Originally appeared in Forum, EOS, TRANSACTIONS, AMERICAN GEOPHYSICAL UNION

VOLUME 79 NUMBER 39 SEPTEMBER 29,1998

A Misuse of Public Funds: U.N. Support for Geomagnetic Forecasting of Earthquakes and Meteorological Disasters

The legitimate scientific community needs to be alerted to the expenditure of considerable public funds for pseudoscientific projects that build false hopes of protection from geophysical hazards. One example of such a project is a recently published, 147-page United Nations (U.N.) document titled "Manual on the Forecasting of Natural Disasters: Geomagnetic Methods." Although the document has been distributed internationally, it seems not to have been subjected to pre-publication, international peer review.

The manual was prepared for a U.N. international workshop recently held in Beijing, China, from February 12 to 18,1998. The meeting was sponsored by the Chinese Academy of Sciences, the State Seismological Bureau of China, the U.N. Global Programme for the Integration of Public Administration and Science of Disasters in the Department for Economic and Social Affairs (DESA), and the U.N. Development Programme (UNDP) in Beijing.

The manual, published by the U.N. DESA (copies available from col@un.org), carries a U.N. endorsement that creates undeserved credibility for a program wholly lacking in scientific merit. The preface is written by JeanneMarie Col of DESA and Arthur N. Holcombe of UNDP. The chapters that follow are authored by X. Zeng, Y. Lin, C. Xu, M. Zhao, and Y. Zhao. An appendix is provided by J.-M. Col and J. J. Chu.

The manual mainly focuses on the prediction of earthquakes but includes forecasts of meteorological disasters." Often the pseudoscientific disaster forecasts mentioned are undocumented, illusionary, or focus upon a single fortuitous event, making it difficult to tie down specifics that can be clearly analyzed (see the U.N. Web page http://www.globalwatch.org/ungp/).

Often in such cases the public is left with the question of whom to believe, the prediction proponent or the attacker. Usually the attacker must apply scientific testing methods to expose the unreliability of such hazard forecasts [Evans, 1997]. Media reports favor the predictor [e.g., Li and Mervis, 1996] because he promises the simple joy of disaster relief (as opposed to the complicated mathematics of the attacking scientist). Fortunately, this U.N. manual provides the rare opportunity to dissect and expose the fuzzy thinking at the foundation of this natural disaster prediction cult.

Document Features

Chapter I sets the stage for the forecasting efforts with the thought that because geomagnetic field variations induce currents in the conducting Earth, the measured surface fields can detect those hypothetical regional changes that precede earthquakes. Mathematical formulations appear in Chapters I and 6; these are essentially copies of standard induction transfer functions and elementary formulae for data averaging, curve fitting, standard deviations, Fourier analysis, etc., with vague references to magnetotelluric (MT) and geomagnetic depth sounding (GDS) induction methods. None of the formulae depict disaster forecasts. Three techniques are described for earthquake prediction: computation of numbers connected to the induction transfer functions for short-period field fluctuations, various combinations and comparisons of the vertical field components from pairs of observatories, and interpretation of the change from stable to unstable field (Z-component) conditions associated with solar-terrestrial disturbances. The latter is said to result from a parallel of the "load-unload" response of earthquakes and geomagnetic storms. The authors show screen displays for their computer programs that forecast the magnitude and location of earthquakes occurring I to 22 months later up to distances of 550 km.

Page 23 of the manual gives particularly good insight into the nature of the earthquake forecasts: 'We have considered the causes of false reports [that is, non occurrence of a predicted earthquake] to be as follows:
1) Instrument troubles such as damp instruments, changed scale values, etc., not explained by the observing staff at station.
2) Changed environment around the station, such as the building of factories and houses near the station, not explained by the staff who submitted the data.
3) Data collected were too scanty, so that our understanding of the normal background and magnetic anomalies related earthquake cases were not abundant.
4) The magnetic anomalies are correspondent to the weather disasters rarely happening in a hundred years (big floods, droughts, high temperatures, freezes, etc.)
5) Phenomenon which we have not yet recognized."

I question the above assignment of the problems only to the "false" reports and not to the authors' "favorable" ones as well; certainly some of their prediction of "corresponding quakes" (occurrence matches forecast) might be also a result of bad magnetic records due to 1), 2), and 3). Reason 4) reveals the writers' distorted understanding of physical phenomena (see paragraph below). I believe that reason 5) may well be the most likely difficulty with all of the magnetic variation studies presented here. The authors need to read an elementary textbook on geomagnetism [e.g., Campbell, 1997] to see what really causes the magnetic anomalies they observed. In disturbed times an irregular distribution of external source currents often produces differing fields at pairs of observatories separated by the reported distances. Typical induction transfer functions can be inapplicable during such conditions.

