Joint Astronomy Centre
Google
Show document only
JAC Home
JCMT
UKIRT
Contact info
JAC Divisions
OMP
Outreach
Seminars
Staff-only Wiki
Weather
Web Cameras
____________________

JCMT home
Telescope
Pointing
DB trawling for EL Pointings errors

DB trawling for EL Pointing errors

Summary: historical pointing data suggest that flexture of the backing structure is the cause of the 'transit elevation error' and that this effect will increase with time if the dish is not re-tightened as in 1994.

Scuba pointings

The DB was searched for all SCUBA pointings ever taken and the average change of the pointing around the following points was calculated:
  • transit (highest EL), either in the north or south
  • around 60°, rising (2nd and 5th quadrant)
  • around 60°, setting (3rd and 4th quadrant)
Since the telescope does not reverse direction in Elevation at the 60° points, these can serve as a baseline for the shifts found at transit. In each case, pairs of pointings were selected where the first pointing was done before reaching the test point and the next one after, with the additional constraints that the pointings were within 2 hours of one another and had EL > 35°. Pointings within 2° of transit, were we expect the shift to already be in progress, were excluded as well as EL shifts exceeding 10 arcsecs. The change of pointing calculated is 'timewise': the pointing before minus the pointing taken after crossing the test position.

Each set consists of ~400-500 pairs covering a period from the fall of 1996 to the present. The results are quite telling:

 <dAZ><dEL>
transit: -0.07" -1.48"
60° rising: -0.05" -0.36"
60° setting: +0.14" +0.54"

From this table it is clear that:

  • the average pointing change in EL typically is much larger than in AZ
  • the average pointing change in EL around transit is much larger than around 60° in any quadrant.
  • the effect around transit has the same sign both in the north as in the south: the AZ motion (opposite) does not matter only the EL motion (same) of the dish.

The above result clearly shows that there is an anomalous shift around transit. I don't know why we do not uncover the full 3.5" effect seen in the high-resolution transit tracking, but the results lend support to Per's assertion that the transit shift is partly compensated in the pointing model, which however introduces extra EL pointing errors everywhere else.

Finally, no difference is found using pointings from before 1/1/1998 only, which indicates that this effect has been present at least as long as SCUBA has been on the telescope. I may try this same analysis using the heterodyne archive to attempt to extend the timeline further back.

Heterodyne pointings

To further investigate the elevation pointing problem, Figure 1 shows the results from data trawling of the heterodyne archive over a 1-year period 1989 through december 1999. As for the Scuba pointings 3 cases were investigated:
  • around transit (green line)
  • rising through 60° elevation (red line)
  • setting through 60° elevation (blue line)
with the condition that the pointings were within 2 hours of one another plus the additional constraints to exclude spurious pointings.

The top frame shows the change in the azimuth pointing error around the test point, the bottom one the change in the elevation pointing error. While there is quite a lot of scatter in the data (remember that the pointing model is part of the error), the change in the elevation error around transit clearly differs from the rest in character. While most of the changes scatter around 0, the elevation transit one appears to show a slope. The two lines are just an eye-ball fit, but a single slope looks consistent.

The most interesting feature though is the break in the pattern around 1994. This coincides with the time that the bolted connections in backing structure were re-tightened and ultimately (July 1994) the cone-bars were welded in an attempt to restore the homology of the dish. Consequently the data suggest that the 'elevation transit error' results from flexture in the dish structure and will grow in time until the structure is re-tightened again. I do not think that the cone-bars are part of the picture, although they should be checked just to make sure.

There are a number of other interesting features in the graph.

  • The large see-saw behaviour of the change of elevation error may be the result in part of successive models alternatively over- and under- compensating for the error. Or better, a feature that was not 'part' of the model. In this context it would be interesting to know whether the pointing model is 'on' when collecting data for a new solution. If not, this explanation is invalid.
  • The see-saw shows up after 1991 and seems to disappear during 1997. The latter coincides with Scuba becoming operational in earnest, perhaps resulting in an overall improved pointing model. An interesting question is what happened in 1991: was a 'better' pointing model adopted?
  • What happened to the AZ pointing at transit during last fall?
  • Over the same period the heterodyne data shows a shift half as large as seen in the Scuba data. I think this is too large to ascribe to statistics. Ian Pain suggests the the different location of the receivers may cause them to be affected by a flexture differently.

    Figure 1: pointstat plot
    Figure 1

    Remo Tilanus
    Last modified: Mon Apr 17 16:08:31 HST 2000
    • Contact: Iain Coulson. Updated: Sat Nov 6 18:00:34 HST 2004

    Return to top ^