How to use the HARPS archive

Quick question: when doing the first part of the data entry (the columns into notepad) must every available data mentioned be used from the HARPS archive on that star?

-M-

I don't think I'm interpreting your question correctly, but unless it is excessively outlying I don't see why you would want to discard a value.

What would you define as excessively outlying? A number that seems inconsistent w. the others?

Right now the star I'm working on has, out of 50 files, four RVC values in the negative, a BIS SPAN value of 0, and one line where the CCF NOISE and CCF FWHM values are the same. What effect does this have on the outcome? Or would it have any?

Thanks for putting up with my very non-technical self and equally silly questions.

-M-

Ah, that sounds like you're dealing with an A/F-star? I've seen these kind of issues several times for early-type stars, which relates to problems using the G2 mask on stars with very different spectra.

I can't really make much of a diagnostic from here, but if you could direct me to the star I could try to judge which spectra are not useful. Based on previous experience, probably more than a few of them…

Quote :

Thanks for putting up with my very non-technical self and equally silly questions.

All for a good cause!^{[citation needed]}

PSI Cap, HD 197692, Pazhan, or whatever else you choose to call it....it's an F5V.

Thanks.

-M-

Alright. First thing, the last three spectra of the 50 are actually of Vesta (as indicated by the name of the files) - the asteroid is often used to take a spectrum of the Sun, and looking in Celestia at the date of observation shows that Vesta was quite close to Psi Cap. So ignore those three.

Now, to preface: literature indicates the radial velocity of the star is about 26 km/s, give or take a couple, and Lagrange et al. found a radial velocity RMS of 30 m/s (0.03 km/s) based on a subset of these spectra (but analysed with their own program)

This is what I got from the 47 spectra, format equal to that in the first post in this thread and split into nights. As you can see, very few of the RVs are in the vicinity of 26 km/s; the first two are definitely unphysical because such high velocities are impossible, while the others are unusually high. Also, the RV scatter is in excess of 50 km/s, which is very inconsistent Lagrange et al.'s result (again, based on the same spectra). The activity indicators are also peculiar; the FWHM is generally less than 10 km/s (though it does correlate with v sin i, and this star is a fast rotator) and the biswidth is always close to 0 km/s and rarely more than a few km/s. The only spectrum which seems to be normal is the second-to-last one, but even then the other one from the same night does not have a convergent biswidth. In all, the series is not representative of the real behaviour of the star.

This is, again, related to problems with using the G2 mask (the earliest that HARPS has) for a spectrum that is significantly different. The problems seen here do seem to stop at some point for other stars; for example, Iota Crateris' spectra from 2006 do not have convergent biswidths, but those from afterwards do - so for the Lagrange et al. sample observations before 2007 are not useful when analysed with the HARPS DRS.

Peculiarly, early-type stars observed by other groups do not seem to have this issue, so it isn't strictly due to the instrument. However, it seems to be a locally persistent problem, so my best advice here is to not bother with any spectra before 2007 for stars observed by Lagrange et al. (you can check the observing group by clicking what's in the "ProgId" column in the archive query results). Their program (which is not publicly available) does not seem to suffer from these issues, but the DRS most certainly does.

So yeah. Sorry for the none-too-positive result, but the DRS was not designed for stars earlier than F7 or so. However, I can assure you that it works wonderfully for any sensible star later than that!

I'm sure that there are plenty of others I can take a stab at...Thanks for your help!

-M-

Another question(s): Are the FWHM and BISWIDTH columns also averaged? Are they also multiplied by 1000?

Thanks (again)

-M-

They are on an absolute scale, so yes, you need to adjust them by their average: they are in km/s, so yes, they need to be multiplied by 1000 (after correcting for the average).

Are you also unable to use Sirius' extractor? It's a lot quicker than doing it manually, as I will never stop saying.

I tried the most recent link and it said the file is no longer available...argh. It looks like it would be most useful for those really big files. Right now I'm working on Gliese 511, and despite the smaller number of files, it's still taking a while.

-M-

I've almost got a new version ready with a few new features (binning, listing offsets, etc). I'm not home right now but when I get home, I will try to get it finished and upload a new version... as well as find a different file host that is less transitory.

Okay. New version - v2.4.

• Programme now lists offset values (one for each time the data has been normalised).
  
• Data can now be binned.
  
• Dataset can now be returned to its original state after being read (effectively undoes datapoint removals, binning, normalisations, etc). So now you don't have to re-read the data to start over.
  
• Fixed a problem where trying to modify non-existent datasets causes divide by zero errors and the subsequent collapse of the Universe.
  

https://www.dropbox.com/s/sr28utt733fh9n0/HARPS_RV_Extract_v2.4.bat?m=

New version 2.5. Some rather minor tweaks.

• Can now produce .csv file.
  
• Offset values cleared if data is re-read.
  
• Fixed problem where programme attempted to draw preview images before any data was read.
  
• Improved preview images:
  - Spacing between datapoints and graph edge.
  - Error bars!
  

https://www.dropbox.com/s/pr563jhkhbh5z72/HARPS_RV_Extract_v2.5.bat

ESO has changed the data access page for HARPS. Now you have to go to this page to download HARPS data, and the download products are a bit messy, with usable RV data mixed with other .fits files. I'm sure there's a better way to download HARPS data such that the only thing acquired is useful RV data, but I haven't yet bothered to find it.

This change (and dreadfully long extraction times) has motivated the writing of a new version of the HARPS RV Extract programme. This version cuts down even further on the amount of manual work on the user's part by allowing them to read the RV data directly from the compressed files within the .tar itself. I've had problems in the past with getting WinRAR to extract large numbers of .tar files to their own folders, so I figured I may as well bypass this step. With this version, you no longer need to extract .tar files to their own directory (though both methods of reading the data are available). It may be that not all downloaded .tar files have usable RV data in them. If this is the case, it will simply skip the file and leave a message in the status box informing you of the filename of the skipped .tar.

Another small change is that I've figured out multithreading in Windows Forms applications (learning this as I go along), so I threw in a status bar to allow the user to see the progress of the data reading. This is especially useful since reading from the .tar files does take some extra time, and one tends to find themselves wanting to poke the programme with a stick to see if it's still alive. Also, the CSV writing was fixed to include a new line after the header (d'oh!).

Simply drop the HARPS_RV_Extract_v2.6 into the same directory as the downloaded .tar files and run it.

Due credit: I've made extensive use of tar-cs but had to make some modifications to get it to work for this task.

Here's the file, I hope someone finds it useful.
https://www.dropbox.com/s/gf42cv1eip1dxjp/HARPS_RV_Extract_v2.6.bat?dl=0