Synthetic Transmission Spectra

• Synthetic transmission spectra are computational predictions that quantify the wavelength-dependent transit depth of exoplanets by integrating atmospheric optical depths.
• They combine detailed radiative transfer, cloud microphysics, and molecular opacities to support target ranking and atmospheric retrieval in exoplanet studies.
• Degeneracies between clouds, composition, and temperature profiles require multi-wavelength diagnostics and tailored retrieval methods to accurately interpret observations.

Synthetic transmission spectra are theoretical or computational predictions of the wavelength-dependent transit depth produced as a planet passes in front of its host star. These spectra encode the integrated effect of atmospheric composition, thermal structure, clouds/hazes, planetary/stellar parameters, and (for dynamic scenarios) non-equilibrium processes. Synthetic spectra are indispensable for atmospheric retrieval, instrument planning, cloud and haze modeling, and the interpretation of observations from current and future facilities.

1. Fundamental Radiative Transfer and Mathematical Formalism

Synthetic transmission spectra are computed by evaluating the wavelength-dependent transit depth, typically expressed as the effective planet-to-star radius ratio squared, $$\delta(\lambda) = [R_p^{\rm eff}(\lambda)/R_\star]^2$$. This follows from integrating the slant optical depth $\tau(\lambda, b)$ along a chord of impact parameter $b$ through a stratified planetary atmosphere:

$$\tau(\lambda, b) = \int_{-\infty}^{+\infty} \sum_i n_i(s) \sigma_i(\lambda, T(s), p(s)) ds$$

where $n_i$ is the number density and $\sigma_i(\lambda)$ the cross-section of absorber $i$. The corresponding transmitted flux is $T(\lambda, b) = \exp[-\tau(\lambda, b)]$, and the aggregate transit light loss is

$$\Delta F(\lambda) = \frac{1}{\pi R_\star^2} \int_0^\infty 2\pi b [1 - \exp(-\tau(\lambda, b))] db$$

The effective transit radius is then determined by setting $$[R_p^{\rm eff}(\lambda)]^2 = R_p^2 + 2 \int_{R_p}^\infty [1 - \exp(-\tau(\lambda, r))] r dr$$.

Analytic isothermal solutions exist for idealized atmospheres. For an isothermal hydrostatic atmosphere under certain assumptions (constant $\tau(\lambda, b)$0, $\tau(\lambda, b)$1, cross-section independent of $\tau(\lambda, b)$2), one finds the canonical relation (neglecting clouds/hazes):

$\tau(\lambda, b)$3

where $\tau(\lambda, b)$4 is a reference radius corresponding to surface or deep atmosphere, $\tau(\lambda, b)$5 is the scale height, and $\tau(\lambda, b)$6 is the slant optical depth at $\tau(\lambda, b)$7 (Jordán et al., 2018, Heng et al., 2017). Corrections for non-isothermal profiles and full numerical integration are standard in modern codes.

In complex models, rays traverse 3D GCM columns, accounting for variable cloud properties, chemistry, and thermal structure as in (Lines et al., 2018).

2. Microphysics: Composition, Clouds, and Opacity Construction

The total opacity in each atmospheric layer arises from gaseous absorption, Rayleigh and Mie scattering, collision-induced absorption (CIA), and condensed-phase clouds or hazes. The extinction coefficient is

$\tau(\lambda, b)$8

where $\tau(\lambda, b)$9 are mixing ratios, and $b$0 is obtained from cloud microphysics.

Cloud opacities are constructed by:

• Extracting local particle size, number density, and composition from GCMs or parameterized models.
• Computing the effective complex index of refraction (e.g., Bruggeman mixing for silicate clouds).
• Applying Mie theory to determine $b$1 and $b$2 for the mean grain size.
• The extinction coefficient per layer: $b$3.

Vertical and horizontal inhomogeneities arise naturally in 3D GCM models, where $b$4, grain size, and composition vary (Lines et al., 2018).

