You are on page 1of 8

NCGT Journal, V. 3, No. 1, March 2015. www.ncgt.

org

21

EARTHQUAKES OCCUR VERY CLOSE TO EITHER 06:00 OR 18:00


LUNAR LOCAL TIME
Giovanni P. GREGORI
IDASC (Istituto di Acustica e Sensoristica O. M. Corbino (CNR) Roma (Italy)
IEVPC (International Earthquake and Volcano Prediction Center) Orlando (Florida, USA)
Giovanni.gregori@idasc.cnr.it; giovanni.gregori@alice.it
Abstract: If an earthquake (EQ) has to occur at some location and on some day, almost always it happens during either
one of two time intervals close either to 06:00 or to 18:00 LLT (lunar local time). This law applies to a98% of case
histories. The procedures are presented that are suited to assess the exact duration of the time lag with a 95% (or higher)
confidence limit.
Keywords: Earthquakes time instant, forecast, lunar local time, crust anisotropy, tidal stress vs. Moons coordinates

Introduction

olvankar (2011) reported about a systematic analysis by studying over 5,000 seismic events with a
magnitude range 2-10, based on the global seismic catalogue NEIC-USGS. He distinguished separate
patterns for different ranges of periods, magnitudes, depths, latitudes and longitudes. He always found a
common law shared by every subset of events, independent of magnitude, depth, latitude, and time, and
displaying only a regular dependence on the longitude M.
He considered the following angles (unless differently stated, all angles are counter-clockwise):
x
SEM (Sun EQs epicenter - Moon)
x
GMT
x
longitude M
x
EMD (EQ epicenter - Moon distance); in addition, let us here also indicate by
x
LT the local time, and by
x
LLT the local lunar time
The definitions of LT and of LLT are analogous to each other, the difference being that the point at Earths
surface with 12:00:00 LT observes the Sun at its maximum elevation, while the point with 12:00:00 LLT
observes the Moon at its maximum elevation above the horizon.
Note that figure 2 in Kolvankar (2011) incorrectly indicates (Kolvankar, private communication, 2015) a
clockwise direction of SEM, while his whole analysis was carried out upon considering counter-clockwise
SEM.
Kolvankar (2011) shows that 98% of all EQs satisfy a linear relationship:
(1)

GMT = EMD + SEM + const

that he shows is satisfied in different regions (one example is here shown in Figure 2), everyone
characterized by approximately the same longitude. Upon considering the 3rd and the last column of table 1
of Kolvankar (2011), it is found that (1) is:
(2)

GMT = EMD + SEM - M

that can be solved (see Figure 1) to get:


(3)

GMT + M = LT = EMD + SEM

(4)

LLT = LT - SEM = EMD

That is, owing to (4), the physical meaning of the empirical relation (2) is that 98% of all EQs happen very
close to LLT = 06 or LLT = 18 (hours of lunar local time).

22

NCGT Journal, V. 3, No. 1, March 2015. www.ncgt.org

This result seems physically plausible as the lunar tidal deformation is maximum at LLT = 12:00 or LLT =
00:00, but the time derivative of the deformation is maximum at LLT = 06:00 or LLT = 18:00, and the
maximum stress on the crust occurs when the time-gradient is maximum of the deformation.

Figure 1. Some relations between angles of time. (a) and (b) are formal definitions, (c) is an empirical evidence. See
text.

Figure 2. "Three earthquake plots for (EMD + SEM) vs. GMT timings for latitude range -35 to -25 and longitude
range -180 to -170 for three different twelve-years period: 1973-1984, 1985-1996, and 1997-2008. The earthquakes
occupy the same strip in these plots and there are no time-dependent variations." Note in the first plot [top left] that
very few events strike along lines roughly perpendicular to the main trend. See text. Figure modified and captions after
Kolvankar (2011).

