The velocity field of our Milky Way outer stellar halo based on DESI DR2

Abstract: Using 64,000 halo K giants from Dark Energy Spectroscopic Instrument (DESI) second Data Release (DR2), we decompose the Milky Way (MW) stellar halo between 3 and 160 kpc into metal-rich (MR) and metal-poor (MP) components via a Gaussian mixture model (GMM). The two populations are nearly equal in number but chemically and kinematically distinct: MR stars occupy highly radial orbits with velocity anisotropy of beta ~0.94 and metallicity dispersion sigma([Fe/H]) ~0.17 dex, without obvious dependence on distance, and are mainly contributed by Gaia-Sausage/Enceladus (GSE) debris. MR component dominates the inner 30 kpc and re-emerges beyond 50 kpc, implying GSE debris can extend to ~70-80 kpc. MP stars exhibit a weaker radial bias of beta ~0.46, decreasing to -0.5 beyond 80 kpc, and with a larger metallicity dispersion of sigma([Fe/H]) ~0.46 dex, showing signatures of multiple minor mergers. Both components exhibit net prograde rotation at ~10-30 kpc with a stronger azimuthal signal in the MP population. The non-equilibrium motions of the outer halo (>50 kpc) are quantified with a dipole-plus-contraction velocity field. We find that the outer halo is simultaneously contracting (~-19 km/s, distance-independent) and subject to reflex motions (increases from -19 to -44 km/s with radius), reflecting the perturbation from the Large Magellanic Cloud (LMC). We also confirm a linear dependence of mean polar velocity for the outer stellar halo on the dipole velocity field, a direct consequence of the LMC and MW interaction. Our results provide a quantitative distance-resolved description of the MW's last major accretion event and its ongoing response to the first infall of the LMC.

• The paper presents a chemodynamical mapping of the Milky Way’s outer stellar halo using 64,000 K giants to detail radial biases and merger signatures.
• It applies Gaussian mixture models to decompose distinct metal-rich and metal-poor populations with varying anisotropies and net rotations.
• The analysis detects a non-equilibrium velocity field marked by contraction and dipole signals, indicating the dynamic influence of the LMC.

The Velocity Field of the Milky Way Outer Stellar Halo from DESI DR2

Introduction and Scientific Context

Understanding the chemo-dynamical structure of the Milky Way (MW) stellar halo provides critical constraints on galaxy formation and the dark matter potential. Recent advances, particularly from Gaia and large spectroscopic surveys, have enabled high-precision phase-space mapping of distant halo tracers, revealing the complex assembly history of the MW and its dynamical state. This paper presents a comprehensive analysis of the outer stellar halo ($3{\rm\ kpc} < r_{\rm GC} < 160 {\rm\ kpc}$) using 64,000 K giants from DESI DR2, with the goal of delineating the kinematic and chemical structure and quantifying disequilibrium features induced by the Large Magellanic Cloud (LMC).

Data and Methodology

The analysis leverages the DESI Milky Way Survey, which provides spectroscopic, photometric, and astrometric information, combined with Gaia proper motions. Stellar parameters and distances are derived via neural network models (SpecDis2), with extensive validation against globular clusters and dwarf galaxies, yielding typical distance uncertainties of $\sim12$–18% for K giants.

Sample selection includes effective temperature, absolute magnitude, metallicity, and spatial cuts to minimize disk contamination and remove known substructures (notably Sagittarius stream, globular clusters, and dwarf galaxies). Velocity uncertainties and the transformation to Galactocentric coordinates are stringently controlled.

The phase-space and chemical distributions are modeled predominantly with:

• Gaussian Mixture Model (GMM): Separating populations into metal-rich (MR) and metal-poor (MP) components, fit in $[V_r, V_\phi, V_\theta, {\rm [Fe/H]}]$ space.
• Dipole velocity field modeling: To capture non-equilibrium features, particularly the LMC-induced reflex motion in the outer halo, a dipole-plus-contraction model is applied to the radial velocity distribution as a function of sky position.

Single-Component Halo Properties

Marginalized posterior distributions from the single-component fit highlight the global properties of the halo K giant population:

Figure 1: Error contours and marginalized posteriors of velocity and metallicity parameters for the single-component model.

The results demonstrate:

• A pronounced velocity anisotropy, $\beta \approx 0.72$, indicating radially biased orbits.
• Modest net prograde rotation with $$\langle V_\phi \rangle \approx -14 \ {\rm km\ s^{-1}}$$ (in disk rotation direction).
• Radial velocity and polar velocity means consistent with zero out to $\sim 50$ kpc, with deviations appearing at larger radii, suggestive of disequilibrium.
• Velocity dispersion profiles indicate a transition from radial to isotropic at large radii, and a potential bias at small radii from thick disk contamination.

Chemo-Kinematic Decomposition: Metal-Rich and Metal-Poor Halo

A GMM-based double-component analysis reveals two distinct components—MR and MP—with nearly equal proportions but sharply different kinematics and metallicity dispersion:

Figure 3: Joint posterior contours for parameters of the double-component (MR and MP) model.

