ORI GIN AL PA PER

Estimation of seismic hazard and risks for the Himalayasand surrounding regions based on Unified Scaling Lawfor Earthquakes

Imtiyaz A. Parvez • Anastasia Nekrasova • Vladimir Kossobokov

Received: 4 April 2013 / Accepted: 26 October 2013 / Published online: 9 November 2013� Springer Science+Business Media Dordrecht 2013

Abstract To estimate seismic hazard, the basic law of seismicity, the Gutenberg–Richter

recurrence relation, is applied in a modified form involving a spatial term:

log N M; Lð Þ ¼ A� B M � 5ð Þ þ C log L, where N(M,L) is the expected annual number of

earthquakes of a certain magnitude M within an area of linear size L. The parameters A, B,

and C of this Unified Scaling Law for Earthquakes (USLE) in the Himalayas and sur-

rounding regions have been studied on the basis of a variable space and time-scale

approach. The observed temporal variability of the A, B, and C coefficients indicates

significant changes of seismic activity at the time scales of a few decades. At global scale,

the value of A ranges mainly between -1.0 and 0.5, which determines the average rate of

earthquakes that accordingly differs by a factor of 30 or more. The value of B concentrates

about 0.9 ranging from under 0.6 to above 1.1, while the fractal dimension of the local

seismic prone setting, C, changes from 0.5 to 1.4 and larger. For Himalayan region, the

values of A, B, and C have been estimated mainly ranging from -1.6 to -1.0, from 0.8 to

1.3, and from 1.0 to 1.4, respectively. We have used the deterministic approach to map the

local value of the expected peak ground acceleration (PGA) from the USLE estimated

maximum magnitude or, if reliable estimation was not possible, from the observed max-

imum magnitude during 1900–2012. In result, the seismic hazard map of the Himalayas

with spatially distributed PGA was prepared. Further, an attempt is made to generate a

series of the earthquake risk maps of the region based on the population density exposed to

the seismic hazard.

I. A. Parvez (&)CSIR Centre for Mathematical Modelling and Computer Simulation, Bangalore, Indiae-mail: parvez@cmmacs.ernet.in

A. Nekrasova � V. KossobokovInstitute of Earthquake Prediction Theory and Mathematical Geophysics, RAS, Moscow, RussianFederation

V. KossobokovInstitut de Physique du Globe de Paris, Paris, France

123

Nat Hazards (2014) 71:549–562DOI 10.1007/s11069-013-0926-1

Keywords Himalayas � Seismic hazard � Unified Scaling Law � Seismic

risk

1 Introduction

Earthquakes constitute the most feared of natural hazards. These occur with no certain

warning and can result in great destruction and loss of life. It is understood that an

earthquake occurs when a propagating rupture suddenly releases the accumulated tectonic

stress in a system of the lithosphere blocks and faults. But it is not known with certainty,

when the rupture will occur and which of the many faults in a region will be involved. The

Himalayas and surroundings have experienced several great earthquakes of magnitude 8

and above in the past, inflicting heavy casualties and economic damage. One way to

mitigate the destructive impact of earthquakes is to conduct a seismic hazard analysis and

take remedial measures. Of course, the errors in the analysis mislead to unexpected eco-

nomic and human losses (Wyss et al. 2012).

Earthquakes are clearly very complex natural phenomena, but in this complexity, there

is also an order. The evident heterogeneity of patterns of seismicity distribution and

dynamics (Mandelbrot 1982; Keilis-Borok1990; Turcotte 1997, 1999) is apparently

scalable according to the generalized Gutenberg–Richter recurrence law that accounts for

the fractal nature of faulting (Kossobokov and Mazhkenov 1988). When epicenters are

uniformly distributed over the surface, their number will be proportional to the area,

whereas when they are distributed along a narrow strip, e.g., along a linear fault zone, the

number will be proportional to the length of a segment. There are other ‘‘fractal’’ pos-

sibilities that cannot be excluded a priori. Seismic activity is by no means a uniform

process. Therefore, the question of spatial and temporal scaling arises necessarily in

seismic hazard and risk assessment, as well as in studies of earthquake sequences prior to

the largest earthquakes aimed at earthquake prediction, which require a transfer of

estimates and/or criteria from one area to another. When the available data permit the

evaluation of the Unified Scaling Law for Earthquakes (USLE) (Kossobokov and Nek-

rasova 2003, 2005, 2007), the estimated coefficients A, B, and C could give a reliable

answer to the question.

