Observational Foundations for Binary Stars

Observational Foundations for Binary Stars

Binary star astrophysics begins with observables, not orbital elements.

A detector records photons. Those photons are converted into quantities such as

• count rate,
• flux,
• magnitude,
• color,
• spectrum,
• sky position,
• observation time.

Those observables must then be calibrated, assigned uncertainties, and attached to a reference system before they can be used to infer stellar or orbital parameters.

For binary stars this bookkeeping is especially important because two stars can contribute to the same image pixel, spectrum, or photometric aperture.

The central observational chain is

(1)

1 What a detector measures

An astronomical detector does not directly measure stellar mass, radius, luminosity, or orbital inclination.

A photon-counting detector measures a number of detected photoelectrons or counts over an exposure.

A simplified count rate can be written

(2)

where N_det is the number of detected photoelectrons and t_exp is the exposure time.

The relationship between incident energy flux and detector counts depends on

• telescope collecting area,
• filter transmission,
• detector quantum efficiency,
• atmospheric transmission for ground-based data,
• wavelength dependence of the instrument,
• aperture and background subtraction.

Calibration is the process that connects instrumental measurements to physically interpretable observables.

2 Luminosity and flux

The luminosity L of a star is the total power it radiates.

Its SI unit is

(3)

If radiation is emitted isotropically, the bolometric flux measured at distance d is

(4)

Flux has SI units

(5)

This inverse-square law is fundamental, but an astronomical catalog often reports a passband flux rather than a bolometric flux.

The measured flux therefore depends on the instrument and wavelength range.

3 Bandpass flux

A real detector has a wavelength-dependent response.

If F_λ is the spectral flux density and S(λ) is the total instrumental response, a simplified bandpass measurement has the structure

(6)

The precise calibration depends on whether one works with energy-weighted or photon-weighted response and on the adopted magnitude system.

The important point is

(7)

A V magnitude, Gaia G magnitude, and TESS magnitude are not interchangeable.

4 The astronomical magnitude scale

Magnitudes encode flux ratios logarithmically.

For two measurements in the same calibrated passband,

(8)

Equivalently,

(9)

A smaller magnitude corresponds to a larger flux.

A difference of five magnitudes corresponds to a factor of one hundred in flux:

(10)

5 Example 1: flux ratio from magnitude difference

Suppose two stars differ by

(11)

Then

	= 10−0.4(2.0)	(12)
	= 10−0.8	(13)
	≈ 0.1585.	(14)

Thus

(15)

A two-magnitude difference is therefore much larger than a factor of two in flux.

6 Combined light from an unresolved binary

If both stars fall inside the same aperture, fluxes add linearly:

(16)

Magnitudes do not add.

Let the component magnitudes in the same band be m₁ and m₂.

Using an arbitrary common zero-point flux F₀,

(17)

and

(18)

Therefore the combined magnitude is

(19)

when the same magnitude zero point is used for both stars.

7 Equal stars and the 0.7526 magnitude result

If two unresolved components have equal flux F,

(20)

The combined system is brighter by

Δm	= −2.5 log 10(2)	(21)
	≈−0.7526.	(22)

Therefore

(23)

This is a useful binary star sanity check.

An unresolved equal-luminosity binary lies about 0.75 magnitude above the corresponding single-star sequence if distance and extinction are the same.

8 Flux ratios and eclipse depth

Suppose star 2 contributes flux F₂ and is completely hidden while star 1 remains visible.

Out of eclipse,

(24)

During the total eclipse of star 2,

(25)

The fractional eclipse depth is

(26)

Thus a photometric eclipse can directly constrain a component flux ratio when the eclipse geometry is sufficiently simple.

9 Color indices

A color index is a magnitude difference between two passbands.

For example,

(27)

Gaia commonly provides

(28)

Color is related to the shape of the spectral energy distribution and therefore to effective temperature, extinction, surface gravity, and composition.

For an unresolved binary, the system color is produced by the sum of the component fluxes in each band.

It is not generally the arithmetic average of the component colors.

10 A binary can change color during eclipse

If two stars have different temperatures, their spectral energy distributions differ.

During an eclipse, the relative contribution of each star changes.

Therefore the observed color can change with orbital phase.

Multi-band eclipse photometry can therefore constrain

• surface-brightness ratio,
• temperature ratio,
• third Light,
• wavelength-dependent limb darkening.

This becomes important later in the eclipsing-binary sequence.

11 Extinction and reddening

Interstellar dust removes and redistributes light.

A simple passband extinction relation is

(29)

where m₀ is the unextinguished magnitude and A_λ is the extinction in that band.

