The Asteroid Impact Scare of Mid-March 1998
by Clark R. Chapman, 5 April 1998 (updated 22 April 1998)
On March 11th, Harvard astronomer Brian Marsden, on behalf of the International Astronomical Union (IAU), issued an IAU Circular to the world's astronomers about a possible very close pass to the Earth of a mile-wide asteroid in the year 2028. That is the size of asteroid that could cause a world-wide ecological catastrophe and kill hundreds of millions or billions of people, according to the best analyses that have been made. Marsden was aware that, in addition to astronomers, science journalists subscribe to the IAU Circulars. Therefore, he prepared a Press Information Sheet, elaborating on his prediction, and stating that there was a "small" chance that the asteroid would actually hit the Earth; that information was distributed to the press by the American Astronomi- cal Society, although the New York Times and other papers were already writing stories based on the IAU Circular alone. The story played prominently on evening news telecasts on March 11th and was banner- headline news around the world the next day.
Impacts by asteroids and comets are a real possibility, even though the probabilities of a civilization-threatening impact are extremely low (about 1 chance in 300,000 per year). The potential consequences, however, would be enormous, even though falling far short of the kind of mass-extinction holocaust recorded at the Cretaceous/Tertiary boundary 65 million years ago. The discovery rate of Earth-approaching asteroids will surely rise, along with future cases of uncertainty about possible Earth impact. Here is a brief report on what really happened a few weeks ago. Simply put, it was all a mistake...there was no basis for making the original announcement in the first place, although the public was led to believe that it was the finding of additional, pre-discovery data that was responsible for retracting the prediction of a possible impact. Of course, it is more complicated than that. What follows is a short summary. If you want more information about the Near Earth Object (NEO) impact hazard, and a detailed -- if incomplete -- history of the events in mid-March, check out the following Web site. It is the latest revision (April 22) of a draft Case Study I prepared for the July 1997 "Workshop on Prediction in the Earth Sciences: Use and Misuse in Policy Making," organized by the Geological Society of America and the National Center for Atmospheric Research, under NSF sponsorship:
There are roughly 2000 Earth-approaching asteroids that approach or exceed the "one mile wide" size that David Morrison and I have established (Nature, 367, 33-40, 1994) as possibly civilization-threatening. Less than 10% of these have been discovered. The job for NEO astronomers is to discover as many of the remaining ones (NASA's stated goal, not yet implemented, is 90% of them in the next decade), and to determine their orbits. Almost certainly, as each one is discovered, it will be demonstrated within days (occasionally as long as a few years) to NOT be on a collision course with Earth. As we go through this process of certifying that asteroids will NOT strike the Earth, we will gradually reduce the probability of impact in the succeeding 30 years from the current 1 chance in 10,000 to 1 chance in 100,000 (due to the remaining 10%).
As of March 11th, observations of the position of 1997 XF11 had been reported to the IAU Minor Planet Center over a three-month period. The good news is that those data, and those data alone, were sufficient to demonstrate that 1997 XF11 would NOT hit the Earth during its approach in 2028 or at any other time in the foreseeable future (more specifically, that the chances of XF11 impacting the Earth could be shown to be MUCH less than the background probability -- 1 chance in 10,000 -- of an unknown asteroid of that size hitting us before 2028).
But, instead of adding 1997 XF11 to the slowly growing list of asteroids certified as not going to hit the Earth, the director of the IAU Minor Planet Center, Brian Marsden, made an announcement to the world (via the IAU Circulars he edits and an associated press statement) that the asteroid was "virtually certain" to pass within 80% of the distance to the Moon and stood a "small...not entirely out of the question" possibility of hitting the Earth. (Marsden has admitted that he used ill-chosen words to convey his results and he has more recently stated, in an April 18th IAU Circular, that he made errors in interpretation and that there never was any chance of an impact. Still more recently, Marsden has disavowed his retraction and has argued in several forums that, prior to obtaining prediscovery observations, the available data were consistent with possible impacts in 2028 and in several later years during the 21st century.))
Marsden had a policy of holding onto the data (for one month) that he receives from the world's astronomers, and a practice of doing his own evaluation of the data before making them available to other astronomers. Therefore, nobody else had the practical capability of making independent calculations on the full dataset available to Marsden. (In fact, people weren't even making independent calculations based on the data that Marsden had made available; had they done so, they would have discovered that the data already available by late December 1997 precluded any impact, for all practical purposes.) Most important, Marsden didn't check his sensational results with colleagues at other institutions before making his faulty announcement to the world. Within hours, Marsden was persuaded to release the most recent data, and within hours of that, several other astronomers (e.g. Paul Chodas and Don Yeomans in Pasadena and Karri Muinonen in Helsinki) had correctly calculated that the probability of Earth impact was essentially zero -- certainly vastly less than the probability of impact from the as-yet-undiscovered asteroids.
Already during the evening of March 11th, two astronomers (from JPL and MIT) requested that Marsden issue a public correction, and tell the anxious world that new calculations by other astronomers showed that the asteroid would not hit. He explicitly refused to do so. Even by noon the following day, Marsden was still telling reporters that the asteroid could hit. If no prediscovery observations were ever located and no subsequent observations of XF11 were ever made, a proper analysis of the Dec. 1997 - Mar. 1998 data -- done by several other astronomers within hours of being provided the data by Marsden -- clearly places 1997 XF11 in the "not dangerous" category.
Marsden finally, late on the 12th, changed his prediction to "no impact". He used, as his basis for changing his prediction, the arrival (and inclusion in his calculations) of prediscovery data from 1990. (Although he could have telephoned several individuals to solicit such data, Marsden believes that issuing the IAU Circular was vital to uncovering such past data and encouraging new data to be obtained.) It is essential to realize, however, that the new data were just "icing on the cake." Every other expert on orbital calculations of NEO's had already concluded that 1997 XF11 was in the "safe" category before the 1990 data became available.
