--- title: 'BlackHoleWeather: Multiscale Accretion & Feedback' url: https://www.emergentmind.com/topics/blackholeweather type: topic --- # BlackHoleWeather: Multiscale Accretion & Feedback BlackHoleWeather denotes a family of weather-based descriptions of black-hole environments. In its principal contemporary usage, it is a physically motivated, multiscale framework for supermassive black hole (SMBH) feeding and feedback in which turbulence, radiative cooling, multiphase condensation, chaotic cold accretion (CCA), and jet regulation are coupled across halo, galactic, and sub-parsec scales [2605.27503][2605.27511]. Closely related work extends the framework to SMBH spin evolution, treating the delivery of three-dimensional torques by cold clouds and filaments as part of the same baryon cycle [2605.27502][2605.27508]. Earlier work used a “weather forecast” analogy for the changing appearance of Sgr A* under an accretion-rate increase [1204.1371], while “Extremal Black Hole Weather” applied the term to weakly non-linear quasinormal-mode dynamics near extremal Kerr [2412.02821]. The term therefore names not a single model, but a set of linked black-hole “weather” programs centered on time-dependent, multiscale structure. ## 1. Definition, scale hierarchy, and condensation criterion In the CCA literature, BlackHoleWeather describes a stratified, turbulent, radiatively cooling atmosphere in which density perturbations condense into warm and cold clouds that “rain” toward the SMBH. This process differs from smooth Bondi inflow by producing multiphase, anisotropic, highly time-variable feeding, with cloud collisions and angular-momentum cancellation playing a central role [2605.27504]. The framework is organized around a radial hierarchy. The **macro-scale** corresponds to the broader hot halo, extending from kiloparsec to tens-of-kiloparsec scales. The **meso-scale** is the \(0.1\)–\(1\,\mathrm{kpc}\) “weather layer,” where condensation, cloud–cloud interactions, jet stirring, and final transport toward the sink are directly coupled. The **micro-scale** is the inner accretion and jet-launching region, resolved to sub-parsec or parsec scales depending on the simulation suite [2605.27503][2605.27511]. A central control parameter is the cooling-to-eddy ratio, \[ \mathcal{C}\equiv \frac{t_{\rm cool}}{t_{\rm eddy}}. \] In the turbulence-driven CCA studies, \(\mathcal{C}\sim1\) identifies the prone-to-condense regime: cooling and turbulent mixing are comparable, so overdensities can nonlinearly condense. The reported scatter is \(0.3\,\mathrm{dex}\). By contrast, \(\mathcal{C}\ll1\) corresponds to cooling-flow-like runaway behavior, whereas \(\mathcal{C}\gg1\) indicates suppression of condensation by efficient mixing [2605.27504]. In the jet-regulated variant, the same criterion is used spatially: \(C\sim1\) is reached mostly outside the jet cone and near the jet–ambient interface, while the excavated hot channel remains at \(C\gg1\) [2605.27511]. The framework also introduces a phenomenological weather lexicon. “Rainy” states denote compact, coherent, centrally connected condensation. “Stormy” states denote extended, filamentary, strongly stirred condensation across the meso- and inner macro-scales. In jet-regulated runs, a later “cloudy” state can emerge when multiphase gas remains present but becomes dynamically decoupled from the sink [2605.27503][2605.27511]. In the spin-coupled extension, a “sunny” hot-dominated center is additionally used for a torque-starved, low-power state [2605.27502]. ## 2. Turbulence-driven chaotic cold accretion without jets The non-jet CCA realization of BlackHoleWeather uses 3D hydrodynamic hyper-zoom simulations of a group-scale halo with radiative cooling and driven subsonic turbulence. One implementation employs a fully GPU-accelerated, adaptive-mesh-refinement hydrodynamics code based on the Athena++/Parthenon/Kokkos framework, with a second-order Godunov scheme, piecewise-linear reconstruction, an HLLC Riemann solver, and a Runge–Kutta integrator. The computational domain is a \(50\,\mathrm{kpc}\) cube with \(128^3\) base cells and 12 nested AMR levels, reaching \(\Delta x_{\min}\simeq0.1\,\mathrm{pc}\) in