Enhancing the detection of low-energy M dwarf flares: Wavelet-based denoising of CHEOPS data

¹ Departament de Física Quàntica i Astrofísica, Institut de Ciències del Cosmos (ICCUB), Universitat de Barcelona (IEEC-UB), Martí i Franquès 1, 08028 Barcelona, Spain
² Institut d’Estudis Espacials de Catalunya (IEEC), 08860 Castelldefels (Barcelona), Spain
³ Departament d’Enginyeria Electrònica i Biomèdica, Institut de Ciències del Cosmos (ICCUB), Universitat de Barcelona, IEEC-UB, Martí i Franquès 1, 08028 Barcelona, Spain
⁴ Institut de Ciències de l’Espai (ICE, CSIC), Campus UAB, c/de Can Magrans s/n, 08193 Bellaterra, Barcelona, Spain

^★ Corresponding author: julienpoyatos@icc.ub.edu

Received: 19 December 2024
Accepted: 29 May 2025

Abstract

Context. Stellar flares are powerful bursts of electromagnetic radiation that are triggered by magnetic reconnection in the chromosphere of stars. They occur frequently and intensely on active M dwarfs. While missions such as TESS and Kepler have studied regular and superflares, their detection of flares with energies below 10³⁰ erg remains incomplete. An extension of flare studies to include these low-energy events could enhance flare formation models and provide insight into their impact on exoplanetary atmospheres.

Aims. This study investigates the capacity of CHEOPS to detect low-energy flares in M dwarf light curves. Using the high photometric precision and observing cadence of CHEOPS, along with a tailored wavelet-based denoising algorithm, we improved the detection completeness and refined flare statistics for low-energy events.

Methods. We conducted a flare injection and recovery process to optimise the denoising parameters, applied it to the CHEOPS light curves to maximise flare detection rates, and used a flare-breakdown algorithm to analyse complex structures.

Results. Our analysis recovered 291 flares with energies ranging from 3.7 × 10²⁶ to 8.9 × 10³⁰ erg for 62 M dwarfs, about 42% of which exhibited complex, multi-peaked structures. The denoising algorithm improved the flare recovery by ∼ 35%, although it marginally extended the lower boundary of detectable energies. For the full sample, the power-law index α was 1.99 ± 0.10, but a log-normal distribution fitted better. This suggests multiple possible flare-formation scenarios.

Conclusions. While the observing mode of CHEOPS is not ideal for large-scale surveys, it captures weaker flares than TESS and Kepler, and thus extends the observed energy range. Wavelet-based denoising enhances the recovery of low-energy events, which enables us to explore the micro-flaring regime. The expansion of low-energy flare observations could refine flare-generation models and improve our understanding of their role in star-planet interactions.