skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: AUTOMATED TRANSIENT IDENTIFICATION IN THE DARK ENERGY SURVEY

Abstract

We describe an algorithm for identifying point-source transients and moving objects on reference-subtracted optical images containing artifacts of processing and instrumentation. The algorithm makes use of the supervised machine learning technique known as Random Forest. We present results from its use in the Dark Energy Survey Supernova program (DES-SN), where it was trained using a sample of 898,963 signal and background events generated by the transient detection pipeline. After reprocessing the data collected during the first DES-SN observing season (2013 September through 2014 February) using the algorithm, the number of transient candidates eligible for human scanning decreased by a factor of 13.4, while only 1.0% of the artificial Type Ia supernovae (SNe) injected into search images to monitor survey efficiency were lost, most of which were very faint events. Here we characterize the algorithm’s performance in detail, and we discuss how it can inform pipeline design decisions for future time-domain imaging surveys, such as the Large Synoptic Survey Telescope and the Zwicky Transient Facility. An implementation of the algorithm and the training data used in this paper are available at at http://portal.nersc.gov/project/dessn/autoscan.

Authors:
;  [1]; ; ;  [2]; ; ;  [3];  [4];  [5];  [6]; ;  [7]; ;  [8];  [9]; ;  [10];  [11];  [12] more »; « less
  1. Department of Astronomy, University of California, Berkeley, 501 Campbell Hall #3411, Berkeley, CA 94720 (United States)
  2. Institute of Cosmology and Gravitation, University of Portsmouth, Dennis Sciama Building, Burnaby Road, Portsmouth, PO1 3FX (United Kingdom)
  3. Department of Physics and Astronomy, University of Pennsylvania, Philadelphia, PA 19104 (United States)
  4. Astronomy Department, University of Illinois at Urbana-Champaign, 1002 West Green Street, Urbana, IL 61801 (United States)
  5. Argonne National Laboratory, 9700 South Cass Avenue, Lemont, IL 60439 (United States)
  6. Kavli Institute for Cosmological Physics, University of Chicago, Chicago, IL 60637 (United States)
  7. Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, CA 94720 (United States)
  8. School of Physics and Astronomy, University of Southampton, Highfield, Southampton, SO17 1BJ (United Kingdom)
  9. Fermi National Accelerator Laboratory, P.O. Box 500, Batavia, IL 60510 (United States)
  10. Department of Physics and Astronomy, University College London, Gower Street, London, WC1E 6BT (United Kingdom)
  11. Kavli Institute for Cosmology, University of Cambridge, Madingley Road, Cambridge CB3 0HA (United Kingdom)
  12. Institut d’Astrophysique de Paris, Univ. Pierre et Marie Curie and CNRS UMR7095, F-75014 Paris (France)
Publication Date:
OSTI Identifier:
22520031
Resource Type:
Journal Article
Resource Relation:
Journal Name: Astronomical Journal (Online); Journal Volume: 150; Journal Issue: 3; Other Information: Country of input: International Atomic Energy Agency (IAEA)
Country of Publication:
United States
Language:
English
Subject:
79 ASTROPHYSICS, COSMOLOGY AND ASTRONOMY; ALGORITHMS; DATA ANALYSIS; DATA PROCESSING; EFFICIENCY; IMAGES; IMPLEMENTATION; NONLUMINOUS MATTER; POINT SOURCES; RANDOMNESS; REPROCESSING; SUPERNOVAE; TELESCOPES; TRANSIENTS

