Reduction of on-the-fly maps using livedata, gridzilla and miriad ----------------------------------------------------------------- First step is to use Livedata. This will read in the raw rpfits files, make a quotient spectrum, using an appropriate reference spectrum, subtract a baseline and write the data out in SDFITS format 1. To start livedata, type "livedata". this will bring up the GUI. 2. From the top of the GUI, select "Mopra", then select input and output directories appropriately. raw rpfits files will then appear in the "Raw files" window. 3. Select whichever raw files are appropriate for your map 4. Click on "Queue selection" 5. Below this area, you need to select both "Bandpass calibration" and "Write data" to be on. You can also select "Monitor output" if you choose, but this tends to slow the system down a bit, so I usually avoid it. 6. In the bottom left panel, you should make sure that only beam 1 is selected in "Input data selection" since there are no multibeams on Mopra (yet). 7. Below this, select which IFs you want to process. Livedata will only allow you to process two IFs at a time. 8. Below this, select the channel range which you want to be saved in the output sdfits file. If you are using broadband mode, you will have 8192 channels per IF. If you are using narrowband (zoom) mode, you will have 4096 channels per IF. 9. In the bottom right hand corner leave "spectral smoothing" and "Spectral pre-scaling" both as "none". 10. All the default options for the section "bandpass calibration options" can be left as is. 11. For "Post-bandpass calibration options", you should select a baseline to fit. I find a baseline order of 1 works OK most of the time. 12. Select "preserve continuum" if you want to have a look at continuum. However, in my experience, the baseline ripple that affects Mopra data tends to make a mess of any continuum images and so there is only limited use of this function. 13. Select which cannels you want to mask out. Usually the end channels are worth masking out, and usually 150 channels seems to get the worst of the noisy channels. So (for example) in narrowband mode I would mask 1-150 and 3950-4097. 14. Choose doppler frame as appropriate: LSRK is the usual one for Galactic work. 15. I have never bothered with the "Validity checking" section. 16. Once you have all the bits filled out, you can click on "Start" at the top to process the data. --------------------------- Gridzilla processing -------------------- Gridzilla is used to regrid data onto an appropriate pixel scale, and to select an appropriate velocity/frequency range of channels to output. 1. start gridzilla by typing "gridzilla" 2. On the top-right corner you should change the "parameter set" button from "General" to "Mopra". This will set a few useful defaults for processing Mopra data, like selecting only beam 1. 3. On the top left, you will see the default button "Velocity Range". If you want your data cubes to be written out with a velocity axis, then leave as is. If you prefer a Frequency axis, eg. if you are processing broadband data with multiple spectral lines, then select "Frequency". 4. Assuming you've chosen velocity: enter an appropriate velocity range over which to make your data cube. 5. Enter your line rest frequency below this. 6. Select the right IFs that you already processed using livedata. 7. If you're not interested in polarisation, the default setting of "Polarization A&B" will add the two polarisations together, so this can be left as is. 8. If you want a spectrum of your line(s), then make sure "Spectral" is selected. If you're interested in continuum (see caveats above), then you can also select this. Since you already baselined data using livedata, you should not need to select "baseline" here. 9. On the right hand side of gridzilla, at the top, you should select whichever reference frame you want your data in. Usual choices are "Equatorial" or "Galactic" 10. Make sure "autosize" is selected below here. The x-scale and y-scale specify the size of the pixels that will turn up in the output file. 11. Below this, you can select bad data to automatically flag out, based either on the Tsys or the range of values for the pixels. I find the Tsys range most useful. At 12mm, you typically wouldn't want to save data that is over 100K, so I use this as an upper cutoff. 12. For gridding parameters, there are a few options here: On the left hand side, you choose the statistic by which overlapping data is combined. The default of "Mean" is a pretty good start, but there are other options that are fun to play around with. It is usually a good idea to weight the data combination by Tsys, so select this. 13. Select a "Beam weighting" of 1 for reasons that are not clear to me. Also enter an appropriate beam size, which is about 2 arcmin at 12mm and about 0.5 arcmin at 3mm. 14. You can play around with the "smoothing kernel a bit here". With no smoothing, you will get the highest resolution data, but the pixel-to-pixel variation will be quite high. using a gaussian smoothing of something like the beam size will degrade your resolution slightly, but will greatly improve the quality of your map. 15. You can leave "blanking level" and "processor host" alone for standard cases. The part "use 50% of xxxMiB" can affect your resultant data cube if it is very big. Basically, if you do not have enough RAM for your cube, gridzilla will automatically break up your cube into multiple files. This is a bit of a pain, so I usually increase the default 50%. But note that a high percentage will hog resources on your computer and may slow stuff down if you are doing other intensive tasks on your computer. 16. For "FITS spectral type", Galactic astronomers will normally choose "VELO-XXX", or "FREQ", if you want your cube in velocity or frequency space, respectively. 17. Select an appropriate output FITS directory and FITS file name 18. Select the appropriate files you want to grid together from the left hand side, then click on "GO" at the bottom to make the cube. ------------------------------------------------ Further reduction in MIRIAD --------------------------- Here are a few other scripts I usually use in miriad which help make something easily viewable with a data cube. First you need to read the data in to miriad format. So type the following on a command line: fits in=gridzilla.fits op=xyin out=IF1 You may not need the entire velocity range in the data cube, so cut it down a bit, with and example like: imsub in=IF1 out=CH3OH-27472 region='images(396,1759)' Make 0th and -2th moment maps, which are useful for a quick look at the data: moment in=CH3OH-27472 out=CH3OH-27472.mom0 mom=0 moment in=CH3OH-27472 out=CH3OH-27472.mom-2 mom=-2 You can then use "kvis" to view the results.