Wednesday, April 09, 2008

Blasted data reduction issues

One of the most boring and yet important parts of astronomical observing is calibration. Calibration means to determine exactly how sensitive our telescope was when we took our images, so we know exactly how much light we were detecting from a star or galaxy. This is important if we want to compare one star or galaxy to another, or even to itself (if we want to look for changes in the amount of light the object is putting out), or if we want to test theories about the star or galaxy. In short, calibration is important, actually crucial and central, to our work.

But calibration is hard. Earth's atmosphere is always changing, and it lets a little more or a little less light through each night, even if the night seems crystal clear. There can be a little extra dust or haze high in the sky. Or maybe the air is a little more turbulent one night than the next, blurring star images a little more. Or maybe there were thin clouds in the sky, but we went ahead and looked anyway. In theory, all of these can be calibrated to some extent. In practice, it is hard.

Even calibrations in space are hard, even though there is no changing atmosphere. The Hubble Telescope has entire week-long workshops every year to discuss calibration. They sound really, really boring, so I've never gone. But, as a user of the Hubble, I'm very glad for the people who agonize over the calibrations. And these calibrations change over time, because radiation and age change the sensitivity of the cameras we put on satellites.

Last night, I was nearing time to go home when I discovered a big problem in some calibrations that I had done for a massive project I'm on. Some of the calibrations are off by 50% -- a huge amount! -- and I don't know why! I know that I was more than careful when doing the calibrations, but that still wasn't good enough. So, now I get to spend time and track down the root cause of the problems. Was the night really partly cloudy and we thought it was clear? Did I make a typo in my programs? Or is there a more subtle problem, like the camera was acting up on one night and had varying sensitivity? This will take some time to track down.

No comments:

Post a Comment