Earthquake prediction research and earthquake preparedness

1
Earthquake prediction research and earthquake
preparedness
A. Peresan G.F. Panza
.
Department of Earth Sciences University of Trieste
2
Outline

• Introduction
• Earthquake precursors
• Premonitory seismicity patterns
• Earthquake prediction
• Algorithms M8 and CN
• Evaluation of prediction results

3

• "Earthquakes modelling involves a detailed
  knowledge of the related physics, which is not
  available at present time. ... A firm and
  complete phenomenological picture should be
  established before any effort can result
  effective, but such a picture is not easy to draw
  due to the long time scales involved. ... The
  analysis of seismicity patterns is useful not
  only for prediction purposes, but it provides
  also the wide set of systematic observations,
  without which any physically based model remains
  a merely theoretical speculation."
• Earthquake prediction the scientific challenge
• (Knopoff, 1996)

4
Characteristics of the lithosphere

• Scale invariance of earthquake distribution in
  time and space
• Self-organization of earthquake occurence

• Non-linear mechanics of earthquake generation
• Statistical features of earthquake sequences

Strongly suggest that the lithosphere behaves as
a non-linear and possibly deterministic chaotic
system (Keilis-Borok, 1990).
5
How we can predict earthquakes?

• DST
• The problem of prediction of destructive
  earthquakes cannot be solved statistically,
  because there have been too few large enough
  events in any small sufficiently area in the past
  century to be able to define probabilities of the
  largest events with the required accuracy
  (Knopoff, 1996).
• Therefore two other strategies remain to attempt
  earthquake forecast detecting precursors or
  modelling the earthquake physics.
• Earthquakes modelling involves a detailed
  knowledge of the related physics, which is not
  available at the present time. Furthermore, if
  the system is demonstrated to be chaotic, then
  deterministic predictions could be impossible
  even provided the physical description.
• On the contrary, precursors are based on
  empirical observations and do not require
  necessarily to know the underlying mechanisms.

Deterministic prediction? NO For complex or
chaotic systems, even an accurate modelling would
not allow for predictions.
Statistical prediction? NO Time scales involved
in the seismogenic process are too long and
observations too limited.
Identification of precursory phenomena.
Possibility
6
Earthquake precursors

• DST
• The precursor strategy seems difficult to follow,
  because earthquakes are too infrequent and each
  phenomenon has its own non-seismic natural
  variations monitoring different phenomena is
  possible, but establishing a precursory
  connection would require observations of many
  cases.

Possible precursors are those phenomena that may
take place in the lithosphere during the
accumulation of stresses. Difficulty to
establish a clear precursory connection, i.e. to
separate the precursory signal from natural
fluctuations. This is due to the lack of
sufficiently prolonged and systematic records.
7
Signals" proposed as earthquake precursors

• DST
• Among the several signals which have been
  indicated as possible precursors relevant to
  earthquake prediction, there are (Lomnitz, 1994
  Geller, 1997 Kagan, 1997) variations in the
  seismic activity changes in the velocity and in
  the spectral content of the seismic waves and of
  the earthquakes sources crustal deformations and
  variations of the stress in the crust changes in
  the gravitational and geomagnetic fields
  geoelectrical precursors anomalous changes in
  the underground water level and chemical
  components anomalies in the atmospheric
  pressure, temperature and heat flow.

• Variations in the seismic activity
• Changes in the velocity and in the spectral
  content of seismic waves and earthquakes sources
• Deformations and variations of the stress in the
  crust
• Gravitational, geomagnetic and geoelectric
  precursors
• Anomalous changes in the underground water level
  and chemical components
• Anomalies in the atmospheric pressure,
  temperature and heat flow

8
IASPEI Preliminary List of Significant Precursors

• DST
• The IASPEI Preliminary List of Significant
  Precursors issued on March 1994 (Wyss, 1997), for
  example, includes only five possible precursors,
  out of the forty precursory observations
  submitted until now one is based on ground water
  chemistry (Wakita, 1988), one is a measurement of
  crustal deformation by ground water levels
  (Roeloffs and Quilty, 1997) and three are
  seismicity patterns. The seismic precursors are
  based on foreshocks (several days before the
  event), on preshocks swarms (several months
  before the earthquake) and on the seismic
  quiescences that can be observed before major
  aftershocks. Four out of these five precursors,
  however, are based on a single example only the
  last premonitory pattern, which should allow for
  the prediction of the strong aftershocks, has
  been observed in several cases. These five
  precursory observations seem to deserve further
  study in the framework of earthquake prediction,
  nevertheless none of them can be considered yet
  as a validated precursor.