the innermost \(\sim10\,\mathrm{pc}\). A spherical sink of radius \(r_{\rm sink}\simeq0.4\,\mathrm{pc}\) absorbs gas onto the SMBH [2605.27504]. The halo is stratified in a fixed group potential with an NFW dark-matter halo, a Hernquist stellar bulge, and a central SMBH of mass \(M_\bullet=2.8\times10^8\,M_\odot\). Turbulence is driven as a purely solenoidal Ornstein–Uhlenbeck acceleration field with correlation time \(t_{\rm corr}=30\,\mathrm{Myr}\) and injection scale \(L_{\rm inj}\approx12.5\,\mathrm{kpc}\) in one suite, and on \(\sim25\,\mathrm{kpc}\) scales in another [2605.27504][2605.27507]. Two endpoint regimes are contrasted: **weak turbulence**, with \(\mathcal{M}\sim0.15\) and \(\sigma_v\approx60\)–\(90\,\mathrm{km\,s^{-1}}\), and **strong turbulence**, with \(\mathcal{M}\sim0.4\) and \(\sigma_v\approx210\)–\(230\,\mathrm{km\,s^{-1}}\) [2605.27504]. Both regimes become thermally unstable and develop a multiphase medium spanning \(8\)–\(10\) dex in temperature and density. The rainy state condenses earlier and remains compact: one reported run forms first cold gas in \(\sim7\,\mathrm{Myr}\) after cooling begins, with most cold gas confined within \(\lesssim100\,\mathrm{pc}\). The stormy state condenses later, around \(\sim30\,\mathrm{Myr}\), but sustains a filament-rich rain pattern to kiloparsec radii, with a tangled network of cold filaments extending up to \(\sim10\,\mathrm{kpc}\) before fragmentation and infall [2605.27507]. At micro-scales, inflow is partly mediated by a clumpy rotating torus, with reported radii of \(\sim5\,\mathrm{pc}\) in rainy runs and \(\sim10\,\mathrm{pc}\) in stormy runs [2605.27507]. The accretion outcome is strongly super-Bondi but predominantly low-Eddington. In one suite, the instantaneous SMBH inflow rate spans \(\dot M_\bullet\sim10^{-3}\)–\(10^{-1}\,M_\odot\,\mathrm{yr^{-1}}\), with \(\sim1\)–\(2\) dex variability. The corresponding Eddington-ratio distributions peak at low \(\lambda\sim{\rm few}\times10^{-4}\), and only \(\sim1\%\) of the time exceeds \(\lambda\gtrsim10^{-2}\), indicating a maintenance-mode state [2605.27504]. In another analysis, \(\dot M/\dot M_B\approx10\)–100, with instantaneous peaks up to \(\sim100\times\) the Bondi baseline [2605.27507]. A recurrent result is that condensed cold mass and SMBH feeding need not scale together. One study reports time-averaged cold-gas masses differing by \(\sim10^3\) between stormy and rainy CCA, yet with similar accretion onto the SMBH. The stated interpretation is that feeding is governed primarily by how efficiently multiphase structures couple to the central inflow rather than by their total condensed mass [2605.27507]. This directly undercuts the common assumption that more cold gas automatically implies stronger fueling. ## 3. Jet-regulated BlackHoleWeather The jet-regulated extension embeds CCA in a self-regulated kinetic-feedback loop. In this picture, cold clouds feed the SMBH; the SMBH responds intermittently with a jet; the jet excavates channels, uplifts cool gas, compresses and entrains condensates, drives turbulence, and resets the conditions for the next precipitation episode [2605.27511]. The problem is therefore no longer pure cooling plus turbulence, but a closed feeding–feedback cycle. The numerical realization uses two hydrodynamical simulations of a turbulent, radiatively cooling galaxy-group atmosphere with self-regulated AGN feedback. One setup employs a \(100\,\mathrm{kpc}\) Cartesian box with 10 levels of static mesh refinement and finest cell size \(\Delta x\approx0.78\,\mathrm{pc}\). The sink radius is \(r_{\rm sink}\approx3.1\,\mathrm{pc}\). A bipolar kinetic jet is injected through cylinders aligned with \(\pm z\), with radius \(r_{\rm jet}\approx2.34\,\mathrm{pc}\), thickness \(d_{\rm jet}\approx1.56\,\mathrm{pc}\), fixed total mechanical efficiency \(\epsilon_{\rm tot}=0.028\), and velocity \(v_{\rm jet}\approx0.25\,c\) [2605.27503]. Jet coupling is anisotropic. The reported mechanisms are **heating**, through a low-density hot cocoon and shocks inside the cone; **compression**, through jet–ambient shear and shock–cloud collisions; **entrainment**, which uplifts cold and warm clumps