Kinematically Coherent Multiphase Galactic Winds in Star-Forming Galaxies Revealed by Unified Radiative Transfer Modeling of UV Emission and Absorption Lines

Abstract

We present PEACOCK, a three-dimensional Monte Carlo radiative transfer (RT) framework designed to self-consistently model rest-frame ultraviolet emission and absorption lines arising from multiphase, clumpy galactic winds. Applied to deep HST/COS spectra of 50 nearby star-forming galaxies, PEACOCK reproduces 220 observed profiles of Ly-alpha, Si II, C II, Si III, Si IV, and C IV spanning absorption, emission, and P-Cygni-like morphologies within a single CGM model. By combining Monte Carlo RT with deep-learning acceleration and nested sampling, the framework enables fully converged multi-line inference at a small fraction of the cost of traditional RT grids. Systematic experiments show that ion column densities, bulk outflow velocities, and turbulent motions leave distinct imprints on line profiles, allowing the underlying gas properties to be constrained with minimal degeneracy. Purely radial accelerating flows often fail to reproduce the observed absorption morphologies, whereas macroscopic velocity dispersion naturally produces the broad asymmetric troughs seen in the data, indicating that turbulent motions are a key component of outflow kinematics. The inferred kinematics reveal strong coherence among low- and high-ionization metal lines in both bulk and turbulent velocities, consistent with a dynamically coupled multiphase wind. In contrast, neutral hydrogen shows weaker correspondence with metals, suggesting incomplete mixing and a distinct kinematic structure. By unifying emission and absorption diagnostics across multiple ions, PEACOCK provides a physically grounded bridge between UV observations and theoretical models of galactic winds.

Figures (46)

• Figure 1: Schematic of the multiphase, clumpy RT model PEACOCK used in this work. A central star-forming galaxy is embedded in a multiphase, clumpy CGM halo. Cool and warm ($T\!\sim\!10^{4}$ – $10^{5}\,$K) gas clumps populate the spherical halo with a radial number density profile. Each clump exhibits intrinsic microscopic turbulence ($b_{\rm D,\,cl}$), macroscopic turbulent motions ($\sigma_{\rm cl}$), and participate in a large-scale radial outflow ($v_{\rm cl,\,out}(r)$). Photons at the wavelengths of Ly$\alpha$ and UV metal lines (e.g., C ii, Si ii, Si iii, C iv, Si iv) are assumed to be emitted by the central galaxy and propagate through the clumpy medium, undergoing resonant scattering (and, for partially resonant transitions like C ii and Si ii, fluorescence), as well as dust absorption and scattering. Note that in the model, clumps are idealized as spheres rather than irregular shapes shown in the schematic.
• Figure 2: Energy level diagrams that illustrate fine-structure splitting of ground and excited states for C ii, Si ii, C iv, and Si iv ions. Starred labels (e.g., Si ii*, C ii*) indicate fluorescent transitions. Left: strictly resonant doublets like C iv and Si iv, which arise from transitions between the ground state $^2S_{1/2}$ and the fine-structure–split excited states $^2P_{1/2}$ and $^2P_{3/2}$. The oscillator strength of the higher-frequency component (the “K” line, shown in blue) is typically twice that of the lower-frequency “H” line (shown in red). Middle and right: partially resonant systems like Si ii and C ii, where the ground term $^2P_{1/2,\,3/2}$ is fine-structure split, allowing both resonant and fluorescent channels to the excited $^2D_{3/2,\,5/2}$ states. In each set of three allowed transitions, one fluorescent line (shown in orange) has a much lower oscillator strength and effectively serves as a bridge between the other two transitions (one resonant and one fluorescent, both shown in blue).
• Figure 3: Overview of the five major steps in our spectral fitting pipeline. (1) A mock spectral library is constructed by performing RT simulations for a number of parameter sets that sample the multidimensional model space. (2) A DNN is trained to learn the nonlinear mapping between model parameters and emergent spectra. (3) The trained DNN model is coupled with a nested sampling algorithm to efficiently fit the observed spectra and locate the global likelihood maximum in parameter space. (4) Posterior-guided adaptive expansion and targeted resampling are applied when the posterior extends to the edges of the explored space, adding new training samples to refine the DNN. (5) The reliability of the best-fit solution is validated by comparing the DNN-predicted spectrum with a direct RT calculation using the same parameters, confirming close agreement between the two.
• Figure 4: Six representative C ii$\lambda1334$ line profiles (black) and their best-fit RT models (red). The galaxy names are indicated in the upper-left corners. Velocity is shown relative to the systemic redshift, and the vertical dashed line marks $v = 0$. The observed spectra are plotted with 1$\sigma$ error bars (gray). The best-fit model parameters associated with these profiles are summarized in Table \ref{['tab:CII_params']}.
• Figure 5: Three representative Si ii$\lambda1260$ line profiles (black) and their best-fit RT models (red). The galaxy names are indicated in the upper-left corners. Velocity is shown relative to the systemic redshift, and the vertical dashed line marks $v = 0$. The observed spectra are plotted with 1$\sigma$ error bars (gray). The best-fit model parameters associated with these profiles are summarized in Table \ref{['tab:SiII_params']}. The Si ii$\lambda1260$ profile of galaxy J0127–0619 shows contamination from S ii absorption at 1259.52 Å, located at $\sim -200\,\mathrm{km\,s^{-1}}$ on the blue side of the line center.