Space Warps CFHTLS

A simulated lens quasar, an example prediction of what we might find in Space Warps.

A simulated lens quasar, the likes of which we hope to find during Space Warps’ first project.

Our first project is to search the 400,000 images of the  Canada France Hawaii Telescope Legacy Survey, or CFHTLS – we’re asking people to spot gravitational lenses in its images, in order to find some new examples, and also to learn how to design automated lens detection systems to use in the future. I caught up with Jean-Paul Kneib, from the Strong Lenses in the Legacy Survey (SL2S) project, and Space Warps co-PI Anupreeta More to ask them to explain a bit more about it.

Jean-Paul, just how big is this survey, and how is it different from the SDSS?
CFHTLS is a survey conducted with the CFHT 3.6m telescope using the Megacam camera. It targeted 4 patches of the sky, adding up to about 150 sq deg. That’s about 60 times smaller than the SDSS-DR8 imaging area, but it goes typically 2.5 magnitude (about 10 times) deeper than SDSS, with higher resolution images. The average seeing was 0.6 arcsec, compared to an average of 1.4 arcsec for the SDSS data. So, in short CFHTLS is a mini SDSS but focussed on the deeper Universe, which means it is a great survey in which to find strong lensing systems!
Sounds good! Was it designed specially for this purpose, or for something else? What was the original idea?
The original design was to measure what we call “cosmic shear” – that is, the tiny deformations that large scale structures produce on the appearance of faint galaxies. This cosmic shear measurement is used to put constraints on cosmological parameters. But similarly the strong lensing systems could also reveal us something about cosmology … but first we need to find them!
Sounds good! Anupreeta, you’ve been thinking about lenses for cosmology lately – how are you thinking of using a sample of lenses from the CFHTLS to say something about the universe as a whole?
Various cosmological models of the Universe predict different numbers of galaxy clusters at various times, and also differences in how concentrated is the mass distribution within these massive structures. Both these factors affect how efficiently the galaxy clusters will produce highly magnified and distorted arcs. That means that the abundance of arcs in surveys like CFHTLS can be used in turn to understand which cosmological model best describes our Universe.
Lens systems allow us to primarily understand the properties -like the mass – of the lensing galaxies. However, it is possible to derive extra constraints from certain types of lenses in order to learn more about the Universe – for example, its age. We see that quasars change their brightness over time; in a lensed quasar system, the different lensed images appear to vary at different times due to the different paths taken by the light rays to reach us. The time delay seen between these multiple images, combined with the speed of light through the lens, allows a measurement of distance to be made.  By measuring these time delays accurately, we can measure distance, compare it with redshift, model the expansion of the Universe, and predict its age.
I know you’ve worked on the CFHTLS data in the past, with the “Arcfinder” code – what kinds of lenses did you find with it, and what do you think it missed?
I mainly looked for arcs in the g-filter since the arcs look brighter in this filter than any other. This helped optimize the arc detection. As the CFHTLS imaging goes very deep compared to SDSS, we found a fainter sample of arcs. In order to contain the number of false positive detections, I had to apply some limits on some of the arc properties such as surface brightness, length-to-width ratio, curvature and area. These limits were essentially decided arbitrarily after some testing on a smaller known lens sample from the CFHTLS. However, it was not known beforehand how this might affect the completeness of the lens sample and the limits on which of the arc properties could be relaxed or made stricter. There are various factors to which a code is sensitive to e.g. a certain arc may satisfy most thresholds, but will go undetected because it happened to be located in an image region with high noise levels or was partially overlapping with a bright galaxy. People are less susceptible to these fluctuations when they look at images, and can cover a wider dynamic range in terms of arc properties and, simultaneously, assess the likelihood of an arc-like image of being a lensed image, given its color, shape, curvature, proximity and alignment with respect to a nearby lensing galaxy in a way which is not currently possible with Arcfinder code.
Raphael Gavazzi wrote his “RingFinder” code to look for galaxy-scale lenses in the CFHTLS. That research programme has been quite successful, we found several dozen lenses and used them to study the distribution of dark matter within the lens galaxies. RingFinder only looks at simple, smooth, bright red elliptical galaxies, and then tries to dig into the lens light looking for blue arcs. We expect it to have missed some red arcs, and also lensed features that are not arc-shaped – like the lensed quasars. One thing I am interested in with Space Warps is making a sample of low-medium probability objects: these will be great for testing tools like RingFinder against: can we make it more flexible, and able to cope with spirals, mergers and other galaxies that look like lenses but are not.
 Anu: have you thought about how the results from Space Warps might be used to improve Arcfinder? That would be cool!
The results from the Space Warps are going to be interesting and exciting in many ways. In terms of improving the Arcfinder, Space Warps will provide a more comprehensive library of lenses – I hope the spotters will find the lenses that Arcfinder missed! By measuring the properties of these new lenses, we will be able to put together a better set of thresholds that would have increased the completeness and purity of the Arcfinder lens sample. It might be possible that some new lens properties that we haven’t thought of yet might prove more useful in terms of getting higher purity. It would  be great to be able to improve the Arcfinder algorithm in this way.
With your help we’re going to find a lot of useful things in CFHTLS, I think!

2 responses to “Space Warps CFHTLS”

  1. JohnF says :

    In the GZ Forum, there are a number of links to Lensing – perhaps they need to be linked to this project.

    This project does not yet appear on

    • Phil Marshall says :

      That’s right – we’re nearly there though! We’ll need some help re-linking useful lensing resources into Space Warps Talk – we’ll get there pretty quickly with enough helping hands! 🙂

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: