Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Update on the North Celestial Pole Transient Search

transientskp
December 03, 2012

Update on the North Celestial Pole Transient Search

Adam Stewart

transientskp

December 03, 2012
Tweet

More Decks by transientskp

Other Decks in Science

Transcript

  1. Update  on  the  North  Celes0al  Pole  
    Transient  Search  
    Adam  Stewart  
    Tom  Hassall,  Jess  Broderick,  Rob  Fender  
    University  of  Southampton,  Southampton,  UK  

    View Slide

  2. Outline  
    •  Reminder  of  NCP  Data  and  Reduc0on  
    •  Updates  to  Strategy  
    – Quality  Control  
    – Peeling  
    •  Current  Image  Examples  
    •  Transient  Searching  
    •  Conclusions/Future  Work  
    03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng   2  

    View Slide

  3. NCP  Data  -­‐  MSSS  
    •  MSSS  -­‐  Mul0frequency  Snapshot  Sky  Survey.  
     
    •  With  each  MSSS-­‐LBA  observa0on,  a  single  sub  band  was  placed  on  
    the  NCP  field.  60  MHz  observa0ons  with  ~200  kHz  bandwidth  with  
    100  deg2  FoV.  
    •  2606,  11  minute,  snapshots.  
    03/12/2012   3  
    Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng  
    •  Observed  December  
    2011  –  April  2012  (~80%  
    complete).  
     
    •  MSSS  LBA  s0ll  at  80%  
    observed  –  so  a  few  
    more  NCP  observa0ons  
    to  come.  
    msss.astron.nl  

    View Slide

  4. Reduc0on  Strategy  
    •  MSSS  data  -­‐>  follow  the  MSSS  reduc0on  strategy.  
     
    •  Features  such  as:  
    –  Automa0cally  flag  bad  sta0ons.  
    –  Automa0cally  clip  bad  amplitude  gains  in  calibrators.  
    –  Provides  diagnos0c  plots  of  results  during  processing.  
    •  Refer  to  MSSS  reports  page  on  the  wiki  
    03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng   4  
    Simultaneously  
    observe  
    calibrator    
    Transfer  gain  
    solu0ons  to  
    target  from  
    calibrator  
    Perform  phase  
    only  calibra0on  
    using  model  
    Image  

    View Slide

  5. Updates  to  Strategy  –  Quality  Control  
    •  Some  datasets  are  
    simply  bad…  
    •  Agempted  to  create  
    a  non-­‐imaging  
    method  to  detect  
    bad  data.  
    •  Read  the  visibili0es  
    from  each  processed  
    dataset  looking  at  
    the  mean  or  median  
    values.  
    03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng   5  
    L49040  (2012-­‐02-­‐27  04:13:00)  

    View Slide

  6. Updates  to  Strategy  –  Quality  Control  
    •  A  quick  method  of  cukng  very  bad  observa0ons.  
    •  However,  could  not  rely  on  method  completely,  and  this  combined  with  
    imaging  each  observa0on  gave  us  a  final  ‘bad  list’.  
    03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng   6  
    Missing  sta0ons  
    Seems  to  be  odd  cases  of  rfi  
    or  poor  calibra0on.  

    View Slide

  7. Quality  Aside  -­‐  Calibrators  
    •  Also  processed  all  the  calibrator  observa0ons  to  check  
    performance.  
    •  Calibrators  seemed  to  show  a  slight  rise  in  flux  
    depending  on  their  eleva0on  in  the  sky.  
    •  When  the  beam  correc0on  is  completely  turned  off  in  
    BBS  –  the  rise  disappears.  
    •  Suggests  an  error  with  the  beam  model?..  Ongoing.  
    03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng   7  

    View Slide

  8. Updates  to  Strategy  -­‐  Peeling  
    •  Adopted  a  peeling  technique  developed  by  Mike  Bell  for  MSSS.  
    –  See  Week  26  MSSS  report  for  more  details  
     
     
     