• Validation criteria for precursor candidates
• the observed anomaly should be related to some
  mechanism leading to earthquakes
• the anomaly should be simultaneously observed at
  more than one site or instrument
• the definition of the anomaly and of the rules
  for its association with subsequent earthquakes
  should be precise
• both anomaly and rules should be derived from an
  independent set of data, than the one for which
  the precursory anomaly is claimed.

9
IASPEI Preliminary List of Significant Precursors

• DST
• The IASPEI Preliminary List of Significant
  Precursors issued on March 1994 (Wyss, 1997), for
  example, includes only five possible precursors,
  out of the forty precursory observations
  submitted until now one is based on ground water
  chemistry (Wakita, 1988), one is a measurement of
  crustal deformation by ground water levels
  (Roeloffs and Quilty, 1997) and three are
  seismicity patterns. The seismic precursors are
  based on foreshocks (several days before the
  event), on preshocks swarms (several months
  before the earthquake) and on the seismic
  quiescences that can be observed before major
  aftershocks. Four out of these five precursors,
  however, are based on a single example only the
  last premonitory pattern, which should allow for
  the prediction of the strong aftershocks, has
  been observed in several cases. These five
  precursory observations seem to deserve further
  study in the framework of earthquake prediction,
  nevertheless none of them can be considered yet
  as a validated precursor.

• Only five possible precursors, out of the forty
  proposed, seem to deserve further study (Wyss,
  1997)
• one based on ground water chemistry
• one based on deformation of the crust
• three based on seismicity patterns

10
PREMONITORY SEISMICITY PATTERNS
11
Premonitory seismicity patterns

Some changes are observed in the earthquakes
flow before a large event.
These changes are akin to the general symptoms of
instability of many non-linear systems before a
catastrophe (Keilis-Borok, 1996).

• In particular, the response to a perturbation
• increases,
• becomes more irregular and
• acts at long distances.

12
Premonitory seismicity patterns

• In the case of seismicity the non-linear system
  is the hierarchical structure made up by the
  lithospheric blocks and by their boundaries (i.e.
  faults).
• The large earthquake is a catastrophic event,
  corresponding to abrupt changes of the system
  characteristics, that may involve a large domain
  of the system.
• The small earthquakes may be regarded as sources
  of perturbation of the system.

13
Premonitory seismicity patterns

• Thus, before a strong earthquake, which
  represents a catastrophic event, we should
  observe
• increase of the seismic activity, clustering of
  the earthquakes in time and space, and spatial
  concentration of sources in other words, the
  increase of the response to the perturbation
• increase of the variation of seismicity and its
  clustering
• long-range interaction of earthquakes, which can
  be interpreted as an increase of the range of
  influence of the perturbation.

14
Single seismicity patterns formally defined as
premonitory

• the burst of aftershocks (Keilis-Borok et al.,
  1980 Molchan et al., 1990), which is associated
  to moderate magnitude events characterised by a
  large number of aftershocks
• the seismic quiescence (Wyss et al., 1992)
• the relative increase of the b-value for the
  moderate events, with respect to smaller events
  (Narkunskaya and Shnirman, 1994)
• the increase of the spatial correlation in the
  earthquake flow and the log-periodic variations
  of the earthquake flow on the background of its
  exponential rise (Bufe et al., 1994).

15
Multiple seismicity patterns formally defined as
premonitory

• M8 (Keilis-Borok and Kossobokov, 1987 Kossobokov
  et al., 1999)
• CN (Gabrielov et al., 1986 Keilis-Borok and
  Rotwain, 1990 Rotwain and Novikova, 1999)
• Mendocino scenario (Kossobokov, Keilis-Borok, and
  Smith, 1990 Kossobokov et al., 1999)
• Next Strong Earthquake (Vorobieva, 1999)

16
EARTHQUAKE PREDICTION
17
What does it means earthquake prediction?

• The United States National Research Council,
  Panel on Earthquake Prediction of the Committee
  on Seismology suggested the following definition
  (1976, p.7)
• An earthquake prediction must specify the
  expected magnitude range, the geographical area
  within which it will occur, and the time interval
  within which it will happen with sufficient
  precision so that the ultimate success or failure
  of the prediction can readily be judged. Only by
  careful recording and analysis of failures as
  well as successes can the eventual success of the
  total effort be evaluated and future directions
  charted. Moreover, scientists should also assign
  a confidence level to each prediction.

