«1 Standard Photometric Systems Michael S. Bessell Research School of Astronomy and Astrophysics, The Australian National University, Weston, ACT ...»
A star overhead in the zenith su ers the least absorption from the atmosphere. At positions away from the zenith, the absorption or extinction is greater as the path length through the atmosphere is longer. The normalized path length is called the airmass. The airmass is 1 at the zenith and is 2 at a zenith distance of 60 degrees (Z=60). The airmass is approximated by sec Z. In the optical part of the spectum extinction is a combination of continuous absorption from Rayleigh scattering of gas molecules and mainly neutral absorption from dust and aerosols. Rayleigh scattering varies with wavelength;4 and is high in the UV and blue. There is in addition, some absorption at speci c wavelengths mainly due to O3, O2, CO2 and H2 O. Ozone is responsible for the atmospheric cuto in the UV and H2 O severely a ects the transmission in the IR. The continous absorption is proportional to airmass, whereas the speci c molecular absorption is non-linear with airmass and in the case of H2 O, can vary with time as well as airmass. The wavelength dependent extinction is quite steady throughout the night and even from month-tomonth so it is usually possible to adopt mean extinction values and simply solve for the neutral component that can and does change nightly. Extinction is generally measured by observing pairs of standard star elds, one in the meridian, the other at high airmass.
And because the extinction varies appreciably across some of the broad photometric bands, especially in the blue and violet, red and blue extinction stars are observed to solve for the color term in the extinction.
Using B as an example, the extinction in the B band kB is therefore generally given as k1 -k2 (B-V) to allow for the fact that in the broad B band, a blue star will su er more extinction than a red star as its light is concentrated toward the high extinction side of the band. The extinction coe cients are normally expressed in magnitudes per airmass. In the optical part of the spectrum the observed instrumental magnitudes (Bi ) are corrected to outside the atmosphere (Bi0 ) by extrapolating to zero airmass and Bi0 = Bi - kB (airmass). However, as will be discussed later, such extrapolation is Standard Photometric Systems 5 not possible for existing IR systems. Strategies to minimize problems with extinction generally involve interspersing observations of unknown objects with standard objects and ensuring that the standards are observed at airmasses similar to those of the program objects. In this way extinction correction is a second-order e ect.
To put one's observations onto a standard system, one derives the di erences between the extinction corrected instrumental values Bi0 and the standard B value. These differences are then regressed against one or more standard colors to see the trends and determine the transformation equation. If an observer's instrumental system is close to the standard system a simple linear relation, with a small color term should be derived.
That is, Bi0 - B = ZP + a(B-V) where a should be smaller than 0.05. ZP is the zeropoint constant that includes any neutral extinction residual, aperture correction and so forth. If there are systematic deviations from a linear t, two lines can sometimes be used or the deviations can be regressed against another color that better correlates with whatever in the spectrum (perhaps the Balmer or Paschen Jump) is causing the deviation. The closer the color term is to zero, the less likely that systematic errors will be made in measuring the magnitude and colors of stars and galaxies whose energy distributions di er signi cantly from those of the ensemble of stars in the standard list.
Finally, if observers are undertaking a large photometric program with the same equipment, it is best to maintain one's extinction corrected instrumental system magnitudes for the duration and not transform onto the standard system nightly (except perhaps for the zeropoint). That way one is less likely to introduce uncertainties and systematic di erences that can occur from using standards with a restricted color range or a subset of less well de ned standards on some nights. Better combine all the standards from all the nights to determine the color term in the transformation equation once.
1.4 The nature of standard systems A standard photometric system is de ned by a list of standard magnitudes and colors measured in speci c bandpasses for a set of stars that are well distributed around the sky. Observed magnitudes are corrected for the attenuation of the Earth's atmosphere away from the zenith and the data is then normally extrapolated to zero airmass (outside of the atmosphere). The method of correcting for extinction is an integral part of the standard system (Cousins & Caldwell 1998, 2001). However, most of the infrared broad bands are di cult to extrapolate to zero airmass because of the non-linear behavior with airmass of the H2 O absorption (Manduca & Bell 1979) compared to the linear behavior of dust, aerosols and Rayleigh scattering, the major components of optical extinction.
New IR bands have been devised to address this problem (Young, Milone & Stagg 1994 Simons & Tokunaga 2002 Milone & Young 2005).
Many astronomical photometric systems have been established over the years by di erent observers with a variety of detectors and passbands. Di erent standard photometric systems usually measure di erent wavelength bands. All photometric systems enable the measurement of absolute uxes, from which can be inferred particular properties (such as temperature, gravity and metallicity) of the emitting object, but di erent systems stake claim to do it more precisely or more e ciently than other systems and some are better suited for hot stars and others for cool stars. Similarly, some are better for isolating various component stars in population synthesis of integrated photometry of clusters and galaxies.
Photometric systems are usually divided into broad-band ( 1000A), intermediateband (70A 400A) and narrow-band ( 70A). Another category, the ultrabroadband, has been recently proposed to encompass those survey systems, such as Sloan and Hipparcos, whose bandwidths are wider than the well known BVRI broad-band system. In the following sections we will discuss many of the well established photometric systems within these three usual bandwidth divisions.
Another important di erence between systems relates to whether they are closed or 6 Bessell open. An open system is one, whose originators encourage others to duplicate the passbands and detector system and to use the originator's standard stars and reduction system for their photometric programs. A closed system is one where a small group of people control the instrumentation and data reduction and only encourage others to use the results but not to attempt to duplicate the system and observe stars for themselves.