Notwithstanding the above-mentioned difficulties, various investigators have

addressed the problem of assessing the seismic hazards in the Himalayas and sur-

roundings (e.g., Auden 1959; Gaur and Chouhan 1968; Kaila and Rao 1979; Khattri

et al. 1984; Parvez and Ram 1997, 1999; Bhatia et al. 1999; Parvez et al. 2003; Parvez

2007; Lyubushin and Parvez 2010). In this paper, we accept the hypothesis that the

seismic process is self-similar, at least locally, and estimates the coefficients A, B, and

C using the Scaling Coefficients Estimation (SCE) algorithm (Kossobokov and Nek-

rasova 2005, 2007; Nekrasova 2008; Nekrasova et al. 2011). We analyze the earthquake

catalog from USGS/NEIC and provide the maps of the A, B, and C coefficients for the

different scales and discuss their likely relationship with the observed continental

deformation and related structure of the Earth. We also discuss the usage of these

parameters in seismic hazard assessment, as well as in describing the seismicity flow

before major events in the region. Furthermore, based on the USLE-derived hazard map

and data on population, we provide some examples of seismic risk estimates for the

Himalayas and surrounding regions.

550 Nat Hazards (2014) 71:549–562

123

2 Methodology

Seismic reality evidences many contradictions to the model assumption of a stationary

Poisson point process with annual rate of N(M), which according to the well-established

Gutenberg–Richter law is parameterized in a log-linear form (Molchan et al. 1997).

Eventually, a response to these leads to complications of the existing hypotheses by

introducing sequences of main events and their associates (fore- and aftershocks) super-

imposed with hypothetical distributions of the associate size, time, and location. Therefore,

estimation of N(M) at a given site of interest remains the basic source of erroneous seismic

hazard assessment (Kossobokov and Nekrasova 2012), as well as the basic source of

inadequate seismic engineering decisions (Davis et al. 2012).

The results of the global and regional analyses (Keilis-Borok et al. 1989; Kossobokov

and Mazhkenov 1994; Kossobokov and Nekrasova 2003, 2005, 2007; Nekrasova and

Kossobokov 2002, 2003, 2006; Nekrasova 2008) imply that the recurrence of earthquakes

at a seismically prone site, for a wide range of magnitudes M and sizes L [ (L-, L?), can be

described as USLE by the following formula:

NðM; LÞ ¼ 10A � 10B�ð5�MÞ � LC ð1Þ

where L 9 L is a square embedding seismic locus and A, B, and C are constants. By an

alternative formulation using the inter-event time distribution, we can denote the rela-

tionship as USLE (Bak et al. 2002; Christensen et al. 2002). The algorithm for Scaling

Coefficients Estimation, named SCE, is described in detail by Kossobokov and Nekrasova

(2003, 2005).

One can use the long-term estimates of the USLE coefficients to characterize seismic

hazard in a rather traditional terms of maximum expected magnitude and/or macroseismic

intensity, peak ground acceleration (PGA), etc. Specifically,

• Consider the values of A, B, and C obtained for grid points of a regular l 9 l mesh of

centers of the L0 9 L0 cells of interest, where L0 [ (L-, L?) is the predefined constant

used in the analysis of sizes.

• Using formula (1) for magnitude ranges from M- to M? with Dm-magnitude step

calculate the expected numbers of events in T years in the cells of interest,

NT M; L0ð Þ ¼ T � N M; L0ð Þ.• For each cell find the maximum magnitude with the expected number NT(M,L0) C p %

and assign the intensity or PGA value that corresponds to this maximum magnitude.

Presumably, the value of hazard was assigned to a cell of interest in such a way

indicates an estimate of the traditional one with the p% probability of exceedance in

T years.

2.1 Data

2.1.1 Seismic data

We consider the territory from 5�M to 40�N and from 60�E to 105�E. The coefficients of

the Unified Scaling Law for Earthquakes, USLE, are evaluated using the earthquakes

with the depth shallower than 100 km from the USGS/NEIC Global Hypocenters

Database System, 1965–2011. The choice of the threshold at 100 km is supported by the

first local minimum of the density distribution of the hypocenter depth in the region

Nat Hazards (2014) 71:549–562 551

123

considered. The magnitude attributed to each seismic event is the maximum of the

average body-wave magnitudes, mb, the average surface-wave magnitude, MS, and the

two authority magnitudes MA1 and MA2 (usually, of the MW type). There are about

5000 of magnitude 5 or larger events considered from the database of the last 47 years.

Figure 1 shows the variation of the annual number of these earthquakes in time. The

annual number is rather stable (about 100–200 events per year) before the mega-thrust

earthquake on December 26, 2004, whose epicenter, but not all aftershocks, is just

outside the region considered. Figure 2 presents the empirical density of the seismic

distribution in space evaluated in a grid of 1/2� 9 1/2� cells (the density total equals to

1). Each cell is color-coded according to the logarithm of the empirical density function.

