Blue Collar Scientist

Last night was the first run of the new control system for the 0.8 meter telescope at Junk Bond (take a crappy video tour here), and everything worked great. That’s really no surprise, because this is probably the best-tested high level control system and observation scheduler that I’ve ever written. It incorporates a number of new features (most of which are strategic secrets), and also has a lot more error condition handling and correction in it. I’ve been putting it through its paces on the simulators for the last 12 weeks, and that work has definitely paid off in a flawless first-night deployment run with a very complicated schedule request from the astronomer.

Another thing I did last night was take time-lapse photos of the night sky. I have a Canon 5D with a circular fisheye, so I get full-sky coverage with the horizon all around, and for quite a while now I’ve wanted to make a dusk-to-dawn animation of the night sky. I did my first experiments along these lines back in February of 2007, but on my trip then I didn’t have the best skies to play with - we had a lot of clouds. So I ended up getting only a single night, and that was interrupted by clouds that came in around 3:00 AM and started raining on the camera. These are obviously sub-optimal conditions for taking pictures of the stars. Still, it was enough to learn what I was doing wrong and figure out some solutions.

The first time out, I set up the camera on a tripod, pointing straight up, with an intervalometer controlling the exposures (an intervalometer is a doodad that takes a picture of a certain duration every so many seconds or mintues). I elected to use 30 second exposures at ISO 1600 with a two-second refractory period. Periodically throughout the night I would swap out CF cards and batteries, which really sucked.

This time, things are way better. First, I’m running the camera off AC power, using a Canon adapter. That means I don’t need to swap batteries all night, and I can get some sleep. Second, the camera is operating tethered, connected to my MacBook. This means that none of the images are stored on the CF card; they are downloaded directly to the MacBook hard drive. I bought Aperture solely for this purpose.

Now, I’ve tethered with my Windows PC’s before, and I’ve even tethered with Linux, and it can be a painful experience. Kinda like the Windows implementation TWAIN, but worse, you have to get a series of drivers and abstraction layers installed and working, creating the correct pipeline, before you start your tethering application. Typically one step of this process involves installing a camera driver from Canon, which you have to download from their disorganized and arcane website¹. If you don’t get it from their website you can install it from the CD that came with your camera - and hopefully you can avoid installing all the other crapware that comes on that disk. Besides this, Canon are not, in my opinion, the best at writing drivers (or supporting them), so the whole thing takes on the appearance of pig-herding exercise.

If you get all this driver software installed, you then need to tell the application you are using to open a tether with the camera. This usually works fine, once you get all the other stuff set up, but it is all just something of a pain in the neck.

But Aperture is a bit lot more elegant. I didn’t install any drivers at all; I just opened Aperture, selected “start tether,” and I was off to the races. Imagine that - two clicks and it works, no system configuration needed.

The upshot is that my camera took 1,400 exposures last night with essentially no intervention from me. Everything ran rock-steady the whole time and nothing blew up. That allowed me to concentrate on some other questions that I had about how to best take these images, namely, how to expose the fading twilight in the evening and brightening twilight in the morning so that the animation looked pretty much natural. It was non-obvious to me how to change the exposure to track the changing sky brightness to come up with a decent-looking movie.

It now appears that I have that figured out as well, so tonight, if it remains clear (and it sure looks like it will), I’ll be executing version 3.0 of my all-sky, all-night time lapse. I’m pretty excited about it, too!

1. One time I downloaded and installed a driver that only spoke Korean, to my dismay. [↩]

Welcome, Mix 103.1 Listeners!

It is often believed that on the equinox, the length of daylight and nighttime are the same. The very term “equinox,” with its similarity to equality, implies it. But in reality, there’s more daylight than nighttime on the date of equinox.

The moment of equinox is when the center of the sun is directly over the equator. If you were standing on the equator, the sun would pass directly overhead. But the sun wouldn’t stay above the horizon for twelve hours, and then set, leading to twelve hours of night. If you had a stopwatch and timed it, you’d find there were several minutes of excess daylight at the equator. And if you timed it in Alaska, the excess would be far greater - more than a third of an hour!

Why would this be?

The biggest factor in messing up this otherwise perfect symmetry is our atmosphere. When looking along the horizon, we are looking through a vast sea of air, which refracts the light from all celestial bodies - including the sun. The result of this refraction is to make everything low to the horizon appear to be higher in the sky than it really is. The effect is so extreme that while the sun is still below the horizon, the atmosphere refracts its image and makes it appear to have already risen. At the horizon, atmospheric refraction bends this image by more than the apparent size of the sun. If we took the atmosphere away the instant that the entire sun became visible over the horizon, the sun would disappear, and it would be a few seconds before it rose a second time.

Of course this effect works in reverse as well. Because the atmosphere makes the sun appear higher in the sky, during sunset the real sun is below the horizon even though the image of the sun is still visible.

How much of a difference this makes depends on your latitude. On the equator, when the sun rises it goes directly up, gaining altitude but remaining over whatever distant mountain or building it rose from behind. So the difference in time between the optical illusion of sunrise and the moment when the actual sun would appear above the horizon if there weren’t an atmosphere is only a few dozen seconds. Similarly, during sunset, the sun plunges straight below the horizon, leading to a similar time difference.

But from more northerly and southerly latitudes, the sun moves to the left or the right during sunrise and sunset. At the latitude of Anchorage, this motion is to the right, and the sun actually moves more to the right than vertically during the course of the day. One consequence of this is that sunrise and sunset are protracted, long-lasting events. When the first bit of the sun appears above the horizon, it does not appear to be in as big a hurry and it doesn’t rise straight up into the sky; it skirts along the horizon, allowing the sunrise to linger for many minutes. This means that the image of the sun appears above the horizon for quite a long while before the “true” sun would appear if there were no atmosphere. And again, the same process works during sunset to further extend the period of daylight.

The end result is that we get far more daylight on the equinox than we do nighttime. Today, sunrise was at 7:56 AM, and sunset will be at 8:19 PM. That’s 23 minutes of excess daylight! For us, here in Anchorage, we had equal daylight and nighttime on Monday (March 17) - when the sun rose at 8:08 AM and set at 8:08 PM.