Chemically, retrieved mixing ratios or explicit compositional parameterizations (e.g., via centered log-ratio for H-poor atmospheres) are forward-modeled within frameworks such as Aurora (Welbanks et al., 2021).

3. Cloud Heterogeneity, Degeneracies, and Observational Diagnostics

Non-uniform clouds and inhomogeneous haze coverage can induce prominent degeneracies. Fractional cloud coverage $b$5 along the terminator introduces a mixed spectrum:

$b$6

Patchy clouds can exactly mimic the muted water and Rayleigh features caused by high mean molecular weight atmospheres in limited wavelength ranges (notably the HST WFC3 window) (Line et al., 2015). Both partially cloudy, solar-$b$7 and clear, high-$b$8 solutions can fit the same synthetic or observed spectra. This cloud–composition degeneracy complicates retrievals and calls for diagnostics outside the principal bandpass.

Physical discriminants include:

• Ingress/egress residuals ($b$9100 ppm) arising from limb inhomogeneities, detectable with sub-second timing precision.
• Rayleigh slopes ($$\tau(\lambda, b) = \int_{-\infty}^{+\infty} \sum_i n_i(s) \sigma_i(\lambda, T(s), p(s)) ds$$0 μm), which differ between patchy clouds and high-$$\tau(\lambda, b) = \int_{-\infty}^{+\infty} \sum_i n_i(s) \sigma_i(\lambda, T(s), p(s)) ds$$1 models.
• Strong molecular bands at longer $$\tau(\lambda, b) = \int_{-\infty}^{+\infty} \sum_i n_i(s) \sigma_i(\lambda, T(s), p(s)) ds$$2 ($$\tau(\lambda, b) = \int_{-\infty}^{+\infty} \sum_i n_i(s) \sigma_i(\lambda, T(s), p(s)) ds$$3 μm), which are present in high-metallicity but not patchy-cloud atmospheres.

Cloud properties (opacity, deck top pressure, composition) fundamentally alter spectral visibility of molecular bands and the magnitude of features (Lines et al., 2018, Komacek et al., 2019).

4. Model Implementations: Grid-Based, Forward, and Machine-Learning Spectra

Synthetic transmission spectra generation occurs through both direct, grid-based forward modeling and advanced retrieval architectures:

• Forward 3D GCM-based: High-resolution, time-dependent runs (e.g., with SOCRATES) generate thousands of spectra incorporating kinetic cloud formation and full radiative transfer (Lines et al., 2018).
• Grid-based parameter sweeps: Multi-dimensional grids explore composition, temperature, cloud properties, and instrument noise impacts (Duque-Castaño et al., 2024). Tools such as TauREx 3 and MultiREx facilitate automated large-scale spectral library generation.
• Atmospheric escape databases ("sunset"): For escape-driven signatures, 1D Parker wind models and NLTE line transfer predict line absorption for nearly all transiting planets using codes like Cloudy (Linssen et al., 2024).
• Advanced retrieval codes: Aurora allows inhomogeneous clouds, Mie forward scattering, refraction, and compositionally agnostic retrievals for both H-rich and H-poor cases (Welbanks et al., 2021).
• Machine-learning classification: Synthetic spectra incorporating stellar contamination and instrument noise are used to train models for biosignature detection at low SNR (Duque-Castaño et al., 2024).

Instrument effects are crucial; realistic noise models (e.g., PandExo for JWST NIRSpec) and convolution with line spread functions are standard.