These cells were at the base of the hierarchy of areas with linear size of 1/2�, 1�, 2�, 4�,

and 8�.

The general geological and tectonic features of the Indian subcontinent are shown in

Fig. 2a. The entire Indian subcontinent can initially be divided into three main subregions

on the basis of general geological and tectonic features. The first subregion is formed by

the Sulaiman and Kirthar mountain ranges in the northwest, the Himalayan Mountains in

the north, extending from west to east for a distance of 2,500 km, and the Arakan-Yoma

mountain ranges in the east, extending from north to south into the island arc system of the

Andaman, Nicobar, Sumatra, and Java Islands. The major tectonic features in the Hima-

layan mountain ranges include, from south to north, the main boundary thrust (MBT), the

main central thrust (MCT), and the Indus-Tsangpo suture (ITS) all along the entire length

of the Himalayas from west to east. The MBT is a series of thrusts that separates the

predominantly pre-Tertiary Lesser Himalayas from the Tertiary Siwalik, composed of

fossiliferous Riphean sediments overridden by several thrust sheets (Wadia 1975; Gansser

1977).

The MCT at the base of the central crystalline zone dips northward separating the High

Himalayas from the Lesser Himalayas (Gansser 1964). The ITS is characterized by the

ophiolite suite on the north and demarcates the northern limits of the Indian plate. The

Arakan-Yoma fold belt consists of a large thickness of Mesozoic and Tertiary rocks

intruded by granitic and ultrabasic rocks (Krishnan 1968). It is a northward continuation

0

1

10

100

1,000

1965 1970 1975 1980 1985 1990 1995 2000 2005 2010

4.0

4.5

5.0

5.5

6.0

6.5

7.0

7.5

8.0

Fig. 1 Annual number of earthquakes of different magnitude in the study region, 1965–2011

552 Nat Hazards (2014) 71:549–562

123

Fig. 2 a The generalized tectonic map and location of great earthquakes of India and adjacent areas(modified after Parvez et al. 2003). b The empirical density distribution of seismic activity in space (basedon the USGS/NEIC Global Hypocenters Database System, 1965–2011)

Nat Hazards (2014) 71:549–562 553

123

into the continent of the Andaman, Nicobar, Sumatra, and Java Island arc system and is

laced with thrust zones and other faults that have been produced by the collision of the

Indian and Burmese plates (Deshikachar 1974). The fault plane solutions indicate a general

northward underthrusting of the Indian plate in the Himalayan front and an eastward under

thrusting in the Arakan-Yoma region.

The second subregion is formed by the vast alluvial plains of the river Sindhu (Indus)

and Ganga (Ganges) with an average elevation of about 200 m. The hills and coastal

plains of the Indian Peninsular shield form the third subregion with an average elevation

of about 600 m, although mountain ranges are present within this region having eleva-

tions in excess of 2,500 m. The Sindhu–Ganga basin (second subregion) is a frontal

depression that is filled by sediments and alluvium. It is also considered the replica of the

trench systems, which are associated with the front of island arc systems at subduction

zones (Khattri 1987). It consists of the autochthonous zone, in which the Precambrian

rocks of the Indian shield are down-warped and plunge under the mountains toward the

northwest, north and east forming the foredeep and marginal depression south of the

Himalayan tectonic zone. Infrequent earthquakes clustered in a few localities occur in

this subregion.

The third subregion, that is the Indian Peninsular shield, is considered to be a stable

landmass and a region of slight seismicity. The geological framework of the region is the

cumulative effect of geodynamics sequences ranging from the early Precambrian crustal

evolution to young volcanism over its northwest segment. This shield is bounded in the

north by the Narmada-Son lineament and Godavari rift system.

Figure 2a also shows the great earthquakes occurred in the past in the region, and most

significant ones are 1819 Kutch earthquake (M = 8.0), 1897 Shillong earthquake

(M = 8.7), 1905 Kangra earthquake (M = 8.0), 1934 Bihar–Nepal earthquake (M = 8.3),

1941 Andaman Island earthquake (M = 8.1), and 1950 Assam earthquake (M = 8.5).

During the last couple of decades, India had several damaging and devastating earth-

quakes, e.g., 1988 Bihar–Nepal earthquake (M = 6.4), 1991 Uttarkashi earthquake

(M = 6.9), 1993 Latur earthquake (M = 6.3), 1997 Jabalpur earthquake (M = 6.0), 1999

Chamoli earthquake (M = 6.8), 2001 Bhuj earthquake (M = 7.7), 2004 Sumatra earth-

quake (M = 9.3), 2005 Pakistan earthquake (M = 7.6), and 2011 Sikkim earthquake

(M = 6.9).

