Understanding Light Curves: Reading the Signals from Distant Stars

Every exoplanet discovered by the transit method started as a pattern in a light curve. Here is how astronomers read these signals, what the data looks like, and how S.O.L.A.R.I.S. automatically extracts planet signatures from the noise.

A light curve is one of the simplest and most powerful tools in all of astronomy. It is nothing more than a record of how bright a star appears over time — but hidden within that record can be the signature of an orbiting planet, a binary companion, stellar pulsations, or explosive flares. For exoplanet hunters, learning to read a light curve is like learning to read a language written in starlight.

What Is a Light Curve?

A light curve is a graph that plots a star's brightness (flux) on the vertical axis against time on the horizontal axis. Each point on the graph represents a single brightness measurement taken at a specific moment. String thousands of these measurements together and you have a time series that tells the story of what is happening to and around that star.

When nothing is happening — no transits, no variability — a light curve is a flat, horizontal line (with some scatter from noise). It is the deviations from this baseline that carry scientific information.

Anatomy of a Transit Dip

When a planet passes in front of its host star, it blocks a small fraction of the star's light, creating a characteristic U-shaped or flat-bottomed dip in the light curve. This transit dip has several measurable properties:

• Depth — How much the brightness drops, expressed in parts per million (ppm) or as a percentage. The depth equals (R_planet/R_star)². An Earth-sized planet around a Sun-like star produces a dip of ~84 ppm; around an M-dwarf, 500–1,500 ppm.
• Duration — How long the dip lasts, from first contact (planet begins covering the stellar disk) to last contact (planet fully exits). Typical durations range from 1 to 6 hours.
• Ingress and egress — The sloped edges of the dip, where the planet is partially covering the star. These slopes constrain the planet's orbital inclination and the star's limb darkening.
• Period — The time between successive dips. If you see the same dip repeat at regular intervals, you have strong evidence of a planet with that orbital period.

Sources of Noise in Light Curves

Real light curves are never perfectly clean. Several types of noise and variability can obscure or mimic transit signals:

Photon Noise

Light arrives as individual photons, and the number detected in any given interval fluctuates randomly following Poisson statistics. Fainter stars produce fewer photons per measurement, so their light curves are noisier. This is the fundamental limit on photometric precision.

Instrumental Systematics

No detector is perfect. Temperature changes, spacecraft pointing drift, and pixel sensitivity variations all introduce systematic trends in the data. TESS's data processing pipeline (the Science Processing Operations Center, or SPOC) applies corrections for many of these effects, producing "Pre-search Data Conditioned" (PDC) light curves with most systematics removed.

Stellar Variability

Stars are not static. Starspots (dark regions on the stellar surface, analogous to sunspots) rotate in and out of view as the star spins, creating slow, periodic brightness modulations. Stellar pulsations cause rhythmic brightness changes in certain types of stars. Flares produce sudden, sharp brightness spikes. All of these astrophysical signals must be modeled and removed before searching for the subtle, periodic dips caused by transiting planets.

How BLS Detection Works

The Box Least Squares (BLS) algorithm is the workhorse of transit detection. Developed by Kovács, Zucker, and Mazeh in 2002, it works by fitting a simple model to the light curve: a flat baseline with a periodic, rectangular (box-shaped) dip.

The algorithm systematically tests a grid of possible parameters:

1. Period — The time between dips (typically tested from 0.5 to 30 days in thousands of steps).
2. Transit duration — The width of the box (tested from 1 to 6 hours).
3. Phase — The timing of the first dip within the observation window.

For each combination, BLS calculates how well the box model fits the data compared to a flat line. The combination that produces the best fit — the highest signal detection efficiency (SDE) — is flagged as a candidate transit signal. An SDE above a certain threshold (typically 6 to 8) indicates a statistically significant detection.

TESS Data: 2-Minute Cadence

NASA's Transiting Exoplanet Survey Satellite (TESS) observes the sky in sectors, each covering 24° × 96° for approximately 27 days. For pre-selected target stars, TESS records brightness measurements every 2 minutes — yielding roughly 19,000 data points per sector per star.

This 2-minute cadence is well suited for detecting transit dips, which typically last 1 to 6 hours (30 to 180 data points per transit). TESS also captures full-frame images at 10-minute intervals, which can be used to extract light curves for any star in the field of view, though at lower time resolution.

TESS data is publicly available through the Mikulski Archive for Space Telescopes (MAST), which is what makes citizen science projects like S.O.L.A.R.I.S. possible. Anyone can download and analyze real NASA satellite data.

How S.O.L.A.R.I.S. Processes Light Curves

The S.O.L.A.R.I.S. pipeline automates the entire light curve analysis process in two stages:

1. Stage 1: BLS Detection — Each TESS light curve is downloaded, detrended to remove stellar variability and instrumental systematics, and then scanned with the BLS algorithm. Candidates exceeding the SDE threshold are flagged for further analysis.
2. Stage 2: MCMC Fitting — For each candidate, Markov Chain Monte Carlo (MCMC) sampling fits a full transit model to the data. This determines precise values and uncertainties for the planet's radius, orbital period, inclination, and equilibrium temperature. The MCMC approach explores the full probability distribution of each parameter, providing robust error bars.

The volunteer software distributes work units — individual TESS light curves — to participants' computers, which run both BLS and MCMC locally and return results to the central server. The software is under 1 MB and available for macOS, Windows, and Linux.

Reading Light Curves Yourself

If you want to explore light curves beyond volunteering with S.O.L.A.R.I.S., TESS data is freely accessible. Tools like Lightkurve (a Python package) let you download, plot, and analyze TESS light curves with just a few lines of code. Look for periodic dips, phase-fold the data at the candidate period to stack multiple transits, and see if the signal holds up. You might spot something the automated pipelines missed.

Every discovery in exoplanet science — from hot Jupiters to potentially habitable Earth-twins — started with someone reading a light curve. The data is waiting.

Help Process Light Curves

S.O.L.A.R.I.S. discoveries are exoplanet candidates based on statistical analysis of TESS photometry. Professional follow-up observations are needed for confirmation. The project is not affiliated with NASA.