s2
= 0 and s1 =s2
= 1/2, the system is a mixture of elements of two different
types in equal shares: some behave according to variant 1, strengthening the
flaws, while the others, according to variant 2, suppress them. Their
critical point is stable, and the probability of the whole system's
faultiness does not depend on the concentration of flaws at the bottom
level, because ; 1. The properties of systems in areas 1 and 2 of the constitution diagram, where there is no critical point, are quite predictable and, consequently, these systems are fraught with no danger. However, when it is in the critical state, a system may be both correct and faulty with nonzero probability. In area 3, it happens only with Fig. 6. The constitution
diagram for the hierarchical system in the s1 –s/ //projection. With s1
< 1/3( areas 2 and 3). a stable flawless state is observed: with s3
< 1/3 (areas 3 and 1). a stable fully faulty state: area 4 corresponds to
the self-organized criticality; black points mark the systems
under review in the article with different receptivity of elements to flaws. a special value p0 =pc,,
while in area 4 it may occur with any . The first case corresponds to the
ordinary critical behavior, when to ensure the appearance of certain of the
system's properties a special tuning is necessary, while the second case
corresponds to self-organized criticality arising owing to the action of
negative feedback, which diminishes the deviation pi from pc,
during the ascent from level to level. The examined model demonstrates the basic
mechanism of the rise of scale-invariant properties and catastrophic
behavior in hierarchical systems. It is also noteworthy that, because in the
critical state the concentration of faulty elements is the same at all
hierarchical levels, their distribution by size has a degree form with the
index. FORECASTING CATASTROPHIC EVENTS For many systems inclined to
catastrophic behavior, it is impossible at the present stage of scientific
development to present detailed mathematical models with forecasting
potentialities; therefore, one has to forecast in another way. Instead of a
detailed description of the background of a catastrophe, we use general
properties of nonlinear dynamic systems. Another approach is to study expert
opinions and decision-making mechanisms. The
starting point in the prediction of catastrophes is the fact that
catastrophic dynamics is not chaotic. Owing to their integrity, critical
systems have a long "memory" of events that have taken place and
"feel" them at long distances from the locations at which they
occurred. This is reflected mathematically in the degree decrease of temporal
and spatial correlations, which. for noncritical systems, usually correspond
to an exponential distribution, meaning that a system quickly forgets its
history and displays a weak interdependence of pans. Let us again take a sand heap as
an example. The local slope of its surface determines its current state. Steep
areas are more inclined to avalanche formation. while flat ones are less
inclined. For a gigantic avalanche to appear, there must be a certain
surplus of sand in the system. However, it cannot concentrate on this or that
separate area without destroying its stability long before the rise of a
gigantic avalanche. Consequently, a considerable part of the heap must have a
large local slope. The corresponding amount of sand can be delivered from
the top to subjacent areas only by means of avalanches, among which there
must be sufficiently large ones. Thus, a gigantic event must have portents— previous
events of a smaller size. This conclusion is also correct for other
self-organized critical systems, although the individual structure of
portents may depend on the properties of the event itself. Reasoning from the universal
nature of catastrophic processes, one may expect that algorithms for identifying
portents and forecasting gigantic events that have been constructed, and
whose effectiveness has been confirmed, on the basis of one material
(seismology), may successfully be applied to others (economy, sociology, and
crime dynamics). Seismology. Instability at the upper levels
of hierarchical systems arises owing to instability at lower ones. In
particular, the reverse cascade in the redistribution of stresses in the
earth's crust as it prepares a large earthquake shows itself in the form of
anomalous seismic activity on smaller scales of energy and size. This allows
for a forecast based on observations of activation and other deviations from
the trend in the behavior of a system, clustering instability acts,
miscorrelations, and so on. At the first step of developing
the algorithm for the forecast of catastrophic events, one creates methods of
