signal-to-noise ratio
/ S-N-R /
Try to hear a friend whisper at a loud party and you meet the core problem of all measurement: is what you detect real signal, or just noise? In astronomy the signal is the light from your target; the noise is the random graininess that always rides along with it. The signal-to-noise ratio measures how confidently the signal stands above that noise.
Noise has several sources — the random arrival of photons (even from a steady source the count fluctuates), the glow of the sky, and the electronics of the detector — but the most fundamental is photon noise, which grows as the square root of the number of photons collected. Because the signal grows with the photon count itself while the noise grows only as its square root, the signal-to-noise ratio rises as the square root of the total light gathered. Collect four times as many photons and the ratio doubles; a detection at signal-to-noise of 10 means the signal is ten times the typical noise wiggle, a solid result, while a ratio near 1 is lost in the grain.
This square-root law governs how astronomers spend their nights. To improve the signal-to-noise on a faint object you must gather more photons — by using a bigger aperture, a more efficient detector, or simply exposing longer, called integration time. But the returns diminish: doubling the signal-to-noise on a faint galaxy means a fourfold longer exposure. This trade-off between depth and time, more than anything else, decides what a telescope can realistically observe.
Hubble's Ultra Deep Field combined nearly a million seconds of exposure — over eleven days of staring at one tiny dark patch — to build the signal-to-noise needed to reveal thousands of galaxies billions of light-years away.
Depth is bought with the square root of time.
Doubling exposure time does not double the signal-to-noise — it improves only as the square root, so four times the time for twice the depth. This diminishing return is why the very deepest images cost so dearly in telescope hours.