Pseudoscience is especially evident in a sentence on page 24: "Apart from very good precursory response to earthquake, the transfer function has obvious precursory response to weather disasters." Then follows, on page 84, "For example, in Fig. 5.16, the increase of the anomalous I A 1, 1 B 1, and az in 1989 at the Sheshan Observatory, apart from appeared the Cahgshou earthquake of Ms5.1 on 1990.2. 10 in Jangsu, they also appeared, in 1990 and after, the heavy floods [which] rarely happened in. hundred years in the Jianghuai Plain and the middle-lower reaches of the Yangtze River, in 1991." This statement is an example of the manual's full Chapter 5, which describes special episodes of local meteorological disasters such as heavy rainstorms, diver flooding, typhoons, extreme temperatures, and serious droughts as having geomagnetic precursors similar to those of the earthquakes. It seems to me that "weather disasters" have been included to make up for a shortage of earthquakes in their correlation window. I know of no meteorologist who will find any scientific reason to expect such multi-month geomagnetic precursors of local violent weather events. It appears that the authors see their predictions" wherever they want to look (or, perhaps, wherever special project funding can be obtained).

Page 22 states, "during these seven years [1990-1996], we have handed in altogether 75 (earthquake] prediction suggestions by the short period geomagnetic transfer function method, 13 of them correspond fairly well with the elements of time, location, and magnitude (See Table 2)." That table and Figure 1. 15 indicate that the word "corresponding" includes earthquakes that occurred from I to 22 months later and at distances up to 408 km away. Then the authors state that "...there were 31 false ones. The rate of false prediction was 41.3%. "My elementary mathematics of 75-13 gives 82.7%; that did not "correspond fairly well"-which I would call "false" predictions.

On page 47, the authors say "...the empirical formula of predicting earthquakes by the three elements cannot be drawn up and the prediction by the above geomagnetic methods- are sti ' 11 in the stage of predicting by experience. "This statement seems to be a euphemism for *the relationship is in the eyes of the beholder." The introduction of a human factor shows that a double-blind evaluation is required. Similar correspondence "statistics" may result from a test in which neither the record supplier nor the forecaster knows whether the purpose of a given magnetic field test record is to predict earthquakes or such obviously unrelated phenomenon such as a "catastrophic" fluctuation in an economic industrial stock index and the distance to the major market disturbance location.

The authors' poor understanding of the Earth's field response is typified by such statements as calling "thunder" an electromagnetic plane wave. It is a sound wave; even the source lightning does not produce a plane wave. Elsewhere an author states that the "ring current" is in the lower atmosphere. The ring current is clearly located in the magnetosphere at about 3-8 Earth radii.

The 20-page appendix is said to demonstrate a method (ESTAPE) for evaluating the accuracy of predictions. This appendix cites none of the recent work on the subject (for example, the special issue "Debate on VAN," Geophys. Res. Let., 23, 1291-1452, 1996; or the special section "Assessment of Schemes for Earthquake Prediction,' Geophys. J. Int, 131, 413-533, 1997). A key point discussed in these references is that evaluations must determine whether the predictions were more successful than would be expected by random chance. ESTAPE does not include provisions for comparing the success rate to random chance. Because earthquake occurrence is clustered due to foreshock-mainshock-aftershock sequences, subtle issues are involved regarding occurrences, although the basic principles are now well understood. Such research is ignored in the manual's appendix. In addition, ESTAPE employs a highly questionable, arbitrary scoring scale to rate the prediction success.

Evaluation

This "Manual on Forecasting" is an excellent illustration of how the threat of natural disasters in populated areas has generated a willingness for funding agencies, including the U.N. (cf., Li and Mervis, 1996, and Web page cited above), to support pseudoscientific forecasting efforts. In fact, the document clearly states, "Especially in recent years, the geomagnetic methods have become one of the important monitoring and predicting means in the earthquake prediction in our country [China]." The described scientific basis for earthquake predictability, as put forth in this document, is an abuse of the reliable studies of electromagnetic induction.

Because earthquakes occur within the same range of field induction depths, the authors have reasoned that the magnetic field changes will also forecast the quakes. That is not so. Although an aeromagnetic survey or a dense array of local field measurements (for GDS or MT) can delineate crustal faults within the electric conductivity substructure, there is no forecasting capability in such measurements. The immediate reduction of resistance to faulting at a specific site, caused by a direct fluid injection through a deep well, was indeed documented by Healy et al. [1968]; however, that extremely localized effect in no way justifies long-term, broad area forecasts with MT or GDS magnetic detection.

