The following should be done separately for each different configuration. You can copy the one line commands into a unix terminal (you do not have to be in the miriad environment). By the way, "help cheat" can be very... helpful. ================================================================================ 1) FITS Read the data into Miriad. The *.FITS file will be in the ../aips/ directory. Just cd into there, run this task, and mv the *.UV folder to the ../miriad/ directory and cd back to the miriad directory. %%%%% cd ../aips/ %%%%% task fits in = *.FITS op = uvin out = *.UV velocity = optbary %%%%% fits in=*.FITS out=*.UV op=uvin velocity=optbary %%%%% mv *.UV ../miriad/ %%%%% cd ../miriad %%%%% ================================================================================ 2) PUTHD Enter the rest frequency task puthd in = *.UV/restfreq value = 1.420405752 We have set this value as $hifreq and you can do value = $hifreq instead. %%%%% puthd in=*.UV/restfreq value=1.420405752 %%%%% ================================================================================ 3) BLFLAG Now do some flagging blflag is a good task to start flagging. Try: task blflag vis = *.UV options = nobase device = /XWIN %%%%% blflag vis=*.UV options=nobase device=/XWIN %%%%% see if there are any high points. If you flag any high points, replot and check to see if the new scaling reveals any new flagging opportunities. ---------- if you plot real vs imaginary, they should be symmetric uvplt will allow you to plot more per page: vis = *.UV device = /XWIN (or whatever else you want) options = nobase axis = real, imaginary %%%%% blflag vis=*.UV options=nobase device=/XWIN axis=real,imaginary %%%%% ================================================================================ 4) UVPLT Plot the UV coverage vis = *.UV axis = uc,vc device = *-UVCOVER.PS/ps options = nobase %%%%% uvplt vis=*.UV axis=uc,vc device=*-UVCOVER.GIF/gif options=nobase %%%%% ================================================================================ When you think your data are nice and clean, you can image them. 5) INVERT Fourier transform the UV data to make a data cube. task invert vis = *.UV map = *.MAP beam = *.BEAM imsize = 128 (512 for B, 256 for C, 128 for D) cell = 8 (2 for B, 4 for C, 8 for D) robust = 2 (see all the faint stuff - we don't care about resolution yet) options = double, systemp %%%%% invert vis=*.UV map=*.MAP beam=*.BEAM imsize=128 cell=8 robust=2 options=double,systemp %%%%% ================================================================================ 6) CLEAN Smooths your dirty map. task clean map = *.MAP beam = *.BEAM out = *.CLEAN niters = 1000 (possibly? 10000?) %%%%% clean map=*.MAP beam=*.BEAM out=*.CLEAN niters=1000 %%%%% ================================================================================ 7) RESTOR Restore your clean components to make a map. task restor model = *.CLEAN beam = *.BEAM map = *.MAP out = *.RESTOR %%%%% restor model=*.CLEAN beam=*.BEAM map=*.MAP out=*.RESTOR %%%%% ================================================================================ Now you should inspect your clean map for problems. Examples of these are: - stripes - these are usually caused by some unhappy UV points here - bad channels - check the specific channel for RFI or weird points Note that the above flagging tools often seem to average your spectral data. Some channels have bad data that get averaged with good data, and it's hard to see these in the visibility plotting programs easily. If you can isolate it to a few channels it's easier. Tasks to flag with: UVFLAG and BLFLAG (plot with UVFLAG) Examples: %%%%% uvflag vis=*.UV select=amp'(5,10000)' flagval=flag %%%%% ##flags all points above amp = 5 %%%%% uvflag vis=*.UV select=ha'(-0.4167,1.00)' line=ch,1,34,1 flagval=flag %%%%% ##flags all points on channel 34 with HA between -0.4167 and 1.00 Another trick to look for problems is to use FFT. YAY FFT. It's a lot of fun, and you can even watch movies of your FFT'd cube to look for problems. If you flag anything, delete the old *.BEAM, *.MAP, *.RESTOR, and *.CLEAN folders and repeat