Galaxy Zoo 2: Return of the Citizen Scientist

The Sloan Digital Sky Survey (SDSS), an observing program taking imaging and spectroscopic data of over 35% of the sky, has imaged over a million galaxies since it began in 2000. This huge sample size provides great opportunities for statistical analyses of galaxy populations. For example, by calculating the frequency of spiral galaxies vs. elliptical galaxies, we can better understand how galaxies evolve. But how do we analyze such a huge dataset? Even just classifying a galaxy as spiral or elliptical is difficult. Computer image analysis isn’t as accurate as visual inspection by a human, but it would take too long for a handful of grad students to sort the images.

The Galaxy Zoo project was created to address this problem using the massive amount of manpower available on the Internet. (See our guest post by Galaxy Zoo co-founder Kevin Schawinski for more information.) Starting in 2007, they asked for volunteers to sort galaxy images as either elliptical, spiral, or mergers. Eventually 250,000 of these citizen scientists produced classifications of ~ 1 million galaxies. Each galaxy was analyzed by multiple volunteers to improve the accuracy of the classifications. Since then, dozens of papers have been written using the original Galaxy Zoo 1 results, including a few that we’ve covered here on astrobites.

Starting in 2009, Galaxy Zoo was updated to include more detailed classifications on the 300,000 brightest galaxies in the Galaxy Zoo 1 sample. The citizen scientists in Galaxy Zoo 2 were asked to look at galaxy characteristics like the presence of a bar and the tightness of the arms in spiral galaxies, and the roundness of ellipticals.  Fig. 1 shows the flowchart followed by a citizen scientist for classifying an image in Galaxy Zoo 2. This paper presents the results of over 300,000 of these newer, more detailed classifications.

Galaxy Zoo 2 Flowchart

Fig 1: The flowchart used by citizen scientists to classify galaxy images in Galaxy Zoo 2.

Each citizen scientists’ votes are weighted by how consistent their classifications are with the majority. The authors of this paper also corrected for various classification biases. For example, fine-detail structures like bars and spiral arms are harder to identify in galaxies at higher redshift (which are dimmer and smaller in the images), so this redshift bias must be corrected in the final results.  Fig. 2 shows the fraction of votes vs. redshift for each response in Fig 1. Note that the vote fractions before the redshift bias correction (solid lines) show a decrease in fine features at high redshift compared with the vote fractions after the redshift bias correction (dashed lines). This can be seen particularly in the plots for “Smooth or features”, “Bar”, and “Spiral structure.”

Vote Fraction Plots

Fig. 2: Vote fractions vs. redshift for each classification step, before (solid line) and after (dashed line) redshift bias correction. The dashed vertical lines indication the redshift sensitivity limit of SDSS (left vertical line) and the lower limit of the correction (right vertical line).

The authors compared the results of Galaxy Zoo 2 with the citizen scientist classifications of Galaxy Zoo 1, recent computerized classifications by Huertas-Company et al. (2011), and expert visual classification by Nair & Abraham (2010) and Baillard et al. (2011). They found that Galaxy Zoo 2 categorization of spiral vs. elliptical galaxies was roughly consistent with Galaxy Zoo 1 and computerized classification (although neither Galaxy Zoo 1 nor the computerized classification attempted to identify fine structure like bars or tightly-wound spiral arms). The identification of strong bars and rings in Galaxy Zoo 2 was consistent with expert classification results, but Galaxy Zoo 2 had trouble identifying fainter bars and rings.

The Galaxy Zoo 2 classification catalogue is now available to the public. You can check it out at If you want to try your hand at classifying galaxies, go to the Galaxy Zoo site at Or take a look at our Citizen Science glossary page, which lists lots of astronomy citizen science projects for you to explore.

About Erika Nesvold

I'm a graduate student in the Physics Department at the University of Maryland, Baltimore County. I do my research at NASA/Goddard Space Flight Center with Marc Kuchner. I'm writing a model of debris disks to understand the way disks and planets interact, which will help us find exoplanets using images of disks. My model includes collisions, so I spend a lot of my day thinking about asteroids smashing into each other.

Discover more from astrobites

Subscribe to get the latest posts to your email.

1 Comment

  1. The stuff is extremely intriguing.|



  1. Galaxy Zoo 2: GÖNÜLLÜ BİLİMADAMLARININ DÖNÜŞÜ | Açık Bilim - Aylık Çevrimiçi Bilim Dergisi - [...] [...]

Leave a Reply