Let me start by congratulating Google for providing us all with a remarkable browser of the Sky...all the functionality of Google Earth (or most of it), with add-ons like various layers of interest with links to additional information. There is a tremendous amount of sky browsing which can be done, to learn about nebulae and galaxies, stars with high proper-motion, variable and multiple stars, and more.
The purpose of this thread, and this compendium, is NOT to collect data problems for the sake of collecting them, or to try to create a comprehensive collection of every single data problem in the GSky dataset, if that were even possible. I'm really not even interested in the data problems, I'd much rather be spending my time learning about the wonders of our universe with the cutting-edge research which is currently in progress.
The only point of this collection is to provide an educational resource to newer users, to demonstrate that the imagery we're looking at in GSky is, in most cases, rather poor, particularly in comparison with professional-quality imagery which is available through other sources on the Web. With a few rare exceptions, GSky has but a single (visible) image of each portion of the sky, with no pedigree to even show when, where, or how the image was taken. The color balance in most of the sky (the DSS area) is badly reproduced. And then there are a variety of image problems from a number of different causes, as you'll see below. These data problems are not at all uncommon, if you spend some time to browse around the GSky imagery.
For an experienced user to find more and more data errors of the same sort, and post them here for me to add to the collection, really serves no purpose. The most common kinds have a typical, characteristic appearance, and once you've looked at a couple of dozen of them, you should be able to recognize them for what they are. An experienced user should see a data problem, recognize it for what it is, and move on without comment. A new user finding the same data problem for the first time has good reason to question what it is he or she is looking at, and in that case, I'm glad to look up the star in SIMBAD or NED, and add it to the collection of problems so that I don't have to look it up again the next time a newcomer finds it. Likewise, if an experienced user finds a data problem which doesn't look like any of the usual suspects, please post it at the bottom for me (and others) to try to figure out what it is.
With permission from PriceCollins, I am modeling my collection after his, using some of his illustrations. His work formed the basis for mine; all I have done is to expand upon his collection of Data Problems in the Google Sky imagery. I have organized the collection into the following categories:
Circuit Boards - There are well over 400 of these in the Google Sky imagery. They are laid out in a regular pattern at 5 degree intervals in the Southern Hemisphere, beginning at declinations between -17 and -47 degrees, and completely circling the Sky in Right Ascension. These are obviously an artifact of the way that the DSS Consortium images are "stitched" together to form a celestial sphere out of flat Sky images. Here are some examples:
Internal Telescopic Reflections - At least 100, ranging from huge internal reflections near the brightest stars in the Sky, to much smaller ones scattered around. The hardware in the inside of the telescope (the "spider" which holds the secondary mirror) is a dead giveaway. Two examples of the kinds of Reflections found in the Google Sky imagery are posted, then two pictures of a telescope "spider" for comparison:
Line segments - perhaps thousands of these in this collection. I've broken them down into Blue, Green, Red, and White folders. A little background: The DSS Consortium imagery, and the SDSS imagery are both based on Palomar Observatory Sky Survey and U.K. Schmidt data, which is typically (at least) three separate pictures taken of the same place in the sky using Blue, Green, and Red filters. The intensity of an object in the various filtered wavelengths provides a lot of information about the spectrum and thus the temperature of the object. Generally speaking, if you find an object which is solid Blue, Green, or Red, that means it may only be a transient object, if it wasn't captured in the other two filtered images. Not necessarily, but it's a good bet.
