«1 Standard Photometric Systems Michael S. Bessell Research School of Astronomy and Astrophysics, The Australian National University, Weston, ACT ...»
The transition from photomultiplier tubes to CCDs is almost complete. It is very di cult now to nd an observatory that still supports photoelectric photometry for visitors. The downside of this transition has been a lowering of the photometric accuracy, in particular for standardized photometry. This is due to several e ects - passband mismatches, lack of faint standards suitable for CCDs, lack of standards with a good range of color and reluctance of observers to observe many individual standards (to cover the full color range) because of slow readout times of CCDs. All these problems are being addressed. More care is now taken with passband matching, new secondary standards appropriate for CCDs are being established and with the availability of CCDs capable of being readout in 10-20 seconds, there is nothing to inhibit taking short exposures of many individual standard stars if necessary.
8 Bessell 1.5.2 Infrared wavelengths The detector revolution has been even more dramatic for IR astronomy. Commencing in the 1960s with single-element, high dark-current, detectors - PbS photocells, InSb photovoltaic and Ge bolometers - pioneering work was done by Johnson (Johnson 1962, 1966 Johnson & Mendez 1970), Glass (Glass 1973, 1974) and Spinrad & Wing (1969). A good summary of IR instrumentation and photometry is given by Glass (1999). The last 10 years has seen a rapid advance in the fabrication of large 2D focal plane arrays in IR-sensitive semiconductors. These photon detection devices have read noises approaching those of optical CCD systems (8-15 e) and dark current levels between 0.1 and 1 e/sec. A useful summary of the latest 2D detector development is given at http://www.nomic.net/ tplagge/norton-arrays.pdf. The original arrays were developed for military applications, but much of the latest IR detector development is underwritten by NASA for future space and ground based large telescope instrumentation (http://irtek.arc.nasa.gov/ARCS&T.html). Details of the latest IR technologies are given in http://www.rockwellscienti c.com/imaging/products.html and http://www.raytheon.com/products/astronomy sensors/.
The main IR standard photometry system is the Johnson-Glass JHKLMN broad-band system, devised to measure the integrated ux in windows in the earth's atmosphere away from the H2 O absorption. A summary of the Johnson-Glass broad band system was given by Bessell & Brett (1988). The original K bandpass has been modi ed in an attempt to minimize the sky and telescope emission in the original band (Wainscoat & Cowie 1992), whereas problems associated with residual atmospheric absorption in all bands have been addressed by Young, Milone & Stagg (1994) and Simons & Tokunaga (2002) by designing new bandpasses. Milone & Young (2005) provide the nal word on this issue. Not only will the new bands permit more accurate photometry but they will enable better extrapolation for outside-atmosphere magnitudes. Standards in the new MKO-IR bands are given by Hawarden et al. (2001)(JHK) and Leggett et al.
(2003)(L0M 0 ).
Space-based instruments have made remarkable contributions to the infrared observations of stars and galaxies. Dedicated experiments such as Infrared Astronomical Satellite (IRAS), Infrared Space Observatory (ISO) and Midcourse Space experiment (MSX), have explored mainly at very long wavelengths inaccessible or barely accessible from the ground. The KAO (Kuiper Airborne Observatory) similarly have concentrated on the longest wavelengths. The HST does near-IR imaging with NICMOS (near infrared camera and multiobject spectrometer) and the Spitzer Space Telescope (formerly SIRTF) is doing 2-30 wavelength IR imaging with large array detectors and 3-700 imaging and spectroscopy with small array detectors.
2MASS (Two Micron All Sky Survey) (JHK) (http://www.ipac.caltech.edu/2mass/overview/abou and DENIS (Deep Infrared Survey of the Southern Sky) (iJK) (http://www-denis.iap.fr/presentation/ are two ground based surveys that are providing a wealth of photometric data on stars
and galaxies. The limiting magnitudes of the surveys are 2MASS - J: 15.8, H: 15.1, K:
14.3 DENIS - i: 18.5, J: 16.5, K: 14.0.
Martin Cohen and collaborators have made immense contributions to the photometric calibrations of all the IR space experiments as well as to ground based IR spectrophotometry and imaging (for example Cohen et al. 1992a,b, IRAS Schaeidt et al. 1996, ISO Fouquet al. 2000, DENIS Cohen et al. 2003c, 2MASS Cohen et al. 1999, IR spectrophotometric standards Cohen 2003b, stellar calibration in the IR.)
