On behalf of the Andromeda Project science team: THANK YOU! We’re (already) done!
Wow. That was unexpected! Before we launched we had expected this second round might take us months to finish. It just goes to show – never underestimate the power of the Zooniverse!
The second round of Andromeda Project took a little more than a week, during which we collected ~750,000 image classifications. This brings the grand total for the project up to ~1.8 million images searched! We had more than 5,000 volunteers hunt clusters this round. In other words, this was a tremendous result that (again!) surpassed our expectations – very impressive!
The science team will get straight to work: over the next few weeks, we’ll combine everyone’s image classifications to create a final catalog of clusters and galaxies that you’ve help identify. Then the real fun begins! We will use the Andromeda Project cluster catalog to study individual, interesting star clusters, and to answer questions about the galaxy-wide cluster population.
And we want to give a special thank you to I F*&king Love Science, who helped bring many of you to us to create this unprecedented spike in image classifications:
Certainly a record number per hour for the Andromeda Project!
While this marks the end of our efforts to search the PHAT data for clusters, we are working on a few ideas for other Andromeda-related projects that we think the Zooites can help us on. Stay tuned here on this blog! We’ll keep everyone updated on our progress over the next few months.
We’re back! After a successful first round last December, we are relaunching the Andromeda Project for a second round of cluster finding in our galaxy neighbor, Andromeda (aka: M31). More than 10,000 volunteers pitched in during Round 1 to search about 80% of the Panchromatic Hubble Andromeda Treasury (PHAT) survey imaging. The Hubble Space Telescope (HST) was able to image the remaining portions of the galaxy over the past year, completing a >3 year effort to survey Andromeda on October 12th. In Round 2, we will complete our cluster search of the remaining PHAT images as well as search other Hubble images of M31, which I will talk about a bit more later on.
Previously on this blog I’ve talked about the results we’ve been able to derive, thanks to the efforts of the AP community, but I wanted to emphasize again how invaluable everyone’s participation has been. Using the first round identifications, we constructed a star cluster catalog containing almost 3000 star clusters, representing a 10x increase over previous catalogs of that portion of Andromeda. The quality of this catalog has allowed us to start answering questions about how clusters form and evolve within the galaxy.
In Round 2, we include two sets of Andromeda images: the final PHAT images and images obtained from the HST archive. The PHAT images we feature this round are 1) those that were observed in the past year (or past month, even!) and 2) images that stretch out to the limits of the survey. You may encounter blank, black portions of images during your searching. No need to worry – these parts of the image represent portions of sky not imaged as part of the survey. Also, you might occasionally encounter a synthetic cluster – an object that we’ve added into the images. As Matt talked about in last week’s blog post, these synthetic clusters are crucial to helping us understand what objects we can and cannot detect.
So, how about the other set of images? What is the HST archive? The Space Telescope Science Institute (STSci), who operates Hubble, hosts a website called the Barbara A. Mikulski Archive for Space Telescopes (MAST) where data from space observatories like Hubble (and other notable missions such as Kepler and GALEX) are stored. This archive is critically important to astronomers, allowing them to find and use images or spectra that were obtained by instruments on Hubble.
Every observation taken by HST is stored and made publicly available, ensuring that they can be used for science in the future. The open availability of these data allow observations to be used by multiple researchers, sometimes in ways that were not originally anticipated. The Andromeda Project is a perfect example of this repurposing: the archival images we include in Round 2 are images obtained as part of an observing program to follow-up on candidate gravitational microlensing events detected within M31. Now we can reuse these images to assist in our study of star clusters!
Remember: these images are publicly available to anyone. That means you! The folks at MAST host a website called the Hubble Legacy Archive (HLA) that provides users an interface to search and view archival images. For an introduction to the HLA, check out this page. Many of our PHAT images are already available for browsing and download through the HLA as well. Try searching for “M31”; some of my other favorites include “M82” or the “Antennae”.
On behalf of the whole Andromeda Project team, I thank you for your interest and assistance. Be sure to sign in when you visit andromedaproject.org and let’s go find some clusters!
For many of you, the most satisfying (or frustrating?) part of Round 1 was clicking on a synthetic cluster. This post explains what these clusters are and how we’re using them. Synthetic clusters help us to calibrate your cluster clicks. These clusters serve a similar purpose to a control group in biological and psychological experiments. We made the synthetic clusters, so we already know their age, mass, and size. We can use these to understand not only what types of clusters we find, but also what clusters we don’t find. For example, say that we find there are no clusters in Andromeda with ages between 100 and 200 million years old. This could either be because there are no clusters at these ages (an interesting science result!!!) or because we just can’t find them because they aren’t easy to detect. Only through using synthetic clusters can we differentiate between these scenarios.
For Round 1, we generated 3100 of these clusters and inserted them in random locations in 3100 selected images. Here’s an example of a portion of a field without and with a synthetic cluster:
We designed the synthetic clusters to be identified at about a 50% rate in Round 1. You exceeded our expectations by identifying ~65% of synthetic clusters! Here’s a plot showing how you did on identifying each of the Round 1 synthetic clusters:
Each dot in this plot is one of the 3100 synthetic clusters. The horizontal axis of this plot is the cluster’s brightness and the vertical axis is the cluster’s Round 1 clusterfrac (the fraction of all viewers that identified the object as a cluster). Color on this plot is used to indicate cluster age. The bright young clusters were the most easily identified, which you can see as the group in the upper left of the plot. On the other hand, old, reddish and dim clusters were the most difficult to find; these are at the bottom right of the plot. You can see that younger clusters tend to be brighter than older clusters; this is because big hot bright stars burn out quickly, and smaller dimmer stars live much longer lives, so clusters get dimmer as they get older. Younger clusters also often contain large well-resolved blue stars, making them easier to identify, while older clusters have fewer big blue stars and often appear as partially resolved orange patches, making them harder to identify.
One of the most important things we can learn from our cluster sample is what mass of clusters we can detect. The completeness is the fraction of clusters at a given mass that were identified. The plot below shows this completeness as we vary the clusterfrac. This plot includes clusters less than 100 million years old.
The horizontal axis is the cluster’s mass in solar masses and the vertical axis is the completeness, or the fraction of all young synthetic clusters that you identified. If you just look at the blue line it tells you that at 1,000 solar masses, nearly 80% of young synthetic clusters were identified by at least 35% of you. The take home message of this plot is that we’re detecting most of the clusters with masses more than a few hundred times the mass of the sun; this is a much lower mass than we can detect in most galaxies outside the Milky Way.
But our search for the truth is not yet complete! There will be more synthetic clusters in Round 2. We’ll be using these to test for consistency between Round 1 and 2 and to correct for biases in the detection of some clusters.