ALMA2014 LBC SVDATA

Overview

This casaguide describes the imaging used for the ALMA 2014 Long Baseline Campaign (LBC) Science Verification (SV) data. The purpose of this campaign was to verify and demonstrate ALMA's performance on baseline lengths of ~10 km in preparation for this very high angular resolution capability to be offered for Cycle 3. As part of the overall LBC testing plan, five science targets were chosen to span the range of properties that would be needed to offer the capability and to demonstrate to the community both the power and caveats associated with long baseline observing. The SV portion of the campaign was carried out from October 12 to November 14, 2014. As described below, the data products for four of the five targets can now be dowloaded from the regional ALMA Science Portals, while the imaging scripts and salient details are provided on this page (the final target 3C138, will be released separately with other SV polarization data in the near future).

Obtaining the Data

To download the data, go to the ALMA Science Portal (SP) Science Verification data page for the region closest to your location:

North America

Europe

East Asia

Choose a LBC SV science target: 13. Juno, 14. Mira, 15. HL Tau, or 16. SDP.81

There you will find gzipped tar files of the form (these files can be unpacked with the tar -xvzf command):

• Target_Band_Freq_UnCalibratedData: Raw, uncalibrated data in ALMA Science Data Model (asdm) format

• Target_Band_Freq_CalibrationScripts: Calibration scripts that go along with the UnCalibratedData. These scripts can be executed once placed in the same directory as the UnCalibratedData

• Target_Band_Freq_CalibratedData: Calibrated and combined uv-data. In the case of continuum datasets, an appropropriate amount of channel averaging has been applied to reduce the data size. Line data sets are provided at their full resolution. If self-calibration was possible, you will also find the final phase-only and amplitude tables in a subdirectory called "calibration". This gives the choice of either performing your own self-calibration or applying the supplied tables. If you are going to apply these tables, be sure to also run any flagdata commands in the imaging scripts. If the self-calibration tables are not present, we deemed self-calibration unfeasible. Note that the imaging scripts are provided on this page, not at the science portal.

• Target_Band_Freq_ReferenceImages: Final reference images with self-calibration applied where possible. For the few cases where primary beam correction is warranted these files are also provided. Continuum images with .pbcor in the name have already been corrected. The line files with .flux in the name are the primary beam images themselves and can be applied to the corresponding line data if desired using the CASA task impbcor.

• README: Summarizes the data available for that science target.

The data have been packaged in this manner to allow a wide range of interaction with the data, ranging from complete re-reduction, to simply having a look at the final reference images. Please only download the data products you expect to use -- we expect the initial load on the servers could be high and the UnCalibratedData, and in the case of targets/bands with spectral line data the CalibratedData, files are quite large.

CASA Version

The data were calibrated and imaged using CASA 4.2.2. At this time, we have no reason to believe that they would not work in CASA 4.3 with a few caveats: (1) The calibration scripts check that the version they were made in matches the version being used. If you want to try them in CASA 4.3 you will need to change this section of each calibration script. (2) Be aware that data imported in CASA 4.3 cannot then be imaged in CASA 4.2.2 due to the necessary addition of a new keyword in the MS format. Please note, however, that only CASA 4.2.2 has been tested.

CASA can be downloaded from: http://casa.nrao.edu/casa_obtaining.shtml The CASA 4.2.2 version is called ALMA pipeline CASA 4.2.2

LBC SV Imaging

For each LBC SV target there is a brief description of the data and imaging, along with links to the imaging scripts. For spectral line targets, a table of the lines observed is also given. The "detection" columns are merely descriptive to give a sense of what the data show, they are not meant to be taken as scientific assessments. The continuum was detected for all four of the science targets at each observed band. For all of the targets the number of antennas employed was of order 30.

The most important imaging take home messages

• UV-Coverage is critically important: For each target/band a scheduling block was designed to run for 60-70 minutes, with about 30 minutes on source per execution. The number of executions carried out depended on the complexity of the imaging required, from short snapshots of the bright compact emission from Juno to to the many executions required for the complex morphologies of HL Tau and SDP.81. Additionally, we found that due to the non-optimal configuration available for the LBC, which had a deficit in uv-coverage for baselines between 200 to 500 m the choice of the robust parameter in clean is a crucial one. Although more naturally-weighted choices (i.e. larger robust parameter) in principle give better sensitivity, they also show poor dirty beam (i.e. psf) characteristics, with a large extended plateau in addition to the more point-like morphology that corresponds to the fitted clean beam. Thus, for full resolution images, we typically employed robust=0.0, though occasionally up to 1.0 was used.

