Editor: Jim Berkland
1175 Chauvet Rd. Box 1926
Glen Ellen, CA. 95442
(707) 935-6512 ::: FAX (707) 935-6639
12 Monthly Issues
Back issues $4.00
$1.49 per minute
QuakeLine is for S.F. Bay Area Only
Let's take the gloves off! Wow, that sounds tough, but why is it any more belligerent than the challenge, "Let's put the gloves on!", which usually precedes a boxing battle. In this case it is a battle of statistics and of statisticians, of which I am not one, thank God.
In past issues of SYZYGY I have pointed out that more than 300 scientific papers in the world literature have dealt with the statistical correlations of tides and quakes. Some say categorically that there is a correlation, and others are equally vehement in maintaining that tidal influences have nothing to do with earthquakes. One only need check with the computer research mechanisms called GeoRef and Dialog in order to obtain the lists of papers and gain further insight into the question. However, if you hear (as I have on so many occasions) some representative of High Science maintain that "No one has shown any evidence of such a correlation" it merely means that the expert is unaware of the wealth of positive literature on the subject, and would prefer not to be informed.
Any examination of such literature will reveal the papers of the eminent seismologist Dr. Leon Knopoff (1964) and Knopoff and Kilston (1983) in which Dr. Knopoff changes from nay to yea in regard to tide-quake correlations. Indeed in his later article with Astronomer Steve Kilston, based upon lunar orbital data alone, he made a prediction for a 6+M quake to strike the San Andreas system in Southern California around November 1987 at the next nodal point of the Moon, especially near dawn or dusk. He had a near miss with the 5.9M Whitter Quake of October 1, 1987, and then scored direct hits with the pair of Superstition Hills earthquakes (6.3 and 6.8M) on November 23-24, 1987. The reaction of Dr. Knopoff at this remarkable success was surprising. He was quoted as saying, "I have just spent the last four years trying to convince my colleagues that I hadn't really made an earthquake prediction, when the damned thing hit." Astronomer Kilston had no such reservations, but felt personally vindicated by the 1987 successes, based upon a talk I heard him deliver to the San Jose Astronomical Society early in 1989. (He also said that, at that time, members of the U.S. Geological Survey were failing to return his calls.)
Also recorded in the literature is a 180 degree change of position by Thomas Heaton, who, as a CalTech graduate student in seismology (in 1975) pointed out a high degree of statistical correlation between tidal forces and quakes in Southern California; however, in a paper seven years later, when he was working as a seismologist with the U.S. Geological Survey, he formally apologized to his colleagues and said he hadn't really understood the results of his earlier statistical analysis. His new insight into the tide/quake problem revealed no correlation at all.
Such flip-flops by brilliant researchers have added to my misgivings about relying upon statistics in attempts to establish "ground truth." I have the 1979 statement from Roger Hunter, of the U.S. Geological Survey in Golden, Colorado, that a computer analysis of my first five years of earthquake predictions (1974-79) had revealed a statistical correlation of 99 percentile, meaning that there was one chance in one hundred that my record was coincidental. Nevertheless, Mr. Hunter felt that I had been "lucky", and if disclosed these results to the media, he would deny it. I kept his secret until I read his final written summary of his four-year study in which he indicated that, "As yet no one had met the statistical test for valid predictions", and ".... no scientist had even bothered to submit any predictions." I knew he was wrong in both of those statements, but can only guess as to why he made them.
I could describe a number of similar problems with statistics and statisticians, but you already know what they say about them. From my own standpoint, I feel comfortable with the following facts: A summary given to me by the U.S.G.S. of all of the +2.5M quakes occurring between 1963-77, and centered within a 70-mile radius of San Jose showed one +3.0M event every 18 days and one +3.5M every 35 days. Thus my Seismic Window of only eight days has less than a one-in-four random chance of happening for a minimum 3.5M quake. In my first year of prediction (1974) I scored six out of eight (75%) and have maintained at least that level of accuracy for anticipating local quakes ever since. For more than twenty years my record has exceeded chance by 300 per cent.