compressing huge amounts of information, presented by monitoring, into a
small set of functionals—time-dependant values that, when aggregated,
characterize the state of the system. Generally speaking, the possibility of
introducing functionals that are fit for forecasting the behavior of a
complex system is not obvious, the specific form they should take being even
less obvious. At the present stage, it is of key importance to ensure the
interaction between forecast specialists and those who have some
understanding of the system dynamics, at least qualitatively, and who are
able to say how the interaction of hierarchical levels in it may be taken
into account. The next step is to determine the functionals
that yield the most information, as well as to construct an alarm algorithm
based on the knowledge of the values of these functionals. The
"teaching" of the algorithm refers to the selection of parameters
used for calculating the functionals; the establishment of thresholds for
them, which, when exceeded, may indicate that the system is entering into a
dangerous state; and the formulation of alarm rules. This
"teaching" is aimed at optimizing the algorithm's capacity for
retrospective forecast (the forecast of catastrophes that have already happened)
based on information about prior activities. Finally, in the last step testing occurs in
real conditions in order to determine the effectiveness of the algorithm. At
this stage, an updating of its parameters is inadmissible. Let us illustrate the described scheme
through the example of an algorithm for the medium-term forecast of large
earthquakes, which is known in the scientific literature as M8 [15-18]. Areas
are considered the sizes of which are determined b\ the magnitude threshold
of the forecast events (to modify the algorithm for forecasting earthquakes
with a magnitude of more than 6.5, 7.0, 7.5, and 8.0 on the Richter scale,
circles with the diameter of 384, 560, 854, and 1333 km are used,
respectively). These areas are situated on seismic belts (globally stretched
areas in which earthquake sites are concentrated) and cover the zone of the
development of the supposed large earthquake. From the stream of events that
have occurred in this zone, the main shocks are singled out—earthquakes that
are not aftershocks. Their sequence is normalized by the magnitude threshold,
which is established in such a way that the average annual number of its main
shocks constitutes a certain quantity (10 or 20: the corresponding
functionals are indicated by indexes 1 and 2). The following functionals are determined on
the basis of the data on the main shocks: • N1 and N2 are the numbers of main shocks in
six years; • LI and L2 are deviations of the N1 and N2
values from their durable trends; • Zl and Z2 are ratios of the average site
size to the average distance between sites; • Â is the maximal number of aftershocks following the main shocks
calculated for the period of one year. Each of the introduced functionals is calculated in a three-year window with a step of six months. As a result, the stream of earthquakes is roughly described by the speed (N), acceleration (L), linear concentration of events (Z), and their grouping capability (B). The value of a functional is considered to be anomalously large if it exceeds the Q percent of its observed values (Q = 75% for  and 90% for the remaining functionals). When six out of seven functionals including  become anomalously large for two neighboring temporal intervals, one should alarm, that is, declare a period of Fig. 7. The example of
the work of the M8 and MSc algorithms for the medium-term forecast of
earthquakes. (a) The forecast map. epicenters, and the first aftershocks of
the earthquake on Shikotan island on October 4. 1994: the alarm area.
diagnosed by the M8 algorithm, is shown by light-grey circles, and the
dark-grey rectangle is the alarm area specified by the MSc algorithm: (b) the
temporal and spatial diagram of the seismic activity in the southern circle
where they alarmed, the ordinate axis corresponding to the projection to the
axis of the seismic belt: (c) the behavior of the functionals. their
anomalously large values being marked by black points and the dark-grey area
corresponding to the period of the increased probability of the earthquake
that started after the tree-year period (light-grey area) when the condition
of alarm had been met. increased
probability (PIP), which is five years long Fig. 7). Areas in which alarm takes place
according to the ^18 algorithm exceed by one or two orders in size the site
of the expected earthquake, which is inadmissible n practice. For this
reason, the corresponding zone for he periods of increased probability is