Because extinction depends on wavelength, dust also changes colors.

A common color excess is

(30)

Binary stars do not avoid extinction merely because the two components are at the same distance.

The advantage is that both components usually experience nearly the same foreground extinction, which can simplify differential comparisons.

12 Right ascension and declination

The standard equatorial coordinates are right ascension and declination.

Right ascension is analogous to longitude on the Celestial Sphere.

Declination is analogous to latitude.

Right ascension is often expressed in hours, minutes, and seconds:

(31)

Thus

(32)

Declination is measured in degrees north or south of the Celestial Equator.

A catalog position is incomplete unless the coordinate frame and reference epoch are understood.

13 Small angular offsets on the sky

For two nearby sky positions with small separations,

(33)

and

(34)

The tangent-plane east-west offset is approximately

(35)

while the north-south offset is

(36)

The small-angle separation is

(37)

This is useful for resolved binary astrometry, but great-circle formulas should be used when separations are not small.

14 Position angle

For resolved binary stars, a relative position is often described by separation

(38)

and position angle

(39)

Position angle is conventionally measured from north through east on the sky.

With local offsets

(40)

toward east and

(41)

toward north,

(42)

The two-argument arctangent is important because it preserves the correct quadrant.

15 Time is a physical coordinate in binary star astronomy

For a binary star, the observation time is part of the measurement.

Orbital phase, eclipse timing, radial velocity, and astrometric position all depend on time.

An observation should therefore be thought of as a pair:

(43)

Poor time bookkeeping can create a false phase shift even when the flux or velocity measurement itself is perfect.

16 UTC, TT, TDB, and barycentric timing

Several time standards appear in astronomical work.

• UTC is the civil time scale used for timestamps and includes leap seconds.
• TT is a uniform terrestrial time scale used in ephemeris-related calculations.
• TDB is a relativistic time coordinate convenient for Solar-system barycentric dynamics.
• Barycentric timestamps account for the changing light-travel time between the observatory and the Solar-system barycenter.

For precision binary star timing, simply recording a UTC date is often not sufficient.

17 Julian Date and Modified Julian Date

Julian Date is a continuous day count.

Modified Julian Date is defined by

(44)

The offset places the MJD day boundary at midnight rather than noon and reduces the numerical size of the date.

The time standard must still be specified.

For example,

(45)

and

(46)

are not identical quantities.

18 Barycentric Julian Date

A common precision-timing quantity is Barycentric Julian Date.

In binary star photometry one often encounters

(47)

It combines a barycentric light-travel correction with a uniform relativistic time scale.

The exact transformation depends on

• observatory location,
• target coordinates,
• Solar-system ephemeris,
• original time standard.

Therefore the phrase “Julian Date” alone can be inadequate for high-precision eclipse timing.

19 TESS time

TESS light-curve products commonly use a truncated barycentric time coordinate.

A frequently used convention is

(48)

The FITS header and data-product documentation should always be checked rather than assuming an offset.

20 Cadence and exposure time

An astronomical time series consists of finite exposures, not infinitely short samples.

Let

(49)

be the exposure time and

(50)

the cadence between reported samples.

If an eclipse ingress lasts only a few exposure times, finite integration can smear its shape.

This matters because ingress and egress durations constrain stellar radii and inclination.

A forward model should ideally be integrated over the same exposure duration as the data.

21 Uncertainty and signal-to-noise ratio

A measurement without an uncertainty has limited scientific meaning.

A simple signal-to-noise ratio is

(51)

where S is a measured signal and σ_S is its standard uncertainty.

For ideal Poisson counting statistics with negligible background,

(52)

so

(53)

Real astronomical data also contain background noise, read noise, calibration uncertainty, contamination, and astrophysical variability.

22 Magnitude uncertainty from flux uncertainty

For

(54)

a first-order differential gives

(55)

Therefore the approximate magnitude uncertainty is

(56)

Since

(57)

(58)

This approximation is best when the fractional flux uncertainty is small.

23 Why many catalogs prefer flux-space errors

A magnitude is a nonlinear logarithmic transformation of flux.

A symmetric Gaussian error in flux becomes asymmetric after transformation into magnitude.

For that reason, modern catalogs can report mean flux and flux error as the fundamental measurement while providing a magnitude derived from that mean flux.

This is the approach used for Gaia mean broad-band photometry.

For rigorous uncertainty propagation, working in flux space is often preferable.

24 Spectra

A spectrum records flux as a function of wavelength or frequency.