Most of the line segments are artifacts from aircraft, satellites, or meteorites, passing between the telescope and the area being imaged. There are indicators which can provide a hint as to which is which, for example, short-lived meteorites may show pointy ends with a brighter middle. Satellites frequently tumble with a regular period, and can show dashed lines. Airplanes will usually show the most solid, uniform line width, but can occasionally show slight wobbles from turbulence in the air. Here are some examples of each:
Blue intermittent line, probably a satellite
Green line, pointy at both ends, possibly a meteorite:
Line across multiple filtered images, most likely an aircraft:
White lines are typically the way that "seams" between rectangular black pictures show up, but there are other causes for them, as well. I found a couple hundred. If you zoom in to very tight magnification, features will show up in the white line which some people have interpreted as having meaning of some sort. I interpret as magnifying the object beyond the useful level of value for the data, and spurious features get introduced because of that. Here is an example:
The SAME white line, zoomed in past the level of usefulness, which introduces spurious "features":
Missing data - There's actually quite a bit of missing data. With as many pieces of imagery that it takes to cover the whole Sky, there are bound to be little bits here and there that get left out, and I found about a hundred or so. There is no conspiracy to hide anything from anybody, if a piece of the Google Sky imagery is missing, you can find the data in a variety of other sources, SIMBAD is my favorite external source, because it provides access to many, many other sources of astronomical imagery. But it's also possible that the missing piece is present in Google Sky's imagery at other wavelengths. In the "Layers", "Featured Observatories", you can examine a piece of Sky at InfraRed, Microwave, UV, and X-Ray wavelengths.
Here's an example of missing data, this one's about 0.6 x 1.25 degrees:
Overexposures - In the SDSS imagery which covers about a third of the Northern Celestial Hemisphere, there are so many stars which are overexposed that I didn't even try to catalog them. You can see "rays" extending out from the star, but there is nothing in the sky, no actual, physical object, associated with those "rays"; they are a simple overexposure.
There are other kinds of overexposures and internal reflections, as well. PriceCollins called this one a "filmstrip", which is fitting, although it has nothing to do with a filmstrip.
Professional astronomical telescopes are so powerful that any bright star, even what would be a dim star to the naked eye, can overwhelm the sensitive light-gathering equipment, and create funny effects. Usually you cannot trust the imagery very close to any bright star. Gamma Hercules is a 4th magnitude star, not exceptionally bright by any stretch, but in the POSS imagery, it looks overwhelmingly bright. I call this one the "French Flag overexposure":
Other - There are lots of oddball blobs, smears, linear features, and things which don't fit neatly into any of the above categories. I suspect the mere process of handling the vast quantity of data, from capturing the images on three separate photographic plates with three different filters, developing and processing those plates, digitizing the photo plates into a digital electronic file, transmitting those files around, "stitching" the images together to form a celestial sphere, then compiling them onto a server and finally presenting them over the Web, there are just too many ways little errors can creep into the data. Usually, if I'm interested, I can check the location against external astro data sources, and verify that whatever we're looking at is a spurious glitch, not some real astronomical object, but I haven't taken the time to research all of the features I've been presented with. Here are some examples:
Scratch in the photographic plate:
Bad data, verified nothing present through external data sources:
Another verified piece of bad data; no corresponding object in the SIMBAD database, and one image out of 21 from an Aladin search, showing this is actually a large hairball or piece of fabric:
Another verified piece of bad data; no corresponding object in 14 images from ALADIN:
Remember, folks, the Google Sky data is a wonderful, amazing collection of astronomical imagery which would have seemed completely unimaginable as little as 50 years ago. But it is a human enterprise, ultimately, with human beings involved at every step of the process. Below is a picture of astrophysicist Dr. Miriam Rengel, who has given me permission to use the picture to illustrate the process of digitizing a photographic plate. You can see her using a paintbrush to clean the photographic plate prior to placing it in the digitizer machine in the background. It's obviously not a "clean room" process, and it is surely possible for lint, dust, fingerprints, scratches, or any of a variety of other extraneous material to get into the digitized image.
This Compendium is a work in progress, so if you have something to add, by all means do so. Either send me a PM with a placemark, or simply add a post to the bottom of this thread, and I'll incorporate your additions to my collection. And thank you in advance.
Good luck, and have fun!
_________________________ Wherever you go, there you are.