18
What does it means to predict an earthquake?

• DST
• Scientific earthquake prediction consists in the
  specification of the expected magnitude,
  geographic location and time of occurrence of a
  future event, with sufficient precision that the
  ultimate success or failure of a prediction can
  be evaluated (Wallace et al., 1984).

• To predict an earthquake means to indicate
• the possibility that an earthquake
• will occur in a given range of

space time magnitude
19
What does it means to predict an earthquake?

• The prediction can miss events or have false
  alarms, but forecasts must demonstrate more
  predictability than a random guess.
• The space-time-magnitude volume considered to
  declare the alarms should be appropriate to
  public needs, i.e. to enable the relevant
  authorities to prepare for an impending
  destructive earthquake.

20
Stages of earthquake prediction

• Term-less prediction of earthquake-prone areas
• Prediction of time and location of an earthquake
  of certain magnitude

21
Space Scale of Prediction
Territorial uncertainty
Predictions
Exact Narrow-range Middle-range Long - range
Earthquake size (Eqs) Two-three Eqs Five-ten
Eqs Up to 100 Eqs
Time Scale of Prediction
Alarm duration
Predictions
Few hours Few days Few years Decades
Immediate Short term Intermediate-term Long-term
22
Middle-range Intermediate-term Prediction
Currently a realistic goal appears to be the
middle-range intermediate-term prediction, which
involves an area with linear dimension about ten
times larger than the linear dimension of the
impending event and a time uncertainty of years.
23
Middle-range Intermediate-term Prediction

• Allows
• To increase preparedness and safety measures
• To define priority for detailed seismic risk
  studies
• Observation of possible short-term precursors
  (generally local)

• Time uncertainty few years
• Space uncertainty hundreds Km

24
Algorithms for middle-range intermediate-term
prediction

• A family of middle-range intermediate-term
  earthquake prediction algorithms, based on the
  identification of premonitory seismicity
  patterns, has been developed applying pattern
  recognition techniques.

25
Algorithms for middle-range intermediate-term
prediction

• Algorithms globally tested for prediction are
• M8 algorithm (Keilis-Borok and Kossobokov, 1987
  Kossobokov et al., 1999)
• CN algorithm (Gabrielov et al., 1986 Rotwain and
  Novikova, 1999)

They identify the TIPs (Times of Increased
Probability) for the occurrence of a strong
earthquake
26
Algorithms for middle-range intermediate-term
prediction

• DST
• MSc (Kossobokov, Keilis-Borok, and Smith, 1990
  Kossobokov et al., 1999),
• Next Strong Earthquake (Vorobieva, 1999)

• The algorithm MSc (Mendocino Scenario,
  Kossobokov, Keilis-Borok, and Smith, 1990
  Kossobokov et al., 1999) can be applied as a
  second approximation of M8. It allows us to
  reduce significantly the area of alarm (by a
  factor from 5 to 20).
• Independently, the algorithm NSE (Next Strong
  Earthquake, Vorobieva, 1999) is applied to
  predict a strong aftershock or a next main-shock
  in a sequence.

27
Algorithms for middle-range intermediate-term
prediction
The algorithms are based on a set of empirical
functions of time to allow for a quantitative
analysis of the premonitory patterns which can be
detected in the seismic flow

• Variations in the seismic activity
• Seismic quiescence
• Space-time clustering of events

28
Functions of seismic flow integrals over
seismic sequences
29

• DST
• 2) Quiescence
• The sign indicates that the sum includes only
  the positive terms therefore only the time
  intervals (t-s,t) where the number of earthquakes
  is less than the average are considered.
• The dotted horizontal line indicates the average
  number of events expected in the time interval of
  length s. The grey areas correspond to the
  periods of quiescence.
• The larger is the value assumed by the function,
  the more marked and prolonged is the quiescence.