Among the several Kolvankars (2011) plots, only figure 2 is here reported, which refers to the Kermadec
trench area. Figure 2a [top row, left diagram] is peculiar, as it displays some lesser amount of events that
strike along a few trends roughly perpendicular to the law (2). They are here indicated by a green shading,
while the orange shading denotes tectonic features and active faults located at some comparatively slightly

NCGT Journal, V. 3, No. 1, March 2015. www.ncgt.org

23

different longitude M. But, the anomalous (light green) feature is observed only during some epoch of time,
not during others. This anomaly is probably associated with a temporal activation of some fault. Therefore
this trend is likely to be explained in terms of a temporary change of the local rheology along that fault. But,
this item has to be better inspected in the future upon considering the detailed location of epicentres relative
to their respective local tectonic setting.
The general result (2) has a most important application. Indeed, whenever it will be feasible by some method
to issue an alert for an eventual possible EQ which can be reasonably expected to hit some given area, by (4)
it is possible to know that, during the 24 hours of every given day - and at every given site - the shock can
occur, with 98% certainty, only during two time windows, each one of some very brief total lag. But there is
a need to assess, with an error bar that specifies e.g. at the 95% confidence limit, the exact duration of either
one of these two time lags, which are close to 06:00 or to 18:00 LLT, respectively, during which an EQ
ought to occur.
Kolvankar (private communication, 2015) claims that when he plotted only events referred to some small
area, no similar clear alignment (2) is found. This is the likely consequence of the approximation (Kolvankar,
2011a) he used in his computation of the location of the epicentre with respect to Moons position.
Indeed, suppose e.g. to use a magnifying glass and to expand some small section of either one of
Kolvankars plots (e.g. of figure 2). Plot on a full page this lesser detail of the former large Kolvankars plot,
and use expanded units on both coordinate axes: the points that in the former plot appear approximately
closely aligned along the line (2) will appear badly scattered. That is, the disturbance with respect to the
linear trend - that is originated by the approximations is negligible in Kolvankars plots, while it is no more
negligible in every enlarged lesser fraction of it.
Therefore we have to reduce the scatter of the points in the Kolvankars plots in order to improve the
precision of the determination of the crucial time lag for seismic alert.
This is the purpose of the present short paper, where algorithms and procedures for data handling are
discussed in detail, although no direct application has as yet been implemented.
A precise computation A first improvement can be achieved by carrying out a direct astronomical
computation for every seismic event, which happens with any given magnitude, location and time. That is,
there is no need to separate seismic events that hit different areas.
The computation of the Moons orbit is known, hence also the instant position of the exact point at Earths
surface where the Moon is located right at the zenith (let us call this point the sublunar point, SLP). This
information derives from some lengthy astronomical computation. Programs and subroutines prepared by
professional astronomers are available on the web. Therefore SEM is known (refer to Figure 1b).
Since for every EQ considered alone we know LT and SEM, also LLT=LT-SEM is known as per figure 1b.
Therefore, it is possible to compute the LLT for every EQ. A histogram can thus be drawn that includes all
5,000 (or more) EQ of some world catalogue, independent of epicentre location and time, e.g. with
magnitude above some given threshold etc.
According to the Kolvankars plots, two Gaussian distributions are to be expected in this histogram, centred,
respectively, at 06:00 and at 18:00 LLT. The width of each one if these two Gaussian distributions envisages
its respective rms deviation: the 95% confidence limit for the probability of occurrence of a possible EQ
around either 06:00 or 18:00 LLT is twice this observed rms deviation (a similar standard definition can
provide with the time lag at every larger confidence limit).
This estimate, however, can be even better improved, physically, upon considering the instant latitudinal
location of the Moon.
Indeed, consider that the Moons orbit lies very close to the ecliptic plane (at an angle a5.1), rather than
close to the Earths equatorial plane such as it occurs for the satellite of other planets of the Solar System.
Hence, during one full 24 hour daily rotation of the Earth, the Moons latitude is almost constant, with
respect to the Earths equator, while the SLP runs along one parallel at the Earths surface. Since the tilt of
the Earths equatorial plane is a23.4, the SLP moves inside a latitudinal belt of ar28.5.