along the cocoon boundary; and **turbulent mixing**, which broadens density and temperature distributions on meso-scales [2605.27503]. Consequently, condensation is suppressed inside the jet channel but survives in the surrounding atmosphere and along the jet–ambient interface [2605.27503][2605.27511]. Three jet-regulated weather states are emphasized. In the **stormy phase** of the stronger-turbulence run, condensation is delayed until \(t_{\rm rain}\approx16\,\mathrm{Myr}\) after feedback onset, then becomes extended, filamentary, and mixed, with a porous cocoon, a broad hot–warm–cold bridge, burst-dominated fueling, and super-Bondi peaks of \(\dot M_\bullet\sim10^2\)–\(10^3\,\dot M_{\rm Bondi}\) [2605.27503][2605.27511]. In the **rainy phase** of the weaker-turbulence run, condensation begins earlier, at \(t_{\rm rain}\approx9\,\mathrm{Myr}\), remains coherent and centrally confined to \(r\lesssim1\,\mathrm{kpc}\), builds a longer-lived inner cold reservoir, and sustains elevated accretion above Bondi [2605.27503]. In the later **cloudy phase** of the high-turbulence run, cold and warm gas remain abundant on meso-scales, but sink coupling weakens: average feeding drops, the central reservoir is depleted or disrupted, and accretion becomes weaker and intermittent [2605.27503][2605.27511]. The variability diagnostics reinforce this distinction. The accretion-rate power spectral density follows a broken power law with low-frequency flicker-noise slopes \(\alpha_1\approx1.1\)–1.3 and high-frequency red-noise tails \(\alpha_2\approx3.7\)–3.8. Stormy epochs have high normalization and steep flicker-noise slopes, whereas cloudy epochs show lower normalization and a flattened low-frequency slope of \(\sim f^{-0.8}\) [2605.27511]. Phase-separated mass-flux measurements show fountain-like recycling in the strongly stirred run, but inner-kpc recycling in the calmer run [2605.27511]. These results support the stated conclusion that jet-regulated CCA is controlled by meso-scale transport, not only by cold-gas production [2605.27511]. ## 4. Spin-coupled BlackHoleWeather The spin-coupled program adds SMBH angular-momentum evolution to the same multiscale cycle. In this extension, the decisive variable is not only how much gas reaches the center, but whether the delivered angular momentum is coherent enough to change the SMBH spin and reorient the jet [2605.27502][2605.27508]. A time-dependent spin vector is defined by \[ \mathbf{a}=\frac{c\,\mathbf{L}_\bullet}{G M_\bullet^2}. \] The preferred “Hybrid” model determines the **direction** of accretion torque from the resolved sink-scale angular momentum, but filters its **magnitude** through a Kerr ISCO closure. The spin is then updated under the combined action of accretion torque and Blandford–Znajek spin-down torque [2605.27502]. This is contrasted with a **Fixed-axis** benchmark, in which the jet axis remains locked along \(z\), and a **Direct** prescription, in which the sink-scale torque is used without ISCO filtering. The reported result is that the Direct model overestimates spin variability and jet-axis wandering, whereas the Hybrid model is bracketed by analytic limits [2605.27502]. The spin-coupled simulations were carried out in a \((100\,\mathrm{kpc})^3\) box with ten levels of static mesh refinement, reaching \(\Delta x_{\min}\simeq0.7\,\mathrm{pc}\) in the central \(\sim20\,\mathrm{pc}\). A Lagrangian sink particle with \(r_{\rm sink}\sim3\,\mathrm{pc}\) represents the SMBH. Four runs compare **Driven-Turbulence** (DT) and **Interrupted-Turbulence** (IT) suites, with low and high stirring amplitudes [2605.27508]. At large radii, all runs show comparable inflow, \(\dot M\sim10^2\,M_\odot\,\mathrm{yr^{-1}}\) for \(r\gtrsim10\,\mathrm{kpc}\). Inside \(r\sim10\,\mathrm{pc}\), however, the IT runs still deliver \(\dot M\sim0.2\)–\(2\,M_\odot\,\mathrm{yr^{-1}}\), whereas the DT runs collapse to \(\sim10^{-3}\)–\(10^{-2}\,M_\odot\,\mathrm{yr^{-1}}\), a drop of \(2\)–\(3\) dex relative to the IT controls [2605.27508]. This split is encoded in the torque-coherence parameter \[ \chi_j(t;W)=\frac{\left|\sum \mathbf{L}_{\bullet,{\rm acc}}\right|}{\sum \left|\mathbf{L}_{\bullet,{\rm