STIS New Spectrophotometric Standards
Final reduction of GO 12813
September 30 2013

Brian Schmidt, Mike Bessell, Australian National University

Ralph Bohlin, Susana Deustua STSci

Sara R Heap, NASA/Goddard Space Flight Center

Elena Pancino, Giuseppe Altaville, Bologna Observatory (INAF)(Gaia)

Nicolas Regnault, Pierre Astier, l’Institut National de Physique Nucléaire et de Physique de Particules

John Tonry, University of Hawaii


Bengt Edvardsson, University of Uppsala

Detlev Koester, University of Kiel

Don Lindler, Sigma Space Corporation

Simon Murphy, University of Heidelberg


Uncertainties in the overall calibration of SN to the fundamental standards is currently the largest single source of systematic uncertainty in the Supernova cosmology experiments. Our goal is to tie photometry in both hemispheres to an absolute spectrophotometric system which will serve as the basis of SkyMapper's photometric system, and would be the basis of the calibrations necessary to undertake supernova cosmology experiments to a higher degree of precision. To achieve this we need a network of stars with precisely determined SEDs (to properly account for passband differences) that are on a common photometric zeropoint scale. The highest precision CALSPEC stars have two drawbacks for SkyMapper, they are extremely hot (Teff > 30,000K), and none are in the southern hemisphere. We therefore propose to obtain STIS spectrophotometry between 200nm and 1020nm with an accuracy of better than 1 percent for 13 stars, 6 northern SDSS standards, 6 southern FG metal-deficient dwarfs plus 1 circumpolar solar twin. These stars have colors similar to the bulk of field stars and galaxies and will better sample the survey bandpasses. All these stars have Hipparcos Hp magnitudes accurate to 0.002 mags providing an independent check on the HST photometric zeropoint. They can also be well modelled enabling longer wavelength flux calibrations to be generated. These data would also be used for current SNLS/SDSSII work, DECCAM, and eventually LSST calibration. Furthermore they will be invaluable to calibrate the bright end of the Gaia spectrophotometry. We are also proposing parallel mode WFC3 observations of adjacent fields in six bands to provide additional calibration fields.


Wide-field CCD surveys underway or planned will provide magnitudes and positions of stars and galaxies with unprecedented internal precision. The first groundbreaking survey in the northern hemisphere was the Sloan Digital Sky Survey (SDSS) which established the now standard very broad u, g, r, i, z passbands and whose excellent photometry continues to be refined through re-calibration (Padmanabhan et al. 2008). Pan-STARRS1 (Tonry et al. 2012), also in the north, is now underway with a g, r, i, z, y survey that extends the photometry deeper and further to the red. The first southern hemisphere survey, SkyMapper (Keller et al. 2007), has commenced. SkyMapper is a 1.3m telescope with a 5.7 sq degree FOV, situated at Siding Spring Observatory in Australia. The imager has 16384x16384 0.5 arcsec pixels and will map the southern sky in 6 bands – u, v, g, r, i, and z. A major driver for all the new surveys is the investigation of Dark Energy.

Our goal with the STIS DDT proposal is to tie photometry in both hemispheres to a precise absolute spectrophotometric system that would serve as the basis of SkyMapper’s photometric system and also allow us to stitch together photometric systems across the entire sky.

Such photometric precision is required for a wide variety of science. These include precision photometric redshifts – used through out cosmology, but specifically for next generation weak lensing surveys. Stellar astronomy also requires precision spectrophotometry for fundamental comparisons with model atmosphere fluxes to substantiate the models’ physical inputs and to use the model fluxes for flux calibration at longer wavelengths.

These stars will also form an essential calibration link for ground based spectrophotometric programs and will help the Gaia satellite tie down their photometric and spectrophotometric systems. It will be the basis of the calibrations necessary to further improve supernova cosmology experiments to a higher degree of precision. Currently, uncertainties in the overall calibration of SN to the fundamental standards are the largest single source of systematic uncertainty in the SN experiments (e.g. Sullivan et al. 2011).

Most ground-based photometric systems are well standardized (Bessell 2005) fainter than 5th mag but ultimately are based on the absolute flux of Vega, measured from the ground in the 1980s and known to only a few percent (Megessier 1995). Most of the uncertainty results from difficulties in measuring the horizontal and vertical atmospheric extinction, and the large temperature difference between the standard lamp and Vega’s photosphere. Although there are plans to improve this situation in the future using modern calibrated Si photodiodes and more sophisticated extinction measurement and modeling, the most reliable photometric calibrations available now are space-based from Hipparcos and HST.