With that update let me present a letter that recently arrived from a critic, journalist Lee Siegel, whom I admire, and with whom I have communicated for several years:
As you know, I have been an interested reader of Syzygy for several years, although I remain very skeptical of your claims of success in predicting quakes.
While I am certainly open to the possibility that tidal forces play a role in quakes, I have yet to seen any convincing evidence.
Your September 1994 issue contained a perfect example of why I---and what you often refer to as "high science" ---remain skeptical of your claims. It disturbed me enough that I felt compelled to write you.
On page 5, discussing August's quakes, you stated:
"In the Los Angeles area the requisite 3.5M was exceeded only by a Simi Valley event of 3.6M just four minutes before my August window opened at midnight on August 19th. Lucerne Valley shook with a 3.4M tremor on August 21st, and there was one of 3.1M near Barstow on August 26th.
I would have to rate myself 98% correct on that Simi Valley shaker."
Jim, from a scientific and statistical standpoint, you were 0% correct in your prediction not 98%. The Simi Valley quake, whether 4 minutes or 4 hours "early," was outside your prediction window and thus the prediction failed.
With a week-long window each month, your predictions aren't very specific to start with and are likely to be correct a certain amount of the time by random chance alone simply because your window is so long and California has so many quakes. In addition, from reading your newsletter so many years, it is clear to me that you often consider your predictions correct or almost correct if an event in your window is almost within the predicted magnitude range, or if a quake of predicted magnitude range happens a short time before or after your window, or if a quake within the predicted window and magnitude range happens nearby but not quite within the predicted geographic area.
Forgive my bluntness, but this isn't science. It's cooking the numbers retroactively to support your theory.
Jim, I hope you know me well enough to know I'm no devotee of "high science" nor am I closed to new ideas. And I do enjoy reading the anecdotes and history in Syzygy. But if you ever hope to convince me or anyone else who has the slightest bit of knowledge of statistics, you need to be more forthright and accurate in tabulating when your predictions succeed and fail.
P.S. Jim, should you choose to publish this letter (and I hope you will, in the interest of fair debate), here's some background: I am currently science editor at The Salt Lake Tribune, a position I took in 1993 after deciding to escape Los Angeles, where I spent a decade covering seismology as a science writer for The Associated Press. I now spend a fair amount time trying to wake up Utah about the quake threat posed by the Wasatch Fault. I am speaking for myself only in the letter above.
Thank you Lee, for your frankness and clarity. Your letter epitomizes much of the double (or triple) standards that exist in the field of earthquake prediction. Where, oh where, do you find scientifically (and politically) correct earthquake predictions? Many knowledgeable authorities point to the February 4, 1975 earthquake in Haicheng, China, where the authorities evacuated most of the residents less than 24 hours before the 7.3M quake destroyed the city. However, the strength had been expected to reach 5.5-6.0 magnitude, not the major level having hundreds of times greater energy. If a correct prediction includes time, place, and magnitude, the Chinese missed it by a wide range.
However, it was far more important that thousands of lives were spared by forewarning and action.
Lee, if you can believe the U.S. Geological Survey, they forecasted the Loma Prieta Earthquake. However, if you closely examine their data, they had similarly forecasted quakes for various segments along the San Andreas system, so wherever one hit, they could claim it. For the Loma Prieta area their confidence was given as "D" (or lowest level), and their expected magnitude was no more than 6.5M, or one-eighth the energy of the actual 7.1 Richter reading. Also, they assigned it a probability of 30% within 30 years (and you consider that my 8-day windows are too long). Anytime a weatherman or an earthquake forecaster assigns 30% for something to happen, that means that the probability is more than twice as great that it won't occur. Since the Loma Prieta Quake did happen, they blew it. I should also point out the 1977 book, EARTHQUAKES by Don DeNevi, with a forward by Peter L. Ward, Seismologist with the U.S.G.S. This book recounts a prediction by James Whitcomb of CalTech for a 6M event within three months in the Mojave Desert. DeNevi called this successful when one of 3.8M struck before the deadline. Also rated correct was a prediction by Columbia University scientists of a 2.5-3.0M quake in 1973 near Blue Mountain Lake, in northern New York. The so-called success was one of 2.5M even though it was part of a prolonged swarm and there had already been quakes there of more than 3.5M.