The reliable estimates of the USLE coefficients were obtained in 1,143 out of the 1,150

cells shown in Fig. 2.

2.1.2 Data on population

We consider the population density map for the territory from latitude 5�M to 40�N and

longitude from 60�E to 105�E. The population data in the present study are taken from

Gridded Population of the World (2005) that is a raster data product compiled at Center for

International Earth Science Information Network (CIESIN), Columbia University, and

Centro Internacional de Agricultura Tropical (CIAT), which renders global population data

at the scale and extent, required for demonstrating the spatial relationship of human

populations and the environment across the globe. The database provides a spatially di-

saggregated population layer that is compatible with data sets from social, economic, and

Earth science fields. We have chosen the population data estimates of the year 2010 to

assess different kinds of risk associated with seismic hazard. Figure 3 demonstrates this

population density distribution of the region.

554 Nat Hazards (2014) 71:549–562

123

3 Results: estimates of seismic hazard

Figure 4 demonstrates the distribution of the estimated values for each of the three USLE

coefficients: Coefficient A represents logarithm of seismic activity; B characterizes the

balance between magnitude ranges; and C corresponds to the estimate of the fractal

dimension of seismic locus. Note that (1) unlike previous applications of formula (1), the

coefficient A is calibrated to recurrence of strong magnitude 6.0 earthquakes

NðM; LÞ ¼ 10A � 10B�ð6�MÞ � LC ð2Þ

and (2) in all the estimates of A, B, and C, the error of determination does not exceed 0.07,

confirming the claimed accuracy of the values plotted on the three maps. If we see the

distribution of coefficient A, the highest values appear in the Andaman region, northwest

Himalayas, and Pamir-Hindukush region. Figure 4b illustrates that the density bulk of the

coefficient of magnitude balance B concentrates mainly between 0.7 and 0.9, and the

higher values of B are observed in the northwest Himalayas, Hindukush, and Indo-Burmese

arc. The estimate of the fractal dimension of the earthquake epicenter locus C has a smooth

density distribution widespread from 0.8 to 1.5 and that can be interpreted as scaling along

a near linear single fault zone and the more complex structure of blocks and faults. The

highest value of C is mainly concentrated in Myanmar and Pamir-Hindukush region once

again. Evidently, the distribution of C corresponds to the level of the Earth crust disin-

tegration: Larger values of C indicate more fractured areas of the widespread zone of the

continental collision.

The USLE coefficients were used for estimation and mapping the expected maximum

magnitude M with 1 % and 10 % probability of exceedance in 50 years. For the purposes

of additional stability check and comparison, the estimation of the USLE coefficients and

the hazard maps was made also at higher resolution, but the same number of hierarchy

Fig. 3 The population density distribution (based on GPWv3 estimate for 2010)

Nat Hazards (2014) 71:549–562 555

123

Fig

.4

Th

ed

istr

ibu

tio

no

fth

eU

SL

Eco

effi

cien

tsin

spac

e(b

ased

on

the

US

GS

/NE

ICG

lobal

Hy

po

cente

rsD

atab

ase

Sy

stem

,1

96

5–

201

1)

556 Nat Hazards (2014) 71:549–562

123

levels (nh = 5) and with seismic data down to magnitude 4 in 1980–2011. In particular,

the reliable estimates of the USLE coefficients (i.e., with the standard error of determi-

nation \0.05) were obtained for 1,574 cells out of the 1,640 cells of 1/4� 9 1/4� size.

Figure 5 displays distribution on the map of the expected maximum magnitude M with

1% and 10 % probability of exceedance in 50 years based on the higher resolution settings

and rich seismic data. The two distributions naturally display evident similarity and dif-

ference. Specifically, for the probability of exceedance of 1 % in 50 years, there are just 67

out of 1,143 cells (i.e., \6 %) where the estimates of the expected maximum magnitude

M differ by one half of the magnitude unit, while this number is 85 cells (i.e.,\7.5 %) for a

less restrictive probability level of 10 %. It is notable that Fig. 5 is in full agreement with

pattern recognition of the sites prone to magnitude 6.5, 7.0, 8.2, or larger earthquakes based

on geomorphologic and geodetic parameters (Bhatia et al. 1992; Kossobokov 1984), in

which validity has been confirmed by the occurrence of strong earthquakes in the region

after publications of the results (Gorshkov et al. 2003, 2012).

