Bayesian Analysis for Remote Biosignature Identification on exoEarths (BARBIE) IV: Analyzing CO2 Detections in the Near-IR to Determine the Long-Wavelength Cut-off for the Habitable Worlds Observatory Coronagraph

TL;DR

This study addresses how to maximize robust CO2 detections on Earth-like exoplanets with the Habitable Worlds Observatory by identifying the optimal long-wavelength cut-off for the coronagraph. Using BARBIE with KEN spectral grids and Bayesian model comparison, the authors evaluate 25 bandpasses from $0.8-2.0~\mu$m across various CO2 abundances and atmospheric compositions, accounting for degeneracies with CO, H2O, and CH4. They find CO detections are strongly hindered by CO and by overlapping H2O/CH4 features, and that longer wavelengths (notably near $1.52~\mu$m and $1.63~\mu$m) enable strong CO2 detections across many scenarios, with an optimal cut-off of $1.68~\mu$m. The results inform the HWO design and observing strategy, suggesting a primary near-$1.5-1.6~\mu$m channel and, if possible, parallel visible/NIR paths to constrain habitability and atmospheric composition while balancing thermal background costs.

Abstract

We present our analysis of how the detectability of carbon dioxide (CO2) on an Earth-like planet varies with respect to signal-to-noise ratio (SNR), wavelength, and molecular abundance. Using the Bayesian Analysis for Remote Biosignature Identification on exoEarths (BARBIE) methodology, we can inform the optimal long-wavelength cut-off for the future Habitable Worlds Observatory (HWO) coronagraph. We test 25 evenly-spaced 20% bandpasses between 0.8-2.0μm, and simulate data spanning a range of SNRs and molecular abundance to analyze the relationship between wavelength and detectability for different planetary archetypes. We examine abundance levels from varying Earth epochs and a Venus-like archetype to investigate how detectability would change throughout the evolution of a rocky planet. Here, we present our results on the planetary conditions and technological requirements to strongly detect CO2. In addition, we analyze the degeneracy of CO2 with carbon monoxide (CO), methane (CH4), and water (H2O). We determine that any abundance of CO does not achieve strong detections and that CH4 and H2O play a pivotal role in the ability to detect CO2. We conclude that the optimal long-wavelength cut-off for the Habitable Worlds Observatory coronagraph should be 1.68μm.

Figures (10)

• Figure 1: Illustrates the balancing act of keeping thermal noise minimal while maximizing the detectability of spectral features like CO2 and CO. The top plot shows the relative noise based on varying telescope temperature between 0.8-2.0$\mu$m. The bottom plot shows spectral features from 0.8-2.0$\mu$m of H2O, CH4, O2, CO, and CO2. We plot a modern-Earth spectra using the Planetary Spectrum Generator (PSG) with four varying CO2 abundances: Modern (4.2$\times10^{-4}$ VMR) in dark blue, Proterozoic-like (1$\times10^{-2}$ VMR) in light pink, Archean-like (1$\times10^{-1}$ VMR) in magenta, and Venus-like (9.6$\times10^{-1}$ VMR) in purple. The figure plots wavelength ($\mu$m) on the x-axis and apparent albedo (I/F) on the y-axis.
• Figure 2: CO2 and CO spectral features at varying abundances from 0.8-2.0$\mu$m. It plots wavelength ($\mu$m) on the x-axis and apparent albedo (I/F) on the y-axis. We plot three CO2 abundances from Table \ref{['tab:abundances']}: Paleozoic-like (3.65$\times10^{-4}$ VMR) in light pink, Proterozoic-like (1$\times10^{-2}$ VMR) in magenta, and a Venus-like (9.6$\times10^{-1}$ VMR) in purple. We plot three CO abundances: Modern (1$\times10^{-7}$ VMR - co-modern) in light pink, Archean-like (1$\times10^{-3}$ VMR - co-archean) in magenta, and the highest abundance (1$\times10^{-2}$ VMR - filler abundance) in purple. The first and last bandpasses are shaded in light blue and light purple, respectively.
• Figure 3: Strength of CO detections at SNRs of 3, 8, 12, 16 & 20 with the highest CO abundance, 1$\times10^{-2}$ VMR, throughout the wavelength range of 0.8-2.0$\mu$m. Light pink circles portray unconstrained detections, magenta diamonds portray weak detections, and there are no strong detections. $\mathrm{lnB}$ classification follows the values in Table \ref{['tab:lnB']}. The 68% credible region is shaded in grey. The true value of the CO abundance is marked by the horizontal dashed black line.
• Figure 4: Strength of CO2 detections at (a) 1.21 $\mu$m, (b) 1.44$\mu$m, (c) 1.52$\mu$m, and (d) 1.63$\mu$m with respect to CO2 abundance and SNR assuming a 20% bandpass. Columns on the x-axis are different CO2 abundances; Bridging, Proterozoic-like, Archean-like, and Venus-like, from left to right. SNR values are on the y-axis. The strength of detection is shown for each combination of CO2 abundance and SNR. As indicated by the color bar on the right side of each heat map, the darker the purple is, the stronger the detection is, and the lighter the pink is, the more unconstrained the detection is, following the same $\mathrm{lnB}$ classifications defined in Table \ref{['tab:lnB']}.
• Figure 5: Minimum SNR required for a strong CO2 detection at each bandpass center for the three highest CO2 abundances that achieve strong detections. Venus-like is plotted in purple stars, Archean-like is plotted in light pink squares, and Proterozoic-like is plotted in magenta circles. The shaded regions represent the spaces in which you could achieve a strong detection for each abundance in their respective colors. The x-axis is bandpass center ($\mu$m) and the y-axis is SNR.