Session 1 : Stellar variability and its impact on EPRV studies
Speaker 1 : Jacob Luhn
Affiliation: NASA Jet Propulsion Laboratory, Caltech
Talk Title: Resolving Oscillations and Granulation Within a Night with EPRV Spectrographs
Abstract: Short-timescale stellar variability from oscillations and granulation induces RV signals that now rival or exceed instrumental noise for Sun-like stars. While p-mode oscillations can be mitigated with carefully chosen exposure times, granulation remains more difficult to model and is assumed to produce resolvable distortions in stellar line profiles—an assumption that has seen limited observational testing. I will present results from an ongoing high-cadence observing campaign with EPRV spectrographs NEID and the Keck Planet Finder (KPF) designed to resolve stellar variability within a single night. The dataset spans subgiants observed with NEID to cool dwarf stars observed with KPF’s fast-readout mode, including several stars of interest for future habitable-planet searches. A pilot study of a subgiant reveals that large-amplitude oscillations manifest primarily as pure Doppler shifts in the cross-correlation function, with only subtle asymmetries that evade commonly used mitigation diagnostics. I will show how physically motivated Gaussian process models separate granulation and oscillation signals in main-sequence stars and how each component manifests in cross-correlation–based metrics and line-by-line diagnostics. These results constrain how short-timescale variability appears in the wavelength domain, test key assumptions behind current mitigation strategies, and illuminate how stellar physics sets fundamental limits for precision RV measurements.
Speaker 2 : Manuel Perger
Affiliation: Institut de Ciències de l'Espai (ICE-CSIC)
Talk Title: Mitigating stellar radial velocity jitter using orthogonal activity indices and a time-aware neural network
Abstract: Current extreme precision radial velocity (RV) instruments are reaching RV precision at the 10 cm/s level, opening the window to the detection of Earth twins. However, exoplanet detection and measurement are limited by stellar contaminations, which introduce a noise floor above the meter-per-second level. I present a novel approach for the mitigation of stellar activity consisting in the extraction of Doppler-shift independent, orthogonal line-shape distortion indices from the cross-correlation function, and in feeding these time-series to the newly developed Convolutional-Attention Network for STellar Activity Removal, or CANSTAR. This neural network combines convolutional layers with transformer encoder operations, allowing the modeling of both short- and long-term correlations in the data. CANSTAR is trained on synthetic data from the StarSim code. This is a simulator that produces high-fidelity photometry and high-resolution spectroscopy data products. As a result, we reach RV precisions of a few cm/s for synthetic StarSim data. We validate our framework using HARPS and CARMENES observations of two active stars, Epsilon Eridani and TZ Arietis. The network effectively mitigates stellar activity, reducing the radial velocity RMS to 53 % and 62% of the uncorrected variability, respectively. This correction enables the improvement of the determination of the planetary parameters of TZ Arietis b in comparison to modelling its stellar jitter with a Gaussian Process model. Our results demonstrate that neural networks incorporating temporal context can outperform state-of-the-art methods. Future improvements on StarSim, incorporating 3D magnetohydrodynamic (MHD) spectra and more complex instrumental modelling, are expected to bridge the performance gap between synthetic and real data, offering a robust pathway toward detecting Earth-mass planets around Sun-like stars.
Speaker 3 : Katlyn Hobbs
Affiliation: Center for Computational Astrophysics, Flatiron Institute; Queen's University Belfast
Talk Title: Tracing the Spectral Fingerprints of Magnetic Activity Using Sun-as-a-star Observations
Abstract: Spectral Differencing Analysis (SDA) is a powerful technique for isolating the spectral signatures of stellar activity. Features arising from the interaction of convection and magnetic fields cause spectral line distortions, introducing noise that can mask planetary radial velocity signals. While numerous mitigation techniques have been developed, the individual contributions of these features remain poorly understood. Characterising the line shape changes due to these active regions is therefore crucial for distinguishing radial velocity signals caused by stellar activity from those produced by orbiting planets. I will discuss how we apply SDA to Sun-as-a-star observations, comparing active and inactive spectra over multiple solar rotation cycles. The resulting difference spectra isolate the spectral signatures associated with solar magnetic activity on the level of individual spectral lines. To separate genuine solar signals from instrumental systematics, we compare contemporaneous solar spectra obtained with HARPS-N, NEID, and KPF SoCal. The use of multiple, independent instruments enables us to confirm subtle activity-driven features. By correlating spectral variability across instruments and further comparing these signals with plage and spot surface area coverage derived from Solar Dynamics Observatory (SDO) images, we distinguish activity-driven variability from instrumental effects and investigate its connection to magnetic structures on the stellar surface.