So far, it is not possible to use USLE estimation of the expected maximum magnitude

M at all sites, which expose seismic activity, due to the issues of spatial and temporal

completeness of earthquake catalogs for the territory under study. Therefore, for the sake

of a more comprehensive estimation of seismic hazard, we complement the final map of

the expected maximum magnitude M based on USLE with the reported maximum mag-

nitude in 1900–2010 at those cells where the USLE coefficients were not evaluated.

(Similar results were obtained using the SCE algorithm applied to the catalog of main-

shocks, although for an obvious reason of smaller sample size, the reliable estimations

were found in a smaller sample of grid points.)

The final map was used to estimate the hazard map of expected ground shaking in

terms of PGA following the semi-empirical relations given by Parvez et al. (2001). They

have used the strong motion data to determine the interdependence among the peak

ground motion, earthquake magnitude, and distance to hypocenter for the Himalayan

region. In order to produce deterministic seismic hazard map as shown in the figure, a

source–receiver distance is estimated at a grid of 0.25� 9 0.25�, and for each source of

different magnitude, PGA is calculated as in (Parvez et al. 2001): The maximum at each

cell is plotted as shown in Fig. 6. We have opted the minimum and maximum epicentral

distance at 10 and 500 km, respectively. The hazard map clearly marks the zones of high

Fig. 5 The expected maximum magnitude M with probability of exceedance in 50 years 1 % (a) and 10 %(b) (based on the USGS/NEIC Global Hypocenters Database System, 1965–2011)

Nat Hazards (2014) 71:549–562 557

123

peak ground acceleration in along the plate boundary of the Himalayas, Indo-Burmese

arc, part of Gujarat, and Andaman Nicobar region; here, PGA is shown over 1 g. In the

foothills of the Himalayas and Indo-Gangetic planes, the PGA varies between 0.06 and

0.2 g.

3.1 Estimates of seismic risks

Any kind of risk estimates results from a convolution of the hazard with the exposed object

under consideration along with its vulnerability—

R gð Þ ¼ H gð Þ � O gð Þ � V O gð Þð Þ;

where H(g) is natural hazard at location g, O(g) is the exposure of objects at risk at g, and

V(O) is the vulnerability of objects at risk. Note that g could be a point, or a line, or some

area on or under the Earth surface and that distribution of hazards, as well as objects of

concern and their vulnerability, could be time-dependent. Furthermore, the convolu-

tion � could be much more complicated operation than usually accepted as product of

terms.

Thus, there exist many different risk estimates even if the same object of risk and the

same hazard are involved. Specifically, it may result from the different laws of convolu-

tion, as well as from different kinds of vulnerability of an object of risk under specific

environments and conditions. Both conceptual issues must be resolved in a multidisci-

plinary problem-oriented research performed by specialists in the fields of hazard, objects

of risk, and object vulnerability. Here, to illustrate this general concept, we perform the

following oversimplified four convolutions of seismic hazard assessment map H(g) with

the population density distribution P:

All the four risk estimates consider the GPWv3 model population data estimates of the

year 2010 as an object of risk and use mathematical product as the convolution law. The

0.625

1.000

0.500

0.250

0.125

0.015

0.031

0.007

PGA

Fig. 6 Final deterministic seismic hazard assessment map

558 Nat Hazards (2014) 71:549–562

123

first estimate in a cell g is based on the constant equal vulnerability of an individual,

Ri gð Þ ¼ H gð Þ �R

gP, where

Rg

P is the integral of the population density over the cell g, i.e.,

the number of individuals within the area of the cell g. The second risk estimate differ-

entiates individual vulnerability in proportion to the population density at a given site,

Rii gð Þ ¼ H gð Þ �R

gP � P. This and the other two assumptions—Riii gð Þ ¼ H gð Þ �

Rg

P � P2

and Riv gð Þ ¼ H gð Þ �R

gP � P3—appear to be rather natural due to specifics of man-made

environment inflicted in the areas of high concentration of individuals, e.g., the number of

floors in a typical residential building changes with population density. The resulting maps

of the four risks in arbitrary units are given in Fig. 7. For illustration purposes, each of the

four risk scales is covering the top seven decimal orders of the risk values, so that the cells

in red are 1,000,000 times more dangerous than those in blue. The collapse of the risky

areas to the region of the densest population appears rather natural demonstrating how

nonlinearity of conditions changes expectation of seismic risk. It is notable that as

expected, the megacities and their agglomerations are at the top of risk distributions.

