Accounting for Absorption Lines in Images Obtained with the Chandra X-ray Observatory
In recent years, new telescopes have dramatically improved the ability of astrophysicists to image X-ray sources. The Chandra X-ray Observatory, for example, was launched by the space shuttle Columbia in July 1999 and provides a high resolution tool that produces images at least thirty times sharper than any previous X-ray telescope. The X-rays themselves are produced by matter that is heated to millions of degrees, e.g., by high magnetic ﬁelds, extreme gravity, or explosive forces. Thus, the images provided by such high resolution instruments help astrophysicists to understand the hot and turbulent regions of the universe. Unlocking the information in these images, however, requires subtle anal-
ysis. The detectors aboard Chandra collect data on each X-ray photon that arrives at the detector. Speciﬁcally, the (two-dimensional) sky coordinates, the energy, and the time of arrival of each photon are recorded. Because of instrumental constraints each of these four variables is discretized; the high resolution of Chandra means that this discretization is much ﬁner than what was previously available. For example, one of the instruments aboard Chandra has 4096 × 4096 spatial pixels and 1024 energy bins. Because of the discrete nature of the data, it can be compiled into a four-way table of photon counts. We refer to this four dimensional data as the image; it is a moving ‘colored’ picture. (Because of the high energy of X-rays the ‘colors’ are not in the visible spectrum.) Spectral analysis models the oneway marginal table of the energy data; spatial analysis models the two-way marginal table of sky coordinates; and timing analysis models the oneway marginal table of arrival times. As we shall see, however, because of subtleties in the instrumentation and the data itself, spatial and spectral analysis cannot be fully separated and both are crucial for a full understanding of the image; in other settings, similar concerns arise for spatial and temporal analysis (see Chapter 12).