Giant Cube: HCNO (12m + 7m + TP)

[keflavich]

This thread is for discussion of the fully-combined HCNO cubes.

As of 11/13/2023, there are the following cubes:

-rw-rw-r--+ 1 adamginsburg adamginsburg 659G Sep 15 16:13 /orange/adamginsburg/ACES//mosaics/cubes/HNCO_7m12mTP_CubeMosaic_20230910.fits
-rw-r--r--  1 adamginsburg adamginsburg 8.2G Oct 11 02:12 /orange/adamginsburg/ACES//mosaics/cubes/HNCO_7m12mTP_CubeMosaic_downsample9.fits
-rw-r--r--  1 adamginsburg adamginsburg 8.2G Sep 15 16:13 /orange/adamginsburg/ACES//mosaics/cubes/HNCO_7m12mTP_CubeMosaic_downsampled920230910.fits
-rw-r--r--  1 adamginsburg adamginsburg 1.7G Sep 17 07:48 /orange/adamginsburg/ACES//mosaics/cubes/HNCO_7m12mTP_CubeMosaic_downsampled9_downsampledspectrally.fits
-rw-r--r--  1 adamginsburg adamginsburg 8.2G Oct  2 05:23 /orange/adamginsburg/ACES//mosaics/cubes/HNCO_7m12mTP_CubeMosaic_downsampled9.fits 
-rw-rw-r--+ 1 adamginsburg adamginsburg 659G Oct  1 11:09 /orange/adamginsburg/ACES//mosaics/cubes/HNCO_7m12mTP_CubeMosaic.fits

The 20230910 versions are deprecated and were just kept as backups; they'll be deleted sometime.

This one is badly broken and being removed

-rw-r--r--  1 adamginsburg adamginsburg 8.2G Oct  2 05:23 /orange/adamginsburg/ACES//mosaics/cubes/HNCO_7m12mTP_CubeMosaic_downsampled9.fits 

The preferred, right now, are:

-rw-r--r--  1 adamginsburg adamginsburg 8.2G Oct 11 02:12 /orange/adamginsburg/ACES//mosaics/cubes/HNCO_7m12mTP_CubeMosaic_downsample9.fits
-rw-rw-r--+ 1 adamginsburg adamginsburg 659G Oct  1 11:09 /orange/adamginsburg/ACES//mosaics/cubes/HNCO_7m12mTP_CubeMosaic.fits

[keflavich]

Downsampled spectrally is fine, but spatially incomplete and needs to be remade:

The full cube is in a good state, but is missing field af

[keflavich]

@ashleythomasbarnes noted edge issues in the peak intensity map:

[keflavich]

Channel 693 is bad:

[keflavich]

Field af is still broken; that's the one field with junk it in in HNCO specifically:
#255 (comment)

[keflavich]

@ashleythomasbarnes the reason for your edge effects is as I said: these are the boundaries between sub-regions:

if you want peak intensity maps without those problems, limit the range to -60 < v < 68 km/s. Or we can try to come up with an edge-excluding mask. You might do that by, e.g.:

data = fits.getdata('HNCO_7m12mTP_CubeMosaic_downsample9.fits')
mask = np.isnan(data)
mask = scipy.ndimage.binary_dilation(mask, iterations=5)
data[mask] = 0
peak = np.max(data, axis=0)

but be wary of the costs of doing that... (even with spectral cube as the backend)

[ashleythomasbarnes]

Also note that there is still some continuum in this images...
I ran very basic continuum identification on the cubes and get the following continuum map.

[d-l-walker]

@ashleythomasbarnes I guess this is related to what Álvaro mentioned during the f2f meeting? That most cubes still appear to have some continuum throughout the mosaic, particularly near Sgr B2/A*. What you're showing seems to confirm that pretty clearly.

[mpound]

HNCO_CubeMosaic_downsampled9.fits - first 381 channels are blank, then intermittent blank channels subsequently.
HNCO_7m12mTP_CubeMosaic_downsampled9.fits - has the checkerboard "didn't finish writing" problem

[keflavich]

Field af has been a mess for a long while, causing the NaNs seen here:

(the contours show a reasonable finite level of the weights)

The solution appears to be this:
astropy/reproject#474