Why can an exoplanet transit appear early at one observatory and late in another data set? A telescope’s UTC timestamp describes a local photon arrival, while a high-precision transit comparison often needs a common barycentric time standard such as BJD_TDB. The observation is not necessarily wrong; the time reference may simply be different.
This guide explains the difference between UTC, TDB and Julian Date; shows why exposure midpoints and barycentric light-travel corrections matter; and uses checkable examples to separate a genuine transit-timing variation from a timestamp error. It is written for curious learners, observers and readers who want the calculation and the scientific limit together.
eduKateSG · The Gold Standard of Astronomy
Choose your reading route
Define the time · Chapters 1–3. Event, exposure and time scale.
Correct and compare · Chapters 4–6. Geometry, cross-observatory checks and residuals.
Test and transfer · Chapters 7–8. FAQs, linked explanations and sources.
Chapter index
Define the time · Chapters 1–3
Correct and compare · Chapters 4–6
Test and transfer · Chapters 7–8
A planet passing in front of its star does not schedule its transit for Greenwich. Its photons travel across space and reach moving observers at different times. Earth turns, Earth orbits the Sun, and the receiving observatory is not always the same distance from the incoming wavefront. That is why a careful transit timing measurement cannot stop at the clock displayed on the telescope computer.
Start by naming the astronomical event, such as the midpoint of the dimming inferred from a fitted light curve. Then name the logged observation time, its time scale and where the observer was located. The event time and the timestamp written by a particular observatory are related, but they are not interchangeable without corrections.
A UTC timestamp is useful and essential for many human records. BJD_TDB is useful when comparing well-timed astronomical events because it expresses a Julian Date on the TDB scale after a light-travel correction to the Solar System barycentre. Those labels are parts of the number’s meaning, not optional decorations.
The basic question is: if two instruments witnessed the same cosmic event, what reference location and time scale make their records comparable? Astropy’s time documentation describes the answer operationally: convert the time scale appropriately and calculate the photon light-travel correction for a known sky direction and observer.
CHAPTER 2 OF 8
2. Separate exposure start, exposure midpoint and fitted transit midpoint
An exposure is an interval rather than a mathematical instant. Suppose a camera begins a two-minute exposure at 20:00:00 UTC and its shutter remains open for exactly 120 seconds. Under a simple uniform-integration convention, the representative middle of that exposure is 20:01:00 UTC. If the reduction software uses its start stamp as if it were the midpoint, the recorded time is biased early by 60 seconds.
This example assumes the exposure-time metadata are accurate, the time reference is UTC and the observing system’s shutter or readout timing creates no additional effective offset. Real cameras may need more precise information, including shutter delay, dead time, rolling-readout behaviour or a measured photon-weighted exposure midpoint.
Do not confuse the midpoint of one exposure with the midpoint of an entire transit. A transit may last hours and contain hundreds of exposures. Its mid-transit time is estimated from the light curve and an explicit fitting model. The timestamps attached to the individual flux measurements must first describe their own exposure intervals correctly.
A practical audit checks whether the archive field stores the exposure start, stop or middle; whether a finite-exposure transit model was used; and whether the clock was synchronised and its uncertainty documented. Otherwise an apparent variation in a planet’s schedule can begin as a logging convention.
The connected finite-exposure averaging guide explains a companion problem: an exposure also averages the signal over time. Correcting its timestamp cannot restore rapid variation already smoothed by integration.
| Recorded item | Illustrative value | Interpretation |
|---|---|---|
| Start stamp | 20:00:00 UTC | Beginning of the 120-second exposure |
| Duration | 120 s | Elapsed open-shutter interval in this toy model |
| Mid-exposure | 20:01:00 UTC | Reference instant of the single flux point |
| Mid-transit | Not yet known | Estimated from a fitted series, not one image |
A time format and a time scale are distinct. ‘Julian Date’ says how a time value is represented numerically. ‘UTC’ or ‘TDB’ says which time scale the value belongs to. Adding ‘B’ to BJD indicates that the light arrival time has been referred to the Solar System barycentre. BJD alone is therefore incomplete when high precision matters: retain the time-scale suffix.
UTC follows the civil atomic-time framework with leap seconds. TAI is an atomic time scale without UTC’s inserted leap seconds, and TT differs from TAI by a defined 32.184 seconds. TDB is a relativistic dynamical time scale appropriate to Solar-System-barycentric work. Do not assume its offset from UTC is always a single universal fixed number over history.
