Astronomy · Finding change in the sky

Light curves and time stamps

Treat each light-curve point as a finite calibrated measurement with an explicit time convention, and separate integrated counts, mean rates and clock corrections.

Continue in the approved chapter order. Necessary vocabulary, quantities and assumptions are explained in the lesson.

Back to the chapter

After reading, you can

Identify the measured quantity, observing band and time metadata of a light-curve point.
Calculate exposure midpoints, a supplied clock correction and expected mean count rates.
Distinguish numerical timestamp precision from calibrated accuracy and unobserved gaps from measured limits.

What is changing on the vertical axis?

A light curve records a measured quantity as a function of time. Its vertical axis might show a calibrated band flux, a magnitude, a count rate or a signed change relative to a reference image. Those quantities are connected by models and calibration; they are not interchangeable labels for the same raw number.

A useful point includes uncertainty and information about the observing band and measurement procedure. A combined aperture can contain the target, neighbouring sources and background contributions. The chosen measurement definition determines what a later change means.

For a common band and magnitude convention, a smaller magnitude represents greater received flux. Near zero signal, a symmetric small-error magnitude approximation can fail. A negative background-subtracted or difference-flux estimate cannot simply be converted into an ordinary logarithmic magnitude of positive source light.

The connecting line in a plotted curve is usually a visual aid or a model, not an observation at every time between points. Before interpreting its shape, inspect how the points were obtained and which intervals were actually observed.

Start, midpoint and effective time are different labels

A detector can integrate light over a finite exposure. Its recorded time might label the start, end or midpoint under the instrument’s convention. A point drawn at one of those times still represents an integrated measurement over an interval.

With uniform temporal weighting, an exposure beginning at t_start and lasting Δt has arithmetic midpoint t_mid = t_start + Δt/2. If throughput or source brightness varies during the exposure, an effective flux-weighted time can differ from that midpoint.

The time scale, numerical origin and arrival-time reference location also matter. A Julian-date-like number by itself does not fully specify them. Observations made at different locations or expressed in different scales may need carefully defined conversions before a physical timing comparison.

Eastman, Siverd and Gaudi explain these distinct timing issues. Their historical instrument examples and software-accuracy claims are not adopted here. This lesson uses a supplied uniform time coordinate and a simple clock offset, without computing an actual barycentric or source-emission time.

More counts can come from a longer exposure

For an original exercise, use seconds in one declared uniform time scale with one origin. Stipulate that the local clock reads 7 s ahead of that reference. The raw timestamps are explicitly exposure starts, and temporal weighting is uniform.

Exposure A starts at raw coordinate 100 s and lasts 20 s, so its raw midpoint is 110 s. Subtract the supplied 7 s offset to obtain corrected midpoint 103 s. Exposure B starts at raw 130 s and lasts 40 s, giving raw midpoint 150 s and corrected midpoint 143 s.

Supply expected background-subtracted signals of 120 detected electron counts for A and 240 for B, with the same calibrated band response. Their expected mean rates are 120/20 = 6 and 240/40 = 6 electron counts s⁻¹. The larger integrated signal alone does not establish a brighter source.

For an optional relative horizontal origin, choose a reference instant whose corrected coordinate is 100 s. The corrected midpoint offsets are then 3 s and 43 s. That reference instant is a declared coordinate choice, not an instruction to treat A’s raw start as already corrected.

Original uniform-weighting exposures in one stipulated uniform time coordinate. Raw times label starts; the clock is 7 s ahead. Expected background-subtracted signals have identical band response. The relative origin has corrected coordinate 100 s.
Exposure	Raw start (s)	Duration (s)	Raw midpoint (s)	Corrected midpoint (s)	Offset from corrected 100 s	Expected signal (electron counts)	Expected mean rate (electron counts s⁻¹)
A	100	20	110	103	3	120	6
B	130	40	150	143	43	240	6

Precision in a number is different from timing accuracy

A timestamp can be printed with many digits while its clock offset, shutter timing or time convention is poorly known. Extra numerical precision can preserve a calculation, but it does not independently calibrate the observation.

For a separate unit conversion only, define one uniform day as 86400 SI seconds. Then 0.001 day equals 86.4 s, while 0.000001 day equals 0.0864 s. These are numerical interval conversions, not a calendar or leap-second calculation.

Timing uncertainty and flux uncertainty are also different. A point can have a well-calibrated mean flux but an uncertain time label, or a precise time coordinate with a noisy flux estimate. A model of rapid variability may need both.

The relevant accuracy depends on the scientific comparison. Keep the original metadata and transformation history so that later analyses can assess whether the chosen convention and corrections are adequate. A corrected observer arrival time is not automatically the time of emission at the source.

Preserve what the observations did and did not measure

A gap can mean that no exposure was taken or that a measurement was unusable. A non-detection within a usable exposure is a different kind of evidence. An upper limit needs its stated band, observing interval and statistical or selection definition.

Neither a gap nor a non-detection should be silently plotted as zero source flux. Even a formally negative signed estimate can contain useful information about uncertainty or change relative to a reference, while remaining different from negative incident photons.

A reproducible light curve therefore preserves the measured quantity and unit, band and calibration, time scale and location reference, exposure convention, uncertainty, quality information and provenance. A clean-looking plot without those definitions can obscure the very distinction needed for its interpretation.

The next lesson examines cadence and aliasing. It will use finite exposure intervals to show both how different signals can produce identical sampled averages and how an event can fall entirely between observations.

Worked examples

Apply the stated start-time and clock conventions

What corrected midpoint represents exposure B under the supplied uniform-weighting model?

Its raw start is 130 s and its duration is 40 s.
The raw midpoint is 130 + 40/2 = 150 s.
The clock is ahead by 7 s, so the corrected midpoint is 150 − 7 = 143 s.

The corrected midpoint is 143 s, representing a finite 40 s exposure rather than an instantaneous sample.

Compare integrated signals at different durations

Does B’s expected signal of 240 electron counts prove twice A’s mean source rate?

A’s expected mean rate is 120/20 = 6 electron counts s⁻¹.
B’s expected mean rate is 240/40 = 6 electron counts s⁻¹.
The signal doubled alongside the duration under the same supplied response.

No. Both supplied expected mean rates are 6 electron counts s⁻¹.

Try the reasoning

Does labelling an exposure by its midpoint make the measurement instantaneous?
Show the answer and reasoning
No.
The value remains integrated over the exposure’s finite time response.
Do more timestamp digits by themselves establish a calibrated clock or correct time scale?
Show the answer and reasoning
No.
Numerical representation, clock accuracy and time metadata are separate issues.
Is a date-like number alone sufficient to compare all precise astronomical arrival times?
Show the answer and reasoning
No.
The time scale, reference location, convention and transformations must also be specified.
Should an interval with no usable observation be treated as measured zero source light?
Show the answer and reasoning
No.
An unobserved gap, a signed estimate and a defined upper limit carry different information.

What to carry forward

A light-curve point needs a quantity definition, calibration, uncertainty and time metadata.
Finite exposure duration changes integrated counts and the meaning of a time label.
Timestamp precision, calibrated accuracy, gaps and non-detections must remain distinct.

References

Teaching resources and research supporting this lesson’s explanations and examples.