• Observing narrow thermal line emission at high resolution is challenging: Except for Mira, which has the advantage of having compact line emission, as well as emission from highly-excited or even masing lines, the imaging of line emission at the very high resolution afforded by the longest baselines is challenging. For example, at 89 GHz with 0.25 km/s channels, 30 antennas, 3.5 hours on-source, and an angular resolution of 70 mas, the expected spectral rms noise per channel is 2.6 mJy/beam but only 83 K in surface brightness sensitivity. As a result, it was necessary to apply a uv taper when making the spectral line images for HL Tau and SDP.81 in order to achieve good detections.

• Extended emission at high angular resolution can best be imaged with multiscale: Cleaning diffuse emission on size scales significantly larger than the beam with the standard delta-function deconvolution method results in the "clean instability": giving the diffuse emission a "pointilated" or "cotton-candy" like morphology. This can be ameliorated by using a range of scales to do the cleaning. In CASA, this technique is accessed through the clean multiscale parameter. The scales are given as multipliers of the cell (pixel) size. It is essential to always use 0 for the first scale as this corresponds to the normal clean beam, multiscale=[0,5,15] is usually a good starting point for experimentation. For more details see: Cornwell 2008 (IEEE Journal of Sig Proc., 2, 793).

• Combining bands can produce amazing results: For sources that have significant spatial changes in continuum spectral index, and wide observed fractional bandwidth (>15%) for which the signal-to-noise is high throughout the frequency range, it is possible to use a technique in clean to simultaneously account for the changing spectral index ([math]\displaystyle{ S_{\nu}\propto \nu^{\alpha} }[/math]) to produce a high fidelity image, and a map of the spectral index along with its estimated error (see Rau, U., & Cornwell, T.J. 2011, A&A, 532, AA71). However, the caveat that the signal to noise must be very good, especially at the frequency extrema, is a strong one. So far we have restricted our tests to nterms=2, though even higher orders are possible. The results of this technique on HL Tau Band 6+7 are dramatic, and it also provides some improvement for Mira at Band 3. However, our test show that for the weaker emission at Band 3 for HL Tau and all bands for SDP.81 at native resolution, this technique does not work well. Bottom line: use with caution, but it can produce amazing results. Since this method automatically masks the resulting spectral index ([math]\displaystyle{ \alpha }[/math]) map where it estimates [math]\displaystyle{ \alpha/\alpha_{error}\lt 4 }[/math] a good rule of thumb is that if the spectral index map is mostly masked in the image signal regions, there is inadequate S/N for this technique to be successful.

Juno - Asteroid

Asteroid (3) Juno was observed in continuum in Band 6 to demonstrate the ephemeris capability. The observations were taken over 60% of Juno's rotation period of 7.2 hours. The data are split into approximately 18min intervals (10min on source) spread across 10 individual datasets and imaged and self-calibrated separately in order to prevent smearing due to the rapid rotation. The angular resolution achieved ranges from 32 x 24 mas to 42 x 36 mas, differing primarily due to the level of phase stability as the observations span the transition from night through dawn and into daytime, but changing uv-coverage also plays a role.

Juno Band 6 Continuum (5 executions): Juno_Band6_Imaging.py

Mean Continuum Frequency: 233 GHz

Mira - AGB Star

Mira (omicron Cet) was observed in continuum and spectral lines in Bands 3 and 6. Angular resolutions achieved for the continuum data are 70 x 60 mas and 34 x 24 mas in Bands 3 and 6, respectively. In order to achieve the best possible dynamic range, the continuum and SiO v=1 line data were self-calibrated independently. The maser emission from the SiO v=1 line was used to derive self-calibration for the line data. In order to obtain high dynamic range on the band 3 continuum, which has a wide fractional bandwidth, it was also necessary to take into account the spectral index with nterms=2.

