• Skip to Content
  • Skip to Sidebar
IU

Indiana University Bloomington Indiana University Bloomington IU Bloomington

Menu

ScIUConversations in Science at Indiana University

  • Home
  • Home
  • About ScIU
  • Write with Us!
  • Contact ScIU
  • The Writers and Editors of ScIU
  • ScIU in the Classroom
  • Annual Science Communication Symposium
  • Search

When human eyes are needed

Posted on January 9, 2018 by Jennifer Sieben

In a world that looks more and more like a science fiction novel, it can be tempting to assume that a computer program or code can do all your tedious work for you. Algorithms suggest videos you might like to to watch on YouTube, plan your road trips to visit friends, and can even frost a cake you ordered for your sister’s birthday with a picture of her face in the center. It doesn’t seem like a leap of faith to assume they can help make astronomical research easier as well.

It should be simple for an algorithm to sort images of blurry dots. But when those dots are galaxies larger than our own, some of which are previously undetected because of how far away they are, it can be hard to train a computer program. Instead, my collaborators and I must visually inspect hundreds of dots and blobs in our images, trying to see if any of them are worth examining in more detail.

A static-like greyscale background with a small faint white dot only a couple pixels in diameter.
An example of the images we have to classify. This dot is one of the bigger dots that may be a galaxy. It might also just be a faint star.

Our project’s goal is to study the star formation rate of galaxies at different times in cosmic history, stretching back to 12 billion years ago. To do this, we take photographs of the night sky, using filters, to only see specific wavelength ranges. These filters help isolate different chemical elements, which are used as clues about how far away a star-forming galaxy is. We also collect more detailed data on how much light is seen at each wavelength. This helps us see the redshift of the galaxy: if a line (a spike in the graph of the spectrum as seen below), like a line indicating the presence of oxygen, has shifted towards a smaller wavelength in order to be seen in our narrow filter, then the object must be very far away. By examining how much a line has shifted, we can determine how far away an object is.

A plot with wavelength on the x axis and intensity on the y axis. The line is mostly flat, with some scatter, with the exception of one thin spike and one thick spike. These indicate the strong presence of certain elements.
A spectrum of the active galaxy NGC 4151. The plot shows the intensity of light at different wavelengths in the ultraviolet range. A large spike indicates emission lines. We expect to obtain similar spectra for the objects selected during the initial classification process. Image credit: NASA/STScI

It’s rather like birthday cards, where different symbols indicate different ages. Cards for toddlers usually have a cute animal with lots of bright colors, those for young adults often depict people drinking, and ones for people over fifty may display gravestones or black and white photographs. No numbers are needed to tell you the age range of the intended recipient.

It may not be a perfect diagnosis, but by using visual cues, like prominent elements, we are able to determine how far away a galaxy is. Hydrogen indicates a galaxy close by, whereas a galaxy with strong Lyman alpha emission indicates a distant and much younger galaxy. There’s only one problem: with only the initial images, we don’t know what element is causing it to show up in our filter. All we can see is a black dot. And this is where human judgement becomes important. We can create a computer code to filter out a lot of things in our images that are obviously not what we are looking for, but the rest require human judgment.

Bright blue beams of light shoot from the center of the image. A dark band of dust is perpendicular to the beam and obscures the center of the image. A translucent blue glow divides the image from top left to bottom right.
An artist’s concept of a quasar surrounded by a ring of gas and dust. Clouds of charged gas orbit above and below this ring. X-rays shoot out from the central region; this is the part the telescopes usually see. In our data, a quasar like this will only appear as a small dot a few pixels across. Image credit: NASA/ESA

Sometimes, the numbers seen by the computer indicate that something is likely to be excluded, but to an astronomer, a different clue is a sign of a possible quasar—an extremely distant galaxy that shoots out energy like a cosmic lighthouse—and we want to keep those. Our code cannot glean that information from the data we give it. It is up to us to look at the comparison images and decide if the dot looks like a camera artifact, an object that is too spread out, or is actually a strong dot that we are searching for.

The work may be tedious, and is sometimes only tolerable when paired with a good playlist or podcast, yet it is necessary. In this case, computer programs are simply not capable of doing the things that we can do ourselves – the job is left to us . By analyzing these images manually, and looking in detail at what elements are being detected, we can determine how far away the objects are and calculate the rate at which they are forming stars. Maybe one day we will be able to train our code to do more, but for now, nothing can quite replace the human judgement.

Edited by Rachel Skipper and Karna Desai

Print Friendly, PDF & Email

Related

Filed under: Cutting-Edge Science at IU, Scientific Methods and TechniquesTagged astronomy, methods

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

Additional Content

Search ScIU

Categories

Tag cloud

#Education #scicomm animal behavior anthropology archaeology astronomy astrophysics Biology biotechnology Black History Month brain cannabinoids cannabis Chemistry climate change conservation coronavirus COVID–19 Diversity in Science diversity in STEM Ecology environment evolution geology history and philosophy of science infectious disease Interdisciplinary Interview Mental Health methods microbiology neuroscience outreach physics Plants primates psychology Research science communication science education Science Outreach science policy Statistics STEM women in STEM

Subscribe

Receive a weekly email with our new content! We will not share or use your information for any other purposes, and you may opt out at any time.

Please, insert a valid email.

Thank you, your email will be added to the mailing list once you click on the link in the confirmation email.

Spam protection has stopped this request. Please contact site owner for help.

This form is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

Current Contributors

  • Log in
  • SPLAT
  • ScIU Guides

Indiana University

Copyright © 2022 The Trustees of Indiana University | Privacy Notice | Accessibility Help

  • Home
  • About ScIU
  • Write with Us!
  • Contact ScIU
  • The Writers and Editors of ScIU
  • ScIU in the Classroom
  • Annual Science Communication Symposium
College of Arts + Sciences

Are you a graduate student at IUB? Would you like to write for ScIU? Email sciucomm@iu.edu


Subscribe

Subscribe By Email

Get every new post delivered right to your inbox.

This form is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

 

Loading Comments...