Earthquake Prediction Model

Posted: January 9, 2008

Presented at the NCGR (National Council for Geocosmic Research) Research Symposium

National Conference: Baltimore, Maryland, March 2007


 The objective of this research was to analyze and investigate correlations between astronomical data and earthquakes, with the intended goal of predicting future earthquakes with a greater advanced warning and higher degree of accuracy than current technology. Specifically, it focuses on severe earthquakes that occurred during the last century, with special emphasis on earthquakes of magnitude 7 or higher.  This research work shows a strong correlation between certain inter-planetary configurations (encompassing the relative geocentric positions and angles of all planets) and the occurrence of strong earthquakes.  However, further research is necessary to build a useful, predictive model that can assess the probability of a given earthquake occurring during a certain time period at a given geographical location on earth.  Predicting earthquakes well in advance of the state of the art will promote, protect, and enhance the world economy, potentially saving millions of lives.


There is absolutely no precedent in predicting an earthquake solely based on planetary configuration. An occurrence of an earthquake is a random event and it can sometimes occur more frequently than other times.  This research began with the idea that planetary positions along the ecliptic, and therefore, their apparent positions as viewed from earth, may potentially correlate with the occurrence of earthquakes. Based on planetary characteristics and a large amount of earthquake data, several hypotheses were tested to see if these correlations actually exist.  The results of this exercise indicate that certain planetary configurations seem to correlate reasonably well with earthquakes. Although the present state of research is primitive, the intent of this paper is to highlight the initial findings on prediction of earthquakes.

Although this paper focuses on earthquake prediction model, since 1993, I have been studying the influence of planetary configurations on natural calamities in general. Starting in 2000, these predictions have been made available to the public on a monthly basis at my website:  While further research is warranted to include the place and type of natural disaster in the predictions, the time periods for the occurrences of natural disasters have been predicted in my monthly columns.

Beginning in 2006, my research on the natural calamities was more focused on the occurrence of earthquakes.  One reason for this was the availability of accurate data on earthquakes from National Earthquake Information Center, United States Geological Survey. 

As a starting point, I chose the top 100 earthquakes by magnitude and their occurrences with corresponding planetary positions using standard statistical techniques. The slow moving outer planets from Mars to Pluto, and the North and South lunar nodes seemed to significantly influence the earthquake occurrences. By refining the method, I obtained better correlations with unique planetary configurations.

  Research Basis – Methodology

As pointed out earlier the bases for this research are the unique planetary positions (longitude measured along the ecliptic) surrounding the earth.  Astronomical data provides planetary positions as a function of time.  I have observed that the geocentric angles of certain magnitudes between some pairs of planets with respect to the earth appear to correlate well with earthquakes.  Correlations between earthquakes of the past and the corresponding planetary angles during those respective periods occur in a statistically significant way.  

These correlations reveal that when increasing number of geocentric angles- when they occur as conjunctions (zero degrees) and in multiples of fifteen degrees all the way to oppositions (180 degrees), the probability of an earthquake becomes greater. In addition, the larger the number of some of these angles, specifically: 180, 165, 150, 135, 105, 90, 75, 45, 30, 15, and zero degrees, the higher the probability of earthquake severity.  

Thus, while oppositions, quincunxes, squares, and sometimes conjunctions between certain planets seem to correlate well with the occurrence of earthquakes, trines, sextiles and conjunctions between specific planets seem to correlate with periods during which earthquakes are least likely to occur. 

The Model  

The model is tested using a simple linear regression technique.  Because every variable influences the earthquake in a specific way, all variables are weighed differently. Thus, in theory there are 55 different pairs of planets (6 outer, 2 inner, Sun, Moon and the North lunar node) and 13 distinct angles (from 0 degrees to 180 in multiples of 15), making a total of 687 maximum possible unique variables that can influence the earthquake occurrence. (Note that the maximum angles between Venus and Mercury, Venus and Sun, and Mercury and Sun are 73.5, 47 and 27 degrees respectively). However, since the Moon’s average daily variation is about 12 to 13 degrees it can form angles with all other planets during a twenty-four hour period of every day. Therefore, the influence of the Moon is assumed to be equal for everyday and is not included in the model.  Since the daily planetary variations on average (excluding the Moon) are with in the orb of one degree or less, the assumption of Moon’s exclusion then allows for Greenwich noontime data to be employed for the Greenwich date when the earthquakes occurred  

Earthquake data of magnitude 7 and higher for the last hundred years were downloaded from the USGS website:  I then computed all corresponding planetary positions and angles. Using an orb of one degree the planetary data pertaining to angles from zero, 15, and multiples of fifteen up to 180 degrees were extracted for all 45 planetary angle pairs. Thus, there are 557unique variables.  A linear model is assumed.  