A simplified one-dimensional spectrum is

(59)

A binary spectrum can contain lines from one star, both stars, circumstellar gas, or blended combinations of all of these.

As orbital phase changes, photospheric lines can shift in wavelength because of the line-of-sight velocity of each component.

This leads to the spectroscopic-binary method developed later in the series.

25 Spectral resolving power

The resolving power of a spectrograph is

(60)

At wavelength

(61)

a spectrograph with

(62)

has characteristic resolution

Δλ	=	(63)
	= 0.025 nm.	(64)

An equivalent velocity scale is approximately

(65)

This is a resolution scale, not automatically the final radial-velocity precision.

26 Doppler shift preview

For nonrelativistic radial speed,

(66)

the wavelength shift is approximately

(67)

Thus

(68)

The sign convention must be stated.

A common convention assigns positive radial velocity to recession.

Full radial-velocity extraction is postponed until BIN07 and BIN20.

27 The FITS format

The Flexible Image Transport System, or FITS, is a standard astronomical data format.

A FITS file can contain multiple header-data units, usually abbreviated HDUs.

A typical file can include

• a primary HDU,
• one or more image HDUs,
• one or more binary-table HDUs.

Each HDU contains a header describing the data and, when applicable, a data block.

Important header metadata can include

• target identifier,
• coordinate information,
• time system,
• units,
• instrument,
• exposure time,
• processing level,
• quality or provenance keywords.

28 Binary tables in time-series FITS files

A light-curve FITS product often stores one observation per row in a binary table.

Typical columns can include

• time,
• flux,
• flux uncertainty,
• background,
• centroid position,
• quality flags.

Mission-specific names must be read from the product documentation.

For TESS, commonly encountered columns include

(69)

and

(70)

The former is based on simple aperture photometry.

The latter applies corrections intended to reduce common instrumental systematics.

29 Quality flags are data

A quality flag is not decorative metadata.

It tells the user that a sample may have been affected by events such as

• spacecraft pointing changes,
• cosmic rays,
• momentum dumps,
• scattered light,
• detector anomalies.

A reproducible analysis must state how quality flags were handled.

Different scientific goals can justify different masks.

30 Gaia mean photometry

Gaia DR3 reports mean fluxes and flux uncertainties in its broad photometric bands, including G, G_BP , and G_RP .

The catalog also provides corresponding mean magnitudes.

A useful observational lesson is that the magnitude values are derived from mean fluxes.

The archive documentation explicitly warns that a symmetric flux uncertainty does not transform into a single symmetric magnitude uncertainty.

This is a concrete example of why one should understand the native measurement space.

31 TESS SAP and PDCSAP flux

TESS light-curve files available from MAST commonly contain both SAP and PDCSAP flux.

SAP flux is the aperture-summed signal after the pipeline’s basic photometric extraction.

PDCSAP flux attempts to remove instrumental trends using cotrending information.

Neither column should be treated as automatically perfect.

For eclipsing binary work, detrending choices can influence eclipse depth, baseline shape, and inferred parameters.

The rawer SAP series and the corrected PDCSAP series are therefore both scientifically useful.

32 Blending and third light

A photometric aperture can contain flux from stars other than the two modeled binary components.

Let the contaminating flux be

(71)

Then

(72)

The contaminating term is often called third light even when it comes from an unrelated background star.

Third light dilutes eclipse depth.

A shallow observed eclipse can therefore be caused by

• a genuinely small blocked flux fraction,
• grazing geometry,
• a faint eclipsed star,
• contaminating third light,
• some combination of these effects.

33 Normalization

Time-series photometry is often normalized by a representative out-of-eclipse flux.

For example,

(73)

Then the out-of-eclipse baseline is near

(74)

A differential magnitude can be defined by

(75)

Normalization is a mathematical convenience.

It should not erase the distinction between calibrated absolute photometry and relative light-curve shape.

34 From archive query to reproducible measurement

A defensible archive workflow should preserve:

1. archive name,
2. target identifier,
3. query or cone-search region,
4. retrieval date,
5. data release,
6. file identifier,
7. units,
8. quality-mask rule,
9. software version,
10. processing choices.

A result is not fully reproducible if only the final plot is saved.

35 Live data and frozen data

The BIN series uses two complementary ideas.

A live query teaches how to interact with a professional archive.

A frozen extract preserves a small subset of the data so that the lesson still works if

• an archive interface changes,
• a catalog is updated,
• network access is unavailable,
• a source is reprocessed.

The frozen data are not meant to replace the live archive.

They provide a reproducible checkpoint.

