Table of contents
- 1. n21dk01 Correlation Report
- 1.1. General information
- 1.2. Status
- 1.2.1.1. Data Status
- 1.2.1.2. Production Correlation Status
- 1.3. A Priori
- 1.3.1. Yebes
- 1.3.2. Effelsberg
- 1.3.3. KVN Yonsei, Ulsan, Tamna
- 1.3.4. DiFX
- 1.4. Notes
n21dk01 Correlation Report
General information
- Project: "Wideband spectral line VLBI study (Dongjin Kim) 22/43GHz 8Gbps"
- Epochs: G, H, I, J, E, K, L
- Station feedback: No further issues found outside the a priori known issues (weather, broken shutter).
- Requested config: PI request was K-band 0.25 MHz/ch 1.0 sec AP, and separately Q-band 0.50 MHz/ch 0.5 sec AP
- Actual config: 0.25 MHz/ch 0.5 sec AP, because after a few unsuccessful split-correlation configuration "hacks" DiFX and recordings turned out to unavoidably require a joint K+Q correlation. The PI will need to use CASA/AIPS to split out K and Q bands, and apply spectral (Q) or temporal (K) averaging to the requested resolution.
Status
what | date |
---|---|
Effelsberg transfer Epochs G and H complete, remainder postponed due EVN Session 2 VLBI (27 May-17 Jun) | 20.05.2021 |
Yebes transfer Epochs G, H, I, J, E complete | 11.06.2021 |
Effelsberg transfer Epochs I, J, E, K, L completed | 23.06.2021 |
Fringes Ef-Ys K-band and Ef-Ys Q-band, epoch G | 10.06.2021 |
KVN started data transfer | 16.06.2021 |
KVN ongoing transfers... 16.08.2021 still transferring... | (not complete) |
Epoch G full correlation K+Q v1 finished | 15.08.2021 |
Epoch I full correlation K+Q v1 finished | 16.08.2021 |
Released (difxexport) version 1 of epoch G and epoch I to the PI. Awaiting feedback. | 17.08.2021 |
KVN finished data transfer, info from DK Jung | 06.09.2021 |
reported back that Ku missing in epoch H, 7 other files 0-byte, retransfer? | 08.09.2021 |
Data Status
Ep | DOY | Ef 2021 | Ys xfer | Kt xfer | Ku xfer | Ky xfer | Correlated | Releasable? |
---|---|---|---|---|---|---|---|---|
G | 124 | 20.5. ok | 8.6. ok | 22.7. ok | 10.7. ok | 1.7. ok | rev 1 | yes; Nov 2021 |
H | 127 | 19.5. ok | 29.5. ok | 25.7. ok | -missing- | 23.7. ok | no | no |
I | 141 | 20.6. ok | 5.6. ok | 5.8. ok | 4.8. ok | 3.8. ok | rev 1 | yes; Nov 2021 |
J | 141 | 21.6. ok | 8.6. ok | 20.8. ok | 17.8. ok | 14.8. ok | no | no |
E | 142 | 20.6. ok | 11.6. ok | 1.8. ok | 29.7. ok | 28.7. ok | started 1.4.22 | no |
K | 143 | 23.6. ok | (not observed) | 28.8. ok; 2 scans 0B | 26.8. ok | 23.8. ok | no | no |
L | 143 | 23.6. ok | (not observed) | 2.9. ok | 1.9. ok; 1 scan 0B | 31.8. ok; 2 scans 0B | no | no |
Production Correlation Status
Ep | DOY | Correlation | Config | Diagnostics | Notes |
---|---|---|---|---|---|
G | 124 | 15.8.2021 v1 | v2d, vex | Residuals pdf, FITS pclist | Ef not on source until ~18:30 UT Ku residual rate +20 mHz |
H | 127 | ||||
I | 1411 | 16.8.2021 v1 | v2d, vex | Residuals pdf, FITS pclist | Ku residual rate +20 mHz |
J | 141 | ||||
E | 142 | (pending) | Ku residual 0 Hz after Ku -0.8e-12 s/s in VEX | ||
K | 143 | ||||
L | 143 |
A Priori
Yebes
- J. Gonzales: rate 1.17e-13 s/s, offsets in logs named n1k01?ys.log on vlbeer.
