; UVLSF ;--------------------------------------------------------------- ;! least squares fit to channels and subtracts from uv data. ;# TASK UV SPECTRAL EDITING ;----------------------------------------------------------------------- ;; Copyright (C) 1995-1997, 1999-2001 ;; Associated Universities, Inc. Washington DC, USA. ;; ;; This program is free software; you can redistribute it and/or ;; modify it under the terms of the GNU General Public License as ;; published by the Free Software Foundation; either version 2 of ;; the License, or (at your option) any later version. ;; ;; This program is distributed in the hope that it will be useful, ;; but WITHOUT ANY WARRANTY; without even the implied warranty of ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the ;; GNU General Public License for more details. ;; ;; You should have received a copy of the GNU General Public ;; License along with this program; if not, write to the Free ;; Software Foundation, Inc., 675 Massachusetts Ave, Cambridge, ;; MA 02139, USA. ;; ;; Correspondence concerning AIPS should be addressed as follows: ;; Internet email: aipsmail@nrao.edu. ;; Postal address: AIPS Project Office ;; National Radio Astronomy Observatory ;; 520 Edgemont Road ;; Charlottesville, VA 22903-2475 USA ;----------------------------------------------------------------------- UVLSF LLLLLLLLLLLLUUUUUUUUUUUU CCCCCCCCCCCCCCCCCCCCCCCCCCCCC UVLSF Least squares fit baseline and subtracts from uv data. INNAME Input UV file name (name) INCLASS Input UV file name (class) INSEQ 0.0 9999.0 Input UV file name (seq. #) INDISK 0.0 9.0 Input UV file disk unit # OUTNAME Output UV file name (name) OUTCLASS Output UV file name (class) OUTSEQ -1.0 9999.0 Output UV file name (seq. #) OUTDISK 0.0 9.0 Output UV file disk unit #. BCHAN Lowest channel to write ECHAN Highest channel to write ICHANSEL Select channels to fit: NOTE this is start,end,increment and IF for each region ORDER 0.0 1.0 Order of fit line (0 -> DC) DOOUTPUT -1.0 1.0 > 0 => write fit baseline as a continuum uv data base CHANNEL 0.0 9999.0 Channel of fit to be written as continuum: 0 -> reference FLUX 0.0 Flag data if residual flux in any fit channel is > FLUX 0 -> 10**20 CUTOFF 0.0 Flag data if RMS of residual in fit channels > CUTOFF 0 -> 10**20 SHIFT Shift in asec to RA, DEC ---------------------------------------------------------------- UVLSF Task: This task does a spectral baseline subtraction. It fits a straight line to the real and imaginary parts of selected channels and subtracts the fitted baseline from the spectrum. Optionally, it can flag data having excess residuals in the channels used for the baseline fitting. This is a good method for detecting and deleting data damaged by RFI. See task FLGIT for a more powerful RFI excision task and UVMLN for a task dealing with multi-source data with FG tables. Adverbs: INNAME.....Input UV file name (name). Standard defaults. INCLASS....Input UV file name (class). Standard defaults. INSEQ......Input UV file name (seq. #). 0 => highest. INDISK.....Disk drive # of input UV file. 0 => any. OUTNAME....Output UV file name (name). Standard defaults. OUTCLASS...Output UV file name (class). Standard defaults. OUTSEQ.....Output UV file name (seq. #). 0 => highest unique OUTDISK....Disk drive # of output UV file. 0 => highest with space for the file. BCHAN......Lowest channel number in the input file to write to the output file. 0=> 1. ECHAN......Highest channel number in the input file to write to the output file. 0=> highest in input data. ICHANSEL...Select up to 20 groups of channels/IF(s) to fit as sets of (Start,end,inc,IF), i.e., ICHANSEL = 6,37,1,0, 92,123,1,0 for two regions applyingto all IFs. The first group for which ICHANSEL(2,i) <= 0 ends the list. Defaults: Any IF having no group assigned to it, gets a group including all channels. ICHANSEL(1,j) defaults to 1, 0 < ICHANSEL(2,j) < ICHANSEL(1,j) defaults to Nchan. ICHANSEL(3,j) < 1 or > ICHANSEL(2,j)-ICHANSEL(1,j)+1 defaults to 1. ICHANSEL(4) <= 0 => this group applies to all IFs. ORDER......Normally the fit should be done with a first order polynomial, namely a DC term and a slope. If only one box is used, the slope may not be adequately defined and a simple DC term for the real and for the imaginary parts would be more reliable. DOOUTPUT...<= 0 => write out only the subtracted uv data. > 0 => write also a data set with the fit baseline as visibilities. CHANNEL....When DOOUTPUT > 0, write the continuum data with the fit evaluated at spectral channel CHANNEL. 