36 Example 2: eclipse depth in flux and magnitudes

Suppose the normalized flux outside eclipse is

(76)

and at mid-eclipse is

(77)

The fractional depth is

(78)

The differential magnitude at mid-eclipse is

Δm	= −2.5 log 10(0.80)	(79)
	≈ 0.242.	(80)

Thus

(81)

A 20% flux loss is not a 0.20 magnitude change because magnitudes are logarithmic.

37 Example 3: timing offset

Suppose a TESS measurement has

(82)

Using

(83)

the corresponding barycentric Julian Date is

(84)

The numerical offset must be restored before comparing to a literature ephemeris expressed in full BJD.

38 Example 4: photon-limited uncertainty

If a detector records

(85)

photoelectrons and photon noise dominates,

(86)

Therefore

(87)

The approximate magnitude uncertainty is

σm	≈	(88)
	≈ 0.00109 mag.	(89)

This is about one millimagnitude.

Real observations can have a larger uncertainty because photon noise is rarely the only noise source.

39 A practical observational checklist

Before interpreting an archive measurement, ask:

1. What quantity did the detector actually measure?
2. What calibration produced the reported flux or magnitude?
3. What passband is being used?
4. Is the binary resolved or blended?
5. What coordinate frame and epoch are attached to the position?
6. What time standard and numerical offset are attached to the timestamp?
7. What are the uncertainty and quality flags?
8. Is the measurement raw, calibrated, or detrended?
9. Is there contamination from neighboring sources?
10. Can the exact archive query and processing steps be reproduced?

40 Common mistakes

1. Adding magnitudes instead of fluxes.
2. Comparing magnitudes from different passbands as though they measured the same quantity.
3. Forgetting that smaller astronomical magnitude means brighter flux.
4. Treating an unresolved binary magnitude as the magnitude of one component.
5. Averaging component colors instead of adding component fluxes band by band.
6. Using right-ascension differences without the cos δ factor in a small tangent-plane calculation.
7. Omitting coordinate frame or reference epoch.
8. Writing “JD” without checking the underlying time standard.
9. Comparing BTJD directly with a full BJD ephemeris without restoring the mission offset.
10. Ignoring finite exposure time near sharp eclipse features.
11. Assuming photon noise is the only uncertainty.
12. Propagating a symmetric flux error into a symmetric magnitude error when the fractional error is large.
13. Treating resolving power R as identical to final radial-velocity precision.
14. Reading a FITS data array without inspecting its header.
15. Ignoring quality flags.
16. Assuming PDCSAP is always scientifically preferable to SAP.
17. Ignoring aperture contamination or third light.
18. Saving a plot without recording the archive query, software, and data release.

41 Practice exercises

1. A binary component is 3.0 magnitudes fainter than its companion in the same passband. Find the flux ratio.
2. Two unresolved stars have magnitudes 8.0 and 9.0 in the same band. Find the combined magnitude.
3. Show that two equal-flux unresolved stars are 0.7526 magnitude brighter than either component alone.
4. A total eclipse removes 30% of the out-of-eclipse flux. Find the corresponding differential magnitude.
5. A source has G_BP = 10.4 and G_RP = 9.7. Find G_BP − G_RP .
6. Convert right ascension 6 h to degrees.
7. Two resolved components differ by 0.80 arcsec in right ascension coordinate at declination 60^∘ and by 0.30 arcsec in declination. Find the approximate tangent-plane separation.
8. Derive the position-angle expression using east and north offsets.
9. Convert MJD 60000 to JD.
10. Convert BTJD 1400 to full BJD using the standard TESS offset adopted in this article.
11. A light curve has normalized flux 0.92 at eclipse minimum. Find the differential magnitude.
12. If SNR = 250, estimate the small-error magnitude uncertainty.
13. A spectrograph has R = 50000. Estimate the velocity resolution scale c∕R.
14. A spectral line at 500.000 nm is observed at 500.050 nm. Estimate the nonrelativistic radial velocity.
15. Explain why the same TESS aperture can produce different eclipse depths if an unrelated contaminating star is added to the aperture.

42 Summary

Flux and luminosity are connected by

(90)

Magnitudes encode flux ratios:

(91)

Unresolved binary fluxes add linearly:

(92)

Precision binary star astronomy requires explicit coordinates and time standards.

For TESS-like barycentric times,

(93)

For small relative flux error,

(94)

Spectral resolving power is

(95)

FITS data products combine numerical data with metadata that are necessary for physical interpretation.

These observational tools provide the measurement foundation for BIN03, where the series turns from measured positions and times to Newtonian two-body dynamics and Keplerian orbits.

References

References