Biosignature detectability on transiting habitable worlds with ELT/ANDES
Abstract
The search for life beyond the Solar System is at a turning point, transitioning from theoretical predictions to observations enabled by next-generation observatories. The Extremely Large Telescope (ELT) will host the ArmazoNes high Dispersion Echelle Spectrograph (ANDES), optimized for visible-to-near-infrared high-resolution spectroscopy. We present a simulation--detection pipeline and evaluate the detectability of CO, HO, and the biosignature gases O and CH in high-resolution transmission spectroscopy of transiting habitable-zone rocky planets with ANDES. Assuming cloud-free, modern Earth-like atmospheres, we model transmission spectra using noise estimates from the ANDES Exposure Time Calculator, based on the latest preliminary instrument design in seeing-limited mode. We introduce a novel Bayesian cross-correlation function (CCF) framework that incorporates molecule-specific kernels and a new autoregressive model to account for correlations in the CCF. We apply our framework to 18 known potentially habitable transiting exoplanets and estimate the number of transits required for a decisive detection (). We find that HO is the most accessible species, with potential detections in 10-19 transits for the TRAPPIST-1 planets and 30 transits for LHS 1140 b. CO, CH, and O are more difficult to detect, requiring approximately 1.5, 3, and 4 times as many transits as HO. These estimates are lower limits that assume favorable observing conditions, perfect detrending, and the absence of systematics, yet still imply large observing campaigns. Alternative approaches, such as reflected-light high-dispersion coronagraphy of nearby nontransiting planets, may offer a promising complementary route for biosignature searches.
Keywords
Cite
@article{arxiv.2607.12017,
title = {Biosignature detectability on transiting habitable worlds with ELT/ANDES},
author = {E. Kurzawa-Ferrandez and A. Bello-Arufe and R. Hu},
journal= {arXiv preprint arXiv:2607.12017},
year = {2026}
}
Comments
31 pages, 8 figures. Accepted by AAS journals