Citation Formats

Goldstein, D. A., Nugent, P. E., D’Andrea, C. B., Nichol, R. C., Papadopoulos, A., Fischer, J. A., Sako, M., Wolf, R. C., Foley, R. J., Gupta, R. R., Kessler, R., Kim, A. G., Thomas, R. C., Smith, M., Sullivan, M., Wester, W., Abdalla, F. B., Benoit-Lévy, A., Banerji, M., Bertin, E., and and others. AUTOMATED TRANSIENT IDENTIFICATION IN THE DARK ENERGY SURVEY. United States: N. p., 2015. Web. doi:10.1088/0004-6256/150/3/82.
Goldstein, D. A., Nugent, P. E., D’Andrea, C. B., Nichol, R. C., Papadopoulos, A., Fischer, J. A., Sako, M., Wolf, R. C., Foley, R. J., Gupta, R. R., Kessler, R., Kim, A. G., Thomas, R. C., Smith, M., Sullivan, M., Wester, W., Abdalla, F. B., Benoit-Lévy, A., Banerji, M., Bertin, E., & and others. AUTOMATED TRANSIENT IDENTIFICATION IN THE DARK ENERGY SURVEY. United States. doi:10.1088/0004-6256/150/3/82.
Goldstein, D. A., Nugent, P. E., D’Andrea, C. B., Nichol, R. C., Papadopoulos, A., Fischer, J. A., Sako, M., Wolf, R. C., Foley, R. J., Gupta, R. R., Kessler, R., Kim, A. G., Thomas, R. C., Smith, M., Sullivan, M., Wester, W., Abdalla, F. B., Benoit-Lévy, A., Banerji, M., Bertin, E., and and others. Tue . "AUTOMATED TRANSIENT IDENTIFICATION IN THE DARK ENERGY SURVEY". United States. doi:10.1088/0004-6256/150/3/82.
@article{osti_22520031,
title = {AUTOMATED TRANSIENT IDENTIFICATION IN THE DARK ENERGY SURVEY},
author = {Goldstein, D. A. and Nugent, P. E. and D’Andrea, C. B. and Nichol, R. C. and Papadopoulos, A. and Fischer, J. A. and Sako, M. and Wolf, R. C. and Foley, R. J. and Gupta, R. R. and Kessler, R. and Kim, A. G. and Thomas, R. C. and Smith, M. and Sullivan, M. and Wester, W. and Abdalla, F. B. and Benoit-Lévy, A. and Banerji, M. and Bertin, E. and and others},
abstractNote = {We describe an algorithm for identifying point-source transients and moving objects on reference-subtracted optical images containing artifacts of processing and instrumentation. The algorithm makes use of the supervised machine learning technique known as Random Forest. We present results from its use in the Dark Energy Survey Supernova program (DES-SN), where it was trained using a sample of 898,963 signal and background events generated by the transient detection pipeline. After reprocessing the data collected during the first DES-SN observing season (2013 September through 2014 February) using the algorithm, the number of transient candidates eligible for human scanning decreased by a factor of 13.4, while only 1.0% of the artificial Type Ia supernovae (SNe) injected into search images to monitor survey efficiency were lost, most of which were very faint events. Here we characterize the algorithm’s performance in detail, and we discuss how it can inform pipeline design decisions for future time-domain imaging surveys, such as the Large Synoptic Survey Telescope and the Zwicky Transient Facility. An implementation of the algorithm and the training data used in this paper are available at at http://portal.nersc.gov/project/dessn/autoscan.},
doi = {10.1088/0004-6256/150/3/82},
journal = {Astronomical Journal (Online)},
number = 3,
volume = 150,
place = {United States},
year = {Tue Sep 15 00:00:00 EDT 2015},
month = {Tue Sep 15 00:00:00 EDT 2015}
}
  • We describe an algorithm for identifying point-source transients and moving objects on reference-subtracted optical images containing artifacts of processing and instrumentation. The algorithm makes use of the supervised machine learning technique known as Random Forest. We present results from its use in the Dark Energy Survey Supernova program (DES-SN), where it was trained using a sample of 898,963 signal and background events generated by the transient detection pipeline. After reprocessing the data collected during the first DES-SN observing season (2013 September through 2014 February) using the algorithm, the number of transient candidates eligible for human scanning decreased by a factormore » of 13.4, while only 1.0 percent of the artificial Type Ia supernovae (SNe) injected into search images to monitor survey efficiency were lost, most of which were very faint events. Here we characterize the algorithm's performance in detail, and we discuss how it can inform pipeline design decisions for future time-domain imaging surveys, such as the Large Synoptic Survey Telescope and the Zwicky Transient Facility.« less
  • We describe an algorithm for identifying point-source transients and moving objects on reference-subtracted optical images containing artifacts of processing and instrumentation. The algorithm makes use of the supervised machine learning technique known as Random Forest. We present results from its use in the Dark Energy Survey Supernova program (DES-SN), where it was trained using a sample of 898,963 signal and background events generated by the transient detection pipeline. After reprocessing the data collected during the first DES-SN observing season (2013 September through 2014 February) using the algorithm, the number of transient candidates eligible for human scanning decreased by a factormore » of 13.4, while only 1.0% of the artificial Type Ia supernovae (SNe) injected into search images to monitor survey efficiency were lost, most of which were very faint events. Here we characterize the algorithm's performance in detail, and we discuss how it can inform pipeline design decisions for future time-domain imaging surveys, such as the Large Synoptic Survey Telescope and the Zwicky Transient Facility. An implementation of the algorithm and the training data used in this paper are available at http://portal.nserc.gov/project/dessn/autoscan.« less
