Minutes: Northern Multibeam Working Group meeting, Mon 21 Oct 2000
Those present: Peter J. Boyce, Michael J. Disney, Marco Grossi, Christine A. Jordan, Virginia A. Kilborn, Robert Hugh Lang, Robert F. Minchin
Item 1. Introductions, composition of the group etc.
It was agreed that, for the time being, those present at this meeting should constitute the sole members of the Northern Multibeam Working Group (NMWG). It was also agreed that PJB should act as Chair of the Group.
The purpose and aims of the Group were debated. It was agreed that the specific functions of the Group included:
Item 2. Present state of the project
PJB briefly reviewed the present state of the project.
It was still not clear whether the data from the first full observing run (in Jan-March 1999) could be reduced in such a way as to make them scientifically useful. An undiagnosed fault with the noise diode had meant that the cal signal throughout this run had been roughly a factor of 5 too high and different in each beam. The very high level of the cal signal and the way in which the gridder median combines incorrectly calibrated data combine to produce very low S/N data. There is a possibility that the correct cal levels in each beam may be recoverable from the manual cals done during the run. However, a way then has to be found of amending the Tsys values in all raw spectra before reduction, a potentially massive undertaking.
At the start of the second run (March-June 2000) the problem with the noise diode was diagnosed and a short term solution found (see Item 4 below). During the second run two distinct surveys were undertaken.
Firstly, a start was made on an all sky survey intended to have similar sensitivity to the HIPASS survey. Data were taken in a strip at dec=70deg to 78deg. PJB reported that of the 6,400 scans needed to complete this strip, around 5,200 has been taken during the second run, leaving only about 1,200 remaining to be done.
Secondly, a 4deg x 8deg area towards the Virgo cluster was observed with a total integration time of 9 times that of the HIJASS data. The intention was that the cluster data itself would be used by Jon Davies' group in Cardiff as part of their multi-wavelength studies of the cluster.
A final, correctly calibrated reduction of the data from the second run has not yet been possible because of the problems with the new version of "gridzilla". However, some of the data have been gridded using the old gridder (and the Parkes beam size). RFM presented a plot of system noise against frequency for a sample HIJASS cube. From this he estimated that the HIJASS data has 20% more noise than HIPASS data. He considered this noise may be mostly due to the higher system temperature at JB. RFM showed a similar plot from the deep Virgo data. This shows that the S/N appears to fall as sqrt(t) to at least this integration time.
PJB reported that he had sorted all the data from the second run into 4deg x 8deg chunks and that he would create a set of 8deg x 8deg cubes as soon as the new gridder was working reliably.
RFM and CAJ gave a brief report on the status of the "drift-scan" experiment which had been running whilst the telescope was stationary for track repair work over summer 2000. This should have surveyed an area about 2deg wide in dec at dec of +54 deg, a total area of about 56 sq degs. It was estimated that this could produce data with an equivalent integration time of about 20 to 30 times longer than HIJASS data. CAJ has given the data from this survey to PJB who will attempt to reduce it and then report further on the quality of the data.
There was some discussion as to the intended use of the Virgo Deep data. MJD expressed the hope that Jon Davies' group would utilise this data as it related to the cluster and that the NMWG would use the data from behind the cluster. Jon Davies' group has multi-colour optical data of the area surveyed and MJD suggested we would be given access to this in return for the HI data. MJD is to discuss this issue with Jon Davies.
Item 3. Software issues
It was agreed that the ideal software situation would be to have working versions of both "livedata" and the latest version of "gridzilla" at Jodrell Bank, Cardiff and Bristol. The latest version of "gridzilla" is particularly important since it enables the user to input beam size (the previous version was hardwired to the Parkes beam size).
At present there are working versions of "livedata" and the OLD version of "gridzilla" at Jodrell Bank. There is a working version of "livedata" at Cardiff. Rodney Smith (Starlink system manager at Cardiff) reported on Thurs 17 Oct 2000 that the new "gridzilla" was working in Cardiff. PJB had verified that this did indeed work and produce a cube from a small amount of data. RFM found that when trying to use it to grid the whole of the Parkes DEEP data, gridzilla crashed. RFM suspects this may be a problem with the memory available on "multi" (the machine used at Cardiff). None of the software is presently installed at Bristol.
RFM agreed to consult with Rodney Smith about the continuing problems running gridzilla at Cardiff. Ant Holloway (JB Starlink manager) has agreed to install the new version of "gridzilla" at JB. PJB is to talk to Rhys Morris (Bristol system manager) about installing the software in Bristol.
VAK's computing needs at JB were discussed. It was resolved that CAJ and VAK should lobby the relevant people at JB with a view to getting for VAK a workstation with at least 500MB RAM and 30GB hard disk space.
