HNCO 7M + TP mosaicking

[d-l-walker]

Feathered 7M + TP cubes are in

/upload/HNCO_comb_fits/7m_TP_feather_cubes/
(all are .image files, except for region af, which is .fits ... ¯\_(ツ)_/¯)

7M weight cubes are in

/upload/HNCO_comb_fits/7m_TP_feather_cubes/HNCO_7M_weights/

@keflavich if you could grab these and run your parallelised mosaicking code to get a weighted mosaic, that would be great! Let me know if there's anything else you need from me.

Context: the motivation for creating this product now is for @ericliang45's project.

[keflavich]

Started it. Latest change in 790804f. This is all being smuggled into #396 because I'm doing too many things in parallel to maintain separate feature branches.

[keflavich]

A pretty good version exists, but there were ~100 channels that were mostly blank; those are rerunning now

[ericliang45]

QA summary (last update 8 Jul 2024)

In general, the cube looks amazing (look at the moment 0 map below!).
Available on Globus: /mosaics/cubes/HNCO_7mTP_CubeMosaic.fits
Beam size: 21.7259" X 16.3434" (0.84 pc X 0.63 pc, using D=8kpc), -79.1339 deg. Channel width: 0.21 km/s or 61 kHz.

A few main points summarised below (details of each point are further down in this thread):
1. A few channels are (almost) blank (i.e. near zero intensities) throughout the whole field of view. (Update: indeed due to an incomplete cube transfer.)
2. The last ~84 channels seem to be damaged. Nothing readable by CARTA or astropy. (ditto)
3. There are some stripe/grid patterns. Real or artefacts? (Update: mostly sidelobes of ACA; see Adam's comments below.)
4. Some negative bowls exist.
5. Missing unit in the header, Jy/beam, mJy/beam, K, or something else? (Update: should be Jy/beam. Update2: The new version has fixed this issue.)
6. Edges of individual fields apparent in 1st & 2nd moment maps (Update: due to differential spectral coverage. Looking for a way to mitigate this based on the mosaic cube. Update2: tentatively, trim the edge before using the cube.)
7. No residual continuum spotted (esp checked Sgr A* and Sgr B2). This is good.
8. When reading the cube, there is a file formality issue. Astropy reports "WARNING: File may have been truncated: actual file length (43008002880) is smaller than the expected size (43008004800) [astropy.io.fits.file]" and CASA reports "WARNING: fits blockio ignoring last 960 bytes." "Physical record 1493337 logical record 14933334" "Error reading from FITS image.". But it seems the use of the cube is not affected both in python and CASA (importfits, etc) (which needs further confirmation). This issue is unlikely due to file transfer; it happens with both the web-based download button and the endpoint sync in Globus.
9. Found a "stitch/scar" feature of spatial discontinuity, probably corresponding to one inner edge between adjacent fields.

[ericliang45]

1. A few channels are (almost) blank (i.e. near zero intensities) throughout the whole field of view.

[ericliang45]

2. The last ~84 channels seem to be damaged. Nothing readable by CARTA or astropy.

[ericliang45]

3. There are some stripe/grid patterns. Real or artefacts?

[ericliang45]

4. Some negative bowls exist.

[ericliang45]

5. Missing unit in the header, Jy/beam, mJy/beam, K, or something else?

[ericliang45]

6. Edges of individual fields apparent in 1st & 2nd moment maps. (Taking all positive pixels. Updated on Jan 18.)

Moment 0:

Moment 1:

Moment2:

[keflavich]

The stripes are artifacts, and they are unrecoverable - they are features of the 7m array.

The edge-of-field effects are because of our limited spectral coverage.

The missing channels are a real issue that should be fixed.

[keflavich]

12m+7m+TP vs 7m+TP comparison for one of the streaky regions. It looks like there are sidelobes from the 7m that show up prominently whenever there's a linear feature.

[ericliang45]

7. No residual continuum spotted (esp checked Sgr A* and Sgr B2)

Sgr A*:

Near Sgr A*, where continuum in CS cube was detected (see WP1 minutes on Jan 17):

Sgr B2 (large aperture):

Sgr B2 (small aperture, only negative bowls seen):

[ericliang45]

A thought about the influence of sidelobe residuals (regarding issue No.3). Since the 7m+TP dataset is used in feathering to create the 12m+7m+TP version, would the sidelobes also affect the latter? Though I don't know details of the feathering process, I would imagine those "fake" emission flux would be folded into the final product (perhaps not in a "spatially anchored" way in the real space, but still likely affecting the flux scale in some way I guess).

I wonder how significant this issue is (if it makes sense at all), and whether we would be motivated to clean 7m+TP deeper to improve the 12m+7m+TP version. @d-l-walker

[d-l-walker]

Hi @ericliang45, I think these features would be carried into the fully combined dataset, yes. I think this is one of the reasons people have been cautious to interpret some of the features in that 12m+7m+TP image (i.e. are all of the 'striations' real? Or sometimes artifacts?).

Re: cleaning deeper, this is the plan for the joint imaging of the 12m + 7m data (#384), which will then be combined with the TP data (which we are not cleaning -- we use the TP images created by the single dish pipeline). The hope is that deeper cleaning, plus other approaches (e.g. better masking) will result in improved images.

Ultimately, though, as Adam said above, such artifacts are likely just inherent to the data, and so they'll always be present at some level.

[ericliang45]

Issue No.9
Found a "stitch/scar" feature of spatial discontinuity, probably corresponding to one inner edge between adjacent fields. The location of this feature can be found in the screenshot.