24

NCGT Journal, V. 3, No. 1, March 2015. www.ncgt.org

The maximum lunar tidal deformation occurs at the SLP, and also at it antipodal point ASLP, and it displays
a spatial trend that can be approximately considered with cylindrical symmetry around the SLP-ASLP axis.
The deviation from cylindrical symmetry is associated with local tectonic heterogeneities.
The maximum lunar tidal crustal stress, however, occurs where the time derivative is maximum of crustal
deformation. Therefore, it is reasonable to expect that the law (2) and consequently also the width of the
two aforementioned Gaussian distributions displays some dependence on the latitude of the SLP at every
instant of time.
This possible - and also likely - effect can be easily taken into account by computing the angle at Earths
surface between the EQ epicentre and SLP: when this angle is zero, the hypothetical EQ occurs right at
the SLP, and when it is 180 the hypothetical EQ occurs at ASLP.

The computation of this angle is straightforward by means of the algorithm explained in the Appendix: it
deals with the transformation of the usual coordinates (latitude and longitude) on every given spherical
surface of some given and pre-chosen radius.
Consider any couple of spherical coordinate systems of this kind, one with North pole at a point A, and the
()
other with North pole at a point B. Every given point P has coordinates, i.e. latitude and longitude, and
()
()
()
in system "A" and and in system "B", respectively. It is possible to compute the coordinates
in either one system A or B when we know the P coordinates in the other system.
The aforementioned computation for every EQ event needs to start from the usual geographical latitude and
longitude, where the North pole is very close to the Pole Star etc. (we can call this the G system, where G
is for Greenwich). The second coordinate system has pole at the SLP, and can be called system M (for
Moon).
The symbol for the angle derives from the fact that = 90 is a colatitude that, as usual, is defined
as the complementary angle of the latitude , while the epicentre is denoted by E, and M denotes the M
system.
It is thus possible to evaluate the width of either one of the two aforementioned Gaussian distributions, and
to plot this width as a function of .

In principle, an additional eventual refinement is not here considered in detail, as at present it is premature. It
could be associated, perhaps, with the additional role played by the solar tide. It is certainly less important
than the effect of the lunar tide. If also the solar tide plays a role, each aforementioned width of the Gaussian
distribution ought to depend on and, up to some smaller extent, also on LT. But, this possibility has to be
assessed a posteriori, and it could result to be different depending on the local tectonic setting. That is, this
possibility is to be inspected empirically.
By this, for every given location and for every given date, it is possible to specify the precise duration - at
95% (or higher) confidence limit - of the time lag around either 06:00 or 18:00 LLT that constraints the
possible occurrence of an eventual EQ.

Appendix - 3D coordinate transformations in Euclidean space


In the present appendix all angles are in radians. A coordinate system is said to be either Cartesian (also
called rectangular) or spherical. When it is Cartesian, it is right handed, with axes , and , or , , and

. When it is spherical, the coordinates are: r for radial distance; for latitude or = for colatitude

(also called polar distance), being + and 0 ; and in addition longitude, which is here

conventionally assumed to be always reckoned clockwise as seen by an observed located at the "positive" or

"North" pole (i.e. at the pole defined by = or = 0).

Hence, in order to avoid confusion, the term clocklongitude is here used, while "longitude" is supposed

