Meeting called for Radio Interest group / Time domain interest group . 2022, Jan 11 Agenda: V.Galluzi : pulsar/transient data at INAF : use cases and status report M.Louys : ObsCore Extension for TimeDomain. Focus on PSRFITS mapping F.Bonnarel : Discovery and Access strategies for radio timeseries. ALL : open questions. Formats/datamodels. Next steps Participants: (please fill your name here, and add your institution ) : JJ Kakelaars (CADC) Marco Molinaro (INAF) Adat Nebot (CDS-ObAS) F Bonnarel (CDS-ObAS) L. Michel (ObAS) Brent Miszalski (AAO, Macquarie University) Frank Fabian Mattia Mancini (AsTRON) Casey Law (Caltech) Mark Lacy , (NRAO) Mark Kettenis , (JIVE) Mark Cresitello-Dittmar (SAO/CfA) James Dempsey (CSIRO) Mireille Louys, (CDS , University of Strasbourg) Peter Teuben (UMD) Tracy Clarke (US NRL) Alessandra Zanichelli (INAF) Severin Gaudet (CADC) Frank Schussler Aard Keimpema (JIVE) Meeting notes V.Galluzzi: Pulsar/transients facilities and observation at INAF Projects : TP,SARDARA, XARCOS, DFB3, ROACH Formats : Baseband data (dada), PSRFITS, Filterbank Archival system (NADIR) /admits evolving datamodels and formats / interface has samp Pulsar ingestion issues : CPU overload for DFB3 (PSRFITS) we have to investigate reasons. we have more to ingest (FILTERBANK) Pulsar data model : PSRFITS main header to gather main keywords used as search criteria for FILTERBANK we produce side FITS headers mapped onto a MYSQL database (datamodel_pulsar) use case : complement data at other wavelength for a known pulsar - search by physical data user stories collected new algorithms developed recently to optimize signal analysis (fir pulsar. for FRBs) cosmological probes gravitation tests Summary of future work Casey (Caltech) : detection of FRBs and repeating FRBs We are providing real time processing in order to identify the appropriate window and process some gating ?) the real time system is triggered by an event, and the data just before, during and after the event are stored (rather than the entire data stream). One such "slice" per ObsCore record. Question : Does Obscore and CAOM fit your data ? --> we may have issues with polarization .. we may need to add period and dispersion measure Mireille Louys Obscore Extension for Time and frequency in radio Extend Obscore for TimeDomain radio data What do I need to describe/discover relevant radio pulsar data? Mapping Obscore to PSRFITS (Nançay) What is missing. Polarization ? JD: how can we define the time extension / radio extension when declaring a TAP service in the registry ML : not set yet but the idea was to use capabilities for radio / timed datasets aligned with the table served Caveat ?: Laurent Michel how can we be sure of the joins we can write when referring to obs_publisher_did as key ? for time domain? for radio data ? M.Kettenis experienced this in his EVN TAP and implemented obs_publisher_did as a join key MCD : would the keys be part of the specification ? Should be explained in the next version of the specification Laurent : natural joins can be an alternative MK: dynamic spectra in radio : time sampling is regular may not be necessary to add the t_obs extension folding period : do we need to mention it ? FB : ? data product dynamic spectrum : is it a good name for designating the Obscore data product type for data in PSRFITS ? FB Discovery and Access : Attaching pulsar lightcurve or dybamil spectra to pulsar records in a catalog -> LINK or DataLink) Discovering counterparts in time and space around an event (FRB ?) -> ObsTAP, SCS with Time, +STMOC, + Extensions ? Discovering all TimeSeries at a given position and spectral range from the whole VO . --> introduce DataSetSAP and send the same query to all servers Access : Full retrieval or SODA-like (for providing excerpts or on the fly transformations?) Format/DataModel ? --> interoperability of retrieval ... format for on line TimeSeries build up ??? what do we look for time residuals variation of period with time single events : FRB differentiate the interest : what is published after data analysis ( results ). Advanced analyses products are not necessarily what an observatory will put on line in the archive BM: In terms of getting some prototype services going, might something to consider be generating dynamic spectra from existing radio data, e.g. you want to search for an FRB in a radio dataset at a particular position and time window. You could supply an STMOC as the input and a list of matches could be returned (+ links to generate each dynamic spectrum). Possible scenario (FB) 1 ) extend SIA to DataSetSAP 2) add a MOC parameter for STMOC searches 3 ) reponse is ObsCore (+time extension ?) 4 ) access_url with Various links in DataLink response including raw data, dynamic spectra,; folded data, light curves ???? Another could just be a radio lightcurve generator, but it would be interesting to gauge what the highest priority service would be (especially in terms of science usefulness). Are there tedious operations that we can automate to unlock more of the potential of these radio archives? --> Sure (FB) : what is the top priority for such services according to the scientific community. VG? :Filterbank is not a standard and not easy to map in a systematic way . need a software (sigproc?) to help for the mapping Filterbank for transients: VLBI FRB data is also in Filterbank Future tracks: proposal to go forward by François Bonnarel : a ) use cases pulsars : study phase diagrams or light curves for secular period changes, periodograms (?) FRB and other events : detection in time series, counterparts in other spectral domain,s and services etc.... repeated FRB ? b ) data we can expose : Pulsar data : full data in PSR fits formats (folded or not) light curves dynamic spectra (without polarization separation) spectra, spectropolarimetric data and for FRB data what is similar and different events, set of time chunks relationship VOEvent/ConeSearch/Obscore c ) description pure ObsCore : mapping PSRFITS and others (FilterBank) extensions : TimeFrame descriptions, frequency, sample exposure time, cadences, phase-folded nature (and period ?), polarization ??? question about dataproductype : sparse cube, dynamic spectra, folded data ? d ) discovery and access : source driven : via catalog of pulsars (LINK or datalink) ObsTAP : ADQL-MOC queries / STMOC in ObsCore table ? ObsTAP :managing time extension how to express capabilities in Registry how to join tables between core and extension Extending Simple accesss protocols to time series Adding STMOC parameter to SCS and DatasetSAP/SIA Access: full retrieval of whole data or DataLink for various flavours SODA -like interface ti generate TileSeries f ) datamodel and dataformat : Cube DM, Mango ? Specific TimeSeries or Light curve format ? combination ?