- Did not observe K and L due to heavy rain and broken shutter.
- Transferred the data of G, H, I, J, E.
- VEX freq block sidebands UUUU UUUU LLLL LLLL rather than the default UL UL ... from SCHED
- Epoch E: scan No0003 (142-0814) file empty or not recorded
Effelsberg
- U. Bach, etransferred epochs G & H. Pending transfer of the remaining epochs.
- Offset +19.3 usec in G
- Logs named ndk01g, ndk01h, and later n1k01i, n1k01j, n1k01k, n1k01l, n1k01e on vlbeer.
- H-maser rate same as in GMVA f211a, i.e., 2.05e-13 s/s.
- VEX freq block sidebands UUUU LLLL rather than the default UL UL ... from SCHED.
- Epoch G: wind and bad weather, first part ot track unusable, not on source until ~18:30 UT
KVN Yonsei, Ulsan, Tamna
- Table of clock offsets, sign for VEX same as in table
- Table of most recently determined site positions
- H-maser rates not available, assuming 0 us/s for Yonsei and Tamna, same 0 us/s for Ulsan but high residual rate 20 mHz so could perhaps use VEX Ku -0.8e-12 s/s
- Waiting for e-transfer for remaining epochs
- Tie of KVN to Ef+Ys : shift KVN VEX clocks by -2.83 us from kvn a priori values
DiFX
Special setup for correlation, because Effelsberg does not have simultaneous K and Q capability.
- Modes 'kband' and 'qband' in Observed VEX are are dual band at KVN, single band at Ef: would end up with FITS with KVN+Ef KQ+K for mode 'kband', and FITS with KVN+Ef KQ+Q and mode 'qband', i.e., KVN KQ in both FITS files (data duplication).
- Initial attempt using v2d "freqIds=..." and a K-band and a separate Q-band v2d file was only partially successful. Issue: any change in array composition (e.g., telescope recorded data in one scan, in another one not) jumbles the DiFX .input FREQ table frequency IDs an e.g nominally K-band -only list of freqIds refers to some Q-band recorded channels as well and misses some K-band recorded channels.
- Production approach uses VEX file where Ef has been made dual K&Q, a v2d file with two Datastreams for Ef (one K, one Q), and a bit of scripting to separate the Ef filelist into two lists one with K-band and the other with Q-band VDIF files
- Zoom bands of 64 MHz added into to KVN 2ch x 512 MHz to match Ef Ys 64 MHz wide channels in USB/LSB
- Zoom bands added also to Ef Ys 64 MHz LSB channels in order to flip these into USB and produce all-USB visibility data; avoids HOPS fourfit issues with mixed USB-LSB visibilities
Effelsberg file list splitting by band, example for Epoch G:
vsum -s /data/n21dk01g/Ef/*.m5a > filelist.ef rm -f filelist.ef.k filelist.ef.q for scanNr in `vexpeek -s n21dk01g.vex.obs | grep kband | grep EF | cut -c 3-6`; do echo "grep kband 01g_ef_no$scanNr" grep "01g_ef_no$scanNr" filelist.ef >> filelist.ef.k done for scanNr in `vexpeek -s n21dk01g.vex.obs | grep qband | grep EF | cut -c 3-6`; do echo "grep qband 01g_ef_no$scanNr" grep "01g_ef_no$scanNr" filelist.ef >> filelist.ef.q done
Notes
The FITS-IDI file of each epoch contains both K and Q band, at a resolution of 0.5 seconds and 0.25 MHz/channel
This is due to the way that the correlation had to be configured because not all scans had all recordings from all stations, hence the DiFX-internal frequency indices were not consistent across all scans, hence selective correlation in two runs using only specific frequency indices and per-band averaging settings never produced consistent K(/Q)-only products per run.
The K and Q bands must be manually split out by the PI from the FITS-IDI file under AIPS/CASA.
For K-band, temporal averaging needs to be applied during the manual splitting to get from 0.5s AP to PI-requested 1.0s APs.
For Q-band, spectral averaging needs to be applied during the manual splitting to get from 0.25 MHz/channel to the PI-requested 0.50 MHz/channel.