Earthquake Magnitude = S Cn * (angle pair)n   + constant         for n =1 to 557  

where Cn is the coefficient of the nth pair.  

All the coefficients were estimated by generalized least squares.  A number of coefficients were so small in magnitude that their influence on the model were deemed negligible.  The corresponding variables were omitted one at a time and the regression was repeated to confirm that their influence on the model indeed was negligible.  

After running several cases, four to five models with between 120 to 420 variables of interest were obtained. Using Greenwich noontime daily planetary positions, each model was then used to predict the earthquakes in the year 2006 and 2007.  The predicted results and the actual dates on which earthquakes occurred are summarized in the following Table-1. Predictions for every month were posted at my website: on the last day of the preceding month.  


Earthquake Predictions since January 2006 of magnitude 7 or higher


Prediction Dates

Actual Dates


January 2006

1-5, 15-21, 25-26

2(7.3, 7.1), 27(7.7)

February 2006

1, 5, 22-28.

22 (7.5)

March 2006

1, 10-12, 24-27


April 2006


7(6.3), 20(7.7)

May 2006

2-4, 7-8, 18

3(8), 16(7.4)

June 2006

19-21, 26-28, 18, 1, 6-7, 10-11, 16, 29, 12, 23-24, 30.

None (failed miserably)

July 2006

20, 22, 26-31

17 (7.7)

August 2006

1-6, 27, 29-31.


September 2006

1-2, 4-6, 15-16, 28

1(6.8), 16(6), 28(6.7)

October 2006

10-11, 13, 15-16, 22-30

10(6), 13(6), 15(6.3), 16(6.5), 22(6.1), 23(6.1)

November 2006

12-14, 17, 23-25, 28

12 (6.7), 15 (7.8, 8.3), 28 (6.1)

December 2006

2-3, 5, 12-13, 17-22, 24-25

1(6.3), 22(6.1), 27(7)

January 2007

8, 12-14, 22, 26, 28

8 (6.2, 6.1), 12 (7.7), 13 (8.2), 22 (6.2), 28 (6), 30 (6.7)

The first two columns in Table –1, list months and the prediction dates for earthquakes of magnitude 7 or higher for the corresponding months. The last column lists the dates on which earthquakes occurred with magnitude shown in the parentheses. 

As shown in Table-1, since January 2006 there were eleven earthquakes of magnitude 7 and higher. Out of these eleven, five were accurately predicted; four missed their prediction by a day, and one by two days. 

The number days predicted for months beginning January 2006 through January 2007 are 13, 9, 7, 3, 6, 18, 8, 10, 8, 14, 8, 9, 13 and 7 respectively.  Discarding the month of June during which the model failed miserably the average monthly predicted dates were about 10.  In particular, the model did progressively well after September 2006.  

Observing the data closely, it is striking that at least four outer planets and the lunar North node form angles that are oppositions, conjunctions, squares, quincunxes, semi-squares or semi-quincunxes. The inner planets and the Sun also can contribute to earthquake occurrences by forming similar angles and enhance the probability of the earthquake severity, but the presence of at least four outer planets and the North lunar node seem to be a necessary condition for an occurrence of earthquake of magnitude 7 or higher. 

Refining the model increased the accuracy of predictions for earthquakes of magnitude six and higher, as demonstrated in Table 1.  Clearly, for the model to be applied for earthquakes of magnitude 7 and higher would require further improvement and therefore, more research work is warranted.  In addition, further research is necessary regarding the locations of earthquakes.




3            “It’s All in Timing,”

Maheshri, J.C., Noble House, Baltimore, Maryland 1997.     


 In addition to natal chart research and interpretation Dr. Maheshri has been researching mundane astrology and provides monthly global predictions at his website: since year 2000. Among other things, he has successfully predicted the presidential victory for George Bush for both terms. His current research interests are weather and earthquake predictions.

As a professional astrologer with over 35 years of practice, Dr. Maheshri teaches Vedic astrology, conducts workshops and seminars. Currently, he is the president of NCGR Gulf Coast Chapter. He could be reached at 281-238-8992 or 832-368-8702 (cell) or email: