It fell to Buie, the person with the most experience massaging Hubble data, to do much of this nitty-gritty work. But after applying three different types of deconvolution and thousands of computer iterations, the images weren’t coming out exactly right. In some cases they even got worse. Features that had shown up clearly in the raw images would disappear. Or, if the computer happened to sharpen the image’s noise instead of real light from Pluto, some spurious feature would pop up out of nowhere. The problem with deconvolution, particularly when you are working with only eight pixels, is that “you don’t know when to quit,” says Buie. “You can just keep iterating and iterating and sharpening and sharpening.”
So they decided to drop deconvolution and go to Plan B. First Buie used a computer to generate an artificial image exactly the size and shape of the ones of Pluto, with a grid overlaying its surface just like the grid of pixels on the Hubble image. Then he tuned the brightness of each pixel in the artificial image to roughly match what appeared in the real images. Next came deliberate blurring—to duplicate the blurring effects of Hubble. The final step was to carefully align, or register, the two pictures (fake and real) and subtract out the difference, leaving—voilà—an idealized but noise-free version of what the telescope actually saw.
Buie repeated this process for each of the 24 images. It was time- and computer-intensive, rewarding in its own way but frustrating too. Every little thing had to be taken into account. At one point he found that even a slight jitter in the telescope at the time the exposures were taken—only 1/10,000 of a degree or so—had degraded some of the images.
Even more frustrating was never being able to work on the problem for more than a short stretch of time. Buie had to fit the Pluto work into an already hectic schedule: observatory visits, meetings, proposal writing, and all the other hassles of the working scientist.
Stern was even busier. In fact, he was in the middle of one of the most frantic years of his life. Shortly after receiving the Pluto data, he and his wife had a third child, and the family moved from San Antonio back to Colorado. During one stretch in the spring of 1995, which was not at all atypical, Stern traveled to the McDonald Observatory in Texas to observe the moon’s atmosphere, made a quick pit stop at home, then flew to NASA’s Marshall Space Flight Center in Alabama to work on an experiment that was flying on the space shuttle. After another guest appearance at home, it was off to Toulouse, France, for a scientific meeting on ices in the solar system, then home for one night, then into the field with a sounding rocket experiment for two and a half weeks, back home briefly, then to California for another meeting. The day after he returned home from nearly six weeks of continuous travel, his taxes were due.
“If I could shut out the world and go to Antarctica, we would do this whole project in two months, “ he lamented.
It wasn’t as if they had forever, either. By STScI rules, any scientist who uses the Hubble gets exclusive access to the data for exactly one year from the day the observation is made. After that, anyone can go into the archives and help himself or herself to the original tapes. At first Stern felt some pressure—what could be worse than another scientist scooping you with your own data before you could publish the results? As the months wore on, though, and it became obvious that the team wouldn’t be able to publish within a year, it started to seem less of a worry. For one thing, it was a damn difficult task. If anyone else thought they could do it better or faster and still get it right, they were welcome to try.
By the time of the annual Lunar and Planetary Science Conference in Houston in March 1995, Stern trusted the pictures enough to begin presenting them to other scientists. In his talk he said that the images showed roughly a dozen albedo regions on Pluto’s surface. One intriguing linear feature might even be a crater ray (later he felt less confident about that interpretation and dropped it). It was always “dangerous to overinterpret” the pictures, announced Stern, who underplayed how far along he and Buie had taken the image processing. This was a “progress report” only.
For months, NASA’s press people had been bugging him to release the pictures. Not until they’re ready, was his standard reply, and NASA always backed off. But magazines were starting to ask too: When are we going to see the pictures?
In the end, it was an external event—an educational project in which schoolchildren got to make their own Hubble observations of Pluto—that forced the team to wrap up the first phase of their work. The original plan had been to publish the Pluto images first as a short paper in a scientific journal, then do the NASA press conference, then follow with a more comprehensive paper comparing the new photos to the old maps. But to avoid having the schoolchildren steal their thunder by releasing Pluto images first, Stern and Buie would have to go public, then publish.