Tierras telescope reports unusually stable starlight measurements over months

What Tierras measures

The Tierras Observatory is a robotic 1.3-metre telescope at the Fred Lawrence Whipple Observatory in southern Arizona. In an arXiv preprint, Juliana García-Mejía and colleagues describe its instrument, automated processing and performance on the sky. Their central result is unusually stable ground-based brightness measurements across months, alongside precise measurements within individual nights.

The relevant unit is a part per thousand, abbreviated ppt: 1 ppt means a brightness difference of 0.1%. Smaller scatter makes weaker changes easier to distinguish. Those changes can include a planet crossing its star, or dark and bright regions rotating across a stellar surface.

The authors describe their observations as the first ground-based light curves with less than 1 ppt night-to-night scatter over months. That priority claim belongs to the paper; the supplied evidence does not independently establish it.

Designing around atmospheric interference

A major obstacle is changing atmospheric water vapour. Tierras compares a target star with other stars in the same image to remove shared brightness changes. But cool, red targets and warmer comparison stars emit different mixtures of wavelengths. Water vapour absorbs those mixtures differently, leaving false variability even after comparison.

Tierras uses a custom filter centred at 863.5 nanometres, near the infrared, with a bandwidth of 40.2 nanometres. It selects a relatively clear atmospheric window while rejecting neighbouring water-absorption features. Simulations predict water-vapour errors below about 0.25 ppt even for an extreme change in atmospheric water content. This is a prediction for one error source, not a demonstrated limit on total measurement noise.

Additional optics expand the view to include many comparison stars. A sensitive cooled detector collects their light, while internal light shields suppress unwanted illumination. The authors report that telescope baffles reduced the median sky background by roughly a factor of 42. A separate lens-edge upgrade in August 2025 reduced a persistent background gradient and enabled effective correction for differences in pixel sensitivity.

Turning images into brightness histories

The automated pipeline measures the light within circular apertures around stars and subtracts a nearby estimate of the sky background. It builds a weighted comparison signal from other stars, iteratively reducing the influence of unreliable or variable references. Dividing each star’s measured brightness by that signal produces its light curve: brightness tracked through time.

Repeatable telescope pointing also matters. Returning stars to nearly the same detector pixels each night helps keep residual pixel-sensitivity errors approximately constant rather than allowing them to mimic stellar changes.

For a single-night test, the team observed the Kepler-1658 field continuously for about 4.5 hours. After excluding sources with estimated contamination from neighbouring stars above 0.1%, they analysed 980 stars. Six showed scatter below 0.3 ppt when measurements were averaged into 20-minute bins. The observed dependence of noise on brightness closely followed their model, without fitting the overall model to match the measured scatter.

The months-long result—and its boundaries

The long-term test used a crowded field containing the nearby M dwarf LEP 1805-1422. The team observed it on 19 nights across more than 70 days, taking 60-second exposures for as long as the field was observable. They removed measurements affected by poor pointing, image quality or cloud cover.

Their stability measure was the standard deviation of each star’s nightly mean brightness. The ten most stable stars all fell below 1 ppt; the best reached 0.52 ppt, or 0.052%.

These were selected best-performing stars, not a typical result for every target. Eight of the ten appear to be evolved stars, so the demonstration should not be read as a direct measurement of equivalent stability for ten nearby M dwarfs. The crowded field also supplied abundant comparison-star light.

Dedicated monitoring is another important qualification. Routine Tierras observations rotate among fields in roughly five-minute intervals. In that mode, the most stable stars typically show night-to-night scatter of 1–3 ppt. The instrument’s 0.25-ppt design target has therefore not been demonstrated as its overall months-long precision.

The long-term campaign preceded the lens-edge upgrade and its images were not flat-fielded. Better performance after that upgrade remains an expectation, not a result established here. The authors also identify detector-calibration uncertainty affecting absolute electron counts, though not the reported relative brightness scatter.

Why it matters

Stable monitoring could help measure slow stellar rotation, interpret atmospheric observations of planets and detect additional planetary transits. Searches for moons and rings are scientific goals, not discoveries reported in this study.

The evidence is an instrumentation and performance report presented as an arXiv preprint. The supplied material does not establish peer-reviewed publication. Its strongest demonstrated result is that carefully designed ground-based observations can achieve sub-ppt stability across months under a dedicated observing strategy.