Color in Astrophysics

(Definition)

Color in Astrophysics: Definition, Color Indices, and Physical Interpretation

In everyday language, color describes a visual sensation.

In astrophysics, color usually has a more precise quantitative meaning:

|---------------------------------------------------------------------------------------------| -astrophysical-color is-a-comparison-of-a-source’s brightness-in-two-different photometric-bands.-

(1)

The most common quantitative measure is a color index, defined as the difference between two magnitudes.

If a source has magnitudes m_X and m_Y in bands X and Y , then

|------------------| |CXY = mX − mY .| --------------------

(1)

Examples include

U − B,		(2)
B − V,		(3)
g − r,		(4)
G_BP − G_RP.		(5)

A color index is therefore a logarithmic measure of the shape of the source’s spectral energy distribution.

It is not merely a descriptive label such as “blue star” or “red star.”

Figure 1. Astrophysical color compares the amount of radiation measured through two different photometric bandpasses.

1 Why magnitude differences measure color

A magnitude in band X can be written schematically as

( ) -FX-- mX = − 2.5log10 FX,0 ,

(2)

where

F_X is the appropriately band-averaged observed flux or flux density,
F_X,0 is the reference zero-point quantity for that photometric system.

Likewise,

( ) -FY- mY = − 2.5log10 FY,0 .

(3)

Subtracting gives

m_X − m_Y =	− 2.5 log ₁₀
	+ 2.5 log ₁₀.	(4)

Thus the color index measures a flux ratio, together with the zero-point convention of the photometric system.

2 The special simplicity of an AB color

In the AB magnitude system, the reference spectrum has constant flux density per unit frequency.

For two idealized bands represented by their band-averaged flux densities,

( ) ⟨fν⟩X mX,AB = − 2.5log10 ------ , fν,0

(6)

and the same reference f_ν,0 is used for both bands.

The zero point therefore cancels:

|----------------------(------)---| mX − mY = − 2.5 log10 ⟨fν⟩X- . | -------------------------⟨fν⟩Y-----

(5)

Hence

|------------------------| |⟨f ⟩ | |--ν-X- = 10−0.4(mX −mY ).| -⟨fν⟩Y-------------------

(6)

Equation (5) is especially useful for understanding the physical meaning of color.

Real broadband photometry still requires the complete system throughput and a precise convention for the band-averaged flux density.

3 What does a smaller color index mean?

Suppose band X is at shorter wavelength than band Y .

mX − mY

(7)

becomes smaller or more negative, the source is relatively brighter in the shorter-wavelength band.

Astronomers usually describe this as bluer.

If the color index becomes larger and more positive, the source is relatively stronger in the longer-wavelength band.

Astronomers usually describe this as redder.

Thus, for a conventional blue-minus-visual color such as B − V ,

|------------------------| |smaller B − V ⇒ bluer,| -------------------------

(8)

while

|------------------------| -larger B-−-V-⇒--redder.-|

(9)

The numerical zero of a color depends on the adopted magnitude system and bandpasses.

4 Color is a property of the spectral energy distribution

A source spectrum is a function such as

f (λ) λ

(10)

fν(ν).

(11)

A photometric band samples a weighted part of that spectrum.

Schematically, a detected band signal has the form

|------∫-----------------| | | |FX ∝ fλ(λ)SX (λ)dλ, | -------------------------

(7)

where S_X(λ) represents the total system response.

Depending on detector and calibration convention, photon-counting factors and normalization terms must also be included.

The important point is that a broadband magnitude is not simply the spectrum evaluated at one wavelength.

A color compares two weighted integrals over the spectral energy distribution.

Figure 2. Hotter and cooler stellar spectra distribute their radiative output differently with wavelength, producing different photometric colors.

5 Color and stellar temperature

For a blackbody,

2hc2 1 Bλ(T ) = --5-----(hc--)----. λ exp λkT − 1

(8)

As temperature increases, the spectral energy distribution shifts toward shorter wavelengths.

Consequently, stellar color is strongly related to effective temperature.

For ordinary stellar photospheres,

|--------------------------------------------------------------------| |hotter stars are generally bluer and cooler stars are generally redder. ---------------------------------------------------------------------

(12)

However, color is not determined by temperature alone.

It is also influenced by:

surface gravity,
chemical composition,
absorption lines and molecular bands,
interstellar extinction,
circumstellar material,
unresolved companions.

Therefore a color-temperature relation is a calibration, not a universal one-to-one law.

6 Color temperature

A color temperature is the temperature of a blackbody whose flux ratio between selected wavelength regions reproduces an observed color.

If an observed ratio satisfies

FX- ≈ BX--(Tcol), FY BY (Tcol)

(13)

then T_col is the inferred color temperature.

For a real star,

T col

(14)

need not be identical to

T . eff

(15)

The distinction matters because stellar spectra contain wavelength-dependent opacity, spectral lines, and deviations from an ideal blackbody.

