An Hα Transit of HD 189733b to Assess Stellar Activity Across the Transit Chord Close to JWST Observations

TL;DR

This paper addresses how stellar activity on HD 189733 can bias transmission spectra of its transiting hot Jupiter, particularly in the JWST era. By acquiring high-resolution Hα spectra with MEGARA on the GTC during a transit, the authors detect a mid-transit decrease in Hα indicating a star-spot crossing and estimate the occulted region’s size as about 3.47 Earth radii. They demonstrate that such activity can shift infrared molecular features by tens of ppm, comparable to JWST’s noise floor, highlighting the risk of stellar contamination in atmospheric inferences. The work argues for concurrent ground-based high-resolution activity monitoring alongside JWST observations to disentangle stellar and planetary signals and improve the robustness of exoplanet atmosphere characterizations.

Abstract

Transmission spectroscopy allows us to detect molecules in planetary atmospheres, but is subject to contamination from inhomogeneities on the stellar surface. Quantifying the extent of this contamination is essential for accurate measurements of atmospheric composition, as stellar activity can manifest as false atmospheric signals in planetary transmission spectra. We present a study of hot Jupiter HD 189733b, which has over 50 hours of JWST observations scheduled or taken, to measure the activity level of the host star at the current epoch. We utilize high-resolution spectra of the H$α$ line from the MEGARA spectrograph on the 10-m GTC to examine the activity level of HD 189733 during a transit. We measure H$α$ becoming shallower mid-transit by an H$α$ index of $δ$ = 0.00156 ${\pm}$ 0.00026, which suggests that HD 189733b crosses an active region as it transits. We posit this deviation is likely caused by a spot along the transit chord with an approximate radius of $R_{spot}$ = 3.47 ${\pm}$ 0.30R${\oplus}$ becoming occulted during transit. Including an approximation for unocculted spots, we estimate that this spot could result in transit depth variations of ${\sim}$17 ppm at the 4.3 micron CO2 feature. Since this is comparable to JWST NIRCam Grism mode's noise floor of ${\sim}$20 ppm, it could bias atmospheric studies by altering the inferred depths of the planet's features. Thus, we suggest ground-based high-resolution monitoring of activity indicator species concurrently taken with JWST data when feasible to disentangle stellar activity signals from planetary atmospheric signals during transit.

Figures (8)

• Figure 1: Example R$\sim$20,000 MEGARA spectrum of HD 189733 after extraction and normalization, with the TAPAS telluric model Bertaux2014Apr overlaid in the Earth rest frame. The solid blue line is the MEGARA spectra of HD 189733, and the solid red line is the TAPAS telluric model. The teal dot represents the telluric peak in the MEGARA data, while the violet dot represents the telluric peak in the TAPAS telluric model. Both have a wavelength of $\lambda = 6580.8$ Å in the Earth rest frame. We also have labeled the H$\alpha$ spectral line in magenta and the Fe I reference line in blue. The Fe I reference line is chosen due to its proximity to H$\alpha$ and its low Landé g factor of $g_{eff} = 0.820$, which was retrieved from the Vienna Atomic Line Database 3 (VALD).
• Figure 2: MEGARA spectrum of H$\alpha$ within its feature window of 6557.9 Å to 6567.6 Å. The larger magenta dots represent the original spectral data points, while the smaller blue dots represent the interpolated data points used to integrate along 0.4 Å window about the line center (see Section 4). The solid orange lines denote this integration window, and the dashed orange line denotes the line center.
• Figure 3: H$\alpha$ index as a function of time in hours, computed using the methodology described in Section 4. The indigo points represent the binned H$\alpha$ indexes with N=6 points per bin, while light blue points represent the unbinned H$\alpha$ indexes. The dashed magenta line shows the transit mid-point and the shaded pink region shows the total transit duration of 1.81 hours. The solid blue line is the Gaussian model that we fit to the data. We measure the increase in H$\alpha$ index by calculating the amplitude of the Gaussian model, which we find to be $\delta = 0.00156 \pm 0.00026$ during the middle of the transit. This is similar in magnitude to the H$\alpha$ index deviations measured by 2016MNRAS.462.1012B. This signals that the line is becoming shallower while it is occulted by the planet, after normalizing each frame. We interpret this to be a star spot crossing, based on the results of our simulated H$\alpha$ transit shown in Figure \ref{['fig:smart']} and Figure \ref{['fig:enter-label']}. This deviation is consistent with the planet crossing a star spot with a radius of $= 3.47 \pm 0.30 R_{\oplus}$. Additionally, we measure the full-width half-max (FWHM) of the Gaussian model of the H$\alpha$ indexes to be 0.57 $\pm$ 0.13 hours (34 $\pm$ 8 minutes), which is comparable to the time it would take for the planet to cross itself, about 24 minutes. This is in line with an occultation of an active region that is smaller than the planet itself, since the planet's diameter then drives the occultation timescale.
• Figure 4: H$\alpha$ index divided by Fe I index. The blue dots represent the binned index ratio, while the pine green dots represent the unbinned index ratio. The pink shaded region shows the transit duration, and the dashed magenta line shows transit mid-point. This figure shows similar behavior to Figure \ref{['fig:Halphadepths']}, which indicates that the source of the signal we observe in the H$\alpha$ index time series is likely not significantly affecting the control Fe I line we observe here. We calculated a Pearson Correlation Coefficient between H$\alpha$ index and Fe I index and find a coefficient of $\rho_{H\alpha,Fe I} = -0.0051$ with a p-value of $p = 0.95$, showing there is no statistically significant correlation in Figure \ref{['fig:HalphaFeIcorr']}.
• Figure 5: Correlation plots of two methods used to check that H$\alpha$ signal is stellar in origin. Top: H$\alpha$ index as a function of Fe I index. We measure a Pearson R Correlation Coefficient of $\rho_{H\alpha,Fe\space I} = -0.0051$ with a p-value of $p = 0.95$ between H$\alpha$ index and Fe I index, and $\rho_{H\alpha, airmass} = -0.067$ with a p-value of $p = 0.45$ between H$\alpha$ index and airmass. No statistically significant correlation is present, which suggests that the origin of the signal variation we observe in H$\alpha$ does not appear to strongly impact Fe I. Since this Fe I line is not magnetically sensitive, the origin of this H$\alpha$ signal could indeed be stellar activity. Bottom: H$\alpha$ index as a function of airmass of the observation. We again do not see any statistically significant correlation, which suggests that airmass or other observational effects are not the origin of this H$\alpha$ variation either.