To avoid misleading counterproductive interpretations, we have to emphasize that the

risk estimates presented for the territory under study are given here for academic meth-

odological purposes only. They do not use complicated procedures that might be more

adequate convolutions of hazard, objects, and their vulnerability, and are used here to

illustrate the general problem-oriented approach. The estimations addressing more realistic

(c)

Riii(g)

(d)

Riv(g)

(b)

Rii(g)

(a)

Ri(g)

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

10

Fig. 7 Oversimplified convolutions of seismic hazard map H(g) with the population density distribution P:

a Ri gð Þ ¼ H gð Þ �R

gP, b Rii gð Þ ¼ H gð Þ �

Rg

P � P, c Riii gð Þ ¼ H gð Þ �R

gP � P2, and d Riv gð Þ ¼ H gð Þ �

Rg

P � P3

Nat Hazards (2014) 71:549–562 559

123

and practical kinds of seismic risks, not presented here, should involve experts in distri-

bution of objects of risk of different vulnerability, i.e., specialists in earthquake engi-

neering, social sciences, and economics.

4 Discussion and conclusion

Timings and places of earthquakes are not predictable with precision of a certain date and

rupture zone. Although it is already proven in the long-lasting global experiment that

intermediate-term middle-range earthquake predictions are reliable, accurate, and can be

effective enough to globally reduce the impact of great earthquakes from M8.0 ? magnitude

and higher ranges in advance (Kossobokov 2012; Davis et al. 2012), there is no link to

implement measures and improve earthquake preparedness in response to them. This is not

done yet, in part due to limited distribution of these predictions and the lack of applying

existing methods for using intermediate-term predictions to make decisions for taking action.

On the other hand, the general locations of earthquakes are well known globally by

earthquake monitoring and geological studies. Strong disastrous earthquakes are low

probability events that happen with certainty (i.e., 100 % probability). Long recurrence or

low probability earthquake information provided by any group as the basis for seismic

hazard assessment leads to false comfort of ignoring such rare eventualities, so that they

will happen again as unpredictable disastrous surprises. Regretfully, a systematic and

quantitative comparison of the global seismic hazard map (Kossobokov and Nekrasova

2012) discloses gross inadequacy of this ‘‘probabilistic’’ product: All 60 earthquakes with

magnitudes of 7.5 or higher were ‘‘surprises,’’ and half of them were ‘‘big surprises’’ for

the map, as well as all twelve of the deadliest earthquakes happened in 2000–2011 with the

total number of deaths exceeding 700,000 people. These and other self-evident short-

comings and failures of the widespread probabilistic approach to seismic hazard assess-

ment prove that the global map, as well as underlying methodologies, is deeply flawed. In

common sense, the map is evidently unacceptable for any kind of responsible seismic risk

evaluation and knowledgeable disaster prevention. Other modern methods for modeling

realistic scenarios of earthquakes allow better advanced seismic hazard assessment (Panza

et al. 2011).

Losses from natural hazards have continued to increase steadily due to the lack of

knowledge and poor understanding by the majority of the scientific community, as well as

by decision-makers and public, of the three components of risk, which are hazard, expo-

sure, and vulnerability. Contemporary science (geophysics and seismology, in particular) is

largely responsible for societal failures in satisfactorily coping with many challenging

changes in exposures and their vulnerabilities inflicted by growing populations, their

concentration, etc. Scientists, for their special education, knowledge, and skills, owe it to

society, which does not possess the same special knowledge and skills. Errors in seismic

hazard assessment propagate nonlinearly into unbearable errors in the expected human and

economic losses (Wyss et al. 2012).

The largest expected earthquake size for earthquake-prone locations can be realistically

estimated with high confidence. Preparing for such hazardous event ensures reduction or

even avoidance of potential disasters. Earthquakes do not kill people, but their conse-

quences, i.e., tsunamis, landslides, collapsed buildings, bridges, and other constructions,

do. The hazard scenarios for such an event estimated by deterministic computations pro-

vide the basis for necessary preparations, from land-use planning and building code reg-

ulation to emergency management. There are many risks generated by earthquakes. These

560 Nat Hazards (2014) 71:549–562

123

should be not ignored in any realistic and responsible seismic risk evaluation and

knowledgeable disaster prevention.

Our study attempts to contribute modestly to an urgent revision of the seismic hazard

maps from the first principles, including (1) background methodologies and (2) imple-

mentation in assessment of seismic risks. Evidently, our contribution at this time (1) takes

into consideration economic and social factors of risk assessment in an oversimplified way,

(2) does not account for the role of site effect due to topography and soils in determining

ground acceleration, and (3) makes use of a point approximation of an earthquake rupture

even for the great and complex seismic events. These essential considerations should be

addressed in the future.