TIP Time of Increased Probability Interval of
time when the probability for the occurrence of a
strong earthquake, within a delimited region,
increases with respect to the normal conditions
Seismic flow Time sequence of the earthquakes
occurred within a delimited region
30
Integrals over the seismic sequence

• Usually aftershocks are not counted so that large
  earthquakes do not dominate the count on
  earthquake sequence.
• The number of aftershocks, however is kept as one
  of the parameters of main shocks i
• ti, mi, bi(e)

31
Aftershocks identification
Windows Method   An earthquake j is an aftershock
of the earthquake i if
time difference distance
between epicentres depth
difference T(M), R(M) and H(M) are numerical
functions of the magnitude M of the main shock.
32
Functions of the seismic flow
DST Variation of seismic activity Is the sum of
the differences between the numbers of
earthquakes in two consecutive time
intervals. It corresponds to the sum of the
differences between the numbers of earthquakes in
two consecutive time intervals, with fixed length
s. The moments belong to the time interval
(t?u,t), where u is multiple of s. Usually s is
equal to one year.
Rate of seismic activity
Function K
the number of earthquakes with M M in time
interval from (t-s) to t, i.e., the number of
events of certain size per unit time, rate of
activity.
increment of activity, acceleration, expressed
by the difference between the number of
earthquakes in the two successive intervals of
time (t-s, t) and (t-2s, t-s).
33
Functions of the seismic flow
DST Variation of seismic activity Is the sum of
the differences between the numbers of
earthquakes in two consecutive time
intervals. It corresponds to the sum of the
differences between the numbers of earthquakes in
two consecutive time intervals, with fixed length
s. The moments belong to the time interval
(t?u,t), where u is multiple of s. Usually s is
equal to one year.
Function L
deviation of activity from a longer-term trend
over the period from t0 to t.
34
Functions of the seismic flow
DST Variation of seismic activity Is the sum of
the differences between the numbers of
earthquakes in two consecutive time
intervals. It corresponds to the sum of the
differences between the numbers of earthquakes in
two consecutive time intervals, with fixed length
s. The moments belong to the time interval
(t?u,t), where u is multiple of s. Usually s is
equal to one year.
Variation of seismic activity
It is a measure of the cumulative changes in
time of the earthquake number
i1,...,n uns
35
Functions of the seismic flow

• DST
• 2) Quiescence
• The sign indicates that the sum includes only
  the positive terms therefore only the time
  intervals (t-s,t) where the number of earthquakes
  is less than the average are considered.
• The dotted horizontal line indicates the average
  number of events expected in the time interval of
  length s. The grey areas correspond to the
  periods of quiescence.
• The larger is the value assumed by the function,
  the more marked and prolonged is the quiescence.

Quiescence
corresponds to the sum of the grey areas for the
time intervals (t-s,t) where the number of
earthquakes is less than the average
is the the average yearly number of events
36
Functions of the seismic flow
DST Variation of seismic activity Is the sum of
the differences between the numbers of
earthquakes in two consecutive time
intervals. It corresponds to the sum of the
differences between the numbers of earthquakes in
two consecutive time intervals, with fixed length
s. The moments belong to the time interval
(t?u,t), where u is multiple of s. Usually s is
equal to one year.
This function estimates different properties
depending on the value of b
Function ?

• the total linear size of earthquake sources
• if bB/3
• the total area of the earthquake sources
• if b2B/3
• the total energy of earthquakes
• if bB
• where B comes from the relation between the
  energy E and magnitude M
• Iog E A BM

the weighted number in time interval from (t-s)
to t and magnitude (MMilt M).
37
Functions of the seismic flow

• Function G
• The ratio of the number of earthquakes from
  two magnitude ranges
• G(t?m1,m2,s) 1 - N(t ? m2,s) / N(t ? m1,s)
• Alternative definition
• G(t?m1,m2,s) log( N(t ? m2,s) / N(t ? m1,s) ) /
  (m2 - m1)
• In such a definition G is the tangent of slope of
  the frequency-magnitude graph.
• The values m1 and m2 may result from inspecting
  the frequency-magnitude statistics of
  earthquakes.

38
Clustering of earthquakes
Functions of the seismic flow

• In case of de-clustered catalogs each main shock
  is associated to its number, bi(e,maft), of
  aftershocks of magnitude M ? maft, which occurred
  in the first e days after the origin time.
• B(t ? m,M',s,maft,e) max bi(e,maft) is the
  maximum calculated over the main shocks with
  mltMilt M' and time interval (t-s,t).

