Image quality is affected by many factors some of which come from the environment. Seeing conditions, the Moon, light pollution, and clouds all affect the quality and accuracy of astronomical observations and images.
Astronomical seeing describes the conditions of the night sky and how suitable it is for astronomical observations. Turbulence and temperature variations in the Earth's atmosphere cause astronomical objects to appear to twinkle and form blurry images, and places a limit on a telescope's ability to resolve stars. These effects can come from anywhere between the air in the telescope itself to air in the high atmosphere. A seeing disk is the angular diameter of a star's image, or the region in which the star appears to be moving, which is spread out because of the motion in the air that the light traveled through to get to the CCD. At the very best astronomical sites in the world, this disk is usually less than 1 arcsec in diameter and sometimes as small as 0.25 arcsec. Bad seeing conditions can have seeing disks of 4 arcseconds or more.
Bad seeing caused by temperature differences between the telescope and the surrounding air is called tube currents. Many observatories attempt to minimize this effect by keeping the observatory air conditioned during the day to the temperature forecast for the time the observatory will open in the evening. The less of a difference between the telescope and the outside air temperatures, the less tube currents will be present. Bad seeing can also be caused by warm ground, such as hot asphalt near a telescope. Many observatories attempt to minimize these effects by planning as few roads and as little development as possible in the surrounding area. Seeing conditions are generally better at high elevations because the atmosphere is thinner, which is why most of the world's great observatories are located on high mountain peaks.
This is a gif "movie" made of 8 individual frames taken from a video of the Lunar crater Clavius. It shows the effect of our Earth's atmosphere on astronomical images.
The Moon, especially when it is full or close to full, fills the sky with light. This extra light fills many pixels on the CCD with light from the moon, making observations of faint objects difficult or impossible. LCOGT's telescopes are often scheduled for maintenance during the few days surrounding the full Moon. Light pollution from cities also makes observations more difficult in highly populated areas for the same reason as the Moon.
Image quality is also affected by various properties of CCDs themselves, as well as the read out process and telescope optics. CCDs are subject to variations in the sensitivity of the individual pixels as well as noise and imperfections coming from the optics of the telescope. These can cause a variety of effects such as in the images below:
Traps are pixels on a CCD that behave irregularly and cause effects that look like dark lines. A pixel behving like a trap only allows electrons to travel through it when it has a certain number itself. In these cases, a CCD will display traps when a short exposure is taken that does not allow enough photons to hit the trapping pixel. If, however, a longer exposure is used, or a star or other bright object appears over the area of the trap pixel, the pixel will behave normally and no dark lines will be seen. The dark lines on the left side of this image are the result of several traps.
Dust on either the filter, the window protecting the CCD, or any of the corrective optics will leave little donut shapes on an image like the one below. They appear as rings because the dust grains lie on optical surfaces above the focal plane so when they cast a shadow on the CCD, it is out of focus. Astronomers can measure the size of a dust ring and tell exactly where in the optics the grain of dust lies. Dust grains on the CCD itself leaves little dark spots. Calibration frames completely remove dust spots and rings from images.
This effect is caused by using an exposure that is too long. As pixels fill up and can no longer hold extra charge, they spill over into adjacent pixels in the column or columns. This is a very overexposed image of Jupiter.
These features are caused by the bars that support the secondary mirror on a reflecting telescope. The shape and angle of the diffraction features depends on the angle and orientation of these supports, which differ from telescope to telescope. In the image below, the x-shaped rays coming off the brighter stars are caused by the secondary mirror supports of the Sedgwick telescope.
High energy particles sometimes hit a CCD during an exposure and leave bright, sharp spots or lines on an image. They are easy to tell apart from stars which cover a wider area of pixels. Some of these cosmic rays come from supernovae, black holes and other objects in the universe, and some come from the decay of radioactive atoms in the optics of the telescope. Astronomers can use computer algorithms to remove cosmic rays from thier images. They use the fact that stars and other objects have smooth edges, while cosmic rays have bright, sharp edges to identify them.
Astronomers use several calibration techniques to compensate for many of the irregularities mentioned above.
A dark frame is an image taken for a similar length of exposure as the planned observation exposures, and the planned flat frames (see below), but with the camera shutter closed so no light can reach the CCD. This frame contains an image of the noise caused by dark current (electrons moving randomly in the CCD) and read out noise. Often the exact length of observations a telescope will be making is not known in advance, so astronomers make a master dark by averaging several long exposure dark frames with the same exposure length, and dividing by the number of seconds of the exposure. In many telescopes the CCDs are kept very cool with liquid nitrogen because dark current is mainly a consequence of heat in the CCD.
These are also commonly known as biases or bias frames. A zero second exposure with the shutter closed records any noise in the CCD caused specifically by the read out process and the computer being used. Some of this noise will be random from exposure to exposure and some of it will have a repetative pattern. It is important to take many zeros and average them to get a useful calibration frame, called a master zero or master bias. This frame won't be able to remove all the noise, but it can lessen the effects of any pattern in the noise. With LCOGT's 2m telescopes, it takes around 22 seconds for each zero to be read out.
Flat frames are images taken of a uniformly illuminated surface such as a section of the sky at twilight or a special screen made for this purpose. The flat frame exposures are timed so that each pixel will be filled to about three quarters of its capacity. This process records variations in the sensitivity of individual pixels, as well as intrinsic variability of the CCD, dust and other obstructions in the light path. It is critical to take flat fields in the same filter as the science data to be processed since different wavelengths of light behave differently as they pass through the telescope's optics and interact with the CCD.
Once all of these calibration frames have been taken, the flat frames will be averaged. The master zero (or master bias) is subtracted first, then the master dark. The resulting frame is called a master flat.
For every image taken with a CCD, a master dark is multiplied by the number of seconds of the exposure of the image. This result is then subracted from the image. The master flat is then divided into the dark-subtracted image, and the image is as accurate and free of noise patters as possible.