In the future, it will be necessary for NEO scientists to do what scientists have traditionally done for centuries, which is to have important results peer-reviewed before they make rash announcements to the public. To facilitate this, all data from the international astronomical community should be disseminated as rapidly as is technically possible, and procedures should be implemented to cross-check and peer-review announcements of potentially hazardous close approaches before such conclusions are published (i.e. made public). Such evaluations may be mandatory (as they are in many scientific organizations) or they could be voluntary. In any case, they should be completed very rapidly. Already, NASA, the IAU, and others are starting to formulate procedures to minimize the chances of future cases of "Chicken Little". I hasten to add that such a procedure is NOT "secretive" -- double-checking and peer review are the normal ways that scientists report results and ensure that the public can have some degree of confidence that the reported results are reliable.
Clark R. Chapman (firstname.lastname@example.org), Institute Scientist, Space Studies Department, Southwest Research Institute, Boulder, Colorado
ADDENDUM, 29 April 1998
In the statement above, I have referred to a most welcome IAU Circular (no. 6879) issued by Brian Marsden on 18 April, making several corrections to his original March 11th announcements. Less welcome, however, was Marsden's subsequent "explanation" of Circular 6879, published in the CCNet Digest (on the Internet) on 24 April. The explanation calls for an objective response.
(1) Marsden begins by owning up to "one inconsequential mistake." (Only he could have the audacity to refer to dominating world headlines for two days as "inconsequential".) Apparently he refers to his statement that passage within 0.002 AU of the Earth in 2028 was "virtually certain." But two more, highly consequential corrections are also contained in IAUC 6879: The most significant is that Marsden now agrees that it was "already evident" (or, at least, should have been, had people been thinking about it) by 23 December 1997 (!) that XF11 had "no possibility" of an Earth impact ("encounter within 0.00019 AU"). That correction (to Marsden's widely publicized statement that there was a "small chance" of impact) is qualified by the words "in the absence of effects that would be highly unusual," which seems to represent a third about-face by Marsden. In his late- March Boston Globe op-ed essay, and at other times, Marsden had emphasized that cometary-like forces or passes near other asteroids could have perturbed XF11 into an Earth impact, so the words "highly unusual" represent at least a change in tone on Marsden's part.
(2) Marsden argues that it "took more than a month" to resolve important XF11 issues (like the above corrections), which -- he says -- does not augur well for those who believe that consensus can be reached, for future cases, within 24 or 48 hours. I worry about how quickly consensus can be achieved, too. But, at least in this case, the record is clear that all orbit-calculators (except for Marsden) reached the conclusion within 24 hours that the chances of impact were inconsequential.
(3) Marsden says that he "never understood why" his wording on the original IAUC was naively interpreted as implying an impact probability of 0.1%. That must make him the ONLY person who doesn't understand why. To see the way typical scientists would obviously react to Marsden's words, read Leif Robinson's editorial and Stuart Goldman's sidebar in the June issue of Sky & Telescope.
(4) Marsden writes: "Nevertheless, as recently as last week, Muinonen was claiming that some of his orbit computations involved earth impact in 2028." This statement, while technically true, is preposterously misleading. Muinonen has been in the process of deliberately forcing hypothetical orbits to go through both the observations *and* a 2028 Earth impact, for purposes of seeing how large the observational residuals would be for an impact. They are, as Alan Harris has explained in CCNet Debates for 28 April, extraordinarily large (tens of arc-seconds for 1997-1998 data alone, tens of arc-minutes for 1990-1998 data) and correlated, thereby demonstrating that Earth-impact is impossible.
(5) Marsden says that "computation of impact probability" after only a few months of observations, as was the case in March for 1997 XF11, is "a meaningless exercise." It is truly difficult to believe that, after everything that has been said, Marsden has written these words. As the corrective IAUC 6879 states, had such calculations been done at the end of last December (or in January, or in February), they would have demonstrated that there was absolutely no practical chance of 1997 XF11 impacting Earth, and the headlines of March 11th and 12th wouldn't have been written. Contrary to what Marsden says, it is vital that accurate, sophisticated, well-checked calculations be done on future PHA's so that they can quickly be placed in the "no-impact-possible" category.
Other elements of Brian Marsden's "explanation" are helpful and insightful, as are the reported changes in operation of the IAU Minor Planet Center. But it will take much more progress in understanding these problems before we can be assured that future impact scares won't happen.
More Information (as of 3 Sept. 1998)
Three abstracts submitted to the 1998 meeting (in Madison, Wisconsin, Oct. 12-16, 1998) of the Division for Planetary Sciences of the American Astronomical Association elaborate on 1997 XF11 and document the conclusions discussed above: Search the 1998 DPS Meeting Program for the following abstracts:
[10.07] Could Asteroid 1997 XF11 Collide with Earth? by P.W. Chodas and D.K. Yeomans
[10.08] Upper Bounds for the Earth-XF11 Collision Probability by K. Muinonen
[7P.16] The Impact Hazard in the Context of Other Natural Hazards and Predictive Science by C.R. Chapman
See also Clark Chapman's written testimony before the House Subcommittee on Space & Aeronautics, for the May 21, 1998, hearings on "Asteroids: Perils and Opportunities".
General information about the impact hazard may be found at the NASA Ames Impact web site.
Available Abstracts, Preprints, Articles.
Clark R. Chapman's Publications.
Clark R. Chapman's Home Page.