Astropy exposes the distinctions explicitly. A Time object can have a ‘jd’ or ‘mjd’ representation while its scale is ‘utc’ or ‘tdb’. Changing the representation is not the same as changing the scale. A bare decimal without those labels is not self-explanatory astronomical evidence.
When sources disagree, ask whether they provide JD_UTC, HJD_UTC, BJD_TDB, a mission-specific relative time offset or a modified Julian Date. A familiar string of digits can hide a half-day convention difference or an epoch offset. Format conversion, time-scale conversion and geometric correction solve three separate problems.
The responsible workflow converts each timestamp to a chosen common system before fitting period changes. The source metadata must specify the original format, reference epoch, scale and observer location, not only the final displayed decimal.
CHAPTER 4 OF 8
4. Calculate a barycentric correction without pretending it is a constant
Astronomy uses the Solar System barycentre as a common arrival-time reference because the Earth’s changing position affects the travel path of photons from a target. To first approximation, a geometric correction follows the projection of the observer’s position relative to the barycentre along the incoming line of sight, divided by the speed of light. The sign depends on the adopted correction convention.
In Astropy the practical operation is to construct the observed time with its original scale and EarthLocation, define the target as a SkyCoord and evaluate the barycentric light-travel time. For a suitable observation, times.tdb + times.light_travel_time(target) gives the barycentric arrival time on the TDB scale. The Earth and target geometry belongs in the calculation.
Use a deliberately invented arithmetic example. Assume a UTC exposure midpoint has been converted to the same instant in TDB, represented here by adding a hypothetical 69 seconds to the displayed clock value. Assume an independent calculated barycentric delay of +240 seconds for that sky direction and observation date. The total illustrative change in the numerical clock representation is +309 seconds, or 5 minutes 9 seconds.
The 69-second term is a pedagogical placeholder, not a live leap-second lookup or a universal UTC-to-TDB constant. A real reduction must use time-scale conversion software and applicable reference tables. The +240-second geometry is equally illustrative. Neither number should be hard-coded into an astronomical pipeline.
Imagine an observation of the same target months later for which the geometric term is −210 seconds. Relative to the first +240-second term, the change in the light-travel contribution is 450 seconds. A period comparison that neglects barycentric correction can therefore introduce apparent several-minute shifts between seasons, even if the astrophysical timing model is stable.
| Component | Toy correction | Where it comes from |
|---|---|---|
| UTC to TDB | +69 s (illustrative) | Time-scale conversion |
| First observing season | +240 s | Observer and target geometry |
| First combined shift | +309 s | Illustrative scale + geometry |
| Later observing season | −210 s | Different observer–target geometry |
| Change in geometric term | 450 s | Difference between +240 s and −210 s |
Separate an annual geometric effect from an observer-to-observer effect on the same night. Earth’s orbital motion can change the barycentric light-travel correction by minutes across seasons. The difference between two Earth-based telescope positions at one instant is constrained by the much smaller size of Earth and is typically measured on a very different time scale, with the actual value depending on the geometry.
For a toy comparison suppose instrument A records an exposure midpoint at 20:00:00.000 UTC with a computed barycentric light-travel term of +240.000 seconds. Instrument B records the same incoming wavefront at 20:00:00.020 UTC and has a correction of +239.980 seconds. After converting both time scales consistently, the illustrative barycentric sums agree: 240.000 seconds for A and 240.000 seconds for B relative to the common clock origin.
That cancellation demonstrates the purpose of the correction, not a promise that two real instruments always match to the millisecond. Both still have uncertainties from their clocks, target coordinates, detector timing, ephemerides and details of how flux measurements were taken.
If the observers’ numbers differ by minutes on the same night, first investigate their timestamp definitions, time zones, mission epoch offsets, clock settings, leap-second handling and JD conventions. It is not sensible to attribute a large discrepancy automatically to ground-station latitude.
The resolution is always to rebuild the time chain from original metadata. A tidy corrected number whose provenance cannot be reconstructed is less useful than a clearly documented measurement with honest timing uncertainties.
CHAPTER 6 OF 8
6. Turn a corrected transit time into an observed-minus-calculated test
Once times belong to the same barycentric system, the next question is scientific: does a particular transit occur when the predicted ephemeris says it should? A simple linear model uses T(n) = T0 + nP, where T0 is a reference mid-transit time, P is a period in days and n is an integer transit count.
