Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.naic.edu/~phil/usrproj/a2127/dataquality.html
Äàòà èçìåíåíèÿ: Mon Apr 24 17:47:05 2006
Äàòà èíäåêñèðîâàíèÿ: Sat Dec 22 11:19:36 2007
Êîäèðîâêà:

A2127: checking the data quality.

feb06

on/off total power repeatability

Intro:

A2127 used the xband receiver to track a source rise to set on 3 separate dates (14feb06,08mar06, and 03apr06). On, off position switching with 5 minute durations and 1 second dumps were done. The wapps were used with 4 100 Mhz bands. Each band had 4096 channels.
There was rain during the 08mar06 observation. The on source position had no continuum in it.

On,off total power versus az.

The total power over each 100 Mhz band was computed and then plotted versus azimuth for each of the 1 second dumps. Since position switching was used, there are two tracks (the on and the off) that cover the same az, (the 300 on samples followed by the 300 off samples). For each on or off there are 4 separate traces that correspond to the 4 frequency bands taken (the 2 polarizations have been averaged). The scan number for each on position scan is also printed. The source rises at az=310 and sets at az=410.

The plots below show the total power versus az for the runs (.ps) (.pdf):

Top 14feb06:.

Black is the on position, red is the off position. The on,offs track pretty well. There is some difference in the 2nd, 5th, and 7th pairs.

Middle 08mar06:

The on,off powers vary a lot. It get particularly bad after transit pairs 6,7,8. In the 6th pair, the off power went from tsys to 2*tsys and back to tsys in the 300 seconds of the off scan. This is when the rain must have started coming down hard.

Bottom 03apr06:

The on off pairs track pretty well . The last 3 pairs show some variation.

Overplotting the on,offs shows that there is considerable total power variation on 08mar06 when it rained.

The rms of the bandpasses, histogram of the data.

Each 5 minute on/off was processed and scaled to kelvins. A baseline was then removed (linear polynomial, 4th order haromonic). The data was then smoothed by 20 channels. The rms across the frequency band was then computed (ignoring 10% of the edges because of filter rolloff).
A histogram of all of the on offs for a day was then made.
The plots show the rms of the on,off bandpasses and a histogram of the data (.ps) (.pdf):

Page 1 14feb06: the top plot is polA, the bottom plot is polB. The rms in Kelvins is plotted versus the onoff pattern number for the day. The colors are the separate frequency bands. The dashed purple line is the expected rms computed as: exprms= Tsys*sqrt(2)/(100e6/4096*1.2*300) The sqrt(2) is from the on/off divide. The 1.2 comes from the filter sinx/x filter widths being wider that a rectangular filter. 30 Kelvins was used for Tsys. The higher values at the beginning and end are because Tsys is increasing at higher za's (i used a constant value for Tsys). Pol B's rms is about 25% higher. This could come from a difference in Tsys or from ripples in the bandpass that were not canceled very well by the on,off processing.
Page 2 top 08mar06: The rms for polB is again higher than polA. The rms for both pols jump at at the 5th pattern when the rain started. It dropped back but stayed higher than the value before 5.
Page 2 bottom 03apr06: The rms's are pretty flat for the entire day. PolB is again higher than polA.
Page 3-5 histograms: Histograms for each days on/off smoothed bandpasses were made. The horizontal axis goes from -10sigma to 10sigma (the expected sigma was used). Black in polA, red is polB. Each page is a separate day. PolB is wider than polA (since the rms was larger). I guess i should fit gaussians to each of these...

The rms's are pretty close to the expected values. The 08mar06 data has trouble after the 5th pattern when the rain started (my bet would be that the pointing was probably off too). None of the other on/off pairs stick out as particularly bad (so there is no reason to not include them in the averaging).

<- page up
home_~phil