简单的介绍SDSS Data的流程

简单的介绍了SDSS Data的流程，个人感觉比较简单易懂。http://cas.sdss.org/dr7/en/sdss/data/data.asp

SDSS Terminology

To understand how data are processed, it helps to understand the terms SDSS scientists use to describe the data:

A scanline is data from a single set of CCDs that sweep the same area of sky. Each set of 5 CCDs is housed in a single dewar: each dewar has 6 sets of CCDs separated by about 80% of the CCD width. The area of sky swept by the 6 CCD columns, or "camcols," is called a strip. A given area of sky is imaged by performing two successive scans, offset by almost a CCD width, to fill in a stripe.

The data stream from a single CCD in a scanline is cut into a series of frames that measure 2048 x 1489 pixels and overlap 10% with the adjacent frames. The frames in the 5 filters for the same part of the sky are called a field.

A run is the set of data collected from one continuous pass of the 2.5 m telescope across the sky, covering one strip. Typically, a run lasts for a few hours. 

Spectra

The purpose of the spectroscopic observations is threefold:
Redshifts: To go from the two-dimensional images to a three-dimensional map of the universe, we need to measure redshifts, or how far the object's spectra has been shifted compared to when the object in not moving relative to Earth. Redshifts allow us to estimate the distances to galaxies and quasars.
Classification: We want to know which objects are stars, which are galaxies, which are quasars, and which are new objects yet to be discovered.
Flux/Wavelength: Spectra can tell us detailed properties of objects, such chemical composition.

The spectroscopic data pipeline is designed to output these important quantities.

Like the imaging data, the spectroscopic data are processed by a large pipeline, which takes the input CCD data and outputs completely processed spectra. The first part of the pipeline applies corrections for detector problems and characteristics. These corrections require a number of other pieces of data:
Flat field images: images that help determine how the telescope optics and spectrograph respond to uniform light.
Arc lamps: emission line spectra of a well-understood excited gas (like the neon in neon signs), which allows us to relate the position on the image to wavelength.
Sky spectra: several fibers on each plate are devoted to blank sky; these allow us to subtract off the background spectrum from the sky.
Standard stars: stars that have known properties, used to relate the intensity we measure to proper flux units.

flat field	arc lamp	science observation

Furthermore, a correction is made to account for the absorption of the Earth's atmosphere (telluric correction) and the Doppler shift due to the Earth's motion around the sun (heliocentric correction).

Once all these corrections are applied, the pipeline extracts individual object spectra, and then produces a one-dimensional spectrum (flux as a function of wavelength) for each object. These one-dimensional spectra must be wavelength calibrated, their red and blue halves must be joined, and then the spectra be identified.

The last task, spectral identification, is important but challenging. The spectra of galaxies can vary greatly, and spectra for stars, quasars, and other types of objects look different. Not only do the intrinsic properties of these objects vary, but they can be at different redshifts, meaning we see a different portion of their spectrum. To make sense of all these spectra, the software first tries to find all the emission lines (spectral features due to the emission of specific wavelengths of light from atoms or molecules) and identify them. Then, the entire spectrum is matched against a set of templates - standard spectra of different kinds of objects - that test how well the spectrum matches each template at different redshifts. The best match tells us what type of object we are looking at, and simultaneously, the object's redshift.

A galaxy spectrum at four different redshifts (0.0, 0.05, 0.10, 0.15, 0.20)