How It Sees

More than just a pretty picture
by Jeff Hester, Arizona State University

Many people have gazed in awe at the Hubble Space Telescope image (left) taken of the Eagle Nebula, a spectacular star-forming region about 6,500 light-years away. Indeed, I'm proud to say—since I "took" the picture with my colleague Paul Scowen at Arizona State University—that the image is arguably the most famous Hubble photograph ever taken. But few laypeople realize just how much effort goes into preparing such images sent down from our orbiting eye on the universe.

In this feature, I'll walk you through the various steps it took to assemble this image from the raw data, then end with just a bit about what the image tells us about the "Pillars of Creation," as this fascinating star nursery has been dubbed.

But first a little background: The Eagle Nebula image was taken on April 1, 1995, by the Hubble's Wide Field and Planetary Camera 2. Like a digital camera, the WFPC2 uses charge-coupled devices (CCDs) rather than film to record images. CCDs have an array of light-sensitive elements called pixels that turn light falling on them into electrical signals. As we'll see, all the hard work of preparing the final image involves knowing how to interpret each of those signals, removing the "fingerprints" of the camera itself, and turning the signals into measurements of how bright the light is at specific points in the sky.

The WFPC2's field of view is about 1600 by 1600 pixels, which makes it roughly equivalent to a 2.5 megapixel digital camera. While the final image is not true color, it is closer to what your eye would see if you went there than the picture you would get with color film.

Before we begin fashioning the image, I need to reveal why the final picture seems to have had a big bite taken out of its upper-right quadrant.

Size matters
The WFPC2 actually has four separate cameras—three Wide Field (WF) cameras and one Planetary Camera (PC), known as PC1. As this illustration shows, each camera images one quarter of the scene being photographed. All except the PC1, that is. It records a magnified view of its portion, which allows astronomers to see finer detail in that corner of the overall image. The final picture would look strange this way, so we need to scale the PC1 image down in size to proportionally match the other three images. (See previous picture for resized PC1 image.) Now you see why all pictures taken by Hubble's WFPC2 instrument have a stair-step shape.

Let's back up and start at the beginning. A warning: the raw image beamed down from Hubble is a far cry from its polished final version.

Raw Deal
I told you it would look different! Here's why: First, and most obviously, we're looking through a blizzard of cosmic radiation and other spurious signals that we'll soon get rid of. Second, we're only looking at an image from the PC1 camera; we'll get to the images from the three WF cameras in a moment. Third, we're only looking at one of four colors of light that each of the four cameras records. (Just as a TV makes a color picture by combining red, green, and blue images, our final color image will combine images taken through different filters—more on that later.)

Finally, we're only looking at one of 32 different images that go into producing the final picture. That's because each of the four cameras takes two images through each of its four filters. Why the duplicate pictures? Let me explain.

Cosmic graffiti
Each of the WFPC2's four cameras takes two images with each of its four filters for a good reason. Most of the "snowflakes" you see here are cosmic rays—atomic nuclei whizzing about space at close to the speed of light—that struck the CCD at the moment the image was made. Each cosmic ray leaves a streak across the CCD. Like scratches on a photographic negative, these streaks interfere with the scene we want to see, namely, the Eagle Nebula. Fortunately, cosmic rays arrive randomly, so the marks they leave differ from image to image. Click between the two "Cosmic Graffiti" images, which were taken within moments of each other. Notice how the cosmic-ray marks differ in each.

Now we need to clear away all that annoying cosmic "snow."

Goodbye to all that
This is the final cleaned-up image. How did we banish the blizzard? First, we flagged sharp features appearing in one image but not the other as cosmic-ray garbage and discarded them. Then we combined the two "snow"-free images to improve the signal-to-noise ratio. We also corrected for a number of instrumental effects with names like "dark current," "flat fields," and "charge transfer inefficiencies." No need to bore you with details of those effects, even though as a member of the team that built the camera I spent a couple of years figuring out how to characterize and remove them!

Despite all these efforts, some cosmic rays and other defects remain that we need to jettison.

Unsavory characters
One way to find the remaining cosmic rays and other defects is to compare the four images taken with the four different filters. Another way is to compare the pictures with the image profile of a point source such as a star. If we see a sharp bit of light that doesn't have the right profile to be a star, it's probably a cosmic ray residual or other defect that we need to flag and remove. This image shows the remaining undesirable bits flagged in white against the desired image, shown for the moment in red to ease visibility.

Now let's delete the remaining "flurries" and see what we've got.