Suppose the model predicts BJD_TDB 2,460,000.500000 for one event, and a fitted, properly corrected observation gives 2,460,000.501389. The difference is 0.001389 day, which is about 120 seconds. The observed-minus-calculated residual is approximately +2 minutes.
The residual is an observation relative to a model. It is not automatically evidence for an additional planet. An out-of-date period, an incorrect cycle count, red noise in the light curve, stellar variability, imperfect exposure integration or a mistaken time standard could also shift the fitted midpoint.
Before invoking transit timing variations, compare several epochs with the same time convention and a model appropriate to their uncertainties. A valid claim also needs careful propagation of the fitted ephemeris covariance because T0 and P uncertainties produce correlated prediction errors at different cycle counts.
Create an audit table with the original stamp, exposure convention, scale, observatory position, sky coordinate, barycentric correction, corrected value, ephemeris version, model midpoint and uncertainty. A surprising residual is now something another researcher can examine rather than merely a number in a caption.
| Quantity | Teaching value | Role |
|---|---|---|
| Predicted transit | 2,460,000.500000 BJD_TDB | Linear ephemeris estimate |
| Observed transit | 2,460,000.501389 BJD_TDB | Fitted and corrected midpoint |
| Difference | 0.001389 d | Observed minus calculated |
| Difference in seconds | ≈120 s | Multiply days by 86,400 |
Is BJD the same as JD?
No. JD is a numerical day-count representation. BJD states that light travel to the Solar System barycentre has been accounted for. Its time scale must still be named.
Does BJD_TDB mean every observatory saw the transit simultaneously?
It provides a common reference framework for the arrival of photons from a given direction. Actual data contain exposure averaging, noise, clock limitations and fitting uncertainties, so measurements need not yield exactly the same recovered midpoint.
Can I convert UTC to TDB by adding one remembered number?
Do not do that as a research method. The time scales have explicit definitions and historical dependencies; rely on a maintained library and the correct input epoch.
What happens if my file contains a start-of-exposure timestamp?
Use the actual integration convention and camera metadata to locate the representative measurement instant. A midpoint offset of half the exposure duration is only the simplest uniform-exposure case.
If a transit is two minutes late, have I discovered another planet?
No. First rule out time-format errors, clock problems, inadequate ephemerides and correlated noise. Repeated, independently checked residuals are necessary before physical interpretation.
Does barycentric correction fix bad light-curve photometry?
No. It fixes a timing reference problem. Sky subtraction, calibration, blending and the validity of the transit model are separate evidence questions.
CHAPTER 8 OF 8
8. Connect timekeeping to photometry, evidence and transferable learning
An astronomical timing claim has two sides. The photometric data must genuinely measure a change in light, and the timestamps must refer those measurements to a common place and time scale. A perfect clock attached to a mismeasured light curve does not produce a reliable planet, and excellent photometry attached to mismatched clocks can produce misleading period changes.
Teach the chain in ordinary language: identify the event, determine what instant was stamped, preserve the source time scale, correct for where the telescope was, fit the event time, and compare it with the prediction plus its uncertainty. Then change one assumption—a longer exposure, a new season or a different time standard—and ask which step must change.
Subject and learning map
The Gold Standard of Astronomy — return to the connected subject hub.
The Gold Standard of Astrophysics — investigate the physical explanations behind measured signals.
The Gold Standard of Education | Learning Transfers into Life — carry a reference-frame habit into a different problem.
How X Works — navigate the wider research and explanation ecosystem.
Build the observational chain
Aperture photometry — establish the source measurements before fitting time.
Difference-image flux and template noise — distinguish new flux from change relative to a reference.
Vetting an exoplanet transit candidate — test astronomical alternatives and instrumental explanations.
Authoritative sources and scope
References reviewed on 22 October 2026. Every numerical timing table in this article is an original simplified teaching exercise, not a real telescope observation. Software-specific correction values require the observation coordinates, epoch, target position and version of the ephemeris.
Astropy: Time and Dates — time representations, time scales and barycentric/heliocentric light-travel calculations.
Astropy: Time API — time objects, observatory position and light-travel correction methods.
NASA Exoplanet Archive — transit ephemeris context and records that require careful attention to their native time systems.