Finally I have personal knowledge of following event: "In a third try not long afterward, Menlo Park scientists monitoring the Survey's network of instruments south of Hollister, California, informally predicted a magnitude 5.2 shock that correctly predicted the time of occurrence, but was slightly off in location. (emphasis added) Yet this was the major American achievement........The 75 or so geologists and guests could hardly believe their ears. They had assembled for an evening of socializing and routine gossip about faults, core samples, and volcanoes. Instead, they were hearing scientific history in the making..... The next afternoon on November 28, 1974, while residents of Hollister were sitting down to their Thanksgiving Day dinner, the earth began to sway and rumble beneath them. The brief 2 to 3 second quake measured 5.2 magnitude and did little damage. But its impact still reverberates through the world of seismology. The accurate forecast of the Hollister earthquake was an astounding demonstration that scientists are on the verge of being able to predict the time, place, and even the size of earthquakes."
A press release by the U.S.G.S. following the quake essentially outlined the same details. However, I attended that meeting of the Pick and Hammer Club with the head of the Geology Department of San Jose State University. The topic was clearly defined as dealing with earthquake prediction, but NO PREDICTION WAS MADE TO THE GROUP AT THAT MEETING!
Despite the overly-dramatic description of that meeting in various magazines, books and reports, there was no prediction, except my written one, predicting a South Bay quake of 3.5-5.5M within one week. I was not allowed to present my data to the group, although photocopies of my forecast were accepted.
Furthermore, Lee, have you forgotten Parkfield so soon? The U.S.G.S. presented their data in 1984 and it was accepted as the only bona fide earthquake prediction by the National Earthquake Prediction Evaluation Committee (NEPEC), with a 95% probability for a 6M event by the end of 1992. All agree that it was an utter failure, although quakes of 3 and 4 magnitude did occur near Parkfield during their 8-year time frame.
I stand by my 98% evaluation for my Los Angeles area predictions for August 19-26, 1989. I am better at pitching horseshoes than at kicking field goals, and close counts for something. Anyone who was alerted to my forecast and was awakened by a shake four minutes before the midnight opening of my "window", would not have been too surprised. Similarly, on January 16, 1994 my Seismic Window closed at midnight, and the horrific Northridge quake struck 4 1/2 hours later. Was that a total miss for me? Not according to many appreciative readers of SYZYGY. Also the 3.4M event of August 21st at Lucerne Valley may well end up as a 3.5 or 3.6 magnitude reading when the final refinements are made for the annual listing of Southern California quakes. Then the discussion of the 4-minute miss at Simi Valley would be moot, as it only takes one hit during my eight-day window for a successful prediction. Consider also that if a quake is reported as 3.5-3.8M by several sources and is later lowered to 3.4M by the U.S.G.S., does that invalidate my successful prediction? Obviously, I do not worship statistics; in fact, I rarely respect them.
On to more pleasant subjects. Following my lecture to the Mensa Asilomar Gathering of September 3-5, 1994, I received a letter from Norma Pezzini, Program Chair. She wrote,
"Your talk was the talk of the weekend. I heard you had a captured audience until 10: PM."
How nice it is to hear such a comment. I do know that my talk was scheduled for 7:30-8:30 PM, and I was still answering questions after 10. (Most of the audience of about 300 stayed to the end, and needless to say, if you want intelligent questions, speak to Mensans.)
JEST FOR FUN
I thought of the following bit of nonsense some years ago, but I would not be surprised if you have come across something like it elsewhere.
( It is so hard to be original.)
Whatever you say about Rocko, you must admit that he is a man of his convictions!
Yes, that's true......................
Arson, armed robbery, embezzlement, earthquake prediction.........
QUOTE OF THE MONTH
"At the present level of technology, earthquakes cannot be prevented.