     
    •   
    While  improving  quality  –  the  method  has  also  been  used  to  
    subtract  3C  61.1.  
    •  Agemp0ng  to  also  peel  the  next  brightest  source,  3C  220.3  ,  
    however  had  issues…  
    03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng   8  
    Subtract  ALL  
    sources,  add  
    back  those  to  be  
    peeled  
    Solve  for  DDGs  in  
    the  direc0on  of  
    the  sources  
    Subtract  peeled  
    sources  and  
    apply  solu0ons  
    Add  back  all  
    unpeeled  
    sources  and  
    perform  phase  
    only  calibra0on  

    View Slide

  9. Updates  to  Strategy  -­‐  Peeling  
    •  Simply  not  enough  data  in  a  single  sub  band  
    for  BBS  to  solve.  
    •  While  some0mes  it  was  successful,  in  others  
    chunks  of  data  would  be  flagged.  
    •  Hence  only  3C  61.1  is  used  for  peeling.  
    03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng   9  
    Before  Peeling   Aqer  Peeling  
    Very  slight  noise  improvement  

    View Slide

  10. Updates  to  Strategy  -­‐  Combining  
    •  Each  block  of  observa0ons  has  a  different  
    Right  Ascension.  
    – Has  the  effect  of  making  the  produced  maps  
    rotate.  
    •  Processed  one  set  of  data  where  the  phase  
    center  has  been  manually  set  using  NDPPP.  
    •  Enables  making  different  combina0ons  much  
    simpler.  
    03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng   10  
    Double  3C  61.1  

    View Slide

  11. Updates  to  Strategy  -­‐  AWimager  
    •  New  version  of  Awimager  –  Thanks  to  Cyril  Tasse  
    •  Speed  of  imaging  the  NCP  has  drama0cally  improved.  
    •  For  example,  a  concatenated  50  observa0on  dataset,  
    imaging  size  of  1024,  40  arcsec  cell  size,  deep  cleaning,  on  
    12  core  Southampton  LOFAR  machine:  
    •  Some0mes  can  s0ll  use  very  small  major  clean  cycles,  thus  
    taking  a  long  0me  –  usually  bad  data.  
    03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng   11  
    Old  AWimager   New  AWimager  
    Average  Memory  Usage   ~25  GB   ~3  GB  
    Time  Taken   11:58:27   00:08:31  

    View Slide

  12. NCP  Images  –  Single  Observa0on  
    •  Single  
    Observa0on.  
    •  Noise  ~400  
    mJy/beam  
    •  310  x  65  
    arcsec    
    (BPA  22  deg)  
    03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng   12  

    View Slide

  13. NCP  Images  –  16  Combined  
    03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng   13  
    •  16  
    concatenated  
    observa0ons.  
    •  Noise  ~170  
    mJy/beam  
    •  118  x  75  
    arcsec    
    (BPA  110  deg)  

    View Slide

  14. NCP  Images  –  50  Combined  
    03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng   14  
    •  50  
    concatenated  
    observa0ons.  
    •  Noise  ~120  
    mJy/beam  
    •  114  x  70  
    arcsec    
    (BPA  43  deg)  

    View Slide

  15. NCP  Movies  
    03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng   15  
    16  Combined  Movie  
     -­‐115  images  
    50  Combined  Movie  
     -­‐36  images  

    View Slide

  16. √N  Improvement?  
    •  Simply,  does  
    combining  
    observa0ons  follow  
    a  √N  
    improvement?  
    •  Shows  noise  
    improvement  from  
    concatena0ng  8,  
    16,  24,  32,  40  and  
    50  observa0ons.  
    •  Can  also  combine  
    in  the  image  plane,  
    using  the  50  SB  
    images  –  not  ideal  
    method.  
    03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng   16  

    View Slide

  17. Transient  Searching  
    •  Believe  image  quality  is  now  where  it  should  be  –  
    concentra0on  going  forward  on  transient  
    searching.  
    •  Search  in  early  stages  –  however  can  begin  to  
    think  about  what  limits  would  be  imposed  
    considering  different  groupings.  
    •  NCP  observa0ons  slightly  sporadic  –  groupings  
    need  to  be  thought  about  carefully.    
    03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng   17  