Currently there are two spectrophotometric catalogs from HST – NGSL, and CALSPEC. We had hoped to use stars in the NGSL catalog as our fundamental standards – but due to irreducible remaining uncertainties in the offsets of the stars from the slit centre, there remain uncertainties of at least a few percent in the relative flux as a function of wavelength. The CALSPEC stars, which used a wide slit, do not suffer from this problem, but have other drawbacks that make the current ensemble difficult to use to achieve our goals. These drawbacks are

• there are very few moderate colored stars which can be tied directly to normal stars in the survey - the very blue stars typically require extrapolation from the bulk of the objects we wish to calibrate;
• there are no appropriate southern stars that we can use to tie down that half of the sky;
• there are very few stars that can be tied to an external photometric system with high precision;
• they are based on a single type of model atmosphere, without cross checks to a wider variety of stellar types.

The Hipparcos Hp magnitudes constitute the most precise set of integrated photometric fluxes. There are 72,300 stars brighter than V=10 mag in the Hipparcos Catalogue with median Hp magnitudes given to better than 0.002 mag. These stars can provide very reliable zero-points for all optical and far-red photometric systems. We have demonstrated that the Hipparcos Hp passband is now well characterized (Bessell & Murphy 2012) so we can reliably synthesize the Hp magnitudes from spectrophotometric fluxes, providing an important check/normalisation on the zeropoint of the HST fluxes.

We have selected stars with precise Hp magnitudes, similar to, and including some of the NGSL extremely metal-poor stars, for which we propose to obtain STIS spectrophotometry to better than 1%. The new northern stars are SDSS standards so that we can refine the overall SDSS calibration. One of the southern stars is circumpolar thus providing good calibration for the far south fields as well as a good monitor of extinction. The remaining southern stars were selected at intervals of 4 hours providing good coverage throughout the year and the ability to measure instantaneous extinction through paired observations of two standards, one at high and one at low airmass. The majority of these stars are 9th magnitude metal-deficient weak-line dwarfs and subgiants like the SDSS fundamental standard BD +17 4708, with little or no interstellar reddening. High resolution spectra are also available for all the stars enabling the stellar parameters to be well defined and their energy distributions (SEDs) to be well fitted by model atmospheres. Fitting model fluxes enables them to be used as spectrophotometric and photometric standards out to beyond 25 microns. Combining these new standards with the existing CALSPEC stars provides us with spectra covering a wide range of colors that will enable us to ensure that our system passbands are well defined.

The stars will be excellent to use as standards for ground-based spectrophotometric programs aiming to establish a network of 13th magnitude white dwarf secondary spectrophotometric standards suitable for PanSTARRS (Tonry private communication). Most importantly, the spectra will be invaluable to use as part of the calibration suite of stars for the Gaia BP and RP prism dispersed, slitless spectrophotometry (Pancino et al. 2012a,b).

Download the Spectra

The following table lists all the stars observed and gives their J2000 equatorial coordinates, V magnitudes, Hp magnitudes and Te, logg, vrad, [Fe/H] (from Simbad) or comment. The spectral files currently provided are final reductions from Ralph Bohlin who has done a new STIS calibration based on new WD models from T. Rauch (see their G191B2B paper in press at A&A). The changes are small <1%. The spectra are available here as .txt files and can be downloaded by clicking on the filename in the last column in the table below. The units are Flambda and Wavelength (A).
The first two lines are labels, the third line gives the number of points.
This .txt data is not presented at equal wavelength spacing or resolution, as it is combined data from observations at three grating settings.
The wavelength spacings are 1.37A from 1667-3060A, 2.75A from 3060-5450A and 4.87A from 5450-10223A.

Here are the fluxes as provided by Ralph with full header information. 

Here also are the spectra smoothed to about 10A resolution, scrunched and written to fits files. 

Target Name

RA (J2000)

Dec (J2000)






01 28 46.4

-24 20 25.3







04 52 9.9

-27 03 50.9







08 40 50.8

-16 20 42.5







12 53 15.1

-18 31 20.0







17 42 49.3

-40 19 15.5







21 06 34.7

-49 57 50.3







20 28 18.7

-87 28 19.9




Solar twin



04 14 35.5

+22 21 04.3







08 19 22.5

+54 05 09.6







10 47 23.1

+28 23 55.9







14 49 2.3

+25 42 09.2







17 39 45.6

+02 24 59.6







19 20 34.9

-07 40 00.0




10650/8.08 DBQ5



05 05 30.6

+52 49 51.9




HST WD standard



05 52 27.6

+15 53 13.7




HST WD standard



Preliminary fits to model atmosphere fluxes


One of the stars, GJ754.1A is a H deficient, C-rich white dwarf. HD185975 is a solar-like star. The others are halo stars with Te between 6500K and 4750K; logg between 4.5 and 2.0; [M/H] between -1 and -3.0. Scaled solar abundances with standard halo alpha enhancements depending on the [M/H] should be appropriate. The two giants HD9051 and HD200654 probably have some additional CN processing.