1.6 Synthetic photometry Synthetic photometry is the name given to magnitudes and colors derived by convolving model atmosphere uxes or observed spectrophotometric uxes with standard passbands.
Synthetic photometry is discussed by Cousins & Jones (1976), Buser (1986), Straizys Standard Photometric Systems 9 (1996) and Cohen et al. (1996). The passband or response function of a standard system is normally obtained by multiplying together the lter transmission, the re ectivity of the telescope mirror, the transmission of the camera optics and the quantum e ciency of the detector used. For ground-based work, this should then be multiplied by the transmission of the Earth's atmosphere for an airmass of (at least) 1.0. Generally, the atmospheric correction is usually only done for broad ultra-violet bands like U, where the extinction is large and varies signi cantly across the passband, and for the far-red and IR bands that contain signi cant molecular (for example, H2 O and O2) absorption. At other wavelengths the atmospheric extinction variation across the passband is so small that it will not cause an e ective wavelength shift but only change the zeropoint. For most purposes the passbands are generally normalized rather than using the absolute values.
Most past computations of synthetic photometry convolved the f spectrum of a star in energy units by the bandpass sensitivity function. That is, the energy measured across a bandpass X is R f( )RX ( ) d where RX ( ) is the response function of the system.
But if photometry is done by counting the number of detected photons across the passband X, that number is: R R (f( )/h ) RX ( ) d = ( f( )/hc) RX ( ) d This in essence weights the uxes by the wavelength. The net e ect is to shift the apparent e ective wavelength of a passband slightly to the red. By rearranging the above equation as follows R R (f( )/h ) RX ( ) d = (1/hc) f( )( RX ( )) d one sees that it is possible to modify the bandpass responses rather than modify the stellar uxes, often making it simpler to compute.
Modern detectors are all photon counting detectors unlike most photomultiplier tubes that were originally used as energy integrating devices, so it is generally correct to modify the resultant passbands to account for these photon-counting observations. As noted above, it is common to use normalized response functions and to obtain the zeropoint of the synthetic photometry from the primary standard star, normally Vega or some other spectrophotometric standard with accurately known standard magnitudes and colors.
The UBVRI system's magnitude zeropoints were set by de ning Vega to have colors of zero. The V magnitude being +0.03 mag means that Vega is 0.03 mag in all bands.
Where there was doubt about Vega having an IR excess beyond K(2 ), Sirius was used (Cohen 1998 see also Price et al. 2003). Setting the zeropoints for the UBVRIJHKL system is discussed in an appendix of Bessell, Castelli & Plez (1998) and by Cohen et al.
When exploring the realization of a standard system it is necessary to have a set of stars that cover a wide color range and have accurate spectrophotometry from which to derive synthetic photometry. It is best for those stars to have established standard colors and magnitudes for the system in question, but if that is not possible, then it is acceptable to compare synthetic photometry relations from observed uxes or model atmosphere uxes with observed relations from standard stars. That is, if exploring the synthetic Sloan Digital Sky Survey (SDSS) g band using the Vilnius uxes (averaged spectral and luminosity types) for which one does not have SDSS photometry, one can regress synthetic g-V against synthetic V-I for the Vilnius stars and compare this with the observed relations for nearby unreddened Population I stars. This presupposes that the V and I passbands, in this case, are well established. One then changes the g passband and recomputes the synthetic magnitude until the synthetic and observed regressions agree.
2 Broad-band photometric systems The passbands of some of the various systems is shown in Figure 1. The e ective wavelengths and FWHMs are given in Table 1.
2.1 The Johnson-Cousins UBVRI system One of the earliest and most used of the standard photoelectric photometric systems is the UBV system. The B band was devised to approximate the raw photographic magnitude (less the UV), while the V band was to approximate the visual magnitude system. The U band provided the important additional band between B and the atmospheric cuto.
