Photometric Passbands and System Response: Throughput, Weighting, and Synthetic Photometry

Photometric Passbands and System Response:
Throughput, Weighting, and Synthetic Photometry

A photometric observation does not usually measure the spectrum at one wavelength.

Instead, a telescope, filter, detector, and often the atmosphere together admit or detect a finite range of wavelengths with wavelength-dependent efficiency.

That wavelength-dependent sensitivity is described by a passband, response curve, or system throughput.

If the source spectral flux density is

(1)

then the measured band signal is constructed from a weighted integral across wavelength.

A schematic energy-weighted form is

(2)

where X labels the photometric band and S_X(λ) describes the system response.

For a photon-counting detector, an additional photon-number weighting appears unless it has already been absorbed into the published response curve.

This article develops what a passband means, what belongs inside a system response, why different detector conventions produce different weighting factors, and how a source spectrum is converted into a synthetic photometric measurement.

1 Why a photometric band is not one wavelength

A broadband filter such as B, V , g, r, or G accepts light over a finite spectral interval.

Even if one associates a band with a nominal wavelength such as

(3)

the detector generally receives photons from a broad neighborhood around that wavelength.

Therefore, in general,

(4)

The measured signal depends on:

• the source spectrum throughout the band;
• the filter transmission;
• the telescope and instrument optics;
• detector sensitivity;
• atmosphere, when relevant;
• the precise calibration and weighting convention.

A representative or effective wavelength can be useful, but it does not replace the full passband integral.

2 Transmission, response, and throughput

The words transmission, response, and throughput are related but should not automatically be treated as identical.

A filter transmission curve,

(5)

describes the fraction of incident radiation transmitted by the filter.

A detector quantum efficiency,

(6)

describes the fraction of incident photons that produce detected charge carriers or recorded events, under an appropriate detector model.

A telescope or optical train has its own wavelength-dependent efficiency,

(7)

For a ground-based observation, atmospheric transmission may contribute

(8)

A simple total throughput can therefore be represented schematically by

(9)

Additional components can be included when necessary.

The product form reflects a sequence of efficiencies: radiation must survive or be detected by every stage of the system.

3 Absolute versus relative response

A response curve may be calibrated absolutely, or it may be normalized to an arbitrary peak such as

(10)

For many synthetic-photometry calculations, only the shape of the response matters because an overall multiplicative constant cancels when the signal is normalized or compared with a reference source.

If

(11)

then

(12)

A corresponding zero point or normalization can absorb C.

However, an absolute detector count-rate prediction requires an absolute throughput calibration together with collecting area, exposure time, and detector gain or event efficiency as appropriate.

4 The weighted-integral picture

Suppose the wavelength axis is divided into narrow bins.

In bin i, the source contributes approximately

(13)

If the system response in that bin is

(14)

the weighted contribution is approximately

(15)

Adding the bins gives

(16)

Taking the continuum limit gives Equation (1):

(17)

Thus the passband measurement is a continuous weighted sum over the spectrum.

5 Energy-measuring and photon-counting detectors

Equation (5) is a useful schematic expression, but the correct weighting depends on what the detector measures.

An idealized energy-measuring detector responds in proportion to radiant energy deposited.

For such a convention,

(18)

A photon-counting detector responds approximately to the number of detected photons.

A photon at wavelength λ has energy

(19)

The incident photon spectral rate per detector area is therefore proportional to

(20)

Thus a photon-counting signal has the schematic form

(21)

The factor

(22)

converts radiant energy into photon number.

6 Why published response curves must be read carefully

Some published passband curves represent physical throughput

(23)

before photon-number weighting.

Other published response functions are defined so that detector behavior is already incorporated into

(24)

Therefore one must not automatically multiply every published passband by an additional factor of λ.

The correct procedure is:

1. identify how the published response function is defined;
2. determine whether it is an energy response, photon response, or dimensionless throughput;
3. use the integration convention specified for that system;
4. use the corresponding photometric zero point.

This convention dependence is one reason that synthetic photometry should be based on the documented system response rather than only on a filter name.

7 Frequency-space form

The same measurement can be written in frequency coordinates.

For an energy-weighted response,

(25)

For photon counting,

(26)

The wavelength and frequency expressions are equivalent when the spectral densities, response functions, and integration measures are transformed consistently.

8 Normalized band averages

Sometimes one wants a representative spectral flux density rather than an unnormalized integrated signal.

For an energy-weighted wavelength response, a simple normalized mean is

(27)

For a photon-counting response described by a physical throughput T_X(λ),

(28)

These are weighted averages.

Different photometric systems can define band averages differently, so the response convention remains part of the definition.

9 A simple top-hat passband

An ideal top-hat response is

(29)

For energy weighting,

(30)

If f_λ is constant,

(31)

then

(32)

This idealized example makes the role of finite bandwidth explicit.