Figure 3. A color temperature matches a selected spectral slope or flux ratio to a blackbody and need not equal the stellar effective temperature exactly.

7 Common stellar color indices

Several systems are widely used.

7.1 Johnson-Cousins system

Examples include

U − B, B − V, V − R, R − I.

(16)

The classic B − V index compares blue and visual-band magnitudes.

7.2 SDSS-like optical systems

A common sequence is

u − g, g − r, r − i, i − z.

(17)

7.3 Gaia

A widely used Gaia color is

|------------| -GBP-−--GRP.-|

(9)

This compares Gaia’s broad blue-photometer and red-photometer measurements.

Colors from different photometric systems should not be treated as numerically interchangeable.

8 Observed color and intrinsic color

The color directly measured by the observer is the observed color.

The color the source would have in the absence of intervening extinction is the intrinsic color.

For example,

(B − V )

(18)

is the observed color, while

(B − V)0

(19)

is the intrinsic color.

The difference is the color excess:

----------------------------------- | | -E-(B--−-V-) =-(B-−-V-) −-(B-−-V-)0.|

(10)

More generally,

|----------------------------------| E-(X--−-Y-) =-(X-−-Y-)-−-(X-−--Y)0.-

(11)

9 Why extinction changes color

Let the intrinsic magnitudes be

mX,0, mY,0.

(20)

Interstellar extinction adds

AX , AY ,

(21)

so that

m_X	= m_X,0 + A_X,	(22)
m_Y	= m_Y,0 + A_Y.	(23)

Subtracting,

m_X − m_Y

= (m_X,0 − m_Y,0) + (A_X − A_Y).

(24)

Therefore

|-----------------------| E (X − Y) = AX − AY . | -------------------------

(12)

For ordinary interstellar dust at optical wavelengths, shorter wavelengths are usually extinguished more strongly.

Thus dust generally makes an object appear redder.

Figure 4. Wavelength-dependent extinction suppresses shorter-wavelength light more strongly and shifts many optical colors toward redder values.

10 Dereddening a color

Once the color excess is known,

|----------------------------------| (X − Y ) = (X − Y ) − E (X − Y ).| ---------0--------------------------

(13)

Similarly, if the extinction in one band is known,

mX,0 = mX − AX .

(25)

Color correction and magnitude correction are related but are not the same operation.

11 The extinction parameter $RV$

A commonly used optical parameter is

|------------------| | ----AV---- | |RV = E (B − V) .| -------------------

(14)

A value near

RV ≈ 3.1

(26)

is often used as an illustrative diffuse Milky Way value, but real sightlines can differ substantially.

One should not treat R_V = 3.1 as universal.

12 Color-color diagrams

A color-color diagram plots one color index against another.

For example,

U − B

(27)

may be plotted against

B − V.

(28)

Such diagrams are useful because:

stellar temperature moves stars along characteristic loci,
reddening often moves sources in a different direction,
unusual spectra can be identified,
quasars, white dwarfs, cool stars, and other populations can separate in color space.

Figure 5. A color-color diagram compares two independent spectral slopes and can help separate temperature, reddening, and unusual spectral energy distributions.

13 Color-magnitude diagrams

A color-magnitude diagram places a color index on one axis and a magnitude or absolute magnitude on the other.

For a stellar population, this produces recognizable structures such as:

the main sequence,
giant branches,
white-dwarf sequences,
binary sequences,
turnoff regions.

A color-magnitude diagram is observationally related to the Hertzsprung-Russell diagram.

Color provides a temperature-sensitive horizontal coordinate, while magnitude provides a luminosity-sensitive vertical coordinate.

Figure 6. Color-magnitude diagrams use a photometric color as a temperature-sensitive coordinate and a magnitude as a luminosity-sensitive coordinate.

14 Color and a power-law spectrum

Suppose over a limited range the spectrum can be approximated by

fν ∝ να.

(15)

For two idealized AB bands centered at frequencies ν_X and ν_Y,

( )α fν,X- νX- f = ν . ν,Y Y

(29)

Using Equation (5),

( ) νX- mX − mY = − 2.5α log10 ν . Y

(30)

Thus

|------------------------| |α = − ----mX--−-mY---- .| -------2.5-log10(νX-∕νY)--|

(16)

A color index can therefore be interpreted as a coarse measurement of spectral slope.

15 Color uncertainty

C = mX − mY

(31)

and the magnitude errors are independent, then

|-----------------| | ∘ -2-----2- | σC = σX + σY. | -------------------

(17)

If the two measurements have covariance,

---------------------------------- | 2 2 2 | σ-C-=-σX-+--σY-−-2-Cov-(mX--,mY-).-

(18)

This is important because the same calibration errors can affect both bands and make the errors correlated.

16 Color of an unresolved binary

Fluxes add before magnitudes are calculated.