We are fitting the observed STIS spectra with theoretical model atmosphere fluxes. These fitted model atmosphere spectra not only provide excellent checks on how well the fluxes from the G230LB, G430L and G750L observations have been spliced together and the overall precision of the flux calibration, but can be used with confidence to extrapolate the fluxes to infrared wavelengths thus providing standard spectrophotometric fluxes out to 40 microns.

Model atmosphere fluxes

We have used two model atmosphere grids: ATLAS and MARCS. The ATLAS grid is from Castelli & Kurucz (2004 A&A419, 725) (3500K – 47500K) plus additional spectra from Castelli ( These spectra cover 2500A – 10500A at 1A resolution.

The other model grid is from MARCS: (2500K-8000K). The fluxes are just monochromatic fluxes sampled with a resolution R=20000. That means that there is no information at all about what happens between the sampling points. If several consecutive samples happen to fall near line bottoms their average will underestimate the real flux, and vice versa if they fall near continuum points. An estimation of these effects were made in Plez (2008): When smoothed to 5-10A these fluxes match the observations remarkably well.

However, Bengt Edvardsson has computed line-by-line spectra for a subgrid of parameters relevant for our sample of stars and these will be used for our definitive fits. These fluxes will extend from 1300A – 40 microns. We note, however, that HD111980 is a triple system and the observed IR flux will be contaminated by its fainter and cooler companion and not well represented by the extrapolated fitted model fluxes.  

Preliminary results

Ralph Bohlin, Sally Heap & Don Lindler, and Mike Bessell & Simon Murphy have independently fitted the spectra using 3 different programs and techniques and with different priors. One approach to fitting is to provide astrophysically consistent parameters with physically likely reddenings. Another is to to get the best possible fits by allowing all parameters to vary independently. However, we will eventually agree on the best fits and errors to the parameters. Adjustments for the radial velocities listed in Table 1 were made prior to fitting. The main differences are in the reddenings used. Revised fits to the final spectra will be available in November 2013.

                   Heap/Lindler                                    Bohlin                                       Bessell/Murphy       

Name           Teff   log g  log Z E(B-V)  rms    Teff  log g log Z  E(B-V) chi   Teff  log g log Z E(B-V) rms

bd02d3375  6207 3.738 -2.468 0.048 0.032    6440  4.40 -2.00  0.092  5.07     6050  3.8  -2.5   0.025  0.0167

bd21d0607  6366 4.138 -1.436 0.031 0.027    6380  4.20 -1.67  0.042  1.89     6150  3.9  -1.75  0.0     0.016

bd29d2091  5877 4.547 -1.650 0.011 0.023    5980  4.50 -1.53  0.032  1.42     5800  4.1  -2.0   0.00   0.015

bd54d1216  6185 4.072 -1.403 0.021 0.022    6140  4.00 -1.53  0.019  1.36     6050  3.8  -1.75  0.0     0.016

hd009051    4873 1.981 -1.480 0.009 0.018    5080  2.50 -1.04  0.080  0.77     4850  2.1  -1.75  0.0     0.014

hd031128    6175 4.387 -1.137 0.032 0.025    6045  4.00 -1.50  0.023  1.45     5900  3.9  -1.75  0.0     0.016

hd074000    6239 3.774 -2.090 0.001 0.021    6480  4.20 -2.00  0.043  2.59     6225  3.8  -2.25  0.0     0.015

hd111980    5832 3.720 -0.939 0.011 0.021    5820  3.70 -1.09  0.009  0.59     5775  3.5  -1.25  0.0     0.016

hd160617    6084 3.571 -1.751 0.021 0.020    6280  3.90 -1.64  0.056  1.87     5950  3.4  -2.00  0.0     0.016

hd185975    5563 3.838 -0.160 0.000 0.025    5760  4.50  0.01  0.037  0.33     5650  3.8   0.00   0.025 0.021

hd200654    5448 2.740 -2.613 0.057 0.016    5520  3.20 -2.00  0.072  3.29     5300  2.7  -3.00 0.025  0.013


The spectrum of GJ754.1A =LDS678B = EG131 = WD1917-077 was fitted by Detlev Koester with specially computed H-deficient and C-rich WD atmospheres. An excellent fit was achieved for Te=10650 and log g=8.08, log H/He = -5.2.
All the fitted model spectra will be provided later.