  • We describe an algorithm for identifying point-source transients and moving objects on reference-subtracted optical images containing artifacts of processing and instrumentation. The algorithm makes use of the supervised machine learning technique known as Random Forest. We present results from its use in the Dark Energy Survey Supernova program (DES-SN), where it was trained using a sample of 898,963 signal and background events generated by the transient detection pipeline. After reprocessing the data collected during the first DES-SN observing season (2013 September through 2014 February) using the algorithm, the number of transient candidates eligible for human scanning decreased by a factormore » of 13.4, while only 1.0% of the artificial Type Ia supernovae (SNe) injected into search images to monitor survey efficiency were lost, most of which were very faint events. Furthermore, we characterize the algorithm's performance in detail, and we discuss how it can inform pipeline design decisions for future time-domain imaging surveys, such as the Large Synoptic Survey Telescope and the Zwicky Transient Facility.« less
  • We describe an algorithm for identifying point-source transients and moving objects on reference-subtracted optical images containing artifacts of processing and instrumentation. The algorithm makes use of the supervised machine learning technique known as Random Forest. We present results from its use in the Dark Energy Survey Supernova program (DES-SN), where it was trained using a sample of 898,963 signal and background events generated by the transient detection pipeline. After reprocessing the data collected during the first DES-SN observing season (2013 September through 2014 February) using the algorithm, the number of transient candidates eligible for human scanning decreased by a factormore » of 13.4, while only 1.0% of the artificial Type Ia supernovae (SNe) injected into search images to monitor survey efficiency were lost, most of which were very faint events. Furthermore, we characterize the algorithm's performance in detail, and we discuss how it can inform pipeline design decisions for future time-domain imaging surveys, such as the Large Synoptic Survey Telescope and the Zwicky Transient Facility.« less
  • We describe the operation and performance of the difference imaging pipeline (DiffImg) used to detect transients in deep images from the Dark Energy Survey Supernova program (DES-SN) in its first observing season from 2013 August through 2014 February. DES-SN is a search for transients in which ten 3 deg(2) fields are repeatedly observed in the g, r, i, z passbands with a cadence of about 1 week. The observing strategy has been optimized to measure high-quality light curves and redshifts for thousands of Type Ia supernovae (SNe Ia) with the goal of measuring dark energy parameters. The essential DiffImg functionsmore » are to align each search image to a deep reference image, do a pixel-by-pixel subtraction, and then examine the subtracted image for significant positive detections of point-source objects. The vast majority of detections are subtraction artifacts, but after selection requirements and image filtering with an automated scanning program, there are similar to 130 detections per deg(2) per observation in each band, of which only similar to 25% are artifacts. Of the similar to 7500 transients discovered by DES-SN in its first observing season, each requiring a detection on at least two separate nights, Monte Carlo (MC) simulations predict that 27% are expected to be SNe Ia or core-collapse SNe. Another similar to 30% of the transients are artifacts in which a small number of observations satisfy the selection criteria for a single-epoch detection. Spectroscopic analysis shows that most of the remaining transients are AGNs and variable stars. Fake SNe Ia are overlaid onto the images to rigorously evaluate detection efficiencies and to understand the DiffImg performance. The DiffImg efficiency measured with fake SNe agrees well with expectations from a MC simulation that uses analytical calculations of the fluxes and their uncertainties. In our 8 "shallow" fields with single-epoch 50% completeness depth similar to 23.5, the SN Ia efficiency falls to 1/2 at redshift z approximate to 0.7; in our 2 "deep" fields with mag-depth similar to 24.5, the efficiency falls to 1/2 at z approximate to 1.1. A remaining performance issue is that the measured fluxes have additional scatter (beyond Poisson fluctuations) that increases with the host galaxy surface brightness at the transient location. This bright-galaxy issue has minimal impact on the SNe Ia program, but it may lower the efficiency for finding fainter transients on bright galaxies.« less