For two unresolved components,

F_X,tot	= F_X,1 + F_X,2,	(32)
F_Y,tot	= F_Y,1 + F_Y,2.	(33)

Therefore the combined color is

C_tot =	− 2.5 log ₁₀
	+ 2.5 log ₁₀.	(19)

The combined color is generally not the arithmetic average of the two component colors.

Figure 7. In an unresolved binary the component fluxes add separately in each band before the combined color index is calculated.

17 Equal-component binary

If the two unresolved stars have identical spectra, then both band fluxes double:

F_X,tot	= 2F_X,	(34)
F_Y,tot	= 2F_Y.	(35)

Their ratio is unchanged:

FX,tot FX ------ = ---. FY,tot FY

(36)

Therefore

|-----------------| Ctot = Ccomponent. | -------------------

(20)

The unresolved system is brighter but has the same color.

This is why equal-component binaries can form a sequence above the single-star main sequence in a color-magnitude diagram without shifting strongly in color.

18 Color variations during eclipses

An eclipsing binary whose two stars have different temperatures can change color during eclipse.

If the hotter star is partially or completely hidden, the combined light can become redder.

If the cooler star is hidden, the combined light can become bluer.

Thus time-dependent color provides information about the relative spectral energy distributions of the components.

This is a direct connection between color photometry and the eclipsing binary geometry developed elsewhere in the binary star sequence.

19 Photometric calibration matters

A scientifically meaningful color requires calibrated measurements.

Important corrections include:

detector bias and dark signal,
flat-field response,
atmospheric extinction for ground-based data,
zero-point calibration,
color terms between an instrumental and standard system,
aperture or point-spread-function consistency,
time variability between exposures taken in different filters.

An instrumental color is not automatically a standard-system color.

20 Why filter definitions matter

A color index belongs to a particular photometric system.

The designation

B − V

(37)

does not mean “flux at exactly one blue wavelength minus flux at exactly one visual wavelength.”

It represents measurements through specified bandpasses, detectors, and calibration conventions.

Changes in filter throughput can produce systematic color transformations.

This is one reason standard-star observations and synthetic photometry are important.

21 Common mistakes

Treating astrophysical color as only a visual RGB description.
Forgetting that color is a magnitude difference and therefore logarithmic.
Assuming a larger blue-minus-red color means a bluer source.
Mixing colors from different photometric systems as if they were identical.
Converting a broadband magnitude to flux using only a nominal central wavelength without regard to the bandpass convention.
Assuming color uniquely determines effective temperature.
Ignoring interstellar reddening.
Treating observed color as intrinsic color.
Assuming R_V = 3.1 for every sightline.
Averaging component colors arithmetically for an unresolved binary.
Forgetting that fluxes add before magnitudes and colors are calculated.
Ignoring covariance when propagating color uncertainty.
Comparing non-simultaneous colors of a rapidly variable source without caution.

22 Connections to BIN02 and luminosity

BIN02 introduced observational fluxes, magnitudes, and broadband measurements.

The present definition article isolates the specific concept of astrophysical color.

The luminosity article addresses the total emitted Power.

Color instead asks how that power or observed flux is distributed with wavelength.

The two concepts therefore answer different questions:

|------------------------------| |luminosity: how much power? | -------------------------------

(38)

|------------------------------------------------------| -color: how-is-the-radiation--distributed-between--bands?-|

(39)

Together, luminosity and color provide much of the observational foundation for stellar classification, temperature estimation, and color-magnitude diagrams.

23 Summary

The fundamental color index is

|C----=-m---−--m--.| --XY------X-----Y---

(40)

In an idealized AB comparison,

|----------------------------| | ( ) | |CXY = − 2.5log10 ⟨fν⟩X- .| --------------------⟨fν⟩Y----

(41)

For a shorter-wavelength band X and longer-wavelength band Y , a smaller color generally indicates a bluer spectral energy distribution.

The intrinsic and observed colors are connected by

|----------------------------------------------| E-(X--−-Y-) =-(X--−-Y-) −-(X-−-Y-)0-=-AX--−-AY-.-

(42)

Color is strongly sensitive to temperature, but also to extinction, surface gravity, composition, spectral features, and unresolved companions.

For an unresolved binary, fluxes must be added separately in each band before the system color is calculated.

References

[1] B. W. Carroll and D. A. Ostlie, An Introduction to Modern Astrophysics, 2nd ed., Cambridge University Press, 2017.

[2] M. S. Bessell, Standard Photometric Systems, Annual Review of Astronomy and Astrophysics, 43, 293–336, 2005.

[3] M. S. Bessell, F. Castelli, and B. Plez, Model atmospheres broad-band colors, bolometric corrections and temperature calibrations for O–M stars, Astronomy and Astrophysics, 333, 231–250, 1998.

[4] D. W. Hogg et al., Astronomical magnitudes and fluxes, pedagogical notes on astronomical photometric conventions.

"Color in Astrophysics" is owned by bloftin.
(view preamble)

Color in Astrophysics