acc}}\right|}, \] measured over a trailing window \(W\simeq10\,\mathrm{Myr}\). The IT runs, especially the low-interrupted-turbulence case, sustain \(\chi_j\gtrsim0.8\)–1 for tens of Myr, while the high-driven-turbulence run often settles to \(\chi_j\sim0.3\)–0.6 [2605.27508]. In the related Hybrid turbulent runs, the reported medians are \(\chi_j=0.92\) for low turbulence and \(0.69\) for high turbulence, with higher mean accretion and larger maximum inclination in the low-turbulence case [2605.27502]. Low-spin SMBHs are also stated to be easier to reorient because a misaligned torque acts on a smaller angular-momentum reservoir [2605.27502]. Jet-axis evolution reflects the same continuity/coherence divide. After the cooling-activation transient, DT runs settle to steering rates of \(\sim10^{-4}\)–\(10^{-3}\,\mathrm{deg\,Myr^{-1}}\), whereas IT runs maintain \(\sim10^{-2}\)–\(10^{-1}\,\mathrm{deg\,Myr^{-1}}\), with brief coherent retrograde episodes reaching a few \(\mathrm{deg\,Myr^{-1}}\) [2605.27508]. The authors describe the meso-scale turbulent state as the primary switch of Black Hole Weather: connected rainy channels deliver high \(\dot M\), high \(\chi_j\), and coherent perpendicular torques, while stormy or cloudy fragmentation erodes both mass continuity and directional memory [2605.27508]. ## 5. Diagnostics, observables, and the Sgr A* forecast BlackHoleWeather relies on a compact set of diagnostics intended to distinguish cold gas that is merely present from cold gas dynamically connected to SMBH feeding. The main tools are the **C-ratio**, the **k-plot**, **phase-separated mass fluxes**, and the **power spectral density** of the accretion history [2605.27511]. The **k-plot** is a projected histogram of line-of-sight bulk velocity \(|v_{\rm los}-v_{\rm sys}|\) versus internal line dispersion \(\sigma_{\rm los}\). In the jet-regulated studies it divides the plane near \(|v|\sim100\,\mathrm{km\,s^{-1}}\) and \(\sigma\sim50\,\mathrm{km\,s^{-1}}\) into quiescent disk/rotation, drifting clouds, turbulent clouds, and high-velocity components. Stormy systems occupy broad, overlapping loci across phases, rainy systems remain tighter and more separated, and cloudy systems move cold gas into elevated-\(|v|\), modest-dispersion regions associated with weakly coupled fountain motions [2605.27511]. The framework explicitly argues that the k-plot must be combined with the C-ratio profile: only gas that is both kinematically coherent and thermodynamically condensation-prone should be treated as genuinely raining onto the SMBH [2605.27511]. Baroclinicity provides an additional diagnostic of turbulence generation in jet-driven halo weather. In the vorticity equation, the baroclinic source is \[ F_{\rm baro}\equiv \frac{1}{\rho^2}\nabla\rho\times\nabla P. \] A FLASH4 analysis of AGN-jet simulations found that baroclinicity is dynamically subdominant for enstrophy amplification on macro-scales, contributing \(\lesssim5\%\) of net enstrophy growth beyond \(r>10\,\mathrm{kpc}\) and \(t>20\,\mathrm{Myr}\). At and below the meso-scale, however, especially within \(r<10\,\mathrm{kpc}\) and during \(t<20\,\mathrm{Myr}\) after outburst, it accounts for \(\sim50\)–100\% of the initial enstrophy injection [2303.02720]. The same study reports that \(F_{\rm baro}\) correlates more strongly with density gradients than with pressure gradients, and that even when the density–pressure misalignment angle is often below \(45^\circ\), jet-boosted gradients suffice to seed fresh turbulence [2303.02720]. An earlier observationally oriented use of the weather analogy appeared in the forecast for Sgr A*. There the accretion rate is parameterized as \(\dot M=f\,\dot M_0\), and a “best-bet” general relativistic MHD plus fully relativistic radiative-transfer model is recomputed for \(f\gtrsim1\) [1204.1371]. The model adopts a Kerr black hole with \(a_*=0.94\), inclination \(i=85^\circ\), \(T_i/T_e=3\), and normalization to a \(1.3\,\mathrm{mm}\) flux of \(3\,\mathrm{Jy}\) at \(\dot M_0\) [1204.1371]. The photon-orbit ring of characteristic radius \(r_{\rm ring}\simeq5\,GM/c^2\) is visible only if the synchrotron photosphere satisfies \(r_{\rm ph}(\nu;\dot M)