Nor for many years to come can such events be effectively and economically predicted in the detail required to safeguard the national economy of a country........Earthquake prediction was a continuous preoccupation for the early soothsayer, astrologer, or prophet, and there are many recorded instances of the forecast of earthquakes. Earthquake prediction is also increasingly practised by some modern seismologists. Neither class of prophet is particularly effective, even when accurate, as they are invariable greeted with skepticism by people who seem strangely reluctant to believe that an earthquake will occur, whether it is forecast by a credited astrologer or by a seismologist. Today earthquake prediction on a scientific basis is making slow but steady progress. At this time, however, there is little that enables the prediction of the timing of earthquakes with any significant degree of certainty, and it is a point of contention among scientists and engineers as to whether an earthquake prediction and warning system can be developed. The question is not only whether the accurate place, time, and magnitude of an event can be predicted, but whether warning the people is likely to be effective......" N.N. Ambraseys, Engineering Seismologist, Dept. of Engineering, Imperial College in the University of London, U.K.
(In: Earthquakes and Volcanoes; U.S. Geological Survey, 1990, Vol. 22, No. 5. p. 204-205.)
(Pet of the Month)
Let us return to the DeNevi book on Earthquakes (op. cit.) for some more monkey business. Under the subheading "Chimpanzees Predict Earthquakes", DeNevi describes the work of Dr. Helena Kraemer, Assoc. Prof. of biostatistics in the Department of Psychiatry and Behavioral Sciences of Stanford University. In 1975 at the primate center near the San Andreas fault, their 15 chimps were observed to show changes in their behavior. Dr. Kraemer reported in October 1976 about happenings of the previous summer:
The animals were more restless than usual. They spent more time on the ground than high in their climbing structures and nesting area.
Their behavior change was so significant it seemed unlikely it was due to change. At the time we recorded animal behavior, we were doing unrelated research. And, a little later, we were unaware that any tremors had occurred.
At that time there was a swarm of 25 small earthquakes nearby, measuring from 1.5-3.1M between June 19-24, 1975 and the most unusual chimp behavior occurred on the day prior to two of the strongest tremors. One of the supporters of these data was Stanford geophysicist, Amos Nur, who had previously conferred with Chinese scientists following the predicted Haicheng earthquake of February 4, 1975. DeNevi reported that recognition of anomalous animal behavior, according to Nur, ..."will comprise the most important single effort in earthquake prediction during the next few years."
I recall newspaper articles covering the announcement and that there was confusion about what earthquakes the chimps had responded to. The articles in October 1976 indicated that the earthquakes had occurred during the previous summer, rather than the previous year. I called the Primate Lab to make sure that they were aware that the earthquake swarm was in June 1975, not June 1976. However, the spokesperson for the Center seemed oblivious of the ambiguity and I received a polite but terse "Thank you." I am such a busybody.
QUAKE OF THE MONTH
Was this the Great October quake that wasn't there? Almost all of the many lists of destructive international earthquakes that I have found include the 1737 Calcutta quake in which 300,000 Indians met their death.
Such a toll was exceeded only by the Shensi, China earthquake of 1556 (830,000+) and the Tangshan, China quake of 1976 (640,000, later politically corrected by the Chinese government to 242,000 dead.)
In October, 1737 some 300,000 persons died in an Indian catastrophe, but the Calcutta quake may not have happened. Part of the mystery lies in the compendium by Jay Robert Nash, DARKEST HOURS (1976, Publ. by Nelson-Hall, Chicago, 812 p.) Under Major Earthquakes Nash lists:
"1737, Oct. 11: Calcutta, India; Quake ruined mass area; 300,000 deaths."
Also in Nash's book under Major Storms he lists:
"1737, Oct. 7: Bay of Bengal, India; Cyclone created monstrous storm wave of 40 ft., 20,000 ships destroyed at mouth of Hooghly River; 300,000 deaths."
This year at the annual meeting of the Seismological Society of America in Pasadena, California. Roger Billham revealed that careful research has shown no evidence for the quake, but that the destructive cyclone was real. (A disastrous Earthquake that Did Not Occur---Calcutta, 1737: Seismological Research Letters, Jan-Mar, 1994; vol. 65, No. 1, p. 66) Perhaps early investigators assumed that the 40 foot wave was a true tsunami and that would have required a major earthquake. However, there was no explanation for the four-day discrepancy in dates, but we know that communication was difficult and confusion was undoubtedly as widespread as the cyclone. There was no geographic problem as Calcutta does lie at the head of the Bay of Bengal.
