Radial Velocity - using echelle

Ask your questions, show your results
Andrew Smith
Posts: 321
Joined: Fri Sep 30, 2011 12:23 pm

Radial Velocity - using echelle

Post by Andrew Smith »

I would be pleased to have any advice on how to do accurate relative or absolute radial velocity measurements using an echelle. Alternatively if you could point me to any good write-ups on the procedures or issues involved. I am after practical aspects rather than theory.

As background I am looking at monitoring the Fe I 5572 line of Rho Cas. This line is in order 40 and does not have any strong atmospheric lines. I have looked at the air lines in order 33 and find I seem to get some systematic shifts compared to the Ar/Ne starter lamp over a few days. I have yet to track the cause down.

I have thermally insulated the echelle and built it on a low expansion ceramic base. I am thus mainly concerned about stability of measurements between sessions (days/weeks/years) rather than within a session but any ideas in this area are still welcome.

Thanks in advance for any advice or pointers.

Regards Andrew
Robin Leadbeater
Posts: 1926
Joined: Mon Sep 26, 2011 4:41 pm
Contact:

Re: Radial Velocity - using echelle

Post by Robin Leadbeater »

Hi Andrew,

How big are the errors? Is this thread any use ?
http://www.spectro-aras.com/forum/viewt ... =882#p4005

Robin
LHIRES III #29 ATIK314 ALPY 600/200 ATIK428 Star Analyser 100/200 C11 EQ6
http://www.threehillsobservatory.co.uk
Andrew Smith
Posts: 321
Joined: Fri Sep 30, 2011 12:23 pm

Re: Radial Velocity - using echelle

Post by Andrew Smith »

Thanks Robin, I had not seen that topic or if I had not appreciated its significance. I posted in the "Beginners" section as I have never attempted RV measurements before and had only just started to give it some serious thought. I have now found some papers/discussions which have helped me form some ideas on the issues.

With my echelle at order 40 I have (according to ISIS) R = 8900 , scale fit 0.0089 A RMS (0.479 K/s at 5772A) and 0.157A per pixel (8.45 km/s). So assuming I could get 1/10 pixel accuracy I would be looking at ~ 1km/s or better. Do you feel this is realistic?

I have reached three conclusions which I would value yours and others thoughts on.

1) Data collection requires the bracketing of target data with calibration data good signal to noise 100:1. To date I have just done it at the beginning or end.

2) Extraction of the line position is key with various methods available. (For relative data cross correlation seems to be preferred but what is best is less clear for absolute line position.)

3) Accuracy and stability of the wavelength calibration is also key. Personally I feel it is good to try to achieve absolute as well as relative accuracy.

I would welcome comment and advice on best practice on these and any other areas people feel important.

As an example of what I have found so far is that if I take 6x10s Ar starter lamp images and process them as the target using a wavelength scale based on 16 Ar I get variations in the position of the Ar 5572.541 spanning the range 5572.542 to 5572.744A i.e up to 1.9Km/s offset as I process 1,2,4,6 images at a time. Comparing Ar lines several hours apart gave shift of ~ 4km/s.

As luck would have it the Fe I line I wish to monitor also at 5572A so if I can get the before and after Ar calibration stable it should provide an accurate reference. However, it also means trying to inject an Ar signal into the telescope during the target capture will contaminate the Fe I line.

I will try moving the fiber and diffusing the Ar lamp more. I am, however, a little concerned that my Burmese cats will fine the moving fiber too much of a temptation when they drop in (literally leaping from the observatory walls) during an observing session!

Regards Andrew

PS I did spot the need for Heliocentric velocity correction!
Robin Leadbeater
Posts: 1926
Joined: Mon Sep 26, 2011 4:41 pm
Contact:

Re: Radial Velocity - using echelle

Post by Robin Leadbeater »

Hi Andrew,

High precision (repeatability) is a matter of getting everything stable, checking calibration often and having repeatable procedures. (For example the particular technique used for measuring the line centre eg gaussian fit or centroid etc is probably not so important as long as you do it the same way each time). Having a high SNR and measuring many lines also improves precision. (This is where cross correlation is so successful - you are using all the data in the spectrum, not just the line positions)

I think getting absolute accuracy is more tricky as you need good agreement between your calibration source lines and star lines, the precise relative position and shape of which can depend on many subtle optical effects. (Tellurics have an advantage here but at highest accuracy even these can vary slightly in wavelength due to atmospheric effects). I think offsets are probably inevitable but provided you have good repeatability these could be measured and compensated for by measuring RV standards (which I suppose is sort of passing the buck but we can take advantage of other people's hard won accuracy :D ). This is what I did for the Convento Deneb RV measurement campaign last year using Vega as a standard so we will see how well it worked when everyone's data is pooled :D

Cheers
Robin
LHIRES III #29 ATIK314 ALPY 600/200 ATIK428 Star Analyser 100/200 C11 EQ6
http://www.threehillsobservatory.co.uk
Robin Leadbeater
Posts: 1926
Joined: Mon Sep 26, 2011 4:41 pm
Contact:

