Galaxy Zoo: Weird and Wonderful

How do I use Talk?

Once you select "Done & Talk", you will be taken to a page with all the 16 subjects that you have viewed on the classify page. This page will let you comment on the collection of all these subjects. To talk about a specific subject, we recommend that you open the specific subject's talk page in a separate new tab. Use (command+left-click) or (control+left-click) to open a talk page on the subject image you want to talk about.

Why don't you tell us what you think is interesting?

This project is an experiment to see if we can find the most unusual and interesting systems in a large dataset. We're investigating a range of techniques for exploring - and it might well be possible to find specific examples of interesting galaxies - and here we want to be surprised. Just go with your instincts, though it will help to have classified a bit on the main Galaxy Zoo workflow first.

Why does finding unusual galaxies matter?

Galaxy Zoo is fundamentally about understanding galaxies through their shapes, which tell us about their histories. So a galaxy with an unusual shape and/or color properties may have had an unusual history or experienced some unique physical process - volunteers have already found many of these, some of which are being followed up by a Hubble Space Telescope program. Distorted images of galaxies might be gravitational lenses, like those studied by the Space Warps project, and other, completely unexpected objects - like Hanny's Voorwerp - may be lurking. Anything could be hidden in the data! Unusual galaxies offer a unique lens through which we can understand galaxy evolution.

Where do the images come from?

The images are selected from the Hyper Suprime-Cam survey. They are from the second public data release in 2019, which contains nearly half a billion objects. We consider a sample of bright galaxies by filtering out some images flagged with known issues, and then choosing a magnitude (brightness) range of 18.5 < i < 21.5. We use the gri wavelength bands to construct 3-color RGB images. You can read more about the fabulous HSC project here.

How do you select which galaxies are shown?

We can ask our trained machine learning based algorithm to provide the most unusual galaxies (top 1%) among a total of 1.5 million galaxies taken from the HSC project. We add to these 15,000 unusual galaxies a randomly chosen control sample of 85,000 "not unusual" galaxies. The Zooniverse system then randomly pools from this combined 100,000 galaxy sample to display 16 galaxies at a time for you to choose the ones that you think are interesting. Remember, though, that your opinion of what is interesting will be different from what the computer thinks is unusual.

Which images are not interesting?

In the field guide, we show you some examples of images which we don't consider scientifically interesting (though they may be useful for other purposes!). These include astronomical objects which might look odd but are uninteresting or common in this sample, such as red compact objects, which are high-redshift galaxies or red dwarf stars, and bright green compact sources, which are potentially stars. The uninteresting things also include artifacts like satellite trails, empty images, and saturated and corrupted regions from bright stars. Another common issue is off-center objects, which are usually due to over-deblending: an algorithm tasked with detecting the objects in an image "shreds" a galaxy, and assigns a chunk of it to be its own object.