For obvious reasons, systematic errors and the quality of the data are better controlled in a closed system than an open system. In the past, the main disadvantage of a closed system was that your particular star of interest was often not in the catalog. However, with the advent of large scale sky surveys to faint magnitudes, it is likely that in the future, photometry for most objects of interest will be provided by closed photometric systems.
Most of the older systems were developed and modi ed over many years as detectors with greater sensitivity and wider wavelength response were used in place of the original detectors di erent lters than those speci ed by the originators have also been used. This has all resulted in the literature containing slightly di erent versions of some standard system, such as the UBV system, which is confusing. But in general, modern versions of most of the well established photometric systems can be homogenized and placed on a rm quantitative footing and provided care is taken with passband matching, precise and astrophysically valid data can be derived by observers for most passbands and for most kinds of stars.
However, some systems, notably the Stromgren system, have proven very di cult to homogenize and the standard system is not really well de ned for some kinds of stars such as supergiants. This has arisen from systematic variations in photometry between di erent observers with di erent photometers due to the great sensitivity of the photometry to the placement and widths of some of the passbands (Manfroid & Sterken 1987,1992).
1.5 Development of standard systems - changes in detectors and wavelength coverage 1.5.1 Optical wavelengths A good summary of the history of photometry is given by Hearnshaw (1996) and the nature of multicolor photometry by Straizys (1992), but a few comments concerning detector advances and the resultant development of standard systems are useful here.
Whitford (1940) was one of the rst astronomers to show the advantages of photoelectric photometry. Important contributions were made by Stebbins, Whitford and Kron in the 1940s at the Washburn and Lick Observatories (Stebbins & Whitford 1943 Kron 1946) using early photocells. But it was the introduction of the cooled blue sensitive RCA 1P21 photomultiplier tube in the mid 1940s and the development of low-noise ampli cation techniques that provided a huge impetus to accurate photometry (eg Johnson 1948 Dewitt & Seyfert 1950, Stebbins, Whitford & Johnson 1950). An informal collaboration was agreed between the major photometrists at the time to establish a sequence of standard stars in the northern Hemisphere but Johnson (Johnson & Morgan 1951, Johnson 1952) preempted this agreement and published his version that formed the basis of the UBV system (Johnson & Morgan 1953, Johnson & Harris 1954 Johnson 1955). The high sensitivity and reliability of the blue photomultipliers was also utilized by Stromgren (1951, 1957, 1966) and McClure (McClure 1976 McClure & Forrester 1981) for the archetypal intermediate-band uvby and DDO system, respectively.
In the early 1950s a red sensitive photocell was used by Kron and collaborators to set up the 6-color system (Stebbins, Kron & Smith 1950) and the Kron RI system (Kron, White & Gascoigne 1953) but the low gain and high dark current of the photocell restricted work mainly to bright stars. But by the end of the 1950s, Kron (1958) reported the usefulness of the rst red sensitive S1 photomultiplier tubes. An S1 tube was used by Standard Photometric Systems 7 Johnson and collaborators to set up the Johnson RI system de ned by the bright star photometry in Johnson et al. (1966). Eggen (1975) also used an S1 tube for his broad band RI photometry as did Oke (1964) and Wing (1967) for red spectrophotometric observations.
Johnson was also one of the early pioneers in IR observations using a lead sul de cell for JK photometry and a germanium bolometer for longer wavelength LMN studies.
Johnson observed many di erent kinds of stars with his UBVRIJKL photometric system and his review article (Johnson 1966) was extremely in uential for stellar photometry and astrophysics. However, the low precision of the RI catalog and the unreliability of the S1 detectors compared to the blue sensitive S5 and S11 photomultiplier tubes inhibited widespread use of RI photometry.
All this changed with the development of the multialkali extended-red response photomultiplier (S25) and more importantly, the GaAs photomultiplier tubes. These tubes, which were more sensitive than 1P21s also had high gain, low dark current and were very reliable. For the rst time wide-band photomultiplier tubes were available that enabled observations from U to I to be done with the same detector (Bessell 1976). Many people took advantage of this (for example Fernie 1974, Canterna 1976, Eggen 1977, Weis 1981, Sandage 1997) but it was Cousins (1976) who revolutionized broad-band photometry by providing lists of extremely precise photometric standards showing that low precision per se was not associated with broad bandpasses.
The precise southern hemisphere E and F region UBVRI standards were an excellent representation of the Johnson UBV system and related linearly to the Kron RI system.
Cousins (1980a,b) also provided photometry of more northern stars and stars covering a wider range of temperature and luminosity useful as secondary standards. Using the Cousins primary and secondary VRI standards, Landolt (1983) set up UBVRI photometric standards around the equator thus opening up the northern hemisphere to accurate UBVRI photometry. Unfortunately, the U and B lters available to Landolt (1973, 1983) were not as good a match to the Johnson system as those used by Cousins, resulting in some systematic di erences in some colors between the Landolt and Cousins UBVRI systems, especially for reddened A stars (Menzies et al. 1991).
CCDs have now almost completely replaced photomultiplier tubes as the photometric detectors of choice. This has resulted in some complications for photometry. As most of the passbands of intermediate band standard systems are de ned by interference lters, their use for o axis imaging puts more stringent conditions on both the interference lters' construction and the camera optics to ensure that the same passband and image quality is provided across the eld. In the case of broadband imaging using colored glasses, the uniformity of passbands is generally not the issue but rather the requirement to mimic the standard photoelectric passbands as closely as possible. This has generally been successfully accomplished except for the U band where large di erences between CCDs makes it very di cult to achieve a good match (Bessell, 1995 Sung & Bessell 2000).
Landolt (1992) extended his standard photometry to close groups of stars suitable for CCD imaging and Stetson (2000) provided homogeneous CCD BVRI data for more than 15,000 standard stars based on the Landolt standards.