5. Physical and Instrumental Factors Shaping Spectral Features

Synthetic transmission spectra encode the interplay between:

• Atmospheric composition (e.g., $$\tau(\lambda, b) = \int_{-\infty}^{+\infty} \sum_i n_i(s) \sigma_i(\lambda, T(s), p(s)) ds$$4, $$\tau(\lambda, b) = \int_{-\infty}^{+\infty} \sum_i n_i(s) \sigma_i(\lambda, T(s), p(s)) ds$$5, $$\tau(\lambda, b) = \int_{-\infty}^{+\infty} \sum_i n_i(s) \sigma_i(\lambda, T(s), p(s)) ds$$6, alkalis, metallicity).
• Thermal/pressure structure (isothermal, non-isothermal, scale height $$\tau(\lambda, b) = \int_{-\infty}^{+\infty} \sum_i n_i(s) \sigma_i(\lambda, T(s), p(s)) ds$$7).
• Cloud/haze opacity (magnitude and altitude), compositional heterogeneity, particle size.
• Physical processes: rainout, gravitational settling, photochemistry, escape, vertical mixing.

Spectral markers include:

• Silicate cloud features (e.g., 8–12 μm band, peaking near 9 μm).
• Pure-component cloud features (e.g., $$\tau(\lambda, b) = \int_{-\infty}^{+\infty} \sum_i n_i(s) \sigma_i(\lambda, T(s), p(s)) ds$$8, $$\tau(\lambda, b) = \int_{-\infty}^{+\infty} \sum_i n_i(s) \sigma_i(\lambda, T(s), p(s)) ds$$9 at 9–10 μm; $n_i$0 blue-shifted to $n_i$18.5 μm).
• Muted alkali and $n_i$2/CO bands under high cloud opacity.
• CO$n_i$3 and CH$n_i$4 bands robust to cloud muting if well-mixed above cloud decks (Komacek et al., 2019, Lines et al., 2018).

In escape regime spectra, synthetic models explore He I $n_i$5 Å triplet, metal UV/optical lines, their dependence on XUV flux, and atmospheric parameters (Linssen et al., 2024, Young et al., 2020).

6. Degeneracies, Limitations, and Validation Against Observations

Intrinsic degeneracies include:

• Reference radius/pressure/abundance: Only the combination $n_i$6 is constrained, absolute abundance or reference radius is degenerate (Heng et al., 2017, Jordán et al., 2018).
• Cloud/mean molecular weight (μ): Fractionally cloudy, solar-$n_i$7 atmospheres vs. clear, high-$n_i$8 are often indistinguishable in broad-band spectra (Line et al., 2015).
• Cloud-top and compositional ambiguity: Apparent feature muting can result from high-altitude clouds or compositional effects; both affect $n_i$9, feature amplitude, and spectral slopes (Lines et al., 2018, Komacek et al., 2019).

Validation against broadband and high-resolution data require comparison of not only feature depths but detailed line profiles and continuum slopes. Forward models validated against, for example, HST WFC3 and ground-based optical measurements, highlight the need for improved cloud physics, inclusion of photochemical/3-D effects, and non-LTE (NLTE) treatments where radiative processes dominate, especially for upper-atmosphere and escape signatures (Young et al., 2020).

7. Applications: Target Ranking, Retrieval, and Strategic Observational Planning

Synthetic transmission spectra are central to:

• Instrumental design and target prioritization: Metrics such as the "He-TSM" (transmission-spectroscopy metric for He 10833 Å) integrate model-predicted feature strength with stellar magnitude for ranking observational potential (Linssen et al., 2024).
• Atmospheric composition retrieval: High-throughput synthetic spectra sets underpin machine-learning and Bayesian retrieval pipelines for rapid spectral interpretation and identification of biosignatures even in low SNR scenarios (Duque-Castaño et al., 2024).
• Bulk composition of disintegrating planets: Synthetic Mie-theory-based transmission spectra of mineral dust tails discriminate interior composition when interpreted with JWST+SPICA joint wavelength coverage (Okuya et al., 2020).
• Theory–observation synthesis: Forward models bridging 3D GCMs and retrieval analyses enable robust interpretation and constrain the limitations of existing atmospheric physics (Lines et al., 2018, Welbanks et al., 2021).

Ongoing development focuses on expanding the range of physical effects (e.g., NLTE, velocity/hydrodynamics, 3D effects), public synthetic spectrum databases, and sophisticated frameworks for planet characterization across a diversity of atmospheric regimes.