Presentation #102.182 in the session Poster Session.
Stellar activity poses a severe limitation to the search and characterisation of small exoplanets with the radial velocity method. This is particularly important for M dwarfs, as they are crucial targets for both ground-based instruments (e.g., SPIRou, NIRPS, CRIRES) as well as space-based missions (e.g., JWST, ARIEL), but can manifest high activity levels over long time-scales. Efficient activity filtering techniques are therefore necessary to disentangle genuine planetary signatures and improve the detectability. In this context, and knowing from previous studies that different spectral lines are affected differently by magnetic activity, we developed a new mitigating technique based on a randomised selection of lines to use in Least-Square Deconvolution (LSD). We benchmarked the analysis on optical spectropolarimetric time series of the active M dwarf EV Lac collected with ESPaDOnS at CHFT, obtaining a reduction of the radial velocity dispersion by at least 50-60%. A similar and consistent improvement was also found when targeting stars of analogous (AD Leo) and lower (DS Leo) activity levels. Finally, we injected synthetic planets with semi-amplitudes between 60 and 120 m/s (i.e. 0.3-0.6 MJup) in our data sets containing moderate (20 m/s) and high (200 m/s) activity levels and we retrieved reasonably unaltered planetary signals, indicating that our technique does not suppress these signals substantially.