10 Example 1: energy-weighted top-hat band

Suppose

λ1	= 500 nm,	(33)
λ2	= 600 nm,	(34)

and

(35)

is constant across the band.

Then

Fband	= fλ(100 nm)	(36)
	= 2.00 × 10−9 W m−2.	(37)

The band flux depends on both the spectral density and the width of the accepted wavelength interval.

11 Basic passband descriptors

Several different numbers are used to summarize a passband.

They do not all mean the same thing.

Useful descriptors include:

• central wavelength;
• wavelength of maximum response;
• full width at half maximum;
• equivalent rectangular bandwidth;
• effective wavelength;
• pivot wavelength;
• isophotal wavelength.

No single number contains all of the information in the full response curve.

12 Wavelength of maximum response

The wavelength of maximum response is simply the location at which

(38)

is largest.

If the response has multiple peaks or a flat plateau, even this quantity may not be unique or especially informative.

It depends only on the response curve, not on the source spectrum.

13 Full width at half maximum

If the response has a single well-defined peak, the full width at half maximum is the wavelength separation between the two points at which the response equals half the peak value.

Symbolically,

(39)

The full width at half maximum is useful for simple, roughly bell-shaped passbands but can be misleading for asymmetric or multi-peaked responses.

14 Equivalent rectangular bandwidth

A useful response-only width is the equivalent rectangular bandwidth,

(40)

If the response is normalized so that

(41)

then

(42)

This is the width of a rectangle with height equal to the peak response and area equal to the area under the actual response curve.

15 Effective wavelength

An effective wavelength is intended to represent the wavelength at which a particular source and response combination is concentrated.

A common energy-weighted form is

(43)

For photon weighting, the corresponding source weighting changes because the detected photon rate contains an additional factor of λ.

For example,

(44)

The key conceptual point is:

(45)

A red star and a blue star observed through the same broad filter can therefore have different effective wavelengths.

16 Pivot wavelength

The pivot wavelength is designed to characterize the passband itself rather than a particular source spectrum.

For a dimensionless throughput T_X(λ), the commonly used definition is

(46)

Unlike an effective wavelength,

(47)

It is especially useful when converting consistently between appropriately defined band-averaged f_λ and f_ν quantities.

17 Example 2: pivot wavelength of an ideal top-hat band

For

(48)

between λ₁ and λ₂,

(49)

Therefore

(50)

For

λ1	= 500 nm,	(51)
λ2	= 600 nm,	(52)

(53)

This value is close to, but not defined simply as, the arithmetic midpoint of the band.

18 Isophotal wavelength

An isophotal wavelength is defined so that the monochromatic spectral flux density at that wavelength equals a suitable band-averaged spectral flux density.

For an energy-weighted convention, one may define λ_iso through

(54)

Because the left side depends on the source spectrum, the isophotal wavelength is generally source dependent.

For some spectra and complex passbands, there can be more than one wavelength satisfying the equality.

Therefore isophotal wavelength should not be confused with a universal center of the filter.

19 Two sources can agree at one wavelength and disagree in the band

Suppose two source spectra satisfy

(55)

at the nominal band center.

If one spectrum rises toward the red while the other falls, then generally

(56)

Thus equality at one wavelength does not imply equality of broadband photometric signal.

20 Synthetic photometry

Synthetic photometry predicts a photometric measurement from a calibrated spectrum and a documented system response.

The general procedure is:

1. obtain or model the source spectral flux density;
2. obtain the appropriate system response curve;
3. place the spectrum and response on consistent wavelength or frequency coordinates;
4. use the detector weighting convention associated with that response;
5. integrate the weighted spectrum;
6. compare the result with the reference spectrum or zero point of the photometric system.

Symbolically, define a detector signal functional

(57)

where W_X includes the response and the detector-specific weighting.

For example,

(58)

for a simple energy-weighted convention, while

(59)

for a simple photon-counting convention.

21 From a synthetic signal to a magnitude

A magnitude is logarithmic.

A general calibrated form can be written as

(60)

where ZP_X is the zero point appropriate to the signal convention and units.

Equivalently, relative to a reference source,

(61)

The reference spectrum and response convention are part of the definition of the photometric system.

22 AB, ST, and Vega-based systems

Different magnitude systems use different reference conventions.

AB system

The AB system is tied to frequency spectral flux density.

Its monochromatic definition can be expressed relative to approximately

(62)

Broadband AB magnitudes require the appropriate passband-weighted treatment rather than evaluation at one arbitrary wavelength.

ST system

The ST system is formulated so that a source with constant

(63)

has constant ST magnitude under the corresponding convention.

Vega-based systems

Vega-based systems use the spectrum of Vega, or a defined realization of that reference, to establish photometric zero points.

