This page was originally created for ALMA Early Science Cycle 0 (16 antennas), and all of the results below the Goals section are for that configuration.
However, please note that the CASA simulator does not account for any sources of dynamic range limitations, except for uv coverage and integration time.
Real results will be worse than these theoretical performance limits. To get an idea of the dynamic range limit for an ALMA snapshot with 31 antennas, see
CSV-553 and the 05-Mar-2014 report which achieves ~5600 in Band 3 on a 5.62 Jy quasar (J0522-364) with 2.5 minutes on source.
A
recent paper in ApJL demonstrates a dynamic range of order 23000 in Band 6 on a 0.5 Jy grid-source quasar (J0635-7516).
To compare this to the theoretical limit, I re-ran the simulation of an 8-hour track using Cycle 5 configuration 3 (with 43 antennas) with the same simulated field
as below. The resulting image reaches a dynamic range of 45000.
Goals
- Explore the performance of imaging quality of an ALMA Cycle 0 configuration that can be obtained by combining several short snapshots at different hour angles.
- Run a case where the image is dynamic range limited, and one where it is thermal noise limited.
Simulated field
- Start with Band 7 (345 GHz), Cycle 0 ES-extended configuration, Dec = -35 deg
- Create a field with 3 bright sources (1, 0.5, 0.25 Jy), and five fainter ones: 10, 5, 2.5, 1 mJy, and a second 0.5 mJy adjacent to the brightest one
- Create a second field without the 3 bright sources
- Simulate both with simdata for an 8 hour track, and flag portions of the data corresponding to different numbers of snapshots
- Include atmospheric noise but not phase noise (since with bright sources, self-cal should remove most of the phase noise)
- Clean interactively like an observer would do, with appropriate small boxes
- Compute the dynamic range as the ratio between max peak flux density of brightest source divided by image rms (computed over a source-free region that represents a significant fraction of the area within the half-power beam diameter)
Results for dynamic-range limited case (8-source field)
Note: These values are computed
without phase noise. Running the same simulations with a realistic level of residual phase noise (about 5um rms) results in
a factor of 2 decrease in all the dynamic ranges.
Total time on source |
|
Split the time into this number of snapshots |
(hrs) |
|
1 |
2 |
3 |
4 |
8 |
20 seconds |
Dynamic range |
120 |
- |
- |
- |
- |
1 minute |
Dynamic range |
178 |
- |
- |
- |
- |
0.25 |
Dynamic range |
410 |
- |
- |
- |
- |
0.5 |
Dynamic range |
500 |
930 |
1130 |
1150 |
- |
Elevation coverage |
|
|
|
|
- |
UV coverage |
|
|
|
|
- |
Image |
|
|
|
|
- |
1 |
Dynamic range |
820 |
- |
1350 |
- |
2000 |
Elevation coverage |
|
- |
|
- |
|
UV coverage |
|
- |
|
- |
|
Image |
|
- |
|
- |
|
3 |
Dynamic range |
2670 |
- |
2800 |
- |
Elevation coverage |
|
- |
|
- |
UV coverage |
|
- |
|
- |
Image |
|
- |
|
- |
8 |
Dynamic range |
5060 |
- |
- |
- |
Elevation coverage |
|
- |
- |
- |
UV coverage |
|
- |
- |
- |
Image |
|
- |
- |
- |
- 480 minutes
- Full 8-hour track (480min) cleaned down to 1mJy
- amp vs. time (ant#0 baselines only), image
- rms of 0.164 mJy/beam: Max = 0.83, dynamic range = 5061, rms is 5 times worse than without the 3 bright sources
- with 10% pwv phase noise: rms = 1.44 mJy/beam, but Max dropped to 0.097
- with 5 um rms phase noise: rms = 0.268 mJy/beam, Max is 0.83Jy
- 180 minutes (rms would be 0.27 scaling the 8-hour track)
- One x 180 minutes cleaned down to 1.3mJy
- rms of 0.31mJy/beam, Max = 0.823, dynamic range = 2670
- Three x 60 minutes spread over 6.5 hours
- rms of 0.30mJy/beam, Max = 0.830, dynamic range = 2800
- 64 minutes (rms would be 0.45mJy scaling the 8-hour track)
- Eight x 8-minute snapshots (64min) spread over 8 hours
- amp vs. time, image
- image rms of 0.41 mJy/beam, Max = 0.83, dynamic range = 2024, rms is 4.4 times worse than without the 3 bright sources
- 5um phase noise gives 0.77 mJy/beam
- Three x 21-minute snapshots spread over 7 hours
- image rms of 0.62 mJy/beam, Max = 0.83, dynamic range = 1351
- One 64min observation at transit
- amp vs. time
- image rms = 1.0mJy/beam, Max = 0.82Jy, dynamic range = 820, rms is 1.39x worse than splitting into 8 chunks
- 32 minutes (rms would be 0.64 scaling the 8-hour track)
- Four x 8-minute snapshots (32min) spread over 6 hours
- amp vs. time
- rms = 0.72 mJy/beam, Max = 0.83 Jy, dynamic range = 1153, rms is 5.5 times worse than without the 3 bright sources
- with 5um rms phase, new rms = 1.18mJy/beam
- Three x 10.67-minute snapshots spread over 4 hours
- amp vs. time
- rms = 0.73 mJy/beam, Max = 0.83 Jy, dynamic range = 1137
- Two x 16-minute snapshots spread over 4 hours (with one at transit)
- amp vs. time
- rms = 0.89mJy/beam, Max = 0.83, dynamic range = 932
- One 32 minute observation near transit
- amp vs. time
- rms = 1.59 mJy/beam, Max = 0.82 Jy, dynamic range = 516
- rms is 2.21x worse than splitting into 4 chunks, 2.18x worse than splitting into 3 chunks, 1.94x worse than splitting into 2 chunks
- 15 minutes (rms would be 0.93 scaling the 8-hour track)
- One 15 minute observation near transit
- rms = 2.0 mJy/Beam, Max = 0.82 Jy, dynamic range = 410
- 1 minute: rms = 4.6 (would be 3.6 scaling the 8-hour track)
- 20 seconds: rms = 6.6 (would be 6.2 scaling the 8-hour track)
Results for thermal-noise-limited case (5-source field)
Time (hours) |
Sensitivity calculator noise (mJy/beam) |
Achieved noise |
8 |
0.038 |
0.032 |
1.067 |
0.104 |
0.094 |
0.533 |
0.147 |
0.141 |
- 480 minutes
- Full 8-hour track cleaned to 0.1mJy (image)
- image rms = 0.032mJy
- Max = 9.3mJy, dynamic range = 291
- 64 minutes
- Eight x 8-minute snapshots spread over 8 hours
- rms = 0.094 mJy
- Max = 8.98 mJy, dynamic range = 96
- 32 minutes
- Four x 8-minute snapshots spread over 8 hours
- rms = 0.127 mJy
- Max = 9.02mJy, dynamic range = 69
- Three 10.67-minute snapshots over 4 hours
- rms = 0.135 mJy
- max = 9.29 mJy, dynamic range = 69
- Two 16-minute snapshots over 4 hours (with one at transit)
- rms = 0.142
- Max = 9.24 mJy, dynamic range = 65
- One 32-minute observation near transit
- rms = 0.141 mJy, which is 1.11x worse than splitting into 4 chunks, 1.044x worse than 3 chunks
- max = 9.23 mJy, dynamic range = 65
--
ToddHunter - 2011-07-06