JPL's assumed that a calibration table sent to them from MCO's manufacturer Lockheed Martin, and used in JPL's navigation calculations, listed the thrust of the spacecraft's small maneuvering thrusters in Newtons (less than 1/4 pound), when in reality it listed the thrust in pounds.

The result of the mix up was that JPL's navigation software miscalculated the spacecraft's position, and so JPL ended up inadvertently ordering MCO to carry out a final TCM (or Trajectory Correction Maneuver) eight days before its arrival at Mars that actually caused it to fly over Mars as low as 57 km. This was deep enough into the Martian atmosphere for air friction to impact the probe adverserly - causing it to either crash or left it drifting in Martian orbit as a charred and crippled wreck.

It does indeed look like a ludicrous error -- but it could perhaps be excused as the sort of very small, subtle individual slipup that can nevertheless ruin any space mission, given the staggering technological complexity of such missions.

Bugs In Space
There have been many such slips before: the missing hyphen that caused the booster rocket for the Mariner 1 Venus probe to veer off course; the erroneous temperature figure in a single manufacturer's memo that -- three years later -- led to the Apollo 13 disaster; the single misplaced decimal point that led to the loss of the billion-dollar Milstar satellite only six months ago.

In a technologically complex world, these things will happen sometimes.

And this seems to be the public impression that JPL and NASA are trying to leave of this accident. But over the past two weeks, it has become increasingly apparent that the loss of MCO was actually due to a bigger and less excusable mistake: namely the failure of MCO's mission control team to heed increasingly clear warnings - during the final days before its arrival at Mars -- that the craft might be seriously off course.

Soon after the "mishap", SpaceDaily was contacted by an aerospace engineer who would prefer to remain anonymous, but whose contacts with JPL have provided him with ongoing reports as to the "true" story.

His conclusions have now been almost entirely confirmed by stories in the Oct. 4 "Aviation Week", and by an Oct. 5 report from science columnist Lee Dye on the ABC News Website.

A Missing Datum Point
The erroneous operations table caused JPL's navigation software to underestimate the strength of MCO's small thrusters during its entire 10-month cruise to Mars, both during MCO's midcourse maneuvers and during hundreds of small thruster firings which it periodically made to allow reductions in the rotation speed of the motorized "reaction wheels" that MCO mostly used to control its attitude.

The errors were small -- especially since MCO carried out its bigger Trajectory Correction Maneuvers with its main rocket engine, which was not affected by the metric error. But they did add up, in a gradual way that was rather difficult to detect.

The errors should not have been impossible to detect, though -- and here is where the possible JPL could face problems.

Mars Climate Orbiter carried out its fourth and last Trajectory Correction Maneuver ("TCM-4") on Sept. 15, eight days before its arrival at Mars; the maneuver was intended to make the craft fly over Mars' surface at 224 km altitude during its Mars orbit insertion burn.

Now --as you might expect -- precisely measuring the position of a spacecraft 200 million km from Earth is not easy. Instead of relying on a single radio tracking measurement to do so, you must average out the results of a whole series of tracking measurements -- often a week or more. Nevertheless, JPL expected to know MCO's actual Mars flyby altitude to within plus or minus 10 kilometers.

But -- to quote Aviation Week -- "Several days after [TCM-4], the navigation calculations had relatively poor convergence [on an estimate of the spacecraft's trajectory]. That is, the size of the estimated errors in the calculation of the trajectory were unusually large. The orbit injection flyby was targeted at 224 km altitude, and the new numbers were trending to 150-180 km -- but with uncertain confidence."

The purpose of the thruster-strength data that was erroneous was to reduce the "residual error" in the trajectory calculations, allowing JPL's computers to converge somewhat more quickly on an accurate and consistent estimate of the spacecraft's position.

Under normal circumstances, JPL's computers should have been able to calculate the spacecraft's flyby altitude with an accuracy of 10 km within a few days after TCM-4.

Indeed, MCO's flight plan called in advance for a possible fifth and last Trajectory Correction Maneuver three days before Mars arrival if the ability to calculate MCO's true position indicated that the craft was dangerously off course.

However, within several days after TCM-4 the radio tracking data was indicating that the craft was 74-94 km off course -- far more than it should have been after such a maneuver -- and even those calculations of its path were far more uncertain than usual.

This should have alerted the flight control team that something was seriously amiss with their ability to calculate MCO's true position -- and, in fact, they did seriously consider carrying out TCM-5 in order to raise MCO's flyby altitude by an additional 40 km as a precaution.

But, to quote Aviation Week: "Project management decided on Sept. 19 to forego TCM-5 [the next day], reasoning that the risks of an added maneuver, such as a possible 'safe mode' halt [which was unlikely, but might have paralyzed the craft for a few days and kept it from carrying out its orbit insertion burn], outweighed the benefits of higher altitude based on shaky navigation data."

The reason for this decision was that, from Sept. 18 through Sept. 21, the "apparent" error in the periapsis calculations dropped back down to the usual 10 km, the calculated periapsis remained at 150-180 km, and "The minimum survivable altitude was 85-98 km, which gave adequate margin," quoted Aviation Week.

Then, during the last two days before Mars arrival, the calculated Mars flyby altitude started to drop slowly. "At 6-8 hours before the orbit injection firing, enough Mars-influenced Earth radio data had been collected to tell that the altitude was going to be a lethal 57 km.

But it took time to process these data and, along with delay induced by a shift change, project managers were not aware of their predicament until an hour before orbit insertion -- too late to do anything except sit and watch."

Two questions spring to mind. First, why didn't the flight managers pay more attention to that data shortly after TCM-4 showing both that the burn had been significantly off target, and that the trajectory calculations were unusually bad?

And secondly, why -- during the crucial period from Sept. 18 to 21 -- did the "apparent" accuracy of the trajectory estimates shoot back up to 10 km, although the actual error was 93-123 km?

(As an added oddity, JPL told Aviation Week that the uncertainties in its estimates of MCO's flyby altitude during this period were "plus-or-minus 10 km" -- but they then said in the same sentence that the total uncertainty was in a range of 30 km, not 20!)

SpaceDaily's source says flatly that the only possible explanation for the last two statements is that, contrary to JPL's public statements, the "apparent accuracy" of plus or minus 10 km during this period wasn't derived from the actual radio tracking data, which -- if properly examined -- would clearly have shown that MCO's position was highly -- and dangerously -- uncertain.

Instead, it was based on a blind, overconfident assumption that MCO must really be on that accurate a course, simply because all of JPL's earlier missions had been that accurate. And this assumption was made in the very face of those earlier indications both that the craft had veered significantly off course during TCM-4 and that the tracking calculations were less accurate than usual.

In short, MCO was lost not just because of that one metric conversion error, but also because of significant overconfidence on the part of its project managers.

Indeed, Lee Dye's report for ABC News indicates that the current tumult within JPL is so great that some sort of major blunder must have occurred. Dye reports that there has actually been one instance of fisticuffs in the halls between a navigator and a senior engineer, and that frantic finger-pointing and mutual accusations are now filling the institution:

"The lab has been rife with rumors that the navigation team determined 10 days before the accident that the spacecraft was on a course that would take it too close to Mars, ample time to make a course adjustment. According to lab lore, the navigators are being blamed to save the brass, which failed to heed the warning, wrote Dye.

"Pat Esposito, the head of the navigation team, declined comment, saying he had been ordered not to speak with the press. According to the rumors, Esposito personally had given warning of the needed course change to Thomas Gavin, deputy director of JPL's space and earth science directorate, which is responsible for the Mars program.

"When I asked Gavin about that, I thought the telephone line was going to vaporize.

"'That never happened,' Gavin said. 'I would have come unglued. I would have called a session and said, "OK, what the hell is going on?" We would have done something,'" reported Dye in his ABC News report.

In short, this was not a subtle or excusable error -- somebody's heads are going to roll at JPL, although it's not yet certain whose. To quote our source: "There are two possibilities: either the navigation team were wimps and didn't make a good case, or they made a good case but the management ignored them."

He suspects the last, on the grounds that many of the flight managers were senior former navigation engineers themselves and had excessive confidence on the basis of JPL's earlier precision during their own tenure. Moreover, since the calculation errors resulting from the multitudes of earlier small thruster firings during MCO's flight to Mars were cumulative, JPL should have noticed some time earlier that the uncertainties in the calculations of its trajectory were increasing abnormally.

At any rate, it seems very likely that the truth will end up coming out -- especially given the fact that, in addition to JPL's own two internal investigation teams, NASA has appointed an outside team which will be led by Marshall Space Flight Center director Arthur G. Stephenson.

In the meantime, hasty measures are being taken to ensure that a similar disaster doesn't strike the Mars Polar Lander, now closing in on Mars for its scheduled landing in the south polar region on Dec. 3.

Spokeswoman Mary Hardin says that there is no danger of this because the Lander does not have reaction wheels for its attitude control -- but despite this, the Lander's next Trajectory Correction Maneuver has been postponed from Oct. 7 to Oct. 18-20.

The final likely conclusion, though, is that MCO's $125 million demise was not so much an example of the disasters which a single subtle error can produce in our complicated technological world, but rather yet another demonstration of a far older principle that has bedeviled the human race from the beginning: pride goeth before a fall.

References