In a recent post I compared the astronomy I do to tending to a herd of dairy cows, since the only way to do it right is to be available for it every day. Presumably some of you were out there thinking, “What the heck does he know about Dairy Farming?” The less generous of you might have been thinking, “What the heck does he know about astronomy?” Steadfastly ignoring these concerns for my breadth of knowledge I would like today to compare doing science to writing poetry, a comparison I make for my students regularly. The goal in each case is something big and beautiful; the daily reality is something entirely different. In science or poetry the only way to get near the meaningful and beautiful is to sweat every detail. Then sweat them all again, at least once more. Astronomy for me has involved spending every night for a month in a windowless basement lab, tediously (and more importantly – carefully) measuring the light transmission efficiency as a function of wavelength for every scintillating fiber available, then modeling the results just as carefully. Long, long weekends of work on photomultiplier bases and pre-amps have led to full weeks of all food tasting of solder flux. None of this is new or novel. It’s just how it gets done. I often have my students read the outstanding book Parallax by Alan W. Hirshfeld so that they might feel all this effort as part of something larger and to encourage them not to lose sight of the grand picture in the difficult details. I hope the students can relate when Hirshfeld describes the painstaking work Bessel completed to measure the parallax of 61 Cygni and he closes with, “John Herschel’s Sole ‘particular’ was whether Bessel had taken into account the effect of temperature on the calibration of the heliometer screw. Bessel, of course, had.”
Because we have so much data of a single field of 1600 stars, the data of a fairly significant temporal span and density, we decided this summer to think about getting into the fray over whether or not long-period pulsating variables occasionally show large outbursts of unknown origin. There have been alternating claims that they do and then that they don’t. See an article by Wozniak et al. for a starting point. A big problem here is that these stars are very red and red stars are difficult to measure with photometric precision. On top of that we shoot almost unheard of unfiltered images that are highly sensitive to the atmospheric variations found at our moist, deep atmosphere site. Still, we have the data and bright students seeking challenging projects so we are compelled to try. The plot below shows bumps in a small slice of a long-period variable light curve. We are trying to determine if these are astronomical in origin or normalization-induced.
We test the photometric quality of each night in our data set by measuring the spread in the values of each star's measured signal on that night divided by that star’s mean signal for the entire season. We restrict to the brightest 50% of the stars in the field and remove known variables. If the spread of these nightly over seasonal values is deemed too large the night is removed. As we started the LPV bump work this summer, one of the first things we noticed was a data column misalignment for some stars in 2009 and 2010, leading to the wrong nights being included or removed. Slowly we went through each star’s data file to check this alignment by eye. This work led us to find that two nights were missing and then to note that a correction factor used to adjust the data to a new camera had not been applied to a handful of nights. That’s how we spent our summer, meticulously hand-checking every night of data before we even think about trying to rule out these “events” by devising the right statistical test. It feels like we have found all the potential problems with the mechanics of the data files but then it usually feels that way.