This is obviously very much a work in progress. Right now it's primarily a note pad where thoughts and questions on the KFPA use cases can be displayed quickly.

Processing and display steps necessary to support the observering steps outlined in Glen's "System Checkout and First Use Case" document as revised 18 May 2009.
  1. K7-Point
    1. Need a way to indicate chosen signal and reference beams.
    2. Is the result of this point to be handled in exactly the same way that the result of the current point is handled?
    3. Just to be clear, although all beams intentionally see the source only the data from one beam is to be used as the source beam (defaults to the center) and one beam is to be used as the reference beam. Pick a default reference beam.
    4. Who's responsible for maintaining the current GFM/Astrid point reduction? Who's going to write the new reduction for this case?
  2. K7-Focus
    1. It's not really described in detail in the document other than to say that it will be carried out in the same manner as is done for the dual beam receivers. Apart from designating a signal and reference beam, is there any difference?
    2. Need the same way to indicate chosen signal and reference beams.
    3. Do we want to do (2) also for the current double beam systems so that the same GFM/astrid focus reduction could could handle both cases with ease?
    4. Seem question as 4 above.
  3. ONOFF/Spectral Band Check
    1. The current GFM display only displays raw data. I think this item is more than that, so correct me if I'm wrong.
    2. It's fairly ambiguous what is desired here.
      1. Is data to be taken with all 7 beams or is this more of a standard dual beam ONOFF?
      2. If this is a standard ONOFF, then is one beam picked to act as the current single beam ONOFF does and move between a signal and reference positions or is this more like a NOD where the user picks a signal beam and reference beam and a nod is done using those two beams. Or should both be supported? Or is it more complicated than that.
      3. What level of "calibration" is desired in the display here?
        1. raw - individual integrations
        2. total-power calibrated out (i.e. use default Tcal values) - indivual integrations, this gets you Tsys values
        3. full ONOFF or NOD calibration - individual integrations
        4. averages over all integrations in cases 2 and 3
    1. This isn't really a scan. I think the idea of demarking the start of a pipeline block is good, but this needs to be folded into the first scan in that block. At least to convey information to the pipeline.
    2. What additional information should be collected here.
      mapping parameters (defaults should be noted in case there's nothing helpful captured from the user)
      1. calibration options - especially smoothing in various places (see the smoothing options in GBTIDL for example)
      2. final desired units in image?
      3. Do we need to capture the type of pipeline block? Or is this always an image-producing pipeline block? Perhaps we need pipeline procedure names that imply the whole block as described in this use case.
      4. Spectral lines to be images.
        1. Channel ranges (or possibly velocity or frequency ranges) to image around each line (may vary with line?)
        2. Channel ranges to use for any baseline fitting prior to imaging
          1. baseline fitting parameters
      5. Statistical flagging options? (we need to discuss at some point what statistical flagging we might want the pipeline to do).
    3. I think the pipeline can't stand around and wait until the end to start processing the entire block. But it can't proceed to making image and weight cubes
  5. ONOFF intensity calibration
    1. In the absence of this step, the default Tcal values should be used.
    2. We'd need some catalog of known spectra or a way to indicate the known spectrum to make use of this later on
    3. Do we need to do this for all 7 beams? How does that generalize to the >> 7 beam case?
  6. Reference Spectrum/TRACK
    1. "display band pass" - same questions as above on spectral band check for similar display
    2. A separate reference for each beam seems necessary. So it's more than a simple TRACK.
    3. How long does this need to be for some desired sensitivity (similar question needs to be asked for the intensity calibration)? Not exactly relevant to the pipeline since it will use whatever is there, but it should be discussed. Ah, I see, 1 minute observations. Is that sufficient in all cases?
  7. Map Region
    1. Initial pass could do total power calibration and save each integration's data. Tcal spectrum is known at this point but reference spectrum is not.
    2. ASTRID, display latest (sig-ref)/ref. Does that mean "ref" as currently known in previous step (which may not be the final ref)? I think that means that the pipeline is going to need to aid ASTRID by doing that processing so that ASTRID can get it from some cache that the pipeline is maintaining.
  8. Reference Spectrum
    1. At this point this is combined with the previous reference to get a composite. Not clear what is intended here. The document says "the mapping software will linearly interpolate the pair of reference spectra so as to compensate for the change in elevation during the observation." Does that mean that the software should consider the reference spectrum to be a function of elevation, i.e. ref(elev) = a + b * elev at each channel and solve for "a" and "b" given the two reference spectra at the two elevations? And then work out ref(elev) for each integration in the map and apply that. Or did you have something else in mind?
    2. Actually, this is "Reference Spectra" I think, since I think you need one for each beam and polarization.
    3. At this point the pipeline makes its first image and weight cubes, corrected for Tcal and the known weather dependent values. What units should the cube be in?
      1. statistical flagging may happen during this step.
      2. baseline fitting and removal happens just prior to imaging.
  9. POINT and FOCUS again. Same issues as above.
  10. Same Reference Spectra, Map, Reference Spectra block.
  11. Pipeline makes second image and weight cubes and combines with first pass. Or, it could just add to the existing image and weight cubes from the first map region. Is this something that should be an option (keep both cubes and final or just go right to the final image and weight)?
  12. Stop Pipeline
    1. Is this necessary? Or perhaps you envision any number of repetitions of the basic internal block and so this would be necessary to mark the end.

-- BobGarwood - 2009-05-19

This topic: KPAF > DirectorsOffice > SWPlanning > KFPADataPipelineUseCasesNotes
Topic revision: 2016-06-08, PatrickMurphy
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding NRAO Public Wiki? Send feedback