|
|||
| Joint Astronomy Centre |
|
|
|
| Show document only |
|
||||||||||||||||||||||
|
____________________
|
||||||||||||||||||||||
|
Investigations into inclinometry data for 1998IntroductionThe object of this work was to see if there was anything to be learned about the recent inclinometry variability from studying the "raw" data. There were no dramatic finds; conclusions are summarized below.Review of difference plotsI looked through the file kept by IMC of differences in models between successive inclinometry plots. This is not exhaustive, but I hoped to get a feel for the kinds of things that change. the wheel joint interactions occur in fourteen clusters, and it is within those clusters that most of the differences are to be seen. What I found was when changes occur they seem to do so in one of two patterns.Pattern A, exemplified by 0613-0610 shows a remarkable symmetry between features in clusters 1, 2, 3 and 8,9,10, with features of opposite sign in cluster 4,5,6 and 11,12,13. Some of the symmetry is due to the symmetric nature of the track, of course. It is striking, but perhaps coincidental, that all of the negative spikes appear in the first interaction in a cluster, whereas all of the positive spikes appear in the third interaction in their clusters. A more extreme example of the opposite sign, in which some of the spikes have broadened to encompass more than one interaction in the cluster, is 0830-0821. Pattern B is rarer and occurs in 0620-0613 and to a lesser extent in 0504-0411. In this pattern, the spikes are in clusters 4,5,6,7,12, and 13. There are three plots showing significant changes outside these patterns. 0611-0610 and 0610-0606 both have a feature between clusters 3 and 4 - presumably some problem with 0610. However, the second of the plots also shows a large and wide feature between cluster 11 and 12. Not so easily explained because it is so large. Since it doesn't appear in the 0611-0610 plot we can assume it is a feature of the 0606 data. See below. Raw data - check on data reduction.I took a set of data from 981107 in the unreduced form. I took the "slope" of the data by taking the difference between means:Slope = ((mean between 30 and 50 degrees)-(mean between 390 and 410 degrees))/360 I divided the voltages by 20, removed an overall mean, and arrived at "corrected" values in arcseconds. I then derived F1, F2 and F3, as follows: F1 = (LY-RY)*1.21
These were identical to the reduced values from the standard data reduction routines. Slope removal - not correlated with variabilityThe slope removal (above) that we do is based on the assumption that any variation in the output of the inclinometers will vary linearly over the duration of the run. I checked how much slope we would expect given the advertised performance of the inclinometers. The original specs that came with the inclinometers give changes in gain of about .7% in 20 C, which ties in with the value quoted on the company's current web site of 0.05% /C. The zero point change with temperature is not given in the supplied literature, but looking again at the web site we see 1.5 microradians/C typical. (1.5 microradians = .3 arc sec). If we were measuring at the limit of the range we use, at about 500mV or 25 arcsec, the change we would expect with a 5 degree temperature change would be 5x25x.05% + 5*.3 = 0.06(slope) + 1.5 (zero point), or roughly 1.5 arcsec. The correlates well with the sort of numbers we actually see (20mV per arcsec). I looked at the slopes in LX and LY for various datasets, the differences in whose models I had already studied. They slopes (from the log generated by the standard data reduction), for datasets chosen because they were involved in interesting difference plots, were as follows:
I could see no obvious correlation between slope values and variability. LY values - comparison with 981214 resultThe model difference plot for 980214 - 971214 showed very small differences, so I assume that they are both, in some sense, "good" data sets and chose 971214 as a rather arbitrary baseline. I obtained the corrected data for all the datasets listed above.Here is the 971214 LY data: 
Here is a plot showing all of the LY data:
There is an interesting trend for the underlying data (ignoring, for
now, the spikes) to rise in the vicinity of 100 degrees. Here is a plot
of the value at xxx degrees (chosen to avoid wheel/joint clusters).
I then tried to see if I could apply that to the other curves and somehow
flatten them all out (apart from the spikes). The hope was that I would
find an underlying change in shape, and that that the spikes would have
a pattern once that trend had been removed. I factored the 4th order sine
curve to fit the data as best I could, set by set. Not a great success:
Finally, I looked at the 980606, 980610 and 980611 data. Interestingly, the large feature at about az 280 appears in the LY data in only one dataset 980610. This is odd, because we see no evidence of it in the model difference plot 980610-980611. This warrants further investigation, starting with study of the RY data. Tidal movements etcI did a quick bit of research on tidal movements in the ground. I spoke to Mike Lebolski (sp?) at HVO, 967 8843. He does a lot of inclinometry on Mauna Loa, and reports typical ground tides of 0.2 microradians (5 microrad = 1 arcsec), measured in a deep borehole. He sees diurnal variations in slope in a 5m deep hole in cinders, induced by temperature effects on the non-uniform ground, of about 5 microrad (1arcsec). Not clear how applicable this would be to MK.Conclusions
|