Acknowledgments The authors acknowledge the financial support from the RFBR-DST Project ‘‘Seismichazard and risk assessment based on pattern recognition: Himalayas and adjacent territories’’ (Grants RFBRNo. 11-05-92691 and DST No. INT/RFBR/P-109). We also would like to thank two anonymous reviewersfor their constructive suggestions and critical reviews to improve the quality of the manuscript.

References

Auden JB (1959) Earthquakes in relation to the Damodar Valley Project. Proc Symp Earthquake Engi-neering, 1st University of Roorkee, Roorkee

Bak P, Christensen K, Danon L, Scanlon T (2002) Unified Scaling Law for Earthquakes. Phys Rev Lett88:178501–178504

Bhatia SC, Chetty TRK, Filimonov M, Gorshkov A, Rantsman E, Rao MN (1992) Identification of potentialareas for the occurrence of strong earthquakes in Himalayan arc region. Proc Indian Acad Sci (EarthPlanet Sci) 101(4):369–385

Bhatia SC, Kumar R, Gupta HK (1999) A probabilistic seismic hazard map of India and adjoining regions.Ann Geophys 42:1153–1164

Christensen K, Danon L, Scanlon T, Bak P (2002) Unified Scaling Law for Earthquakes. Proc Natl Acad Sci99(suppl. 1):2509–2513

Davis C, Keilis-Borok V, Kossobokov V, Soloviev A (2012) Advance prediction of the March 11, 2011Great East Japan earthquake: a missed opportunity for disaster preparedness. Int J Disaster Risk Reduct1:17–32. doi:10.1016/j.ijdrr.2012.03.001

Deshikachar SV (1974) A review of the tectonic and geological history of eastern India in terms of ‘platetectonics’ theory. J Geol Soc India 15:137–149

Gansser A (1964) Geology of the Himalayas. Interscience, New York, p 289Gansser A (1977) The great suture zone between Himalaya and Tibet: a preliminary account. In: Jest C (ed)

Himalaya: Science de la Terre, Colloq. Int. CNRS, 268: 181–192Gaur VK, Chouhan RKS (1968) Quantitative measures of seismicity applied to Indian regions. Bull Indian

Soc Earthq Technol 5:63–78Gorshkov A, Kossobokov V, Soloviev A (2003) 6. Recognition of earthquake-prone areas. In: Keilis-Borok

VI, Soloviev AA (eds) nonlinear dynamics of the lithosphere and earthquake prediction. Springer,Heidelberg, pp 239–310

Gorshkov A, Parvez IA, Novikova O (2012) Recognition of earthquake-prone areas in the Himalaya:validity of the results. International J Geophys 2012, Article ID 419143, p 5. doi:10.1155/2012/419143

Gridded Population of the World, Version 3 (GPWv3) (2005) Palisades, NY: Socioeconomic Data andApplications Center (SEDAC), Columbia University. Available at http://sedac.ciesin.columbia.edu/gpw. 2012.05.29

Kaila KL, Rao M (1979) Seismic zoning maps of Indian subcontinent. Geophys Res Bull 17:293–301Keilis-Borok VI (1990) The lithosphere of the Earth as a nonlinear system with implications for earthquake

prediction. Rev Geophys 28(1):19–34Keilis-Borok VI, Kossobokov VG, Mazhkenov SA (1989) On similarity in spatial distribution of seismicity.

In: Theory and algorithms of interpretation of geophysical data. Nauka, Moscow, pp 28–40 (Com-putational Seismology 22, in Russian)

Khattri KN (1987) Great earthquakes, seismicity gaps and potential for earthquake disaster along theHimalaya plate boundary. Tectonophysics 138:79–92

Nat Hazards (2014) 71:549–562 561

123

Khattri KN, Rogers AM, Perkins DM, Algermissen ST (1984) A seismic hazard map of India and adjacentareas. Tectonophysics 108:93–134

Kossobokov VG (1984) General features of the strongest (with M C 8.2) earthquake-prone areas in the non-Alpine zone of the Transasian seismic belt. In: Keilis-Borok VI, Levshin AL (eds) Logical andcomputational methods in seismology (Comput Seismol 17). Nauka, Moscow, pp 69–72

Kossobokov VG (2012) Earthquake prediction: 20 years of global experiment. Nat Hazards. doi:10.1007/s11069-012-0198-1

