User Tools

Site Tools


Second workshop stellar parameters

The purpose of these exercises is to simulate finding astrophysical parameters from Gaia data. We will focus on using the low resolution BP/RP data for this, but will include the information from astrometry and possibly also high-resolution RVS data. The exact aim depends on which challenges are suggested and which are popular. There may arise some constraints connected to our ability to simulate particular datasets.

Package is organized by Rene Andrae and Kester Smith


Default Data Package

NOTE: updated 14/10/2014

There was an error in the first version of this data set. Please download the new version.

Here is the first set of mock data. This has been made as follows;

  • Obtain a sample of objects from the Gaia Universe Model Snapshot catalogue. Coordinates are l,b=0,80 (so a sparse halo field). There are just over 2000 objects.
  • Use Gog to produce BP,RP, RVS spectra. Note: there are few RVS spectra
  • Gog results are held in the file gums_0_80_v1.dat. Data include position, proper motion, parallax, Radial velocity, Gaia band fluxes, and basic astrophysical parameters.
  • BP, RP and RVS spectra are in separate files (identify with the sourceId included in the filename) The columns are wavelength,flux, error
  • BP,RP wavelengths are currently in 'pseudo-wavelengths' (~pixel numbers). I will try to provide an approximate true wavelength scale soon.
  • I will provide a mean instrument model (line spread function + sensitivity function) soon.

In the meantime, please have a look at this data and let me have any comments re: ease of use (or more likely the opposite), any suggestions etc etc.

The data set is in the mpia public ftp site (, in pub/smith/gaiachallenge, name gums_0_80_v1.tar

Also available;

  • bprpWavelengths.dat , which are the wavelengths in nm
  • bprpTransmission.dat , which is the total transmission function (column 1: bp, column 2: rp)
  • bprpLsfs.tar , a set of LSFs, one for each pixel. The wavelengths are indicated in the filename. The offsets are in pixel coordinates.


Challenge 1

Two datasets are available from the mpia public ftp site (, in pub/smith/gaiachallenge/andrae, name andraeChallengeV1.tar.

Unpack the tar archive to obtain;

* nominal-grid-MARCS.csv - basic APs and noise-free spectra for the a grid of known stars * test-data-MARCS-reduced.csv - basic data (position,astrometry,photometry) fro a set of test sources * test-spectra/* - files containing noisy BP/RP data for the test data.

The challenge is to estimate the APs (Teff, logg, Fe/H, AV) for the test spectra.

Challenge 2

Set of reference spectra supplied by Paula Jofre.

These are available at the MPIA anonymous ftp site. cd to pub/smith/gaiachallenge/jofre and pick up the tar file.

The file contains bp and rp spectra for 32 stars, plus the file jofre_v1.dat which contains the positions, astrometry, photometry, RV and VsinI for each source.

UPDATE Also available jofre_v1_rvs.dat, containing rvs spectra for all stars.

This sample has been selected from observations taken with high resolution from the Ground. These stars are candidates to be reference stars for Gaia in the near future. The high-resolution spectra were then convolved and transformed to BP-RP spectra using GOG. The challenge is to try to determine their parameters, in a totally blind way, and then we can during the workshop compare with high-resolution ground based analyses.

Challenge 3a

Data for several moving groups was run through GOG. The results are available on the ftp site in directory murphy in file movingGroupsKnown.tar

The challenge is just to look at the data!

From: Andy Casey

See how well the working group can recover astrophysical parameters for a sample of N (where N is ~“large”) stars. The spectra could be interpolated from model spectra, doppler shifted, resampled and convolved with some profile which would roughly represent the Gaia RVS, and with noise + jitter added. Making it a blind test would be much more fun.

From: Andy Casey

Fitting low-resolution BP/RP spectra and high-resolution (AMBRE grid from Gaia-ESO, not RVS from Gaia) spectra simultaneously. Does it outperform fitting each individually? High-resolution model grid is from MARCS and BP/RP are provided in MARCS, too.

From: Maria Bergemann

the performance of our new Bayesian pipeline (Schoenrich & Bergemann 2013) on the Mock Gaia photometry+RVS+parallax data samples.

From: Simon Murphy

Finding *new* young moving groups in the solar neighbourhood (<1 kpc say) using a combination of BP,RP photometry, astrometry and RVS velocities.

From: Kester Smith

Identify the red clump.

From: Kester Smith

Identify the blue horizontal branch.

From: Kester Smith

identify extreme metallicity stars (ultra low, or very high alpha enhanced, or some other extreme)

From: Rene

validation of HR diagram: do all relevant types of stars exist in test data? do they end up in the right region of the HRD?

Useful references

tests/astropars/astroparsii.txt · Last modified: 2015/06/16 11:47 by glennvdv