GALEX was launched on 28 April 2003, and was retired on 28 June 2013. The primary objective of the mission was to survey the sky in two different UV bands: an FUV band (1350-1750 Å) and an NUV band (1750-2750 Å), which it could observe in simultaneously. It had a circular, 1.2 square degree diameter field of view, and had both imaging (5-6 arcseond resolution) and spectroscopic (via a grism with resolution R ~100-250) capabilities. It used microchannel plate detetors to record the times and positions of the UV photon events with accuracies of ~0.005 seconds. Most of the survey was split into a variety of sub-surveys: a wide and shallow All-Sky Imaging Survey (AIS), and deeper but narrower Medium Imaging Survey (MIS), an ever deeper and narrower Deep Imaging Survey (DIS), a Nearby Galaxy Survey (NGS), and various Guest Investigator programs (GII). During the final ~18 months the spacecraft was controlled by CalTech, which allowed for new areas of the sky to be explored, including a new type of observing mode known as "scan mode".


I'm a member of the gPhoton project (on GitHub), an open-source project that archives more than a trillion photon events detected during GALEX's lifetime. The database is hosted at MAST, and with our accompanying Python software, users can create calibrated images, lightcurves, and image cubes (sets of image slices in time) with user-specified coordinates and time resolutions. Since the basic data products are the photon events themselves, users no longer need to download multiple tiles from the different sub-surveys: just specifiy a coordinate and aperture size, and all the available photon data will be there for you to use.

So what sort of science am I doing with gPhoton? Mostly looking for transients and variables that occur over timescales within a GALEX visit (~30 minutes). This is where gPhoton really shines. Because gPhoton can create lightcurves with time resolutions of seconds, short duration events like stellar flares can be identified and studied, whereas the standard image tile at MAST would only contain the total flux for the star during that visit. Since flares can only last a few minutes, the entire event might go off during the single eclipse. The two image frames below are taken just minutes apart; the video covers just 20 minutes!
I'm interested in studying these flares because the UV is a good place in the electromagnetic spectrum to search for them: the contrast between the flare and the underlying photosphere is greater in the UV than at longer wavelengths. In particular, you can find smaller flares (not just the giant ones). I'm using gPhoton to study the frequency and statistics of these smaller flares. I'm interested in knowing how a stellar environment with smaller, but more frequent, flares affects the habitability of orbiting planets, to complement studies that examine single, very large flares.
Of course, there's plenty of other things that go bump in the night besides stellar flares. gPhoton is a great tool to find and study those variables: eclipsing binaries, pulsating white dwarfs, hot subdwarfs, supernovae, ephemeral transients...the list goes on and on. As an example, here's an AGN that was observed by GALEX off and on over a seven year baseline. By creating these postage stamp images across the various tiles, we can see that the AGN had a substantial decrease in brightness sometime between 2008 and 2010. While this variability occurs between visits, we can use the same idea to look for variability and transients during visits thanks to gPhoton.

Project Blacklight

Project Blacklight is an effort to create a catalog of every object that varies within a GALEX visit using gPhoton. This project involves two main components:
  1. A search for variability of known GALEX sources (from the mission-produced source catalogs) over timescales of < 30 minutes. There are tens of millions of known sources to examine.
  2. A search for new objects ("transients", which are not in the catalog of known sources, but show up briefly during a GALEX visit).
To create the catalog, I'd like to use two complementary techniques:
  1. Machine learning algorithms to detect and characterize the light curves of known sources, including additional data from optical surveys for color information, etc.
  2. Citizen science participation to help identify transients and challening variables from image frames.