February 19, 2004 Short Warning Times
Following is information on the small asteroid known last January 13-14 as AL00667. A preliminary analysis of the discovery data for this object yielded a possible impact with Earth in less than 2 days time -- a situation not encountered previously in the Spaceguard Program. Although we knew at the time that such a prediction of imminent impact was improbable, a collision could not be ruled out. And if a possibility of an impact in 2 days existed, what should we do about notifying governments or the public? The story of this situation on January 13, 2004, is included as part of a paper by Clark Chapman (Southwest Research Institute) presented on February 22 at the Planetary Defense conference of the AIAA (American Institute of Aeronautics and Astronautics). Several paragraphs taken from this paper are reproduced below. Following these quotes from Chapman's paper are additional quotes from a letter Brian Marsden (Minor Planet Center) wrote to CCNet on 14 January on the same subject. Finally, there is a statement posted on the website of the IAU (International Astronomical Union) discussing what lessons we should draw from the story of AL00667, and how such a situation might be better handled in the future.
Asteroids never cease to surprise us. We may never encounter a situation just like this again, but we are fairly sure to have other crises as the rate of discovery of NEAs continues to increase.
FROM CLARK CHAPMAN'S AIAA PAPER "NEO IMPACT SCENARIOS"
presented February 22, 2004
"Just last month (January 2004) perhaps the most surprising impact prediction ever came and went, this time out of the view of the round-the-clock news media. It illustrates how an impact prediction came very close to having major repercussions, even though -- with hindsight -- nothing was ever, in reality, threatening to impact. It is a story of success in that the impact prediction was nullified in record time, less than half-a-day, but the success was accomplished through a set of ad hoc, unofficial, and often unfunded activities and relationships, although assisted in major ways by the official infrastructure, such as it exists (the LINEAR Project, the IAU Minor Planet Center, and the NASA NEO Program Office).
"About 36 hours before President Bush's planned speech at NASA Headquarters on future American space policy, the Lincoln Near Earth Asteroid Research (LINEAR) observatories in New Mexico routinely recorded four images of a moving object. Half a day later, on Tuesday, January 13th, these data were sent (as part of the daily submission of data) to the Minor Planet Center in Cambridge, Massachusetts. Just before going to dinner, MPC research Tim Spahr ran the data through standard software to generate a nominal ephemeris for the new object. These are posted on the publicly accessible NEO Confirmation Page (NEOCP) so that amateur and professional asteroid astronomers around the world might be able to follow up on the LINEAR observations that night. It is through such follow-up astrometry that NEO orbits can be refined so that the object is not permanently lost. Spahr posted the ephemeris, based on LINEAR's four detections, on the NEOCP under the designation AL00667, along with ephemerides for several other recommended targets. Less than an hour later, a European amateur astronomer, Reiner Stoss, went to the NEOCP and noticed a curiosity: AL00667 was predicted to get 40 times brighter during just the next day, meaning that it was going to be six times closer to the Earth! He expressed his amazement on Yahoo's MPML (Minor Planet Mailing List) chatroom on the internet.
"Professional asteroid researcher Alan Harris happened to be monitoring the chatroom and noticed the strange posting about a "bogie" (military slang for an unidentified, potentially enemy aircraft). Indeed, he found on the MPC's web site, with no comment at all, what he recognized to be an alarming prediction. He quickly calculated that an asteroid following this nominal ephemeris would strike the Earth just one day hence. He immediately alerted several of us, including NASA Ames Research Center's David Morrison (who chairs the IAU's Working Group on NEOs) and JPL's Don Yeomans (who heads NASA's NEO Program Office). His message was sent at 5:09 pm (MST, used hereafter, is the time zone of LINEAR and of the amateur astronomer who later laid this bogey to rest).
"Yeomans and his associate Steve Chesley telephoned to the MPC to try to find out more about the asteroid from Spahr. Forty-five precious minutes had already elapsed since Harris' email, when all that Yeomans could report back to his colleagues was, "We've got a call into Tim Spahr to see if we can get the astrometry itself but Tim is not yet at home." About half-an-hour later, they reached Brian Marsden, director of the MPC, who was working late that evening. (Harris also alerted Marsden by email.) By policy, LINEAR and other single-night asteroid data (termed one-night-stands) are kept private by Marsden until they have been verified and linked with observations on other nights. But once contacted by Yeomans and Chesley, Marsden agreed to provide the data to the JPL researchers and did so about 20 minutes later; then, about 1 hour 45 minutes after being notified by Harris, they got to work trying to understand this anomalous asteroid. (Other asteroid orbit experts, in Arizona and Italy, later complained that they never had access to the data on AL00667.)
"The MPC faced the embarrassing fact that they had effectively made the first-ever prediction of a near-term asteroid impact without even realizing it themselves. Marsden hastily tried to fix the web page. Supported by no new observations, he posted a new, non-impacting (actually receding) trajectory, which was also consistent with the data. An hour later, Spahr -- having finished dinner, gone home, logged in and discovered what was happening, and raced back to the MPC -- replaced Marsden's post with yet another trajectory, once again showing the asteroid headed toward the Earth, but this time narrowly missing an impact. None of the later postings reflected new data: Spahr and Marsden were simply frantically trying to figure out for themselves what the data meant and what was politically correct to display on their web site. With hindsight, it is clear that the highest priority should be to search for "virtual impactors" -- that is, the subset of asteroid trajectories allowed by the uncertainties in the fit to the data that would result in an impact; if no asteroid is found in the patch of sky that meets these criteria, then there is no longer a threat of impact. The second priority should be to find the NEO, wherever it might be within the spread of uncertainty, so that it isn't lost. Another priority, of course, is not to confuse, mislead, or frighten people by leaving an effective impact prediction posted on the web site (without appropriate caveats, especially for non-experts who might suddenly be alerted to this web page's existence). With hindsight, we can surely imagine better solutions than any of those implemented on the NEOCP in unplanned crisis-mode that night. But the chief blameworthy error is lack of thorough planning by the NEO community for such a contingency, not in the spur-of-the-moment decisions actually made".
. . . . . . (several paragraphs omitted)
"What can we learn from this case? How could there have been an official, if unmonitored and obscure, posting by the MPC based on a calculation implying a major asteroid impact the following day, without the MPC even realizing it? How could the data, on which the calculation was based, be kept private so that many of the world's asteroid experts could not evaluate the situation, long after the threat was being debated in a public chatroom? How could the JPL Sentry system and the parallel NEODys system in Italy have failed to post the relevant information on their own official asteroid impact web sites? Why were the LINEAR data worse than usual for this particular "one-night-stand"? Were the computer programs used by the MPC and JPL that evening truly state-of-the-art and, if not, did that contribute to the scary predictions? How could one JPL expert calculate something like 1-chance-in-four of a near-term impact disaster, when in fact the asteroid never passed within millions of miles of our planet? Just how big was the nominally calculated impacting body, where would it have hit, and how much damage might it have caused? Did this event merit the unexpectedly high value of 3 on the Torino Scale (designed to educate the public about the seriousness of an impact prediction)? How did this potentially most dramatic of all asteroid impact predictions fail to be noticed by the news media? How close did astronomers come to issuing another false alarm, this time with the potential for embarrassing not only NASA but the White House? What things went wrong, and what things went right during the evening of January 13? Finally, what can be learned from the events so that a more reliable treatment and analysis of Spaceguard Survey data can be accomplished next time?"
THE SAGA OF ASTEROID AL00667 = 2004 AS1
Brian G. Marsden (from CCNet, 15 January 2004)
"That this latest PHA should have generated so much heated discussion on numerous mailing lists and the internet on the basis of four observations covering a time interval of one hour on the morning of Jan. 13 is surely quite amazing. On the routine arrival of the night's LINEAR data at the Minor Planet Center at 5:15 p.m. EST that day, the usual computations on them were quickly done, and, within a matter of minutes, five of the objects were placed on the MPC's WWW "NEO Confirmation Page" as being of potential NEO interest, predictions of the expected positions and their uncertainties being provided in the hope of securing early confirmation from observers in Europe. It was evidently cloudy over most of the continent, however, and the only follow-up observations immediately forthcoming were in fact from a single observer in the U.K. Also according to usual procedures, on the receipt of these U.K. observations, the predictions on the WWW could be quickly and significantly refined, well in time for further observations to be presumably made from North America. There was in fact also rather extensive cloud cover that night over North America, particularly over the numerous professional and amateur observatories in the frequently blessed Southwest.
"At 7:25 p.m. the LINEAR team withdrew one of the day's crop as not real, and this object was removed from the NEOCP. Another object was far to the north, and the single U.K. observer was unable to observe it because at that high declination his CCD camera fouls the fork mounting of his telescope. By that time this particular object, temporarily called AL00667, had attracted the interest of the MPML folk, however, because--curiously--the "nominal" prediction adopted for use on the NEOCP showed that the object would collide with the earth in little over 24 hours! This circumstance had not in fact been noticed by the MPC staff member maintaining the page (who had a dinner date with a visiting professional colleague), although it was of rather limited consequence because he had also delineated the moderately large sky-plane uncertainty area over which follow-up observers could expect to carry out their searches. Unfortunately, as already noted, very few follow-up observers were able to do anything.