    View Slide

  18. 03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng   18  
    10−4
    10−3
    10−2
    10−1
    100
    101
    102
    103
    104
    10−7
    10−6
    10−5
    10−4
    10−3
    10−2
    10−1
    100
    101
    Laz2010
    Mat2009T
    Flux Density S [Jy]
    B2010T,V
    B2011T
    LOFARC (10 Epoch)
    Croft2010T
    LOFARF 30 MHz (10 Epoch)
    Ban2011V
    Ban2011T
    Bell2011T
    GY2006T
    LOFARF (10 Epoch)
    B2007T
    (1 Week)
    F2003V
    B2007T
    (2 Month)
    C2003V
    B2007T
    (1 Year)
    Bk2010V
    Snapshot Rate [deg−2]
    Ban2011 − Bannister et al. (2011)
    Bk2010 − Becker et al. (2010)
    Bell2011 − This work
    B2007 − Bower et al. (2007)
    B2010 − Bower et al. (2010)
    B2011 − Bower & Saul (2010)
    C2003 − Carilli at al. (2003)
    Croft2010 − Croft et al. (2010)
    F2003 − Frail et al. (2003)
    GY2006 − Gal−Yam et al. (2006)
    Laz2010 − Lazio et al. (2010)
    LOFARC
    − LOFAR commissioning
    LOFARF
    − LOFAR full array
    Mat2009 − Matsumura et al. (2010)
    V − denotes variable rate
    T − denotes transient rate
    ? − denotes upper limit
    NCP  Single  Obs  
    ~2500  snapshots  
    NCP  16  Combined  
    ~115  snapshots  
    NCP  50  Combined  
    ~35  snapshots  
    NCP  50  Combined    
    ~35  snapshots  1.4  GHz  
    α=-­‐1  
    NCP  50  Combined    
    ~35  snapshots  5.0  GHz  
    α=-­‐1  
    More  epochs  
    α<-­‐1  
    Extrapolate  limits  to  
    higher  frequencies  
    assuming  α=-­‐1:  
     
     
       
    Log  N  Log  S  
    120  mJy  -­‐>  5.14  mJy  at  1.4  GHz  
    120  mJy  -­‐>  1.44  mJy  at  5.0  GHz  

    View Slide

  19. TraP  
    •  Of  course  the  transients  pipeline  will  be  the  
    main  tool  for  analysis.  
     
    •  Light  Curves,  source  posi0ons  etc.  
     
    •  Data  processed  in  August,  imaged  with  CASA,  
    was  partly  passed  through  –  no  transients.  
    •  Image  sets  are  ready  to  be  passed  through.  
    03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng   19  

    View Slide

  20. NCP  Conclusions/Future  
    •  NCP  processing  in  a  very  good  state  –  any  addi0ons  would  
    be  small  improvements  –  closely  monitor  MSSS.  
     
    •  Producing  very  nice  maps,  especially  with  the  new  
    AWimager.  
    •  Huge  poten0al  for  transient  searching,  is  now  really  the  
    main  focus  in  the  immediate  future.  
    •  Dataset  also  proved  to  be  useful  simply  in  terms  of  LOFAR  
    commissioning/MSSS  
    •  Next  steps:  
    –  Find  transients  (?)  
    –  NCP  Field  hopefully  con0nued  to  be  monitored.  
    03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  LOFAR  TKP  Mee0ng   20  

    View Slide

  21. Deep  NCP  LBA  Image  
    •  What  if  we  image  everything?  
    03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  TKP  Mee0ng   21  
    20  –  40  
    mJy/beam  
     
    Confusion  
    limited  
    ~40  mJy/
    beam  

    View Slide

  22. 03/12/2012   Searching  for  Transients  at  the  NCP  –  Amsterdam  TKP  Mee0ng   22  


    View Slide