Kossobokov VG, Mazhkenov SA (1988) Spatial characteristics of similarity for earthquake sequences:fractality of seismicity. Lecture notes of the workshop on global geophysical informatics with appli-cations to research in earthquake prediction and reduction of seismic risk (15 Nov–16 Dec., 1988),ICTP, Trieste, p 15

Kossobokov VG, Mazhkenov SA (1994) On similarity in the spatial distribution of seismicity. Computa-tional Seismology and Geodynamics. Am. Geophys. Un., The Union, Washington, DC, 1:6–15

Kossobokov VG, Nekrasova A (2003) Generalized Gutenberg-Richter recurrence law. Geophysical researchabstracts, 5, abstracts of the contributions of the EGS-AGU-EGU Joint Assembly, Nice, France, 06–11April, 2003 (CD-ROM): EAE03- A-06597

Kossobokov VG, Nekrasova AK (2005) Temporal variations in the parameters of the Unified Scaling Lawfor Earthquakes in the eastern part of Honshu Island (Japan). Doklady Earth Sci 405:1352–1356

Kossobokov VG, Nekrasova A (2007) Unified scaling law for earthquakes: implications for seismic hazardand risk assessment. In: IUGG2007, July 2–13, 2007, Perugia, Italy. Abstracts, SS002–65

Kossobokov VG, Nekrasova AK (2012) Global seismic hazard assessment program maps are erroneous.Seism Instrum 48(2):162–170. doi:10.3103/S0747923912020065

Krishnan MS (1968) Geology of India and Burma. Higginbothams, Madras, p 536Lyubushin A, Parvez IA (2010) Map of seismic hazard of India using Bayesian approach. Nat Hazards

55:543–556Mandelbrot BB (1982) The fractal geometry of nature. Freeman, New York, p 488Molchan G, Kronrod T, Panza GF (1997) Multi-scale seismicity model for seismic risk. Bull Seism Soc Am

87:1220–1229Nekrasova A (2008) Unified scaling law for earthquakes: application in seismic regions worldwide. PhD

Thesis, M.V. Lomonosov Moscow State University, Moscow, p 200 (in Russian)Nekrasova AK, Kosobokov VG (2006) General law of similarity for earthquakes: evidence from the Baikal

region. Doklady Earth Sci 407A(3):484–485Nekrasova A, Kossobokov V (2002) Generalizing the Gutenberg-Richter scaling law. EOS Trans AGU

83(47), Fall Meet Suppl, abstract NG62B-0958, 2002Nekrasova A, Kossobokov V (2003) Generalized Gutenberg-Richter recurrence law: global map of

parameters. Geophysical research abstracts, 5, abstracts of the contributions of the EGS–AGU–EGUJoint Assembly, Nice, France, 6-11 April, 2003 (CD-ROM): EAE03-A-03801

Nekrasova A, Kossobokov V, Peresan A, Aoudia A, Panza GF (2011) A multiscale application of theUnified Scaling Law for Earthquakes in the Central Mediterranean area and Alpine region. Pure ApplGeophys 168:297–327

Panza G, Irikura K, Kouteva-Guentcheva M, Peresan A, Wang Z, Saragoni R (eds) (2011) Advancedseismic hazard assessment. Pure Appl Geophys 168(1–4):752

Parvez IA (2007) On the Bayesian analysis of the earthquake hazard in the north–east Indian peninsula. NatHazards 40:397–412

Parvez IA, Ram A (1997) Probabilistic assessment of earthquake hazards in the north–east Indian Peninsulaand Hidukush region. Pure Appl Geophys 149:731–746

Parvez IA, Ram A (1999) Probabilistic assessment of earthquake hazards in the Indian subcontinent. PureAppl Geophys 154:23–40

Parvez IA, Gusev AA, Panza GF, Petukhin AG (2001) Preliminary determination of the interdependenceamong strong motion amplitude, earthquake magnitude and hypocentral distance for the Himalayanregion. Geophys J Int 144:577–596

Parvez IA, Vaccari F, Panza GF (2003) A deterministic seismic hazard map of India and adjacent areas.Geophys J Int 155:489–508

Turcotte DL (1997) Fractals and chaos in Geology and Geophysics, 2nd edn. Cambridge University Press,Cambridge

Turcotte DL (1999) Seismicity and self-organized criticality. Phys Earth Planet Int 111:275–294Wadia DN (1975) Geology of India. Tata McGraw-Hill, New Delhi, p 508Wyss M, Nekrasova A, Kossobokov V (2012) Errors in expected human losses due to incorrect seismic

hazard estimates. Nat Hazards 62(3):927–935. doi:10.1007/s11069-012-0125-5

562 Nat Hazards (2014) 71:549–562

123