additionally checked by means of the "Mendocino scenario" (MSc)
algorithm [18, 19], which specifies the place of the future large earthquake
and decreases the zone of alarm ) to 20 times over (see Fig. 7). This
algorithm is based on the search for associated areas (clusters) of relative
seismic calm, which testify to an accumulation of .tress, in the temporal and
spatial zone of alarm. In order to construct algorithms
for medium-term earthquake forecast, different statistical tests and
proce-dures were used, allowing for the estimation of forecast authenticity,
quality, and stability. Since 1985, algorithms M8 and MSc have been checked
by a "forward" "forecast, and since 1992 official testing has
been carried out. The results of the "forward" monitoring of large
events are presented in the table [18]. These and
other algorithms for earthquake forecasting, methods of determining risks
and damages, and forecast decision-making strategies have been developed in
recent decades. One of the main reasons behind the active development of
ideas and principles in connection with forecasting catastrophes was the
well organized long-term local and world services for collecting routine
data, which make the whole world a testing area for investigations. The
long-term series of observations obtained substantially reduce the necessity
of using statistical-analysis and hypothesis-checking methods. Economy. In order to forecast recessions, US scientists
employ an approach that is in many respects similar to the one described
above [20]. In a recent case, they examined the period from 1959 through
1997, which contained six recessions (large abatements) in the US economy. Eight
monthly temporal series were analyzed: the income of the population without
social transfers; the number of jobs in private companies; the number of
requests for unemployment benefits; the costs of reserves at storehouses in
production and commerce; the rate of the treasury bill for three months; and the
difference between the government bond rate over ten years and the interest
rate of the US Federal Reserve. Six functionals were derived from these
series, which made it possible to predict the 2001 recession (Fig. 8). Since 1995, the Russian stock index AK&M
and the American Dow Jones Industrial index have been chosen for forecasting
socioeconomic crises in Russia [4]. The series of the original indexes are
nonstationary, but there may be stable long-term associations between them,
the violation of which testifies to an imbalance in the economy. It was
discovered that, to forecast Russian crisis events, it is important to
account for their miscor- Fig. 8. The forecast of
recessions in the United States from 1959 through 2003. The black rectangles are
periods of recessions, the grey ones are periods of forecast alarms, the
total length of the alarm being 38 months, that is. 9.3% of the period under
examination; the last alarm by this algorithm was in May 2001. while the
recession was declared in the United States onlv in November 2001, its
beginning being assigned to April 2001. relation:
when the American index goes up on average, the Russian one goes down. A
functional was constructed on this basis, presenting the cumulative sum of
relative AK&M increments during the periods when it drops; and the values
of the indices under review, leveled in the 70-day sliding window, are
negatively correlated. Since 1995, when Russian stock indices first
appeared, the following crises have taken place on the financial market: the
bank liquidity crisis (August 25, 1995); the socioeconomic crisis of
confidence in the government (June 3 through July 16, 1996); the stock market
collapse (October 24, 1997); and the crisis of August 17, 1998, which was
simultaneously a foreign exchange crisis, a bank crisis, an investment
crisis, and a crisis of foreign debt. The proposed functional retrospectively
predicts the crises of 1995, 1996, and 1998 (Fig. 9). The omission of the
1997 crisis is due to the fact that it was caused by an event external to the
Russian economy—the collapse of the Asian stock markets. The
developed methodology of forecasting allows for an interesting observation:
two out of the three predicted crises take place after the forecasting
functional has returned to the zero value. This probably reflects the
specificity of Russian financial markets. Lacking a direct connection with
the national economy, they are fictitious to a certain extent and aim not at
real activities but at its simulation. Therefore, having signaled a structural
crisis, they quickly plot to the changed situation. No doubt, this
circumstance should be taken into account when developing forecasting methods
in the economy. Social
systems.