"Dinnerless myself, I was still in office (working on the SOHO comets) around 8:30 p.m., when I was alerted to the MPML "crisis" by a telephone call from one of the NASA NEO Program Office staff members. I quickly verified the complaint about the impact, computed several orbits myself that would have the object receding from the earth rather than approaching, supplied the observations to the NASA colleague so that he could do some independent orbit computations (which essentially agreed with those of the MPC), contacted a few potential observers in the U.S., advised my MPC colleagues that "we had a problem", and replaced the offending nominal orbit by one that was less threatening. Of course, what we now urgently needed were _follow-up observations_. All we got, unfortunately, was ever more unnecessary and often quite uninformed "talk" (notably by professionals!) in the MPML." (several additional paragraphs omitted)
INTERNATIONAL ASTRONOMICAL UNION WORKING GROUP ON NEOs WGNEO STATEMENT ON DEALING WITH SHORT-TERM IMPACT PREDICTIONS (posted on WGNEO website 19 February 2004) On the night of January 13/14, 2004, the international asteroid science community faced an unprecedented situation, with an initial prediction of an impact of a small asteroid on the Earth within a few days. The asteroid in question, bearing the temporary designation AL00667, had been discovered the previous night by the LINEAR telescopes of the Spaceguard Survey. Routine and largely automatic processing of these observations led to the posting (by the IAU-supported Minor Planet Center in Cambridge MA) of an orbit that intersected the Earth within the next 36 hours. Although the observations were limited and the orbit solution was highly non-unique, there initially appeared to be a distinct possibility, verified by orbital calculations carried out at the NASA NEO Program Office at JPL, of an impact in the northern hemisphere.
The estimated size of asteroid AL00667, diameter 30 m, was small enough that no serious consequences were expected, but large enough that significant ground damage or possible injury could not be ruled out. The very limited data (from one night's observations) and the preliminary nature of the orbit at no time allowed an estimate of a possible impact site. The situation remained uncertain for several hours, until amateur astronomer Brian Warner, with a 20-inch aperture telescope in Colorado, searched the area where JPL calculations showed that the asteroid would have to be if it were on an actual collision course. Warner's observations were negative, showing that the true orbit for AL00667 did not intersect the Earth. Thus, as Clark Chapman of Southwest Research Institute writes in a paper presented at the American Institute of Aeronautics and Astronautics workshop on protection of the Earth (February 23), "instead of waking up to headlines and TV news specials of an asteroid strike about to happen in the next day or two, life went on as usual around the world." Only a handful of astronomers (professional and amateur) were involved, and there were no scare stories in the press.
Several members of the IAU Working Group on Near Earth Objects were involved in the activity surrounding the possible impact orbit of AL00667, and the Working Group on NEOs seeks to draw some lessons from these events.
First, we stress that there were no significant errors in either data or interpretation made that night. The observations were real, the object did exist (although it turned out to be much farther away than originally thought), and there were orbital solutions that predicted an impact with an apparent probability of at least 10%. Additional observations were required in order to verify that the object was not on a collision course. These observations were of the sort used previously to falsify a dangerous orbit by looking at the place where the asteroid would be if it were on a collision course (sometimes called a process of eliminating a virtual impactor). The IAU WGNEO congratulates those who worked to eliminate this particular threat in a highly professional way, and did so within a few hours, and without negative publicity.
Second, we note that the Spaceguard Survey was not designed to detect small asteroids a few days before collision. It does not have either the equipment or the resources to function as a warning system for imminent impacts. Its purpose is to carry out a long-term survey, with emphasis on discovering and calculating accurate orbits for asteroids larger than 1 km -- those with a possibility of triggering a global catastrophe if they collided with Earth. The Spaceguard goal is to discover 90% of the near-Earth asteroids larger than 1 km by the end of 2008. This approach should provide decades of warning for any impact by these larger asteroids.
Third, the probability that a small asteroid will be discovered within a few days of hitting the Earth is extremely small. While many asteroids might masquerade as impactors based on discovery observations spanning a single night, the great majority of these will be "false alarms."
Finally, however, we note that it is possible for the survey telescopes to pick up a small asteroid on its final approach, and the interested community should decide how cases like AL00667 should be handled in the future. The IAU WGNEO suggests to its members that we consider what actions should be taken to facilitate the determination of a more accurate orbit, and the ways this information should be made available to the public and to interested official groups. The current mode of verification for larger asteroids, including the provision for an optional IAU peer review of orbital calculations, does not apply when there is a possibility of an impact within a few days of discovery. The WGNEO therefore suggests that we should discuss these issues among ourselves and with other interested parties to determine protocols for dealing with such cases in the future. There should be planning at an international level of what action to take with respect to notifying governments and the public of a possible near-term impact. This planning must recognize that it is likely that in all such cases new observations will quickly eliminate any possibility of impact rather than confirm it.
In the meantime, the WGNEO suggests the following two interim guidelines: (1) If any orbital solution for a newly discovered asteroid includes a possible near-term impact, every effort must be made to pursue that possibility (even if it is not the most likely of several possible orbits). The raw data should be available immediately to the community so that experts on asteroid orbits all over the world can be involved in the interpretation and can freely check each other's calculations. (2) Observatories, amateur and professional, should be informed of the high priority to be given to checking the possibility of an impact by searching for the virtual impactor and (hopefully) showing that it is not on a collision course. Searching positions corresponding to a virtual impact orbit should be given priority, as it is more important to eliminate the virtual impactor than to recover the asteroid and compute its correct orbit - the latter will happen routinely within a few days.
David Morrison Chair, IAU WGNEO