Re: Radial Velocity - using echelle

Post by Robin Leadbeater »

Here is my long term stability for Vega. Mean 13.8, SDev 0.52 km/s (Christian's eShel results on exoplanets were an order of magnitude better than this :o )

LHIRES 2400, measured with ISIS using mean of 2 Si II lines at 6347.100 , 6371.360.
Calibrated with superimposed Ne/Ar lines from a Filly lamp mounted on the edge of the telescope aperture in line with the slit run during star exposures

Cheers
Robin
Attachments
RV_Vega_LHIRES2400_THO.png
RV_Vega_LHIRES2400_THO.png (32.04 KiB) Viewed 8212 times
LHIRES III #29 ATIK314 ALPY 600/200 ATIK428 Star Analyser 100/200 C11 EQ6
http://www.threehillsobservatory.co.uk
Andrew Smith
Posts: 321
Joined: Fri Sep 30, 2011 12:23 pm

Re: Radial Velocity - using echelle

Post by Andrew Smith »

Thanks Robin, very helpful as always. I am making some progress with much better results from the simple procedure of bracketing a set of target images with calibration images. I have done a quick on-line search and found a number of sources of standard stars for RV measurement - some of which are bright enough for me to use practically.

I did consider tellurics but there are none near order 40 and even in order 34 which includes Ha they are not always obvious. I also feel I need to treat each order on it's merits and not project accuracy from one to another.

I agree the method of line determination may not matter too much as long as you are consistent for stable lines but not for ones that evolve with changing RV. I found this paper Comparison of Selected Methods for Radial. Velocity Measurements here http://journals.cambridge.org/download. ... c5c0bfb3b2 which describes a "Mirroring method" which while having a subjective element to it allows one to measure different aspect of a line (e.g. wings or core). Just need Christian to code it in to ISIS!

Thanks again for your help, I feel I can now make some good progress.

Regards Andrew

PS I have an old electric focuser motor that will make a good fiber shifter although it will need to be under a cover to keep the cats off!
Andrew Smith
Posts: 321
Joined: Fri Sep 30, 2011 12:23 pm

Re: Radial Velocity - using echelle

Post by Andrew Smith »

Rho Cas in 2015 and 2010
Rho Cas in 2015 and 2010
Rho Cas.png (70.47 KiB) Viewed 8207 times
Robin - The attached shows one of my Rho Cas observations around Ha compared to one made bt Christian Buil in Sept 2010. I think the "Mirror Method could work well for complex lines like these allowing the red and blue shifted peaks to be both measured.

Regards Andrew

In my defense my observation had one half the exposure!
Robin Leadbeater
Posts: 1926
Joined: Mon Sep 26, 2011 4:41 pm
Contact:

Re: Radial Velocity - using echelle

Post by Robin Leadbeater »

Hi Andrew,
Andrew Smith wrote: I agree the method of line determination may not matter too much as long as you are consistent for stable lines but not for ones that evolve with changing RV. I found this paper Comparison of Selected Methods for Radial. Velocity Measurements here http://journals.cambridge.org/download. ... c5c0bfb3b2 which describes a "Mirroring method" which while having a subjective element to it allows one to measure different aspect of a line (e.g. wings or core). Just need Christian to code it in to ISIS!
In cases like this I would say there is no single "right answer" for the radial velocity as the line shape represents an evolving distribution of velocities so cannot be adequately described by a single number. I think it is probably important to consider what is physically causing the line profile to change and how this relates to whatever property of the system you are attempting to measure. This can then guide you towards the most suitable parameter. (exoplanet measurements for example are ultimately limited by stellar pulsations which move and distort the lines). It can be useful to separate out the different components due to different physical phenomena eg the star photosphere absorption and the disc emission in the case of Be stars) There was recently a discussion on another forum as to how the expansion velocity of a Nova was derived from the P cygni profile. It turns out apparently that this is defined as the difference (in velocity) between the emission component maximum and the blue edge of the absorption component.

Cheers
Robin
LHIRES III #29 ATIK314 ALPY 600/200 ATIK428 Star Analyser 100/200 C11 EQ6
http://www.threehillsobservatory.co.uk
Andrew Smith
Posts: 321
Joined: Fri Sep 30, 2011 12:23 pm

Re: Radial Velocity - using echelle

Post by Andrew Smith »

I agree Robin, thanks again for your help.
Regards Andrew
Robin Leadbeater
Posts: 1926
Joined: Mon Sep 26, 2011 4:41 pm
Contact:

Re: Radial Velocity - using echelle

Post by Robin Leadbeater »

BTW SPLAT VO mentioned in the paper can be downloaded as a JAVA program here
http://star-www.dur.ac.uk/~pdraper/spla ... at-vo.html
I have used it occasionally and is quite a nice program for playing with line profile fitting, RV and EW measurement and the like, going beyond ISIS and VSpec

Cheers
Robin
LHIRES III #29 ATIK314 ALPY 600/200 ATIK428 Star Analyser 100/200 C11 EQ6
http://www.threehillsobservatory.co.uk
Post Reply