39
Correlation dimension vs. Clustering

• A correlation dimension is inversely related
  to the degree of spatial clustering, as the slope
  of the straight segment decreases when the number
  of event pairs with relatively small inter-event
  distances increases. Smaller values for CD
  indicate higher degrees of spatial clustering.

40
Long-range interaction of earthquakes

• A. Prozorov (1982) introduced a term-less
  precursor, which follow shortly a major
  earthquake but on a large distance from it. He
  concluded that long-range aftershocks mark the
  location of a future major earthquake.
• The two new phenomena which represent the
  long-range correlation were found first on a
  synthetic catalogue.

41
Normalization of functions
Functions of the seismic flow

• Normalization of earthquake sequences is
  necessary to ensure adequate uniform application
  with the same set of adjustable parameters in
  regions of different seismic activity.

• We use the normalization by minimal magnitude
  cutoff Mmin, defined by one of the two
  conditions
• Mmin M0 C, C constant
• Mmin such as N(Mmin) A, A constant rate of
  activity

42
Percentile cutoff

• The values of functions are normalized to their
  empirical probability distribution functions,
  that is performing a non-linear transformation of
  different ranges from different seismic
  environment to 0,1.

43
Functions of the seismic flow (magnitude range)
 
 
CN algorithm
 
 
m1(a3.0) m2(b1.4) m3(c0.4)
M8 algorithm
44
Algorithm CN

• DST
• G-R distribution All events with Mgt3, including
  aftershocks, occurred within the Northern Italy
  region since 1950

Choice of Mo
Area 5L-10L (L is the source linear
dimension) Magnitude Yearly average number of
events with magnitude above completeness must be
gt 3
Return Period for events with MMo not lower than
6-7 years
45
CN algorithm the learning procedure

• Fix Mo (average return period not lower than 6-7
  years)

• Definition of the classes
• - D "dangerous" periods (2 years before each
  strong earthquake)
• - N "non dangerous" periods
• - X periods not classifiable as D or N
• (3 years after a strong event 1 year
  before each D period)

Definition of the characteristic properties and
of the discretization intervals (X objects are
excluded from this stage of the analysis)
46
CN algorithm the recognition procedure

• Selection of the objects for recognition with a
  fixed time step (2 months)
• Discretization and coding of functions
• TIPs diagnosis

47
ALGORITHM M8

• The algorithm M8 is applied on a global scale for
  the prediction of the strongest events

Magnitude 8.0 Area 667 Km radius Magnitude
7.5 Area 427 Km radius
48
(No Transcript)
49
(No Transcript)
50
Reproducible intermediate-term earthquake
prediction algorithms M8 and MSc

• The algorithm M8 uses traditional description of
  a dynamical system adding to a common phase space
  of rate (N) and rate differential (L)
  dimensionless concentration (Z) and a
  characteristic measure of clustering (B) to
  determine middle-range predictions.
• The algorithm MSc reduces the area of alarm
  outlining such an area where the activity, from
  the beginning of seismic inverse cascade
  recognized by M8, is continuously high and
  infrequently drops for a short time. The
  phenomenon might reflect the narrow-range
  intermittence of the seismic premonitory rise
  near the incipient source of main shock.

51
Performance of the algorithmsresults obtsined
worldwide in the last 20 years(updated at
January 2004)
Prediction made within 170 overlapping circles
of 1333-km diameter. The space-time volume has
been determined by the most conservative measure
that allows for various seismic activity of the
territories.
52
Evaluation of prediction results

• The quality of prediction can be defined by using
  two prediction parameters
• N on/N the rate of failures-to-predict
• T ot/T the rate of alarm times (Molchan,
  1997).
• N is the number of strong earthquakes occurred
  during the time period T covered by prediction
• The alarms cover altogether the time t and they
  have missed n strong events

53
Evaluation of prediction results

• DST
• Molchan, 1997).