Therefore the same instrumental bandpass can produce different numerical magnitude zero points depending on the adopted magnitude system.

23 Color indices

An astronomical color is a difference of magnitudes:

(64)

Since each magnitude is derived from a different weighted passband integral,

(65)

the color measures how the source spectrum is distributed relative to the two response functions.

This explains why astronomical color is not equivalent to assigning a single visible wavelength to the source.

24 Atmospheric response

For ground-based photometry, the atmosphere can be an important part of the effective system response.

Transmission depends on:

• wavelength;
• airmass;
• molecular absorption;
• aerosol content;
• water vapor;
• time and observing conditions.

Thus one may distinguish:

(66)

from

(67)

A photometric standard system may then require transformations from the natural instrumental system to a defined standard system.

25 Red leaks and out-of-band response

A response curve need not vanish perfectly outside its nominal wavelength range.

A small unintended transmission feature at longer wavelengths is often called a red leak.

Likewise, unwanted sensitivity can occur on the blue side or elsewhere.

Even a weak leak can matter if the source spectrum is much brighter in the leak region than inside the nominal passband.

Therefore

(68)

can produce a non-negligible signal.

This is another reason to use the full measured system response rather than an idealized central wavelength and width.

26 Example 3: same central flux, different broadband signals

Consider an ideal top-hat band from

(69)

to

(70)

Let the nominal center be

(71)

Suppose two spectra have the same value

(72)

at 550 nm.

Let

(73)

and

(74)

If the top-hat band is perfectly symmetric about 550 nm, the linear positive and negative slope contributions cancel in an energy-weighted integral:

(75)

This special cancellation is a useful warning: the outcome depends on both the source spectrum and the response shape.

For an asymmetric real passband, or for nonlinear spectral curvature, equality at the nominal center does not generally imply equal integrated signal.

For photon counting, the extra factor of λ can also break the symmetry even for this simple linear example.

27 Example 4: photon weighting shifts the mean wavelength

Consider the same ideal top-hat band from 500 to 600 nm and a source with constant f_λ.

For energy weighting, the effective wavelength is the arithmetic mean:

(76)

For photon weighting,

(77)

Therefore

(78)

Longer-wavelength photons carry less energy individually, so a fixed spectral energy flux corresponds to more photons at longer wavelength.

28 Calibration and zero points

A real photometric measurement does not end with the raw integral.

Calibration must account for quantities such as:

• detector gain;
• collecting area;
• exposure time;
• aperture corrections;
• atmospheric extinction;
• instrumental zero point;
• standard-star calibration;
• time dependence of system throughput.

The response integral tells how the source spectrum maps into an instrumental signal.

The calibration procedure then maps that instrumental signal into a reproducible photometric quantity.

29 Common mistakes

1. Treating a broadband measurement as the spectrum evaluated at one wavelength.
2. Using only the filter transmission while ignoring optics, detector sensitivity, or atmosphere when the system response requires them.
3. Multiplying by an extra factor of λ when the published response already includes photon-counting weighting.
4. Omitting the photon-energy factor when using a physical throughput with a photon-counting detector.
5. Confusing a relative response curve with an absolute detector efficiency.
6. Assuming effective wavelength is a source-independent property of the filter.
7. Confusing effective wavelength with pivot wavelength.
8. Treating full width at half maximum as a complete description of a strongly asymmetric passband.
9. Ignoring out-of-band leaks.
10. Mixing f_λ and f_ν without the appropriate Jacobian transformation.
11. Using a magnitude zero point that belongs to a different response convention.
12. Treating AB, ST, and Vega-based magnitudes as though they used the same reference spectrum.
13. Comparing synthetic and observed photometry without confirming that the same passband definition is being used.
14. Forgetting that ground-based atmospheric transmission can be part of the natural passband.

30 Connections to other PhysicsLibrary articles

The article Electromagnetic Waves and Wavelength establishes wavelength and frequency as spectral coordinates:

(79)

The article Spectral Flux Density defines

(80)

and

(81)

The present article adds the instrument response and produces the weighted band signal:

(82)

The article Color in Astrophysics then compares two such band measurements through

(83)

Thus the full conceptual chain is

(84)

31 Summary

A photometric passband describes wavelength-dependent sensitivity over a finite spectral interval.

A total system throughput can combine several multiplicative components:

(85)

For an energy-weighted convention,

(86)

For a photon-counting convention based on a physical throughput,

(87)

An effective wavelength is generally source dependent, while the pivot wavelength

(88)

is a property of the response curve itself.

Synthetic photometry can be summarized by

(89)

followed by a calibrated magnitude relation such as

(90)

The central lesson is that a broadband photometric measurement is an integral over the source spectrum and the complete system response, not a monochromatic measurement at a single wavelength.

References

References