Spanking clean
This is the final, fully cleaned-up version of the PC1 image—or at least one of the PC1 images. Remember that a picture was taken with the PC1 through each of four filters. Each filter records light emanating from the target object—in our case the Eagle Nebula—in very specific parts of the visible-light spectrum.

Different types of atoms, it turns out, emit light at very specific wavelengths—that is, very specific colors. This allows astronomers like me to take images that only show the light coming from specific types of atom. For the Eagle Nebula image, the WFPC2 used filters that took images in the light of three kinds of atoms: hydrogen atoms, sulfur atoms with one electron removed (sulfur ions), and oxygen atoms with two electrons removed (doubly ionized oxygen). A fourth filter sees only starlight, which lights up the dust that is mixed in with the nebula's gas.

The PC1 image that we've been manipulating was taken with the oxygen filter. The first step in getting the PC1 image together with its Wide Field mates is to obtain cleaned-up versions of the three WF images also taken with the oxygen filter.

Number two
This is the cleaned-up version of the oxygen-filter image taken by the WF2 camera, that is, the camera that took the image of the upper left quadrant of the final image. Note that we have reoriented this image and the next two from the way the WFPC2 actually took them, in order to make them "upright."

Number three
This is the cleaned-up version of the oxygen-filter image taken by the WF3 camera, that is, the camera that took the image of the lower left quadrant of the final image.

Number four
This is the cleaned-up version of the oxygen-filter image taken by the WF4 camera, that is, the camera that took the image of the lower right quadrant of the final image.

Let's now combine the four images into a single portrait of the Eagle Nebula.

Come together
Each of the four images has geometrical distortions similar to those you see when you look through a magnifying glass. To combine the images of the four different cameras into a single, accurate picture of the nebula, we need to map and remove those distortions. Once we've done so, the four images fit together nicely, giving us this single mosaic image of the Pillars of Creation, showing light coming through the oxygen filter.

Now the picture is beginning to resemble its final self. But who wants those ugly seams? Let's erase them.

Wrinkle free
Okay, we've gotten rid of those unsightly lines. But remember, this image only shows light originating from doubly ionized oxygen atoms.

Let's have a look at similarly processed images taken in the light of hydrogen atoms and sulfur ions. As you go through the images, notice the subtle differences between them. Comparing the light coming from different types of atoms is what allows us to turn what otherwise might be just a pretty picture into detailed knowledge of the physical conditions from region to region within the nebula. More on that in a minute.

A striking couple
Here are images of the Eagle Nebula taken in the light given off by its hydrogen atoms (1) and its sulfur ions (2). Click between the two pictures to compare how the nebula looks in the two images.

If you're curious as to where these atoms' so-called emission lines fall on the visible-light spectrum, see the graph below. The wavelengths of the light picked up by the WFPC2's hydrogen and sulfur filters are 6563 and 6725 angstroms, respectively. (One angstrom is equal to one ten-billionth of a meter.) The corresponding wavelength for light coming through the oxygen filter is 5007 angstroms.

We now have three images, each showing us light from a different type of atom. Now, the best way to see the differences among these three images is to make each of the three a different color, then use our color vision to tell us how the nebula's properties are changing from place to place.

Blue sky
Here, we have assigned the color blue to the oxygen-filter image.

Heavenly hues
We've cast the hydrogen-filter image in green and the sulfur-filter image in red. Compare the three color images to see how the colors enhance the subtle structural variations that the two pictures reveal in the nebula.

Now for the grand finale—combining all three processed color images.

The Eagle Nebula
Here is the final, fully processed photograph that NASA released in 1995. Now, what does the image tell us about what we're seeing? Basically we're looking at dense clouds of molecular hydrogen gas and dust that form just one part of the Eagle Nebula. This is a large area: From top to bottom, the pillar at left extends about four light-years, or the distance light travels in four years going at roughly 186,000 miles per second!

Parts of the clouds, particularly the finger-like projections you can see at various points along the pillars, are dense enough to collapse under their own weight, forming young stars. These embryonic stars continue growing as long as they can draw mass from the surrounding clouds. But ultraviolet light from massive newborn stars (unseen off the top edge of the picture) is eating away at the pillars, essentially evaporating the gas in these clouds. Such "photoevaporation" gives shape to the pillars and illuminates the fog of gas burning off their edges.

If you could go back 4.5 billion years and watch as our sun and solar system formed, you would probably see a region very much like what you see in the Eagle Nebula today. There's a lot more science having to do with understanding the detailed physics of the gas and what images like this tell us about how stars form. But now you have at least a rough idea of both what we're learning from this single snapshot from the Hubble Space Telescope and what it takes to produce all those gorgeous images taken in orbit. For more on how the Hubble works, see the official Hubble Web site.