Blends, Blends Blends

The science team is working on incorporating data from the Hyper Suprime-Cam (HSC) survey into Comet Hunters.  We started with the archival Suprime-Cam data first to get a better understanding of what are the false positives and challenges for identifying Main-Belt Comets (MBCs) in data from 8-10-m class telescopes. We’ll continue with both datasets as there’s more Suprime-Cam asteroids, but when we have the chance we’ll move to reviewing the new HSC observations hopefully a few days after they’re taken.Most previous asteroid detection surveys are using 1-3-m class telescopes, so there are bound to be surprises that we wanted to know about before we developed the decision tree for the HSC snapshots on to the site. So we launched Comet Hunters with the archival Suprime-Cam images first. Now that things are going smoothly, we can turn our attention to the HSC data.

We combined your classifications from the first batch of Suprime-Cam images and had 125 candidates in need of further vetting. Thanks to volunteer Tadeáš Cernohous who on Talk went through our list comparing repeat images of the asteroid at slightly different positions in the same batch of subjects. What we learned that all of the candidates are unfortunately blends with stationary background sources. There are lots of faint background blobs that the asteroid moves on top of overlapping in the images creating very tail-like features. All of these images the science team would have had said has a tail.

A few examples are below (all blends with faint background sources):

Looks like a candidate tail, but a blend with a background source. We’d want this marked as ‘has a tail’ in the classification interface.

Looks like a candidate tail, but a blend with a background source. We’d want this marked as ‘has a tail’ in the classification interface.

Looks like a candidate tail, but a blend with a background source. We’d want this marked as ‘has a tail’ in the classification interface.

Looks like a candidate tail, but a blend with a background source. We’d want this marked as ‘has a tail’ in the classification interface.

There’s a lot more blends than we had anticipated given some of the team’s past experience with 2-m asteroid survey data. It’s still very much worth digging into the rest of the Suprime-Cam archive to look for MBCs. There might be many blends, but there could still be undiscovered MBCs too! Knowing that the background blends rate is much higher  because of the increase in the photon collecting bucket is extremely useful. From the candidates, we could see the blends are faint blobby structures that would be likely hard to get a source extractor to pick up in all cases.  Because of the quality of the HSC data and the repeat observation cadence we can try and take this into account possibly by checking the image of the asteroid and the repeat image of the same position take later on in the same night (not all Suprime-Cam images will have that and are taken in all types of sky conditions).

Now the Comet Hunters team is thinking about how best to develop a classification interface for the HSC data to include this. In the meantime, there are new Suprime-Cam images in need of review at http://www.comethunters.org if you have a minute or two to spare.