Forecasting in socioeconomic spheres, as in seismology, is based on
functionals that are sensitive to the precritical states of the system followed
by a catastrophic event. Materials of pilot studies show that it is possible
to develop a new approach in socioeconomics—"the sociology of fast
response." This may be important in order to improve the stability of
the activity of large cities and of the country as a whole, as well as to
prevent social instabilities. Let us
consider the problem of forecasting the dynamics of criminality as an
example. This example is based on the records of accidents in Yaroslavl
oblast from February 3, 1993, through June 25, 2001. The events were divided
into five groups: (1) car thefts, road traffic accidents, sudden deaths, and
so on; (2) hooliganism, theft, and so on; (3) bodily damage, blackmailing,
fraud, and so on; (4) discovery of bodies, suicide, and robbery; (5) murder,
gross bodily injury, rape, robbery-related assault, and disappearances. The
object of the forecast was a jump in felonies (group 5), which is determined
as the above-threshold excess of their number over one week compared to the
average for five weeks (including the current week and the previous four). In the
forecast, we used an analogy with the development of large earthquakes
through a reverse cascade, that is, the migration of activity from the bottom
levels to the upper ones. The dynamics of such processes is reflected in the
change of the slope angle in the log-log plot of the distribution of events
by size. In order to construct this plot, the number of the group was used as
a magnitude—the logarithmic measure of gravity. This implies that different
categories of crime correspond to different levels on an "asocial
hierarchy." The remnants—the difference between the current plot's
slope angle and its average time value (Fig. 10)—were taken as the forecast
functional. With the thresholds specified, 30 out of 34
objects are forecast (88.2%). The sum of all alarms is 118 weeks out of 392
weeks of monitoring (30.1%). These 118 weeks include one false alarm, the
average alarm time being about a month. Thus, the described forecast methods
may be applied even in the sphere where not only models describing the
dynamics of events but even rough ideas about the interaction of different
processes are absent. Let us summarize. Past failures in the
attempt to organize many programs for the forecasting and monitoring of
dangerous phenomena and processes have provided negative examples of how to
create a program. Such programs cannot proceed with the object of generating
research papers, maps, or economic effects. These are only tools for
obtaining the main result, that is, a system that ensures the forecasting and
monitoring of dangerous and crisis phenomena. This system requires an
interdisciplinary approach and cooperation amongst investigators,
administrators, and risk managers. It is of fundamental importance that the
monitoring system becomes a By order of the Russian Security Council and the
Russian State Council Presidium, the Russian Academy of Sciences has
prepared the Interdepartmental complex program "Basic, Exploratory, and
Applied Research Ensuring the Protection of Hazardous Facilities and
Population." This program has been sent for coordination to a number of
interested departments. Based on the unprecedented problems that the program
poses, it is hoped that it will become the "program of a new type"
discussed in the present article.part of the management of the country. Fig. 10. The forecast of the jumps in grave offences in Yaroslavl
for 1993- 2001. The week dynamics of grave
offences; (b) the difference between the current angle of the slope of the
criminality graph and its average value by time: the horizontal line corresponds
to the alarm threshold, the vertical lines, to the jumps in the number of
grave offences, and the black rectangles on the abscissa axis. to the alarm moments. By order of the Russian Security
Council and the Russia State Council Presidium, the Russian Academy of
Sciences has prepared the Interdepartmental complex program "Basic,
Exploratory, and Applied Research Ensuring the Protection of Hazardous
Facilities and Population". This program has been sent for coordination
to a number of interested departments. Based on the unprecedented problems
that the program poses, it is hoped that it will become the "program of
a new type" discussed in the present article. It is of fundamental importance that the development of nonlinear dynamics and synergetics has provided the scientific base and principally new possibilities for analyzing and forecasting disasters and catastrophes. We have discussed the general approach and have provided a series of examples in the most complicated cases in which there are no models to make a forecast; there are no commonly approved macroindicators that could help forecast dangerous phenomena; and there
are no (except for earthquakes) established structures to monitor and
organize the corresponding informational streams. Nevertheless, owing to the
universal nature of the properties of many locations where catastrophes
occur and the developed methods of monitoring and forecasting, a forecast is
possible. Moreover, considerable work has been carried out in order to
systemize and describe many risks and threats [21], which makes it possible
to advance to a higher level of analysis and forecast. In our opinion, there exists at present a
serious research base for the fulfillment of the order of the Russian
president concerning the forecast and prevention of catastrophes and
instabilities in the natural, anthropogenic, and social spheres. The work
that lies ahead will obviously make it possible to utilize most fully and
develop the potential of the Russian Academy of Sciences under the present
historic conditions. REFERENCES 1. V. A. Vladimirov,
Yu. L. Vorob'ev, G. G. Malinetskii, et al.. Rise Control. Risk, Sustainable
Development, Synergetics (Nauka, Moscow, 2000) [in Russian]; http:/www.keldysh.ru/papers/2003/source/book/gma-lin/risk.htm. 2. S. P. Kapitsa, S. P.
