130 likes | 271 Views
Oversampling mode. Roi Alonso , Pierre-Yves Chabaud Christian Surace, Raphael Cautain, P. Barge. Planned procedures . Inputs N1 data from the LESIA pipeline Scientific criteria Additional programs Specific algorithms in various steps Output List of targets (XML files) SEF/CMC.
E N D
Oversampling mode Roi Alonso , Pierre-Yves Chabaud Christian Surace, Raphael Cautain, P. Barge
Planned procedures • Inputs • N1 data from the LESIA pipeline • Scientific criteria • Additional programs • Specific algorithms in various steps • Output • List of targets (XML files) SEF/CMC
The various steps • Preprocessing of the data • To filter residuals of the SAA and orbital perturbations • To remove disturbing low frequencies of Stellar Variability (moving box, Fourier & wavelets, … multiplex-PCA) • Detection of possible transits • Two complementary algorithms running in parallel • MID (Morphological Detector) + EPF (Periodicity Finder) • BLS (Box fitting Least Square) • Estimate of a confidence level for each detected event • Discrimination/priority • Building a common list of detected events • Use simple tools to remove ambiguities (SOS) • Management of the lists • EXOP/AP
Merging of the detection lists Filtered LCs MID BLS MID+EPF List with Epoch of best detections List with P and E of best detections List with P and E of best detections Trapezoidal fitting of the phase-folded LC Period P and Epoch E Single list of outputs Sorting - no repetitions Input for the discrimination tools
Priority: removing ambiguites To discriminate planetary transits from: Eclipsing Binaries, variable stars, false detections (noisy features) • Identification of EB ligth curve • Search for secondary transits in the signal folded at 1xPeriod around epoch + 0.5 • Looking at differences (depth & position) between even and odd eclipses in the signal folded at 2xPeriod around epoch • Search for out of eclipse modulations (peak in the spectrum) • Use of colours (trapezoidal fitting in each channel) Reduced list of candidates (sorted by increasing order of the sum of 4 significances) Exoplanet candidates for oversampling (ordered in priority after visual inspection - SOS)
Exo - Transit Detection Exo - Detection list Tools: Discrimination/Priority EXODAT Exo - Sorted list Visual Inspection - SOS Combine lists Exo - Candidate list (core) Exo - Initial list (core) Exo-CCD list AP-targets list Exo-CCD - XML file AP - Variability analysis
How things are presently working • Inputs • Raw N0 data instead of N1 (N1 only available since ~ 1 month) • Raw data are put in the N1 format • Algorithms • Automated procedures not at work • LC analysis made using visual inspection • Management of the oversampling lists • Variability analysis by the AP at work • Output • List of targets (XML files) SEF/CMC (OK)
Results from the alarm mode • Two alarm candidates during IRa01 • One confirmed planet • No alarm candidates during SRc01 (no available data) • 19 alarm candidates during LRc01 Intense Follow Up effort during the summer • 2 confirmed planets • ~ 5 eclipsing binaries • FU observations are still going on Only the most obvious candidates were detected The alarm pipeline must be improved !
Information released for FU operations • The lists of « alarms » are accessible to the Science Team only for the purpose of Follow Up operations (periods, depths and durations of the events + the contaminating background stars) • The coordinates of the targets are accessible only to the members of the FU coordination group (leaders of the FU working groups + some experts)
Encountered difficulties • Due to the production of the N1 data • Long delay in the data delivery (up to 1 month) • Some gaps introduced in the N1 products • Non homogeneous level of the corrections • Pipeline corrections introduced in the data: • Reversed or negative « hot » pixels • Some additional discontinuities • Hot pixels: • Detection and modeling (L. Jorda & R. Cautain) -> corrections • Filtering using wavelet transform (R. Alonso)
N0 N1 Incomplete corrections and “mirroring” of hot pixels
N0 ? N1 Discontinuities introduced in the N1 data
Necessary evolution To be effective the alarm software needs: • Regular delivery of data all along the run • Data corrected in an homogenous way (continuity of the corrections) A possibility is to produce data specifically processed for the alarm mode (only simple corrections performed on the raw data in a regular and continuous way) This possibility is under study