From Detector Counts to an Interpretable Light Curve
6 min read
Lesson objectives
- Explain from detector counts to an interpretable light curve and its evidence-bound distinction.
- Apply from detector counts to an interpretable light curve to a new case and justify its inferential limit.
Please sign in to use all the tools
Once you sign in, the reading tools are yours: larger or smaller text, typeface, line spacing, focus mode, highlighting and bold, and notes saved to your account and gathered in My study.
A light curve plots a measured brightness against time; it is not an image of the object's path across the sky. The horizontal axis carries observation time with a stated unit; the vertical axis specifies detector counts, calibrated flux, or a ratio to baseline. These units are not interchangeable. A label such as 0.98 needs a definition: is it 98% of baseline or something else? Before searching for a dip, inspect timestamp order, missing exposures, filter and background correction. In a simple classroom example, normalized flux is the target value divided by a reference outside the event. With a baseline of 1000 counts and a measured 980, the normalized value is 0.980 and the drop is 0.020, or 2%. That is a descriptive calculation only. A star image can move across pixels of unequal response, and atmospheric absorption can vary, so compare apparently stable stars in the same frames and inspect their scatter. Do not turn every low point into a planet transit or eclipse. Test whether a dip persists across multiple samples, returns toward baseline and repeats if a periodic claim is made.
A useful calibration starts with what the file actually records: exposure time, background, measurement filter and the apertures used for target and comparison counts. If a comparison star in the same frame drops 2% at the same instant, a shared cause such as atmosphere or detector response merits correction before interpreting the target. In a simplified example, raw target counts fall from 1000 to 980 while comparison counts fall from 500 to 490. Both drop 2%. Target/comparison is 980÷490=2, equal to baseline ratio 1000÷500=2; no relative dip remains in this model. If instead the comparison stays 500 while the target becomes 980, the ratio is 1.96 against baseline 2.00, a 2% relative decline. These figures illustrate an initial correction, not a cure for blended sources or color-dependent atmospheric change. Retain raw measurements and transformation steps so the result can be checked. A polished curve without underlying counts or timestamps cannot support review of the feature's origin. If a point is removed as an outlier, document the exclusion rule beforehand rather than dropping a point simply because it conflicts with a favored hypothesis.
Read a light curve at its actual scale before choosing an explanation. A dip that begins and ends over several points can fit a transit, but intrinsic stellar variability or an instrumental effect may resemble it. In a simple model of a small opaque planet crossing a uniformly bright stellar disk, transit depth is approximately (planet radius ÷ stellar radius)². This approximation does not yield planet radius from depth alone when stellar radius is unknown or extra light from another source dilutes the signal. One low point also cannot establish orbital period: repeated events and their time separation are needed. In reporting, separate 'a normalized 2% decrease in this window' from 'we discovered a planet'. The first describes checkable data; the second proposes a physical cause needing more evidence. If sample spacing exceeds event duration, the event may be missed or represented by a lone point with no measured start or end. Set cadence before collecting data, and do not claim time precision finer than the sampling allows. End each analysis with approximate uncertainty and its source, alternative explanations and a specific follow-up decision.
Lesson summary
A light curve plots a measured brightness against time; it is not an image of the object's path across the sky. A useful calibration starts with what the file actually records: exposure time, background, measurement filter and the apertures used for target and comparison counts.