It was noted that both VAK and RFM had produced an "automatic galaxy finder" for use with HIPASS/HIJASS data. It was noted that a third finder (the "Swinburn finder") had been developed by other member of the HIPASS group. It was decided that VAK's and RFM's galaxy finders should be tested using the Virgo deep data.
Item 4. Hardware issues
It was reported that the problems encountered when the telescope scans through sources at Elevations of 38deg or 65deg are still present and will need to be taken into account in drawing up an observing strategy for the next observing run.
RHL reported that he had checked the observing room hardware and this seemed in good order.
It was found at the start of the second observing run that the noise diode power level increases with switching frequency from about 5Hz to about 50Hz. Beyond this it stays at a level around 5 times higher than its specified level. Since the MB software switches the diode at 500 Hz this gave us a very high cal signal during the 1st run. During the second run a set of annenuators were placed between the diode and the recievers to bring cal to an accetpable level.
RHL reported that the technical staff at JB are trying to determine the cause of the fault in the noise diode. Unfortunately, John Edgely, the technican with responsibility for this issue, was absent on the day of the meeting. RHL himself had tested the diode and found it now seems to respond differently to how it did during the last run. It increases in power level from 20Hz to a maximum at 140 Hz. At 280Hz it appears to fail altogether.
RHL reported that Colin Baines know that it is imperative that the diode be fixed before the next observing run. RHL will remain in contact with CB and JE.
MJD queried what our strategy should be if we were to lose one channel/beam during the observing run. RHL pointed out that the answer may depend to a large extent upon the cause of the problem and hence the likely time required to fix it. RFM pointed out that on the last run we had operated a strict policy that we did not consider it worth observing unless the whole system was working since the drop in system sensitivity was too great to make this worth while. PJB pointed out that there is sometimes flexibility in the scheduling in that pulsar time can sometimes be brought forward and further MB time made available at the end of the run in compensation. PJB undertook to determine the drop in S/N from the loss of one beam.
Item 5. Next Observing run
It was reported that the MB receiver will be put on the telescope on 22 December 2000 and that we shall be observing from 3rd - 28 Jan 2001.
RHL said he would be at JB from a few days before 22 Dec 2000 and would be around during the run for as much time as his teaching commitments allow. He will be available in emergencies.
VAK and CAJ will be around at JB for the entire period of the run. It was agreed that MR would help with the observing. PJB volunteered to do a few days observing. RFM would prefer not to have to observe during this run.
There was a lengthy discussion about the observing strategy for the upcoming run.
PJB pointed out that, taking into account PIs and losing time to wind, a reasonable estimate is that we will have 24 days observing time during the run. Making reasonable estimates for observing overheads, it would take about 9 more days to finish the dec=74deg strip. This would enable us to make the full 12.5 8deg x 8deg cubes needed to cover the whole strip (rather than the 7 we can make with the data we presently have). It was generally agreed that this should be done.
The remaining 15 days observing time would be sufficient to observe between 5 and 6 further 8degx8deg HIJASS cubes. PJB suggested that a possible strategy would be to stick to observing scans within the HIJASS dec strip strategy, but to select 8degx8deg chunks which cover potentially interesting areas of the super-galactic plane. Possibilities include finishing off the M81 Group area in the +66deg strip and/or surveying the Ursa Major cluster area. MJD suggested that areas which will be surveyed by the SLOAN survey in the near future may make high priority targets.
MJD suggested that some or all of the remaining 15 days could be spent surveying a 4degx4deg area around the A1367 cluster. This cluster is known to contain many gas rich galaxies and MJD feels it may be the best place to search for giant low surface brightness galaxies. MJD proposed that we conduct a "deep" (i.e. long integration time survey of this area). RFM pointed out that at the distance of the cluster (cz=6,500 km/s) even large Malin-1 type galaxies will be smaller than the beam size. Hence, the only advantage of longer integration time will be to sample further down the HI mass function. MJD suggested this may be necessary in order to detect giant LSBGs at this distance. It was generally agreed that A1367 may make an interesting target. However, there was some concern as to what sort of integration time would be necessary to detect giant LSBGs. MJD agreed to consider this matter further and report his conclusions to the Group.
PJB pointed out that the results of the final reduction of the data from the last observing run, including the Virgo deep data and the drift scan data, may have an impact on what we wish to observe in the next run. A final decision on what to observe was deferred.
Item 6. Website
PJB suggested that the Group should have a public website along the lines of the HIPASS group's website. Whilst this idea was generally supported, it was also felt that it would be prudent to wait until we have produced publishable results before producing such a public site.
Item 7. AOB
It was agreed that PJB would write up the minutes to the meeting and make them available on the web to members of the Group.