The performance of the prediction algorithm is
characterized by its error curve G, which
demonstrates how far from a random guess are the
resulting predictions. To make use of prediction
and optimise its benefits one needs to specify a
cost-benefit function L, i.e. the cost of safety
measures imposed by prediction minus the cost of
the damage which the measures prevent. The point
where G and L touch each other determines both
the minimum loss and the optimal set of
parameters in the prediction algorithm to be used
for prediction.
54
Contingency table for CN predictions
Contingency table for the algorithm CN, for a
time period of 100 years. The yearly base-rate of
earthquakes is the probability of occurrence of a
strong event during one year, and it is obtained
considering the average return period of six
years (condition for the applicability of the
algorithm). The accuracy in prediction of
earthquakes and the percentage of total alarm are
drawn from global results in retrospective and
forward analysis, while the other quantities have
been calculated in the hypothesis of yearly
duration of TIPs. The accuracy here refers to
prediction within a single class of objects
(dangerous or non-dangerous) and gives a measure
of their predictability with algorithm CN. The
conditional probability of predictions is given
by the ratios "true alarms/total alarms" and
"true no alarm/total no alarm".
The yearly base-rate of earthquakes probability
of occurrence of a strong event during one year,
obtained considering the average return period of
six years. The accuracy in prediction of
earthquakes and the percentage of total alarm are
drawn from global results in retrospective and
forward analysis, in the hypothesis of yearly
duration of TIPs. The conditional probability of
predictions is given by the ratios "true
alarms/total alarms" and "true no alarm/total no
alarm".
55
Seismic Roulette
How to estimate the effectiveness of predictions?

• Consider a roulette wheel with as many sectors
  as the number of events in a sample catalog, a
  sector per each event.
• Make your bet according to prediction determine,
  which events are inside area of alarm, and put
  one chip in each of the corresponding sectors.
• Nature turns the wheel.
• If seismic roulette is not perfect
• then systematically you can win! ?
• and lose ?
• If you are smart enough and your predictions are
  effective ------
• the first will outscore the second! ? ? ? ? ? ?
  ? ? ? ?

(V. Kossobokov)
56
Seismic Roulette
(V. Kossobokov)
57
IASPEI Preliminary List of Significant Precursors
vs. M8 and CN

• DST
• The IASPEI Preliminary List of Significant
  Precursors issued on March 1994 (Wyss, 1997), for
  example, includes only five possible precursors,
  out of the forty precursory observations
  submitted until now one is based on ground water
  chemistry (Wakita, 1988), one is a measurement of
  crustal deformation by ground water levels
  (Roeloffs and Quilty, 1997) and three are
  seismicity patterns. The seismic precursors are
  based on foreshocks (several days before the
  event), on preshocks swarms (several months
  before the earthquake) and on the seismic
  quiescences that can be observed before major
  aftershocks. Four out of these five precursors,
  however, are based on a single example only the
  last premonitory pattern, which should allow for
  the prediction of the strong aftershocks, has
  been observed in several cases. These five
  precursory observations seem to deserve further
  study in the framework of earthquake prediction,
  nevertheless none of them can be considered yet
  as a validated precursor.

58
IASPEI Preliminary List of Significant Precursors
vs. M8 and CN

• DST
• The IASPEI Preliminary List of Significant
  Precursors issued on March 1994 (Wyss, 1997), for
  example, includes only five possible precursors,
  out of the forty precursory observations
  submitted until now one is based on ground water
  chemistry (Wakita, 1988), one is a measurement of
  crustal deformation by ground water levels
  (Roeloffs and Quilty, 1997) and three are
  seismicity patterns. The seismic precursors are
  based on foreshocks (several days before the
  event), on preshocks swarms (several months
  before the earthquake) and on the seismic
  quiescences that can be observed before major
  aftershocks. Four out of these five precursors,
  however, are based on a single example only the
  last premonitory pattern, which should allow for
  the prediction of the strong aftershocks, has
  been observed in several cases. These five
  precursory observations seem to deserve further
  study in the framework of earthquake prediction,
  nevertheless none of them can be considered yet
  as a validated precursor.

