Joint EMU/POSSUM Discussion

• 10.30am-1.30pm AEDT, Thu 2 Dec 2010
• Education Seminar Room 323, The University of Sydney
  • Grid ref H13 at this map

Attendees

• Bryan Gaensler
• Tom Landecker
• Ray Norris
• Lisa Harvey-Smith
• Ettore Carretti
• Tim Robishaw
• Jeroen Stil
• Shea Brown
• Paul Hancock
• Larry Rudnick (via Skype)

Agenda

NB: No formal presentations expected. Names below are suggestions of people to lead the discussion.

Minutes

Coordination Structures

There are essentially 2 pipelines: uv data --> images and images --> data products. Second pipeline will be implemented by CSIRO, but only for components involving purely ASKAP data. Things that need external info (e.g., cross IDs, redshifts) will need to be done by us, although CSIRO may host these products once we generate them (perhaps on an ASKAP-hosted VO server).

Larry: really 4 steps:

1. creation of maps by CSIRO (although SSPs will want to have input)
2. maps to catalogues (SSPs responsible for providing algorithms)
3. added value products (SSPs entirely responsible for implementing/generating)
4. VO/archive (CSIRO to host)

For step 2, people will want to initially tweak data on minute by minute basis. Difficulty if this always has to go through Tim. Is there a parallel path?

Ettore: should be possible to have our own version of these algorithms during BETA testing. Probably need to code everything anyway to test the algorithms. Challenge will be putting this code together, given that each module will be written by different people in different languages; if things don't work when glued together, may require rapid response from ASKAP computing to resolve. Or this sort of integration could be focus of 'busy weeks'.

Ray: We can use CASA on our own laptops. Fairly similar to ASKAPsoft. Nothing to stop us taking uv data from BETA & ASKAP, processing some part of it ourselves and play with algorithms. Can then bring these back to Tim.

POSSUM/EMU coordination: should set aside some part of each monthly POSSUM and monthly EMU meetings for discussion of the other survey.

ACTION: Ask Andrew Hopkins, Emil Lenc, Julie Grant, Paul Hancock to join POSSUM

WG1: Shane O'Sullivan, Russ Taylor and Emil Lenc should talk WG2: Jeroen Stil and Andrew Hopkins have had some contact WG3: Shea Brown and Rainer Beck have had good discussion as needed WG4b: Ettore Carretti & Ray Norris (plus continuum commissioning postdoc to be appointed) WG5: Lisa H-S and Ray Norris should make sure they talk

EMU/POSSUM dependencies

Larry: milestones should be shared between groups

• EMU tasks required/expected by POSSUM

1. Source finding algorithm
2. Flux calibration (EMU aim is 1% flux accuracies)
3. Radio catalogue (position, flux, extent, decomposition into components, spectral index, curvature)
4. Added value products: components grouped into sources, cross-IDs, multi-wavelength SEDs, redshifts

• POSSUM tasks required by EMU

1. Polarisation leakage calibration
2. Fractional polarisation or upper limits
3. Polarised source finding
4. Polarised source fitting / polarisation characterisation
5. Single-dish data (EMU doesn't need this for main/first set of data products)
   • 1300-1800 MHz covered by STAPS south of equator; but 1150-1300 MHz not currently covered, nor northernmost extent of ASKAP survey area. Ongoing discussion in WG7 as to how to obtain these data

• Joint tasks

1. Source classification (e.g., FR-IIs are expected to be double sources with polarised lobes)
2. Beam weighting scheme to optimise leakage / sensitivity / FoV / survey speed or some subset thereof
3. Survey pointing centres (also coordinated with WALLABY)
4. Quality control
5. Extended/Galactic emission

Coordinated approach to simulations & source finding

Emil Lenc is coordinating simulation analysis for EMU.

Input catalogue for EMU and POSSUM simulations have so far been the same. However, EMU and POSSUM simulations have different pixel resolutions so sources with same RA/Dec don't fall on same pixels.

Also, simulations don't currently contain the sort of sources that are simple in Stokes I / complicated in poln that we will need to handle.

Outcomes/tasks:

• Decisions on how to fill in information on polarisation for every EMU source
• Identify common set of simulated sources, beyond ASKAP simulation (which doesn't have enough sources for POSSUM's needs)
• Let Paul Hancock's source finding efforts proceed as planned for now, with POSSUM connection through Bryan Gaensler who is involved in Paul's work

Common calibration/imaging issues

Main issues:

• The need to calibrate polarisation will be a significant time overhead. For DRAO ST and ATCA, have done this carefully, but only need to do it once, unless physical structure changes. Here beam pattern is dominated by electronics - how stable will this be?
• Will polarisation performance degrade at edges of field? If so, POSSUM may need more closely packed mosaic pointings than EMU would like. Won't know the answer to this until we begin getting data.
• Optimal weighting of the PAF beams. WALLABY, EMU and POSSUM will need to agree on this if they are commensal.
• EMU doesn't really need any information from ionospheric calibration. Correcting for ionospheric RM wlll improve dynamic range in Q and U, but probably won't make much difference for Stokes I. And phase variations in Stokes I will be taken out via self-calibration.
• Do we need to worry about solar system objects (moon, Jupiter)? Probably want to stay at least 30 degrees away from the Sun.

CSIRO's focus will likely be WALLABY and EMU. Important for POSSUM to specify polarisation tests that they want performed with BETA.

Does EMU want to spend some of its time at 700-1000 MHz? E.g., WALLABY wants 8 hours per day, EMU wants 12 hours per day. Could spend 4 hours a day at 700-1000 MHz. Or could piggyback on FLASH (25000 deg^2, 2 hours per pointing). Or FLASH could use 4 hours per pointing instead of 2 hours per pointing?

Coordinated approach to catalogues and public products

Seven levels of data products. Levels 1-6 handled/hosted by CSIRO. SSPs are responsible for level 7 (enhanced data products).

Stacking isn't hosted/offered by VO. But can use VO tools to pull out data that you want to stack.

Lisa: won't be a POSSUM catalogue as such, but will be POSSUM-tagged entries. This will all be hosted by ASKAP.

• Coordinated data releases or publications between EMU and POSSUM?
  • Ray: EMU paper 1 will be a data description
  • Data releases: need to put a line under the data-set at some point and call this "data release 1" (underlying data-set will be continually evolving)
  • But if catalogues are being released / updated daily, then what is a data release? Data are not likely to be re-processed, and releases will be continuous rather than staged. So key paper will describe survey and catalogue, but won't correspond to a particular data release.

Coordination of reporting and milestones

Certainly EMU and POSSUM (and other SSPs) should share their draft list of milestones and reports. Need to identify overlaps and dependencies in this list. Should make sure WGs 1-5 of both EMU and POSSUM see the other survey's WGs' respective milestones.

Date for next joint discussion

14 Feb, 10.00am-12.30pm, at ATNF