Mira Band 3 Continuum and Lines (3 executions): Mira_Band3_Imaging.py

Mean Continuum Frequency: 94 GHz

Line	Transition	El (K)	Rest Freq (GHz)	detection
SiO v=0	2-1	2.1	86.8470	weak maser? + thermal
SiO v=1	2-1	1771	86.2434	strong maser + thermal
SiO v=2	2-1	3523	85.6405	thermal + absorption
29SiO v=0	2-1	2.1	85.7590	weak maser? + thermal

Mira Band 6 Continuum and Lines (2 executions): Mira_Band6_Imaging.py

Mean Continuum Frequency: 231 GHz

Line	Transition	El (K)	Rest Freq (GHz)	detection
H30alpha	-	-	231.90093	none
H2O v2=1	5(5,0)-6(4,3)	3450.7	232.68670	thermal
SiO v=0	5-4	20.84	217.10498	weak maser? + thermal
SiO v=1	5-4	1789.8	215.59595	strong maser + thermal
SiO v=2	5-4	3541.7	214.08854	thermal + absorption
29SiO v=0	5-4	20.58	214.38576	weak maser? + thermal

HL Tau - Protoplanetary Disk

HL Tau was observed in continuum and spectral lines in Band 3, and (only) continuum in Bands 6 & 7. Angular resolutions achieved for the continuum data are 85 x 61 mas, 35 x 22 mas, and 30 x 19 mas in Bands 3, 6, and 7, respectively. It was necessary to uv-taper the spectral line data so the resulting angular resolution is coarser (ranging from 0.25arcsec to 1.1 arcsec). In addition to the more standard image products we also provide an example of combining the Band 6 and Band 7 continuum data using nterms=2 to derive the spectral index and create a high fidelity image.

HL Tau Band 3 Lines (7 executions for each of two tunings: 89 and 115 GHz): HLTau_Band3_89GHz_Imaging.py , HLTau_Band3_115GHz_Imaging.py

Line	Transition	El (K)	Rest Freq (GHz)	detection
HCN	1-0	0.0	88.63185	absorption
HCO+	1-0	0.0	89.18853	outflow, disk, absorption
CN	1-0	0.0	113.49097	absorption
12CO	1-0	0.0	115.2712	outflow, confused

Note only the frequency of the strongest hyperfine (HF) transition is listed for HCN (3 HFs) and CN (4 HFs).

HL Tau Band 3 Continuum (combined continuum from both line tunings): HLTau_Band3_Cont_Imaging.py

Mean Continuum Frequency: 101.9 GHz

HL Tau Band 6 Continuum (9 executions): HLTau_Band6_Cont_Imaging.py

Mean Continuum Frequency: 233.0 GHz

HL Tau Band 7 Continuum (10 executions): HLTau_Band7_Cont_Imaging.py

Mean Continuum Frequency: 343.5 GHz

HL Tau Combined Band 6+7 Continuum (nterms=2): HLTau_Band67_Cont_Imaging.py

Mean Continuum Frequency: 287 GHz

SDP.81 - Lensed Galaxy

The lensed galaxy SDP.81 was observed in continuum and spectral lines in Bands 4, 6 & 7. Angular resolutions achieved for the continuum data are 60 x 54 mas, 39 x 30 mas and 31 x 23 mas in Bands 4, 6, and 7, respectively. It was necessary to uv-taper the spectral line data so the resulting angular resolution is coarser (up to 169 mas). This source lies at a red-shift of ~3, accounting for the large offset in rest frequency compared to the observing band. In addition to the more standard image products we also provide an example of combining the Band 6 and Band 7 continuum data using nterms=2 to derive the spectral index and create a high fidelity image. However it is important to note that this procedure did not work well at full resolution, as evidenced by the almost complete masking of the resulting spectral map -- this is an indication that the spectral index and hence resulting flux-densities are not well constrained. However, using a bit of taper (larger beam) improved the signal enough for the technique to work.

SDP.81 Band 4 Continuum and Lines (12 executions): SDP.81_Band4_Imaging.py

Mean Continuum Frequency: 151 GHz

Line	Transition	El (K)	Rest Freq (GHz)	detection
CO	5-4	55.3	576.26793	detection

SDP.81 Band 6 Continuum and Lines (9 executions): SDP.81_Band6_Imaging.py

Mean Continuum Frequency: 236 GHz

Line	Transition	El (K)	Rest Freq (GHz)	detection
CO	8-7	154.9	921.79970	detection
H2O v=0	2(0,2)-1(1,1)	53.4	987.92676	weak detection

SDP.81 Band 7 Continuum and Lines (11 executions): SDP.81_Band7_Imaging.py

Mean Continuum Frequency: 290 GHz

Line	Transition	El (K)	Rest Freq (GHz)	detection
CO	10-9	248.9	1151.98544	detection

SDP.81 Combined Band 6+7 Continuum (tapered, nterms=2): SDP.81_Band67_Cont_Imaging.py

Mean Continuum Frequency: 268 GHz