Identifying Exoplanets with Deep Learning VI. Enhancing neural network mitigation of stellar activity RV signals with additional metrics

TL;DR

The paper tackles stellar activity as a limiting factor in radial velocity exoplanet searches. It extends prior work by training a neural network on six years of HARPS-N solar data with a broad set of activity indicators, including chromatic CCFs, TSI and its derivative, unsigned magnetic flux, and line EWs. The model shows that several indicators improve RV prediction—unsigned magnetic flux, TSI, TSI time derivative, S-index, Hα EW, chromatic CCFs, and CCF shape metrics—while some, like the BIS and Na D EWs, provide no significant gain beyond white-light CCFs; the RV scatter reduction on a held-out test set is from $147.1 \,\mathrm{cm\,s^{-1}}$ to $93.3 \,\mathrm{cm\,s^{-1}}$, aligning with supergranulation noise levels. This work emphasizes that identifying effective tracers of (super)granulation is crucial for advancing RV jitter mitigation and enabling Earth-like exoplanet detections.

Abstract

The measurement of exoplanet masses using the radial velocity (RV) technique is currently limited by stellar activity, which introduces quasiperiodic variability signals that must be modeled and removed to enhance the sensitivity of the RV measurements to exoplanet signals. Neural networks have previously been demonstrated effective in modeling stellar activity signals in HARPS-N solar data using white light cross correlation functions (CCFs). Building on this work, we train a neural network on six years of HARPS-N solar data with additional parameters commonly associated to stellar activity, including chromatic CCFs, line shape metrics, spectral activity indicators, total solar irradiance (TSI) light curves from SORCE and TSIS-1, and TSI time derivatives. Our results show that parameters such as the bisector inverse slope and Na D equivalent widths do not significantly improve the neural network's ability to predict activity-induced RV variations compared to using the white light CCFs alone. However, parameters such as unsigned magnetic flux, the TSI and its time derivative, S-index, H-alpha equivalent width, chromatic CCFs, contrast, and full width at half maximum do improve the neural network's ability to predict RV scatter. Our new model reduces the RV scatter in a held-out test set from 147.1 cm/s to 93.3 cm/s, consistent with supergranulation noise levels reported in previous studies. These results suggest that finding effective tracers for (super)granulation will be critical to train models capable of further mitigating RV jitter, and necessary for characterizing Earth analogues.

Figures (1)

• Figure 1: The overlapping region between the two data sets of TSI values. Pink points represent measurements from SORCE, while blue points represent measurements from TSIS-1. There is an offset between the two, so we shifted the SORCE data to match the level of the TSIS data to merge the datasets. The shifted SORCE data are shown in orange. Small variations can be seen between the TSI values from the two data sets.