The Milky Way from Surbiton, de-rotated
Peter, on first seeing it: "that is so beautiful."
What it is
astrocam is a fixed camera pointed north from a garden in Surbiton, taking a one-minute exposure all night. It does not track, so the sky turns through the frames. On 7-8 October, a clear night with no Moon, it took 533 usable frames.
Each frame was turned about the pole so its stars landed where they were at 00:05 BST, the city's sky glow was subtracted, the trees and the edge of the roof were masked, and the frames were added. That needs the camera's geometry known to under a binned pixel across the whole frame (a binned pixel is a 2 x 2 set of photosites, about 2 arcminutes here), and it is: stars come out as points from edge to edge. Each frame was also spread onto a grid twice as fine (drizzling), so the picture is at the full resolution of the sensor, not of the binned pixels the geometry works in.
When the Milky Way was there
Not all night, on either night. astrocam looks north, and as the night turns, Cassiopeia and Perseus climb towards the zenith and out of the top of the frame. About 40 degrees of the band were in view from dusk to about 02:15 BST, less by 02:30, and none from 03:00 until dawn. Early in the evening it was in view but the sky was 60% brighter (last figure), so it only stands out from about 22:00.
The stack counts each patch of sky only in the frames that actually saw it. Polaris is in all 533; the Double Cluster in 298, from 20:51 to 01:54 BST; Mirfak in 207 and Caph in 206. So the Milky Way here is four to five hours of sky, not nine.
What is in it
- The Milky Way, the disc of our own galaxy seen edge-on from inside. Here it is the autumn stretch, from Cassiopeia down through Perseus, and it follows the galactic equator closely.
- Dark lanes along the band near Cassiopeia. These are dust clouds in the disc blocking the stars behind them. The darkest runs just beside the middle line of the disc, and it is in the same place on both nights while it turned with the stars through each one, so it is the sky and not the camera.
- The Double Cluster in Perseus and the Alpha Persei cluster around Mirfak, two groups of young stars, both obvious knots.
- Polaris, Kochab in the Little Dipper, Errai in Cepheus, Algol.
How deep
The faintest stars picked out are about magnitude 9: binocular depth, about 100 times fainter than the faintest star the naked eye sees from here (about magnitude 4, under a sky of 18.1 magnitudes per square arcsecond).
Adding 533 frames should have gone over 3 magnitudes deeper than one frame. It went about 1. The reason is in the picture: at 2 arcminutes a pixel, every pixel already holds several faint stars too close together to separate. That grain of unresolved stars is what makes the Milky Way look like the Milky Way, and it is also the floor. More nights will not go deeper with this lens; only a longer one will.
A second night, with shorter frames
On 9 October astrocam switched from one-minute frames to 15 seconds, and that night was dark and clear too, so here it is the same way. A star crosses the sensor while the shutter is open; in a minute it draws a dash, in 15 seconds a quarter of one. Turning and adding the frames cannot undo the dash, so the shorter frames give rounder stars.
They also go deeper where it is crowded. Measured against the Gaia catalogue in the same patches of sky on both nights (magnitude at which half the catalogue stars are picked out):
| patch | 7-8 Oct, 60 s | 9-10 Oct, 15 s |
|---|---|---|
| Milky Way, Cassiopeia | 7.96 | 8.35 |
| Milky Way, Perseus | 8.14 | 8.59 |
| centre of the frame | 8.97 | 9.00 |
| below the pole | 9.02 | 9.10 |
| near the pole | 9.44 | 8.92 |
In the Milky Way the 15 second night goes about 0.4 magnitudes deeper, with no more sky time on the band and in Cassiopeia much less. That is where stars are packed tightest, so a tighter star stands out from its neighbours. Away from the band it is a draw. Near the pole stars barely move in a minute, so shorter frames gain nothing there; this night also had less time overall (7.0 hours against 8.9) and stronger rings from the cover glass around Polaris.
So shorter frames help, but they do not break the floor: the faintest stars are still about magnitude 9, set by the size of the pixels.
What else shows
- Faint rings around Polaris, top left. These are the cover glass, not the sky: anything fixed to the camera is swept into circles about the pole when the frames are turned. See Newton's rings on astrocam.
- Stars near the edges are slightly stretched. That is the minute-long exposure's own trail, not an error in the geometry.