justify Vhelio value in fxcor
vega wrote on Feb 14, 2011
Dear all,
I have problem in justify the right value for Vh of variable stars which calculated by fxcor. I listed the trial on measuring the radial velocity of H-alpha. The changing values of parameters: shift to Verr are due to the small changes on osample and rsample region selected. The filterpars are the same. What parameters should I consider to evaluate the right value of Vhelio?
IMAGE SHIFT HGHT FWHM TDR VOBS VREL VHELIO VERR
Ha20100221dcorr.fits -4.282 0.83 304.51 25.87 -19.4915 -19.5653 -29.3111 8.901
Ha20100221dcorr.fits -4.526 0.83 307.29 33.58 -20.6031 -20.6770 -30.4228 6.980
Ha20100221dcorr.fits -5.858 0.83 294.49 33.26 -26.6884 -26.7623 -36.5080 6.752
Ha20100221dcorr.fits -4.301 0.83 314.80 25.87 -19.5779 -19.6518 -29.3975 9.200
Ha20100221dcorr.fits -3.526 0.85 274.85 53.86 -16.0346 -16.1085 -25.8543 3.935
Ha20100221dcorr.fits -2.561 0.87 237.19 174.90 -11.6276 -11.7015 -21.4472 1.059
Ha20100221dcorr.fits -4.552 0.82 328.80 22.89 -20.7223 -20.7962 -30.5419 10.808
Ha20100221dcorr.fits -5.692 0.93 291.97 28.99 -25.9296 -26.0035 -35.7492 7.647
Ha20100221dcorr.fits -3.529 0.95 290.21 60.11 -16.0508 -16.1247 -25.8704 3.730
Ha20100221dcorr.fits -3.526 0.85 274.85 53.86 -16.0346 -16.1085 -25.8543 3.935
Hope for an advice and thank you very much.
Regards,
Vega
I have problem in justify the right value for Vh of variable stars which calculated by fxcor. I listed the trial on measuring the radial velocity of H-alpha. The changing values of parameters: shift to Verr are due to the small changes on osample and rsample region selected. The filterpars are the same. What parameters should I consider to evaluate the right value of Vhelio?
IMAGE SHIFT HGHT FWHM TDR VOBS VREL VHELIO VERR
Ha20100221dcorr.fits -4.282 0.83 304.51 25.87 -19.4915 -19.5653 -29.3111 8.901
Ha20100221dcorr.fits -4.526 0.83 307.29 33.58 -20.6031 -20.6770 -30.4228 6.980
Ha20100221dcorr.fits -5.858 0.83 294.49 33.26 -26.6884 -26.7623 -36.5080 6.752
Ha20100221dcorr.fits -4.301 0.83 314.80 25.87 -19.5779 -19.6518 -29.3975 9.200
Ha20100221dcorr.fits -3.526 0.85 274.85 53.86 -16.0346 -16.1085 -25.8543 3.935
Ha20100221dcorr.fits -2.561 0.87 237.19 174.90 -11.6276 -11.7015 -21.4472 1.059
Ha20100221dcorr.fits -4.552 0.82 328.80 22.89 -20.7223 -20.7962 -30.5419 10.808
Ha20100221dcorr.fits -5.692 0.93 291.97 28.99 -25.9296 -26.0035 -35.7492 7.647
Ha20100221dcorr.fits -3.529 0.95 290.21 60.11 -16.0508 -16.1247 -25.8704 3.730
Ha20100221dcorr.fits -3.526 0.85 274.85 53.86 -16.0346 -16.1085 -25.8543 3.935
Hope for an advice and thank you very much.
Regards,
Vega
Mike Fitzpatrick wrote on Feb 14, 2011
The VHelio value is computed from the shift value in a straightforward way, however there are many thing that could affect the derived shift: the points used to fit the peak, the width of the peak, continuum flatenning, sample regions, weights of the fit points, peak fitting function used, the accuracy of the dispersion solution, and so on.
If you're primary interest is in the shift I would recommend the 'center1d' function rather than a Gaussian fit to the peak. Your table of results may be reasonable for the data, without seeing the images I can't really comment further. I'd suggest you try this again with center1d and see if you get more stable results.
If you're primary interest is in the shift I would recommend the 'center1d' function rather than a Gaussian fit to the peak. Your table of results may be reasonable for the data, without seeing the images I can't really comment further. I'd suggest you try this again with center1d and see if you get more stable results.
vega wrote on Feb 14, 2011
Dear Fitz,
Sorry for the delayed reply. Anyway thank you for the suggestion.
Yes I interested in the line radial velocity measurement. I have changed the function from gaussian to center1D but the problem is the error, Verr cannot be calculated. I read in the help manual that we need to set the background value. I try give some value for the background like 0.0 or 0.1 etc. but the Verr still INDEF. How should I change the value of background to get the error value?
Thank you for the helps.
Regards,
Vega
Sorry for the delayed reply. Anyway thank you for the suggestion.
Yes I interested in the line radial velocity measurement. I have changed the function from gaussian to center1D but the problem is the error, Verr cannot be calculated. I read in the help manual that we need to set the background value. I try give some value for the background like 0.0 or 0.1 etc. but the Verr still INDEF. How should I change the value of background to get the error value?
Thank you for the helps.
Regards,
Vega
Mike Fitzpatrick wrote on Feb 14, 2011
The center1d function doesn't produce an error estimate (see the 'help center1d' page), the Verr value is just the error of the center parameter times the velocity dispersion per pixl and so can't be defined. In this case you would need to estimate the error from repeated observations if possible.
None of the values produced by the task are absolute, i.e. experiment with different fitting functions and then for each tweak the points used in the fit, the background, weighting etc and you'll see that there is a lot of variation. It is up to you to be sure you understand how the fitting is being done and what that does to the data (e.g. do you simply fit everything above some CCF value or are you careful to set the same number of points left-and-right of the expected peak?).
None of the values produced by the task are absolute, i.e. experiment with different fitting functions and then for each tweak the points used in the fit, the background, weighting etc and you'll see that there is a lot of variation. It is up to you to be sure you understand how the fitting is being done and what that does to the data (e.g. do you simply fit everything above some CCF value or are you careful to set the same number of points left-and-right of the expected peak?).
Last post on Feb 14, 2011