Reducing photometric polarimetry data

Reduction of photometric polarimetry is not yet available in the ORACDR pipeline, and requires considerably more interaction than jiggle-map polarimetry data. Firstly, you must log onto a SUN machine - at the JAC, the recommended machine is called "ulu". You will need access to the file called sit.e - if it is not locally installed you can download it here (right click, save link as ...). To start, run up SURF and set up useful variables based on UT date:

• surf
• scusetenv yyyymmdd

Do the basic reduction (take out chop, flatfield, correct for extinction), e.g.:

• scuquick analysis=samp -sub long -remsky -tau 0.2 #

where # is the observation number. You do need to supply an approximate opacity for the extinction/airmass correction to work properly. The -sub long option reduces only the LW array. The analysis=samp parameter is important for statistics (gives more samples).

The next step is to make an ascii file with the 16 signals and their errors, plus some header information. Type e.g.:

• polphot.pl -s lon #

(defaults to 8s integrations and 850 micron data; for different integrations put the number of on+off seconds at the end of the line, and use e.g. -s p13 for 1350 micron data)

You are now ready to run the reduction package by typing:

• ./sit.e

(the program name stands for step-and-integrate true, which is the observing mode).

First, select option 4 from the main menu, and set the directory name to the area where your reduction will be done. Note that there must be a final backslash (i.e. /home/bob/ not /home/bob).

• select option 1 and enter a filename for the polarization results (any format, but something descriptive like orion_850 is useful ; note that long filenames that extend beyond the line end may cause errors).
• select option 2 and enter the observation number (#; the file read is the output from polphot.pl, called obs#.dat ).
• check the analysis method is set to fit (not subtract).
• decide on the analysis options - fit the whole waveplate cycle, or two half-cycles? -– despike the data? (half cycle, no despike is generally good, but the S/N may be improved by trying the other options).
• select whether or not to plot the fits (yes is recommended).

Now all the options are set, press <return> to see the fits. You can repeat the above steps for several observations before proceeding -– generally you will only need to change the observation number.

Next select option 4 to return to the main menu (you may need to hit return twice). It is very easy here to enter a return before the 4, in which case you'll get another identical fit in the output file! These should be edited out before proceeding (use option 5 in the main menu to exit from sit.e).

If it crashes, you may not have a defaults.h file, which is needed to store the parameter settings. The file can be obtained from here. It is also possible to corrupt the defaults.h file, e.g. by entering a number somewhere a character is expected. In this case, do not panic! – keep re-entering the option numbers and it will sort itself out.

• select option 1 and enter the name of the file containing the fits (e.g. orion_850)
• select option 2 and set the object type - source or instrumental polarization.
• check option 3 is set to angle_track FALSE (the current observing mode with SCUBA).
• set option 4 to standard average normally, unless a weighted fit is really appropriate (several fits with similar errors, which can be checked from the fitted parameters file).
• set K-S test to yes and tracking file to no, also correct for rotations should normally be yes
• for IP data, correct for IP in option 8 defaults to no; for source data, set it to yes and enter the IP %, error on IP% and zero-elevation IP-position angle in option 81 (these have standard values; consult commissioning reports or your support scientist).

Now press <return> to run - lists of intensities (I) and the magnitudes of the two orthogonal polarization components (Q,U) will come up. The polarization values are fractions (multiply by 100 for percentages). For a quick answer, hit <return> again and the p,theta from the co-add of the fits will be displayed.

Note that the second p is the one you want - it is corrected for a bias effect that can be significant in low S/N data. (It results from squaring and adding Q,U which forces a positive result.) Also, there is a bug in the program - if the polarization S/N in the first line is less than 1, the true p should be 0!

The S/N can often be improved by removing small numbers of very anomalous fits (often caused when a sky variation dominates over the source polarization signal). As noted above, the I,Q,U parameters are listed before the averaging is done - they are all in increasing order, so entries on one line do not usually correspond to the same observation. To remove values, enter Q or U and then the corresponding line number (all parameters of that fit will be removed); repeat as needed and then hit <return> to get the final p,theta . Anomalous fits can be found by examining the lists of Q,U for unusually large values or large gaps between values. For example, 0.08 would be anomalous if the final p is 0.01 (1%), and 0.07 would be anomalous in a list of 0.01, 0.015, 0.02, 0.025, 0.07 (a gap of several %).