steps 5-7. This may have to be repeated several times. %%%%% rm -r *.BEAM *.CLEAN *.MAP *.RESTOR %%%%% Once you've tracked down all your problems and have a nice image (with continuum), you can move on to continuum subtraction. Be sure to copy all header information into the reduction script for the final *.RESTOR image. It is also advised to copy the final theoretical rms noise into the script given in the INVERT task. ================================================================================ You can use the restor map (with continuum) to search for line free channels, using kvis. You might need to adjust the histogram so you can actually see the galaxy appearing. %%%%% karma_start %%%%% ##loads the kvis environment Open kvis, load in your map, and watch the movie. Record which channels are line free. Leave some buffer between where you think the galaxy is and the channels you pick. Pick some at the beginning (planes z1 to z2) and some at the end (planes z3 to z4) ================================================================================ 8) UVLIN Make a continuum subtracted UV file. Make a line and continuum file. LINE: Task uvlin vis = *.UV chans = 1,33,65,98 out = *.UVLIN mode = line %%%%% uvlin vis=*.UV chans=1,33,65,98 out=*.UVLIN mode=line %%%%% ---------- CONTINUUM Task uvlin vis = *.UV chans = 1,33,65,98 out = *.UVCONT mode = cont %%%%% uvlin vis=*.UV chans=1,33,65,98 out=*.UVCONT mode=cont %%%%% ================================================================================ Now use your restored map with continuum to determine the noise, and hence the level that you can safely CLEAN to. Open the restored map in kvis, and measure the noise in a continuum- AND line-free section of the map. We will always need this number, so make a note of it somewhere. Determine the noise level in the .RESTOR (line free) image. Do this by making a box and typing 's'. Check the rms in a few channels. Now that you know the noise, set the cutoff in clean to 4 times the rms value. ================================================================================ Now you can reimage the data. 9) INVERT Fourier transform the UVLIN and UVCONT data to make a data cube. UVLIN: task invert vis = *.UVLIN map = *.LINMAP beam = *.LINBEAM imsize = 256 (1024 for B, 512 for C, 256 for D)? cell = 8 (2 for B, 4 for C, 8 for D) robust = 2 (see all the faint stuff - we don't care about resolution yet) options = double, systemp %%%%% invert vis=*.UVLIN map=*.LINMAP beam=*.LINBEAM imsize=256 cell=8 robust=2 options=double,systemp %%%%% Theoretical rms noise: # ---------- UVCONT: task invert vis = *.UVCONT map = *.CONTMAP beam = *.CONTBEAM imsize = 256 (1024 for B, 512 for C, 256 for D)? cell = 8 (2 for B, 4 for C, 8 for D) robust = 2 (see all the faint stuff - we don't care about resolution yet) options = double, systemp, mfs %%%%% invert vis=*.UVCONT map=*.CONTMAP beam=*.CONTBEAM imsize=256 cell=8 robust=2 options=double,systemp,mfs %%%%% Theoretical rms noise: # ================================================================================ 10) CLEAN Smooths your dirty map. LINE: task clean map = *.LINMAP beam = *.LINBEAM out = *.LINCLEAN niters = 1000000 cutoff = 4*rms noise %%%%% clean map=*.LINMAP beam=*.LINBEAM out=*.LINCLEAN niters=1000000 cutoff=4*rms noise %%%%% ---------- CONT: task clean map = *.CONTMAP beam = *.CONTBEAM out = *.CONTCLEAN niters = 1000000 cutoff = 4*rms noise %%%%% clean map=*.CONTMAP beam=*.CONTBEAM out=*.CONTCLEAN niters=1000000 cutoff=4*rms noise %%%%% ================================================================================ 22) RESTOR Restore your clean components to make a map. LINE: task restor model = *.LINCLEAN beam = *.LINBEAM map = *.LINMAP out = *.LINRESTOR %%%%% restor model=*.LINCLEAN beam=*.LINBEAM map=*.LINMAP out=*.LINRESTOR %%%%% ---------- CONT: task restor model = *.CONTCLEAN beam = *.CONTBEAM map = *.CONTMAP out = *.CONTRESTOR %%%%% restor model=*.CONTCLEAN beam=*.CONTBEAM map=*.CONTMAP out=*.CONTRESTOR %%%%% ================================================================================ This should be your nice pretty image!