0 => reference channel. FLUX.......If the residual flux in any channel used to fit the baseline exceeds FLUX, then the spectrum for that time, IF, and polarization is fully flagged. <= 0 => 1.0E20 CUTOFF.....If the RMS flux in the channels used to fit the baseline exceeds CUTOFF, then the spectrum for that time, IF, and polarization is fully flagged. <= 0 => 1.0E20 SHIFT......Shift in arcseconds during fitting process. The data are shifted, fitted, flagged and then shifted back. These are shifts in RA, DEC (i.e. NOT on the tangent plane) and are the same as in UVFIX. ---------------------------------------------------------------- UVLSF: Task which subtracts continuum from channels in UV-plane DOCUMENTOR: H.J. van Langevelde (Sterrewacht Leiden) RELATED TASKS: UVLIN, UVBAS PURPOSE UVLSF will estimate the continuum visibilities and subtract these from a specified range of channels, hopefully leaving only the information about spectral features in the output UV-file. It fits, by least squares, a straight line to the to the real and imaginary parts of the selected channels. This baseline is subtracted from the spectrum. This can only work properly if the UV coverage is the same for all spectral line channels. It can, however, deal with frequency dependent flagging. PHASE SHIFTING Since UVLSF works perfectly for a single point source at the phase center, one should shift any dominant point source to the phase center prior to the fitting process. Note that UVFIX now does this correctly since it does the relative phase shift between channels. The shift option is offered here for convenience and since THE SHIFT IS UNDONE AFTER THE FIT. The sense of the shift is the same as in UVFIX. The code for computing the shift is: RA = RA + SHIFT(1) / 3600. DEC = DEC + SHIFT(2) / 3600. DTORAD = 3.14159265358979326D0 / 180.0D0 DXC = SIN (DTORAD * (SHIFT(1)/3600.)) * * COS (DEC * DTORAD) DYC = COS ((DEC - SHIFT(2)/3600.) * DTORAD) * * SIN (DEC * DTORAD) - SIN ((DEC - SHIFT(2)/3600.) * * DTORAD) * COS (DEC * DTORAD) * COS ((SHIFT(1)/3600.) * * DTORAD) DZC = SQRT(1.0D0-DBLE(DXC)**2-DBLE(DYC)**2)-1.0D0 For small shifts, the necessary values for SHIFT can be measured from an image and adjusted by dividing the RA value by cos(declination). COMMENTS This tasks has proved to be powerful in problems where bandwith is small and extended continuum emission has to be removed. Its accuracy is limited by the fact that the visibilities in the uv plane change over the bandwith. The errors made in the approximation of the continuum visibility --and thus in the subtraction-- can be estimated as: d V d u d V L D V = --- * --- * D v = --- * --- * D v L d u d v d u c Where D V is the error, D v the bandwith over which we try to do this and u a coordinate in UV-plane. L is the baseline length specified in the same units as those used for c, the speed of light. The formula tells us that UVLSF should do an accurate job on short baselines and/or small bandwith, provided there is a reasonably smooth signature in the UV-plane. That means that it will generally not work for a field dominated by discrete (point) sources. It will work better if the dominant point source is shifted to the filed center. Use UVFIX which now does a correct shift for spectral data. In comparison with the method of cleaning the background and UVSUB the components from the spectral line channels, this method has the main advantages that it is 1) much faster, 2) will work when the background is difficult to model with clean components. When point- sources are the main source of continuum emission the UVSUB method is perfectly suited. In some cases a hybrid method may be advantageous. In comparison with averaging maps to estimate the continuum this method is again faster and more reliable, since there will be no sidelobs of the continuum in the map. The output data set can also be a powerful diagnostic tool. In principle, the data can be used to apply selfcal on your spectral line data. UVLSF differs from UVLIN in a number of ways. UVLIN does its flagging only on the peak residual in the channels selected for fitting and scales the peak by the square root of the weight before deciding on flagging. This is correct for peak residuals due solely to thermal noise (and reflected correctly in the data weights), but is an added confusion to the user. In any case, you should set the FLUX cutoff at a conservative level (6 - 8 times the expected sigma in the integration time typical for the current data set). The CUTOFF test is better for spectra that do not fit the linear model or are just excessively noisy for some reason. UVLSF offers the option of writing the fit continuum as a UV data set which can be used for self-cal, TVFLG editing, etc. UVBAS fits the linear baseline in amplitude and phase rather than real and imaginary. If the emission is well off the center of the field, the amplitude and phase are more likely to be nearly linear than the real and imaginary parts. However, the amplitude does not have Gaussian noise statistics and UVBAS is therefore suspect in low signal-to-noise ratio cases. REFERENCES See Cornwell, Uson and Haddad (Astron. Astrophys. 258, 583; 1992) for a detailed discussion which includes estimates of errors.