Kurdyumov, and G. G. Malinetskii, Synergetics and Future Prediction,
3rd ed. (Editorial URSS, Moscow, 2003) [in Russian]; http://iph.ras.ru/~mifs/kkm/Vved.htm. 3. G. G. Malinetskii,
"Scenarios, Strategic Risks, Information Technologies," Inform. Tekhnol.
Vychisl. Sist., No. 4 (2002); http://www.keldysh.ru/e-biblio/ij/s r/jst.htm. 4. G. G. Malinetskii,
V. I. Osipov, D. S. L'vov, et al., "Crises of Modem Russia:
Scientific Monitoring," Vestn. Ross.
Akad. Nauk, No.
7 (2003); http://www.keldysh.ru/departments/dpt
17/k.html. 5. G. G. Malinetskii
and S. P. Kurdyumov, "Nonlinear Dynamics and Prediction Problems,"
Vestn. Ross. Akad. Nauk, No. 3 (2001) [Herald Russ. Acad. Sci. 71,94-113 (2001)]; http://www.keldysh.ru/departments/dpt
17/neldim.htm. 6. S. P. Kurdyumov, G. G. Malinetskii, I. G. Medvedev, and N. A.
Mitin, "Nonlinear Dynamics and Problems of Forecasting," Bezop. Evrazii,
No. 2 (2001). 7. "A New Outlook on Synergetics: Glance into the Third
Millennium," in Informatics: Unlimited Possibilities and Possible
Limitations, Ed. by G. G. Malinetskii and S. P. Kurdyumov (Nauka, Moscow,
2002) [in Russian]; http://www.keldysh.ru/book/ns.html. 8. N. N. Moiseev, Mathematical Tasks of Systemic Analysis (Nauka,
Moscow, 1981) [in Russian]. 9. 0. I. Larichev, Decision Theory and Methods (Logos, Moscow,
2003) [in Russian]. 10. EM-DAT: The OFDA/SRED International Disaster Database (as of
Oct. 24, 2003) (Universite Catholique de Louvain-Brussels-Belgium);
http://www.cred.be/emdat 11. P. Bak, How Nature Works: The Science of Self-Organized
Criticality (Springer, New York, 1996). 12. D. Dhar and R. Ramaswamy, "Exactly Solved Model of Self-Organized
Critical Phenomena," Phys. Rev. Lett. 63 (16) (1989). 13. M. G. Shnirman and E. M. Blanter, "Mixed Hierarchical Model
of Seismicity: Scaling and Prediction," Phys. Earth Planet. Inter.
Ill, 295-303 (1999). 14. M. G. Shnirman and E. M. Blanter, "Scale Invariance and
Invariant Scaling in a Mixed Hierarchical System," Phys.Rev.E60 (5)
(1999). 15. V. I. Keilis-Borok and V. G. Kosobokov, "Periods of Increased
Probability of Emergence of the Strongest Global Earthquakes. Mathematical
Methods in Seismology and Geodynamics," Vychisl. Seismol., No. 19
(1986). 16. Long-Term Earthquake Prediction, Ed. by M. A. Sa-dovskii
(Nauka, Moscow, 1986) [in Russian]. 17. "Intermediate-Term Earthquake Prediction: Models, Algorithms,
Worldwide Tests," Ed. by V. I. Keilis-Borok, Phys. Earth Planet. Inter.
61 (1-2) (1990) (Spec. Iss.). 18. Nonlinear Dynamics of the Lithosphere and Earthquake
Prediction, Ed. by V. I. Keilis-Borok and A. A. Soloviev (Springer,
Heidelberg, 2003). 19. V. G. Kossobokov, V. I. Keilis-Borok, and S. W. Smith,
"Location of Intermediate-Term Earthquake Prediction," J. Geophys. Res.
95B (12) (1990). 20. V. Keilis-Borok, J. H. Stock, A. Soloviev, and P. Mikha-lev,
"Pre-Recession Pattern of Six Economic Indicators in the U.S.A," J.
Forecasting 19, 65-80 (2000). 21. Natural Hazards in Russia, Ed. by V. I. Osipov and S. K.
Shoigu (KRUK, Moscow, 2003), Vols. 1-6 [in Russian]. |