LAST MONTH'S QUAKES
The Seismic Window for September 5-12, 1994 was statistically and actually successful.
My prediction for a 3.5-6.0M event within 70 miles of San Jose was met by a 4.1M surprise on September 7th near Interstate 5, about 35 miles due east of this city. It was a real shocker for the Modesto-Patterson area, as they are not used to such things. It may have been on the Vernalis Fault, or a branch of the Tesla-Ortigalita fault, that parallels Highway 5. So far this year I have predicted all of the four local quakes that have exceeded 4.0 in magnitude. Who else can make that statement and verify it in print? To summarize, there was a Watsonville quake of 4.3M on January 11th (New Moon +0 days), a Pinnacles quake of 4.0M on May 19th (New Moon eclipse +9 days), and a 4.0M Tres Pinos shaker on August 27th (NM+6.)
I had also called for a +3.5M quake within 140 miles of Los Angeles and that prediction was fulfilled by one of 3.6M near Hemet on September 11th, neatly meeting the time and space limits. A similar prediction for Seattle was also completely satisfied by a rare 4.1M quake east of Tacoma on September 10th, the first in that area in several years.
I must admit my timing was off a bit for the 7.2M major quake southwest of Eureka on September 1st. It hit four days before my September "Window" opened, but it was the only major quake anywhere in the world between July 21 and today (Sept. 26, 1994,) so it was a least in the pit, if I were pitching horseshoes.
Also my non-scoring MOSS prediction (Monthly Outright Seismic Speculation) for a +5.5M to hit Central America during the first half of September was at least interesting. On September 16th, northernmost Columbia was rocked by a 6.0M tremor, and I am sure it was felt strongly in Panama (Historically, if it weren't Teddy Roosevelt's "need" for a Panama Canal, that country would still be part of Columbia.)
Furthermore, let me not overlook the series of shockers east of Lake Tahoe, with the mainshock of 6.1-6.3M hitting at 05:23 a.m. PDT on the last day of my "Window" on September 12th. This quake was clearly felt from Fresno to Chico to San Francisco to Reno and, except for the earlier Eureka event, would have constituted the strongest quake in California in more than a year. I hope that my October record for predictions will match those for September.
The Seismic Window of October 4-11, 1994 looks to be quite potent for the Bay Area, even if we were not approaching the fifth anniversary of the World Series Earthquake. (It is now certain that we will not have such an event again this year, as the World Series has been cancelled for lack of interest and lack of principle. ($$) However, the quake may not be cancelled. The Mogi Doughnut of encircling epicenters has been defined again and again with action to the north, east and south, leaving us ominously quiet. We should all remember that October ranks in the "Big Three" for months with damaging Bay Area quakes and several other factors are closing in. It seems that a possible physical connection has now been established between faults in the East Bay and West Bay, enhancing my 1990 projection for a large East Bay "echo quake" within six years. (Others have referred to this pattern of earthquakes as "pairings," but no matter what you call it, history bears it out.)
Between October 4-11, 1994 I predict, with 80% confidence that there will be: (1) An earthquake of 3.5-6.0M within 70 miles of San Jose; (2) A similar quake within 140 miles of Los Angeles; (3) A similar quake within 140 miles of Seattle; and (4) A major quake of at least 7.0M somewhere in the world, probably within the Pacific Ring of Fire.
I would like to add that last month I ignored Salt Lake City in this section, and an unusual swarm of earthquakes up to 3.7M struck on September 9-10th, east of Price, Utah, within my usual 140 mile range. (That will teach me to overlook the Mormon's, who really emphasize earthquake preparedness.)
In addition for this month I have a MOSS prediction (Monthly Outright Seismic Speculation) for a quake of at least 5.0M to strike Africa during the first half of October.
(A MOSS may involve rolling stones??)