NCGT Journal, V. 3, No. 1, March 2015. www.ncgt.org

25

counterclockwise. In standard geographic coordinates it is customary to use different terms and to distinguish
explicitly west-longitude or east-longitude. In addition, as a simple mnemonic rule, the term local time
refers to a longitude, while hour angle refers to a clocklongitude.
Consider the transformation between any two given coordinate systems.
One of the best known case histories of this kind is concerned with a spherical surface of given radius. This
is an elementary topic. The algorithms of spherical trigonometry are involved, which date back to the
Alexandrinian Greeks (mainly Hypparcus of Nicea [II century BC], Ptolemty [Claudius Ptolemaeus, AD
100-168 or 178], Menelaus [aAD 98]; e.g. Kline, 1972).
The problem is not elementary, maybe, when dealing with rotations between two spherical coordinate
systems, when the classical Euler angles ought to be used, and substantially more involved algorithms are
implied suited to define a more general approach. This can be treated either by matrix algebra (e.g.
Gantmacher, 1959 and 1959a, or Lanczos, 1956 and 1961), or by tensor calculus (e.g. Finzi and Pastori,
1961), or by group theory (e.g. Wigner, 1959, or Namark, 1962, or Gel'fand et al., 1963). All these
algorithms, however, are not here of concern.
Only a short practical "guide" is here given. It refers to easy transformation formulas, whereby two spherical
systems can be transformed into each other simply by knowing three angles: the latitude of each polar
direction referred to the other frame of reference - i.e. their respective latitude, and relative clocklongitudes
[i.e. of pole (2) in system (1), and of pole (1) in system (2)]. That is, only 3 degrees of freedom are required
to define a new coordinate frame (like the 3 Euler's angles). Recall, however, that Russell (1971), instead,
prefers to use matrix formalism.
Therefore, let us simply consider a change of frame of reference over a given spherical surface of fixed
radius. Suppose that we deal with a reference system "1" having North pole at point "(1)" (figures 3 and 4),
and a reference frame "2", having North pole at a point "(2)". Latitude, colatitude, and docklongitude are
called , and in system (1), and , and in system "2 ", respectively. Thus, point "(2)" has in
()
()
()
()
()
system "1 " coordinates , and , while point "(1)" has in system "2 " coordinates , and
()
()
()
()
. Given any arbitrary point P of coordinates and , in system "1 ", and given , and , it
is possible to evaluate and , which are the coordinates of P in system "2".

The formulas for this transformation are as follows. Apply the Euler formula to the spherical triangle (1)(2)P,
and the sinus formula, and get:
(A.1)
(A.2)

()

sin = sin

()

sin + cos

sin =

()

cos cos ( )
()

sin ( )

NCGT Journal, V. 3, No. 1, March 2015. www.ncgt.org

26

Figure 3. Computation of the clocklongitude M2 when 180 d M1 -

Figure 4. Computation of the clocklongitude M2 when 0 d M1 -

M 1(2 )

M 1(2 )

d 360. See text.

d 180. See text.

that solves the problem. This formulation can be found e.g. in Chamberlain (1961, p. 66). A different
procedure was used by McNish (1936) and Chapman and Bartels (1940, p. 646). This same problem was
considered also by Schmidt (1918 and 1926) and by Hunten (1958).

Some care has to be taken when evaluating and from (A.1) and (A.2). In fact, since 0 || , no

ambiguity is raised by (A.1) when evaluating while a somewhat more lengthy analysis is needed for
(A.2). In the case of Figure 3 it is:
(A.3)
the angle at pole (2) is:

()

NCGT Journal, V. 3, No. 1, March 2015. www.ncgt.org


()

27

()

= +( )

(A.4)
or

()

(A.5)
while in the case of figure 4 it is:

0
()

(A.6) )
the angle at pole (2) is:

()

()

= ( )

(A.7)
or

()

(A.8)

()

Consider the case of Figure 3. Since it is known that sin( ) 0 , there is need to assess whether

()
()
()
, or > , or, When it is:

(A.9)

()

()

and when > draw the great circle from P perpendicular to the great circle drawn through

points (1) and (2) and consider the arc M shown in figure 3. It is:
()

tan = cot cos( )

(A.10)
being:

(A.11)
If

(A.12)

>

(A.13)

()

it is

()

it is

()

()

>

This method works as long as the perpendicular is uniquely drawn from P to the great circle through points
(1) and (2). This always occurs except when the triangle (1)(2)P has three right angles. In this case it can be
easily shown that:
(A.14)

()

()

Consider the case of figure 4. From (A.8) it follows:


()

(A.15)
(A.16)

when

()


()

and when

<

it is

()

consider again the arc M given by (A.10) and (A.11). It is:

NCGT Journal, V. 3, No. 1, March 2015. www.ncgt.org

28

(A.17)

when

(A.18)

when

>

that fully solves the problem.