IASPEI the observed anomaly should be related to
some mechanism leading to earthquakes M8 and CN
the equations to describe the dynamics of the
lithosphere are still missing thus M8 and CN are
algorithms based on empiricism guided by the
concept of deterministic chaos.
59
IASPEI Preliminary List of Significant Precursors
vs. M8 and CN

• DST
• The IASPEI Preliminary List of Significant
  Precursors issued on March 1994 (Wyss, 1997), for
  example, includes only five possible precursors,
  out of the forty precursory observations
  submitted until now one is based on ground water
  chemistry (Wakita, 1988), one is a measurement of
  crustal deformation by ground water levels
  (Roeloffs and Quilty, 1997) and three are
  seismicity patterns. The seismic precursors are
  based on foreshocks (several days before the
  event), on preshocks swarms (several months
  before the earthquake) and on the seismic
  quiescences that can be observed before major
  aftershocks. Four out of these five precursors,
  however, are based on a single example only the
  last premonitory pattern, which should allow for
  the prediction of the strong aftershocks, has
  been observed in several cases. These five
  precursory observations seem to deserve further
  study in the framework of earthquake prediction,
  nevertheless none of them can be considered yet
  as a validated precursor.

IASPEI the anomaly should be simultaneously
observed at more than one site or instrument M8
and CN give statistically significant results at
global scale.
60
IASPEI Preliminary List of Significant Precursors
vs. M8 and CN

• DST
• The IASPEI Preliminary List of Significant
  Precursors issued on March 1994 (Wyss, 1997), for
  example, includes only five possible precursors,
  out of the forty precursory observations
  submitted until now one is based on ground water
  chemistry (Wakita, 1988), one is a measurement of
  crustal deformation by ground water levels
  (Roeloffs and Quilty, 1997) and three are
  seismicity patterns. The seismic precursors are
  based on foreshocks (several days before the
  event), on preshocks swarms (several months
  before the earthquake) and on the seismic
  quiescences that can be observed before major
  aftershocks. Four out of these five precursors,
  however, are based on a single example only the
  last premonitory pattern, which should allow for
  the prediction of the strong aftershocks, has
  been observed in several cases. These five
  precursory observations seem to deserve further
  study in the framework of earthquake prediction,
  nevertheless none of them can be considered yet
  as a validated precursor.

IASPEI the definition of the anomaly and of the
rules for its association with subsequent
earthquakes should be precise M8 and CN are
formalized published algorithms regular
workshops are held at ICTP every second year.
61
IASPEI Preliminary List of Significant Precursors
vs. M8 and CN

• DST
• The IASPEI Preliminary List of Significant
  Precursors issued on March 1994 (Wyss, 1997), for
  example, includes only five possible precursors,
  out of the forty precursory observations
  submitted until now one is based on ground water
  chemistry (Wakita, 1988), one is a measurement of
  crustal deformation by ground water levels
  (Roeloffs and Quilty, 1997) and three are
  seismicity patterns. The seismic precursors are
  based on foreshocks (several days before the
  event), on preshocks swarms (several months
  before the earthquake) and on the seismic
  quiescences that can be observed before major
  aftershocks. Four out of these five precursors,
  however, are based on a single example only the
  last premonitory pattern, which should allow for
  the prediction of the strong aftershocks, has
  been observed in several cases. These five
  precursory observations seem to deserve further
  study in the framework of earthquake prediction,
  nevertheless none of them can be considered yet
  as a validated precursor.

IASPEI both anomaly and rules should be derived
from an independent set of data, than the one for
which the precursory anomaly is claimed M8 and
CN prediction criteria have been defined
globally, using information on past seismicity to
predict new strong earthquakes.
62
IASPEI Preliminary List of Significant Precursors
vs. M8 and CN

• DST
• The IASPEI Preliminary List of Significant
  Precursors issued on March 1994 (Wyss, 1997), for
  example, includes only five possible precursors,
  out of the forty precursory observations
  submitted until now one is based on ground water
  chemistry (Wakita, 1988), one is a measurement of
  crustal deformation by ground water levels
  (Roeloffs and Quilty, 1997) and three are
  seismicity patterns. The seismic precursors are
  based on foreshocks (several days before the
  event), on preshocks swarms (several months
  before the earthquake) and on the seismic
  quiescences that can be observed before major
  aftershocks. Four out of these five precursors,
  however, are based on a single example only the
  last premonitory pattern, which should allow for
  the prediction of the strong aftershocks, has
  been observed in several cases. These five
  precursory observations seem to deserve further
  study in the framework of earthquake prediction,
  nevertheless none of them can be considered yet
  as a validated precursor.

None of the precursors from the IASPEI list has
been subject to forward prediction testing, while
M8 and CN are routinely globally tested.