Measuring Radial Velocity with Alpy and ISIS

[Robin Leadbeater]

The stability of my Alpy600 is so good (<1um movement in the spectrum due to pointing or temperature) that I though I would test the precision of RV measurements made using Alpy and using the ISIS cross correlation function.

I measured two G0v Miles stars (HD157214 and HD160693) with significantly different RV (The measurements were made on two different nights) Checks on the calibration lamp spectra confirmed no detectable (ie <1/10 pixel) calibration shift between the two observations. The spectra were normalised to remove the continuum (This may not be necessary but it may help the cross correlation function by removing any large long wavelength features)

Click on the graphic to see the spectra and the ISIS screen shot

RV_measurement_Alpy_ISIS_THO.png (53.81 KiB) Viewed 15219 times

The results were excellent with the measured RV difference of 107 km/s only 12km/s in error from from the published value (Note there appears to be a sign error in ISIS)

Despite the low resolution of Alpy, its high stability and wide wavelength coverage make it ideal for using the cross correlation technique to measure wavelength shifts to a remarkable level of precision of the order of 1/40th of the spectrograph resolution.

Cheers
Robin

[Valerie Desnoux]

Robin,

Very impressive.
Do you crop the spectrum to certain wavelength region before running the CCF ?
Wish you clearer sky than here...

Valerie

[Martin Dubs]

Hi Robin,

this is a remarkable result. You might add that it is achieved with very careful recording, keeping everything as constant as possible. I am looking forward for a confirmation by more spectra, two do not make good statistics. But I do not want to downplay your result, it is very encouraging.

Congratulations, Martin

[Robin Leadbeater]

Thanks Valerie,

Do you crop the spectrum to certain wavelength region before running the CCF

For this test I used the full spectrum shown, cropping the strong telluric lines at the red end and stopping before the steep drop towards the Balmer jump at the blue end. On reflection it might have been better to stop at the blue side of the Na D lines to avoid all telluric contamination and potential interference from IS Na lines.

Cheers
Robin

[Robin Leadbeater]

Hi Martin,

You might add that it is achieved with very careful recording, keeping everything as constant as possible.

This is of course good practise in measurements of this sort and I was prepared for the process of beating down the various potential sources of error but in this case it was very straightforward with no special attention needed, though nothing was disturbed between the two observing sessions. The key I believe is the exceptionally high instrument stability. There was <1um shift in the lamp lines between observing sessions (confirming the results from earlier stability tests) which meant I could use the same calibration equation for both stars. (Main sequence G stars also make good targets for this sort of work)

I am looking forward for a confirmation by more spectra, two do not make good statistics.

Agreed. I had hoped to get individual RV for the two stars by comparing with the Miles spectra as references (which appear to be RV corrected) but the Miles wavelength calibrations for these two stars seem to have some issues at this sort of level with what look like small non linearity differences which need a closer look.

I had also planned to do more stars but was thwarted by the weather and with little better weather forecast, I decided it was worth publishing something preliminary to perhaps encourage others to try. (The results are consistent with the measured instrument stability though.) This is the result of just two short gaps in the clouds over two nights and the SNR of even these two stars is not that good. The CC technique though is very forgiving. For example I worked on some results from the CfA speedometer a while back and could not believe the way it generated good RV data even with very low SNR



Cheers
Robin

[Robin Leadbeater]

A repeat measurement on the same two stars on 2013-06-24 gave an error of +33km/s relative to the published figure compared with -12km/s for the original measurement. The mean of the two measurements is 129km/s compared with the published figure of 119km/s. Checks on the calibration lamp spectra again confirmed the intrinsic spectrograph stability with a shift equivalent to <2km/s between the two spectra.

Robin