Characterizing Background Signals in Red Giant Power Spectra

by Dr Anohita Mallick (DAA, TIFR)

Tuesday 11 Aug 2026, 16:00 → 17:00 Asia/Kolkata

AG-66

The background component of red giant power spectra, comprising granulation and mesogranulation signals, is commonly modeled using scaling relations that assume these signals behave self-similarly. Using ∼9,300 Kepler red giants with stellar parameters from a crossmatch with APOGEE, we model the background components and infer empirical scaling relations for the characteristic frequencies and amplitudes of both signals as functions of surface gravity, effective temperature, and metallicity within a Bayesian framework, testing for deviations from simple scaling behavior using flexible, data-driven models.

We extend this characterization of the background signal to test whether it also carries diagnostic information on stellar multiplicity, as a complement to traditional  detection methods. We construct synthetic asteroseismic power spectra reproducing Kepler observations, with binaries modeled as flux-weighted combinations of two stellar power spectra spanning merged, overlapping, and widely separated oscillation components. A gradient-boosted tree classifier is trained on 40,000 such spectra to identify binaries via global power spectrum morphology and is validated against known classifications in literature.

Together, these two analyses use detailed modeling of the background signal in red giant power spectra to place new constraints on convective processes and to develop an extendable framework for identifying unresolved binaries, with applicability to larger Kepler samples and future PLATO data.