Extended Data Fig. 2: Effect of different data reduction methods on the transmission spectrum of LP 791-18 c.

a. Zoom on the saturated region of the transmission spectra of the sub-Neptune LP 791-18 c (median values with 1σ error bars) obtained when considering the standard STScI saturation flags (red) or when expanding the flags to previous groups and neighboring pixels (grey). The number of groups (N) used as a function of wavelength is shown by the grey regions (in the standard STScI flags case). b. Transmission spectra of the sub-Neptune LP 791-18 c (median values with 1σ error bars) obtained when running the complete analysis from a standard stage 1 reduction with (blue) and without (orange) the NIN correction step. Residuals in the transit spectrum between the reductions with and without the NIN correction are shown in the bottom panel. The NIN correction step rectifies a wide trend in the transmission spectrum which is most important around 2 μm and decays towards longer wavelengths. The effect is similar to that observed in the case of TRAPPIST-1 g (Benneke et al., under review). c. Transmission spectra of the sub-Neptune LP 791-18 c (median values with 1σ error bars) obtained when using the light curves produced by the Eureka! (orange) and ExoTEDRF (green) reductions. Residuals in the transit spectrum between the Eureka! and ExoTEDRF reductions. The NIN correction is turned off in the Eureka! reduction (to isolate the difference between the two reduction pipelines). Both transmission spectra are in agreement.