Key numerical and conceptual results:

• MR component: $\left< \rm [Fe/H]\right>\sim -1.15$ dex; $\sigma_{\rm [Fe/H]} \sim 0.17$ dex; $\beta_{\rm MR} \sim 0.94$, indicating highly radial, kinematically coherent debris, associated with the Gaia-Sausage/Enceladus (GSE) event.
• MP component: $\left< \rm [Fe/H]\right>\sim -1.58$ dex; $\sim12$0 dex; $\sim12$1, with increased isotropy/tangential bias at larger radii, and metallicity dispersion indicative of multiple minor accretion events.
• Both components exhibit net prograde rotation ($\sim12$2–$\sim12$3), but the amplitude is larger in the MP population—a notable, somewhat counterintuitive, kinematic feature.

The radial dependency of MR fraction highlights a dominant MR halo within 30 kpc, a sharp decrease from 30–50 kpc due to dropping GSE contribution, and a mild resurgence beyond 50 kpc.

Radial Trends in the Velocity Field and Anisotropy

The distance-binned analysis of MR and MP components shows pronounced differences in mean velocity, dispersion, and anisotropy:

Figure 2: Radial trends in mean velocity, dispersions, and anisotropy for MR (top) and MP (bottom) populations. Right panels compare with literature K giant, BHB, and RR Lyrae results.

• The MR component maintains a high $\sim12$4 ($\sim12$50.9) to 50 kpc; only beyond this does anisotropy weaken.
• The MP component transitions from moderate radial bias in the inner halo to isotropy and tangential bias at $\sim12$6 kpc.
• The observed net prograde rotation in the MP population is robust, present beyond the regime where disk/stream contamination is significant.

Crucially, the superposition of these two populations with different radial profiles and proportions naturally reproduces the heterogeneous anisotropy profiles measured in independent studies using different tracers.

Non-Equilibrium Features: Reflex and Contraction Motions from LMC Interaction

The velocity field of the outer halo reveals systematic non-equilibrium motions, quantified via the dipole-plus-contraction model:

Figure 6: Mean radial velocity profiles for northern/southern halo at various radii; comparison with other works and N-body predictions.

Figure 4: Posterior contours for the dipole velocity field model parameters describing the outer halo reflex and contraction motions.

Figure 5: Mollweide projections of observed (left) and best-fit (right) mean radial velocity for $\sim12$7 kpc, demonstrating the hemispheric asymmetry attributed to LMC perturbation.

Notable findings:

• The outer ($\sim12$8 kpc) halo shows simultaneous contraction ($\sim12$9) and a dipole reflex motion ($[V_r, V_\phi, V_\theta, {\rm [Fe/H]}]$0 increasing from $[V_r, V_\phi, V_\theta, {\rm [Fe/H]}]$1 to $[V_r, V_\phi, V_\theta, {\rm [Fe/H]}]$2 with radius), consistent with the signature of the LMC infall scenario.
• Hemispheric asymmetries in mean $[V_r, V_\phi, V_\theta, {\rm [Fe/H]}]$3 (stronger in the south) are not fully matched by N-body models, motivating further investigation of the contraction origin.
• The relationship $[V_r, V_\phi, V_\theta, {\rm [Fe/H]}]$4 directly links observed polar velocity in the outer halo to the LMC-induced kinematic dipole, as anticipated from simulation.

Systematic Effects: Impact of Distance Uncertainties

Distance uncertainties induce Eddington bias, particularly for large-radius and low-number-density bins. Quantitative correction shows outer-halo MR fraction and velocity anisotropy are modestly overestimated at large radii, but not enough to overturn key conclusions.

Figure 7: Impact of 18% and 30% relative distance uncertainties on MR fraction and MP velocity anisotropy as a function of radius.

Astrophysical Implications and Theoretical Outlook

• The dichotomy between the MR (GSE-dominated) and MP (minor merger-dominated) populations encodes the hierarchical assembly and last major accretion event in the MW's halo.
• The highly radial, chemically narrow MR population corroborates the GSE impact, while the more isotropic, metal-poor component provides a fossil record of numerous minor mergers.
• The robust net prograde rotation in the MP component, and retrograde features at the most extreme radii, are at tension with naive expectations; potential mechanisms include bar-halo coupling, anisotropic accretion, or disk-induced angular momentum transfer, though these remain incompletely understood.
• The clear detection of a distance-independent contracting motion in the outer halo, on par with the LMC-induced dipole reflex, is physically notable. The underlying cause remains uncertain—candidate explanations include genuine time-dependent halo response to LMC infall, lingering unrelaxed substructure, or limitations in current N-body models.
• These dynamical signatures, when interpreted in the context of simulations, provide stringent constraints on the MW mass profile, dark matter halo triaxiality, and the dynamical timescales and orbits of major accreted structures.

Conclusion

This analysis of the DESI DR2 K giant sample yields a comprehensive, distance-resolved kinematic and chemical map of the outer MW stellar halo. The data robustly identify dual chemo-dynamical halo populations whose radial trends and anisotropies reflect their assembly history. Observed non-equilibrium features, especially the dipole reflex and contraction motions, provide direct evidence of ongoing dynamical response to the LMC, with clear observational-theoretical synergy and outstanding theoretical challenges. These results, including the strong, somewhat counterintuitive net rotation in the metal-poor halo and the robust LMC perturbation signature, establish a new empirical baseline for galactic archaeology, halo modeling, and the study of time-dependent processes in massive galaxies.