Session 2 : Connecting convection and magnetic activity
Speaker 1 : Varghese Reji
Affiliation: Tata Institute of Fundamental Research, Mumbai
Talk Title: Modeling the vertical velocity gradient to disentangle stellar activity from exoplanet signal
Abstract: The radial velocity (RV) method is one of the most used methodologies in planet detection. Modern instruments have achieved cm/s stability; however, that hasn’t translated to the discovery of Earth-like planets around sun-like stars. Below a few m/s, the Doppler shift of spectral lines due to stellar activity will start dominating planetary signals. A planetary radial velocity signal should be consistent across all the heights and all spectral lines, while a stellar activity-induced photospheric velocity could be different at different heights of the stellar atmosphere. Based on this idea, we developed a method to disentangle stellar activity signals and planetary signals in radial velocity data. In our model, we treat the rising and falling lanes of granulation separately. We first calculate the ‘granulation contrast’, and then determine the velocity profile of both raising and falling lanes. To test our model, we fit this synthetic spectrum with multiple epochs of disk-averaged solar spectra observed with NEID. The stellar activity parameters of our model are the coefficients of the velocity gradient polynomial. Because our approach uses the full observed spectrum, this methodology is readily extendable to M-dwarfs as well. In this talk, I will present our model and its effectiveness in disentangling planetary signals from stellar activity signals.
Speaker 2 : Valeriy Vasilyev
Affiliation: Max Plank Insitute for Solar System Research
Talk Title: Sensitivity of spectral lines to granulation: from the Sun to K-type stars
Abstract: Granulation is no longer just background noise: at the ~1 m/s level, it is a major barrier to detecting Earth analogs with radial velocities. We present a physics-based way to charaterize sensitivities of spectral lines to granulation. Building on our solar study, we extend a line-by-line granulation-sensitivity diagnostic to late-G and early-K dwarfs using 3D time-dependent MURaM simulations and high-resolution spectra synthesized with MPS-ATLAS. The diagnostic quantifies how each line’s Doppler shift and strength respond to convective velocity and thermodynamic fluctuations, while using spatial variability across a single granulation snapshot as an efficient proxy for temporal variability. As effective temperature decreases, weaker convection and a changing ionization balance produce a clearer split between line families: Fe I lines become less velocity-sensitive and more stable in strength, while Fe II lines show the opposite behavior. Using cumulative contribution functions, we further connect spectroscopic RV jitter to characteristic line-formation temperature. The practical implication is clear: solar-optimized line selections do not generally transfer to cooler dwarfs. Spectral-type-aware masks and line weights are therefore essential for mitigating granulation and separating it from magnetic activity in next-generation RV surveys.
Speaker 3 : Ryan Rubenzahl
Affiliation: Flatiron Institute
Talk Title: Pixel-by-pixel response to granulation seen in EPRV solar spectra
Abstract: (Super)granulation remains the dominant source of astrophysical spectral variability lacking reliable spectral indicators. Because convective velocities vary with height in the stellar atmosphere, each spectral segment—which arises from its particular distribution of photons over height in the stellar atmosphere—is sculpted by the net RV of those layers. Hence, high-resolution spectra encode in the line shapes the convective RV. However, directly measuring this via individual line bisectors demands resolving powers beyond what modern planet-hunting spectrographs achieve (>150k). In this talk, we discuss how we can trace the convective radial velocity (RV) using the full spectrum. We jointly model pixel-by-pixel solar RVs from KPF, EXPRES, NEID, and HARPS-N with a multicomponent exposure-time-aware state space Gaussian process to remove instrumental drift and other solar signals (e.g. p-modes, active regions), leaving just the granulation RV. We then model each pixel's contribution function to compute a contribution-weighted RV across the spectrum, yielding an RV timeseries per layer of the solar atmosphere. We show that spectral region ‘families’ with common responses to granulation can be identified with this method, and statistics from differencing the per-layer RVs may be used as a granulation indicator compatible with existing instrumentation.