Perceived problems with the original system were the short wavelength cuto of the U band being provided by the atmosphere (and the glass optics of the photometer and the 1P21 glass envelope), while the long wavelength cuto of the V band was provided by the detector. This resulted in the atmospheric extinction becoming an integral part of the U band and the V band cuto was a function of the temperature of the photomultiplier tube and the speci c photomultiplier selected. Additional problems arose as di erent observers used di erent kinds of glass lters with di erent thicknesses and S11 photocathodes rather than the S5 photocathode of the 1P21. The di erent natural systems thus de ned resulted in poor standardized photometry and systematic di erences for unusual stars or for stars of a kind (such as highly reddened or metal-de cient) not represented in the standard lists.
There are good contemporary photoelectric catalogs of UBV standards. Cousins (1973, 1983, 1984) established accurate UBV photometry in the E-regions and the equatorial regions by carefully transferring the original Johnson standards to the south. Landolt (1973, 1983) independently established equatorial UBV standards from the Johnson standards. Cousins (1984) noted di erences between contemporary northern hemisphere U-B systems and that of the Cape. These systematic di erences need not a ect derived astrophysical quantities as long as observers identify if they are using Landolt or Cousins standards. Unreddened stars can accurately be transformed between the two systems at the (generally) extreme color ranges where they di er. Menzies et al. (1991) also commented on these systematic di erences resulting from di erent lters used and di erent transformation techniques. Systematic di erences can be introduced into standard U-B and B-V photometry by whether or not the transformation equations use a B-V term instead of or as well as a U-B term.
The UBV system originated with the 1P21 and although the B and V bands have been well duplicated using redder and more sensitive detectors, the U band has provided di culties. The quartz windows of many of the substituted phototubes permitted too much UV light compared to the glass envelope of the 1P21 and the e ective wavelength moved too far to the UV (see Landolt 1983 gs.4,5,6,7), while most CCDs, having low UV response, often produced an e ective wavelength too far to the red (eg. Sung & Bessell 2000) when the same photoelectric lters were used. Menzies (1993) details the Standard Photometric Systems 11 transformation procedures and the linear and non-linear relations involved in converting from the natural photometry system (in UBV and RI) to the standard Cousins system at Sutherland.
There have been several attempts to realize the UBV passbands (Matthews & Sandage 1963 Azusienis & Straizys 1969 Buser 1978 Bessell 1986:U, 1990:UBVRI).The agreement for B and V are good but there is still some disagreement and uncertainty associated with the U band, the major e ect of which is to require adjustment of the U zeropoint for di erent temperature ranges (see Appendix 3.1 in Bessell, Castelli & Plez 1998) The red bands could not be measured with the 1P21, so a variety of photomultiplier tubes were used with mixed success and limited usefulness until the extended-red S20 and GaAs tubes were developed as discussed in Section 1.3. Bessell (1983) discussed in detail di erences between the original Johnson RI system developed using a S-1 phototube and later observers claiming to be observing on the Johnson system. One of the problems was the fact that Johnson published R and I passbands for his system that were about 200 A further to the blue than indicated by the colors of stars in his catalog (Johnson et al.
1966). The low precision of many of the Johnson RI catalog stars made it time consuming to observe a su cient number of standard stars to ensure that the same standard system was de ned each observing run and as a result it often was not.
Independently, Kron & Smith (1951) and Kron et al. (1953, 1957) established the Kron RI system using a Continental Electric CE25A/B tube that did not extend as far to the red as the S-1 tube and as a consequence, the R and I bands of Kron were not the same as the Johnson RI bands. As the extended-red S20 and GaAs tubes also did not extend as far to the red as the S1 tube, their natural photometric systems were more closely related to the Kron system. Weis (eg. Weis 1983, 1996) developed a precise contemporary version of the Kron system using a GaAs tube (Bessell & Weis 1987).
Weis's version is an excellent representation of the Eggen-Kron RI system (Eggen 1975).
Cousins (1976) established a stand-alone RI system in the E-region using the GaAs tube and noted (Cousins 1980b) that it was similar to the Kron and Eggen system for the hotter stars but diverged for the redder stars. Bessell (1979) also compared the new Cousins VRI system photometry with Johnson et al. (1966), Kron et al. (1953, 1957) and some other systems and provided transformations between the di erent systems. Bessell & Weis (1987) updated and re ned the transformations between the Kron system and the Cousins system.