As pointed out by Main [ 1997], "...modern theories of earthquakes hold that they are critical, or self-organized critical, phenomena, implying a system maintained permanently' on the edge of chaos', with an inherent random element and avalanche' dynamics with a strong sensitivity to small stress perturbations." An international meeting, "Assessment for Schemes for Earthquake Prediction," convened by the Royal Astronomical Society and the Joint Association for Geophysics, was held in London from November 7 to 8, 1996. Papers from that meeting appeared in the GeophysicalJournalIntemadonal (op.cit.). Geller [ 1997) reported the results of that meeting: "The overwhelming consensus of the meeting was that earthquake prediction, in the popular sense of deterministic short-term prediction, is not possible at present. Most of the participants also agreed that the chaotic, highly nonlinear nature of the earthquake source process makes prediction [time, place, and magnitude of individual quakes] an inherently unrealizable goal." Although some earthquake forecasts have been widely publicized, they have not withstood scrutiny. Letters in the June, 1998, issue of Physics Today refute claims of geoelectric earthquake prediction in Greece (see also Pham et al., 1998; Schneider, 1998).

The "Manual on Forecasting" stretches the imagination with its prediction of meteorological disasters. Rather than using this manual's summary for a background on the solar effect upon climate and weather, I would advise serious scientists to read the recent book by Hoyt and Schatten [1997]. Some long-term changes in climate are ascribed to the solar irradiance output balance between sunspot regions and faculae regions and to some variable solar ultraviolet output. The fact that there is a similar solar-cycle change in magnetic field disturbances (resulting from the mass ejection from the Sun) does not create a catastrophic-weather forecasting ability by magnetic records.

A geomagnetic storm, as a part of a solar-terrestrial disturbance, results from a variety of magnetospheric, field-aligned, and ionospheric currents. Thermospheric heating by -field-aligned currents is part of a geomagnetic storm; there is an indication of a barely detectable auroral region source of infrasonic pressure waves at ground level, and a small modification of the polar-region atmospheric pressure patterns may occur. These well-established physical relationships provide insufficient reason for geomagnetic forecasting of future localized catastrophic weather events as are described in this "manual;" such events have always been clearly traced to well-understood meteorological causes with no geomagnetic precursors.

The forecasting methods described in this United Nations manual were provided by selected Chinese authors only. No other nations contributed. There are no internationally recognized geomagneticians, seismologists, nor meteorologists among the authors. The "mathematical" section of the manual presents no scientific details of any physics connecting changes in geomagnetic recordings and subsequent localized disastrous earthquakes or weather events.

The manual should be read by scientifically astute leaders in the world community. However, I believe that the publication has no scientific value as a document on the use of geomagnetic fields for the "forecasting of natural disasters." Instead, the manual provides clear and reliable evidence of the misdirection of public funds for pseudoscientific nonsense. Main [ 1997] stated a better direction for earthquake forecast funding: "Finally, in the absence of reliable prediction methods, we should concentrate on hazard mitigation based on a better understanding of earthquake source mechanisms, their statistical properties, the propagation of seismic waves, and the response of individual sites, buildings, and infrastructure to seismic vibration." The California earthquake monitoring program described by Mori et al. [ 19981 is one example of the proper use of public funds.-Wallace H. Campbell, World Data Center A, NGDCI/NOAA, Boulder, Colo., USA

Acknowledgments

I thank seismologists James W. Dewey of the U.S. Geological Survey, Robert J. Geller of the University of Tokyo, and Ian G. Main of the University of Edinburgh for their assistance in the preparation of this article.

References

Campbell, W. H., Introduction to Geomagnetic Fields, Cambridge University Press, New York, 304 pp., 1997.

Evans, R., Assessment of schemes for earthquake prediction: Editor's introduction, Geophys. J. Int., 131, 413A20, 1997.

Geller, R. J., Earthquakes: Thinking about the unpredictable, Eos, Trans. AGU, 78, 63, 1997.

Healy, J. H., W. W. Rubey, D. T. Greggs, and C. B. Raleigh, The Denver earthquakes, Science, 161, 1301-1310, 1968.

Hoyt, D. V., and K. H. Schatten, The role of the Sun in climate changes, Oxford University Press, Oxford, 279 pp., 1997.

Li, H., and J. D. Mervis, China's campaign to predict quakes, Science, 273, 14841486, 1996.

Main, I., Long odds on prediction, Nature, 385, 19-20, 1997.

Mori, J., H. Kanamori, J. Davis, E. Hauksson, R. Clayton, T. Heaton, L. Jones, A Shakal. and R. Porcella, Major improvements in progress for southern California earthquake monitoring, Eos, Trans. AGU, 79, 217, 221, 1998.

Pham, V. N., D. Boyer, G. Chouliaras, J. L. LeMouel, 1. C. Rossignol, and G. N. Stravakakis, Characteristics of electromagnetic noise in the loannian region (Greece); a possible origin for so called Seismic Electric Signal (SES), Geophys. Res. Lett., 2~ 2229-2232, 1998.

Schneider, D., On shaky ground, Scientific American, 23-24, April 1998


POSTSCRIPTS