We are nearing completion of a survey of galaxy distances using the method of Surface Brightness Fluctuations. The sample consists of most early-type galaxies within about 3000 km/s -- rather similar to the Shapley-Ames catalog of galaxies. Being part of a survey, not all data are of equal quality, but we expect the median distance error to be about 8 percent.
The results reported here are preliminary, since we will not have performed our final photometry reconciliation until later this year, but it is unlikely that anything will change by more than 10 percent.
THE SBF SURVEY
>5000 CCD frames E. Ajhar
>1600 Summed images J. Blakeslee
>800 Galaxy/filters A. Dressler
~400 Galaxies J. Tonry
Observations 100% Distances:
Reductions 85% d < 2000 km/s: ~complete
< d < 2800 km/s: substantial
Distance errors: < d < 4000 km/s: some
5% - ???
median ~ 8%
The plot below shows the distribution of galaxies for which we have data on the sky. The open symbols are galaxies for which we have data but which has not yet been reduced.
The SBF method derives distances by first measuring the point-to-point variance in a galaxy image, correcting for non-stellar contamination, and writing the result as an effective stellar flux or magnitude: mbar. The second step is to assign an absolute magnitude to the stellar population in the galaxy and thereby determine a distance. We can use theoretical models to find the absolute magnitude as a function of galaxy color, but we mainly rely on an empirical determination. We observe 26 galaxies in Fornax covering a large range in luminosity and type, and knowing that they are all the same distance, find that the absolute Mbar is a one-parameter family which depends on galaxy color.
We see the same dependence on color in many groups, and we then use Cepheid distances to set a zero point. Until recently the zero point was local group galaxies and the bulges of M31 and M81. With the advent of HST measurements of more distant galaxies in Virgo, we find very good agreement with our original zero point. We will be doing a final photometry reconciliation this fall, and it is possible that our zero point could change about 0.10 magnitude, but by then distances to Leo, Virgo, Fornax, and the N1023 group should all be available, so we will be very securely tied to the Cepheid distance scale.
Theoretical models confirm our empirical findings about Mbar and stellar populations. Work originally done by Tonry, Ajhar, and Luppino has been greatly improved and extended by Worthey. He finds that to a large extent Mbar in the I-band is a one-parameter family, because the effects of age, metalliticity, and IMF variations are pretty much degenerate. The zero point from Worthey's theoretical models matches the empirical determination from Cepheids to about 0.10 magnitude.
SBF is not a cheap way to get distances, but it is perhaps the most accurate and reliable method around. From the ground it is possible to measure distances to perhaps 4000 km/s in 0.5 arcsecond seeing (the distance goes inversely as the seeing) in about 2 hours on a 2.4-meter telescope (such as we have at Michigan-Dartmouth-MIT Observatory) with a very good CCD. Using HST, the improved imaging along with the worse quantum efficiency of WFPC2 leads to a requirement of about 5 orbits at that distance. It is possible to measure distances to more than 10000 km/s with HST, with exposure time going as the square of the distance.
Thus SBF is a vastly cheaper way to get distances than measuring Cepheids, but is in turn much more expensive than Tully-Fisher or Dn-sigma.
When SBF distances are compared to Dn-sigma we find a good correlation, with the errors in Dn-sigma pretty much given by 20 percent (as indicated by the dashed lines). However, as the case of NGC3641 and NGC3640 illustrates, the errors in Dn-sigma are not Gaussian. SBF finds these two neighboring galaxies with similar redshifts to have the same distance to 5 percent. Dn-sigma finds distances that differ by almost a factor of four. Therefore anyone who wants to calculate Malmquist selection bias corrections to Dn-sigma distances should consider the possibility that the error distribution may have extended tails as bad as a Lorenzian.
This comparison with Tully-Fisher (TF) is intended to look bad. I deliberately matched SBF distances for early-type galaxies with TF distances for late-type galaxies according to loose "group" criteria. This matching purely by location on the sky and redshift leads to substantial distance discrepancies which I believe to arise from non-physical groups. When stricter grouping criteria are used, such as insisting that there be several spirals and ellipticals intermixed in a substantial overdensity, the agreement between TF and SBF becomes much better, with TF achieving its advertized accuracy of about 20 percent. It is therefore important to be cautious about assigning galaxies to groups.
Both of the TF and Dn-sigma distances are uncorrected for Malmquist bias, and reported in km/s according to their tie to galaxies whose velocities are dominated by the Hubble flow. We can therefore determine a preliminary Hubble constant from the SBF distances in Mpc, and the result in both cases is about 85 km/s/Mpc.
I would like to shift to displaying galaxy positions in supergalactic coordinates. There is a North-South band across the sky running through Ursa Major, the Virgo cluster, and Centaurus which contains a large fraction of the nearby galaxies. Supergalactic coordinates put that band in the SGX-SGY plane, so equatorial N/S is +/- SGX and E/W is +/- SGZ. The galactic plane is almost in the SGY-SGZ plane. Virgo is found near the +SGY axis; Fornax is near the -SGY axis; Centaurus is in the middle of the second quadrant; etc. The points here are from the SBF survey; redshifts are not used in placing these galaxies. Therefore any "Finger of God" features are either distance errors or real strings along the line of sight.
Up:Tonry
Next:Tonry: Section 2
Home:Workshop Home Page