()

it is

()

it is

()

0
()

But, one can also more simply apply the Euler formula to the spherical triangle (1)(2)P in either figure 3 or
figure 4, and get:
(A.19)

()

cos =

()


()

Note that, while implementing the computing software, a well-known practical rule, which is suited to avoid
the uncertainty while evaluating an angle from its arctangent, is to refer to both its known values of sine
and cosine.
References
Chamberlain, Joseph W., 1961. Physics of the aurora and airglow. 704 pp., Academic Press, New York.
Chapman, Sydney, and Julius Bartels, 1940. Geomagnetism. 2 vol., 1049 pp., Oxford Univ. Press, (Clarendon), London
and New York.
Finzi, Bruno, and Maria Pastori, 1951 and 1961. Calcolo tensoriale ed applicazioni, (I ed., 1951), 427 pp. (II ed., 1961),
510 pp., Zanichelli Editore, Bologna.
Gantmacher, F. R., 1959. The theory of matrices. Vol. 1, 374 pp., and vol. 2, 276 pp., translated by K. A. Hirsch,
Chelsea Publ. Co., New York.
Gantmacher, F. R., 1959a. Applications of the theory of matrices. 317 pp., translated by J. L. Brenner, D. W. Bushaw
and S. Evanusa, Interscience Publishers, Inc., New York etc.
Gel'fand, I. M., R. A. Minlos, and Z. Ya. Shapiro, 1958, 1963. Representations of the rotation and Lorentz groups and
their applications, 366 pp., Pergamon Press, Oxford, etc. First published in Russian by Fizmatgiz in 1958.
Hunten, Donald M., 1958. A nomogram for solving spherical triangles, Scientific Report No. Br-8, Phys. Dept. Univ.
Saskachewan, Saskatoon, Sask., Canada.
Kline, Morris, 1972. Mathematical thought from ancient to modern times. 1238 pp., Oxford Univ. Press, Oxford and
New York.
Kolvankar, Vinayak G., 2011. Sun, Moon and earthquakes, New Conc. Global Tect. Newslett., no. 60, p. 50-66.
Kolvankar, Vinayak G., 2011a. Methodology to check correlation between Earth tide and earthquakes, New Conc.
Global Tect. Newslett., no. 61, p. 108-111.
Lanczos, C., 1956. Applied analysis. 539 pp., Prentice Hall, Inc., Englewood Cliffs, N.J. (USA).
Lanczos, C., 1961. Linear differential operators. 564 pp., Van Nostrand Company Limited, London etc.
McNish, A. G., 1936. Geomagnetic coordinates for the entire Earth, Terr. Magn. Atmos. Electr., v. 41, no. 1, p.3743;
doi: 10.1029/TE041i001p00037.
Namark, M. A., 1962. Les reprsentations linaires du groupe de Lorentz, 371 pp., Dunod, Paris. Originally published
in Russian by Les Editions d'Etat de Litrature physico-mathmatique, Moscow.
Russell, Christofer T., 1971. Geophysical coordinate transformations, Cosmic Electrodynamics, 2, 184-196.
Schmidt, Adolf, 1903-1928. Archiv des Erdmagnetismus (Herausgegeben von A. Schmidt). Heft 1-4 (1903-1926)
contained in Band I. Heft 5, 6 and 7 appeared in Abhandlungen des Preuss. Meteorol. Inst. Berlin, J. Springer,
Berlin: Heft 5 as vol. 8, no. 2 (Veroeff. Nr. 332, 1925), Heft 6 as vol. 8, no. 11 (Veroeff. Nr. 354, 1927), Heft 7 as
vol. 9, no. 1 (Veroeff. Nr. 357, 1928).
Schmidt, Adolf, 1918. Geomagnetische Koordinaten. In Schmidt (1903-1928), Heft 5 (or 3 ?), p. 14-.
Schmidt, Adolf, 1926. [Tables of the geomagnetic coordinates for various observatories]. In Schmidt (1903-1928), Heft
4, pp. 35-38.
Wigner, Eugene P., 1959. Group theory and its applications to the quantum mechanics of atomic spectra, 372 pp.,
Academic Press, New York etc.

You might also like