A 2026 direct numerical simulation (DNS)–Lagrangian model (Iwashima and Onishi) that resolves all warm-cloud microphysical processes in a vertically developing cloud, with homogeneous isotropic turbulence (HIT) added as a frozen snapshot stacked vertically, showed that turbulence promotes autoconversion early and accretion later, advances surface precipitation onset by about 270 s, and enlarges the first surface-reaching raindrops. The work leaves a coherent set of unresolved issues: the turbulence was a non-evolving snapshot, so applicability to realistic time-evolving atmospheric turbulence is untested; a single kinematic KiD warm-1 configuration with maritime aerosol, unit coalescence efficiency, and no breakup was simulated, so sensitivity to thermodynamic environment, aerosol population, collision efficiencies, hydrodynamic-interaction models, and breakup regimes is unknown; and the observed mismatch between the constituent-droplet volume ratio (1.1) and the total-number-inferred mean-volume ratio (1.37) is attributed to, but does not quantitatively resolve, the feedback by which turbulence-altered droplet size distributions modify subsequent condensation growth. The open problem is to complete this framework — with evolving turbulence and realistic microphysics — so that the two-way coupling between turbulence-enhanced collision-coalescence and condensation growth, and the robustness of turbulence-induced precipitation acceleration, are quantitatively established.

Background

Warm clouds form rain without ice through a chain of droplet-level processes: activation of cloud condensation nuclei (CCN, the aerosol particles on which water vapor condenses to form droplets), growth by condensation, growth by collision–coalescence (droplets colliding and merging; collisions of similar-sized droplets are called autoconversion, collection of small droplets by much larger ones accretion), and sedimentation (falling under gravity). Direct numerical simulation (DNS, resolving turbulence down to the viscous dissipation scale) coupled with Lagrangian particle tracking — following every droplet individually along its trajectory — is the most explicit way to model these processes, since it avoids the sampling and closure assumptions of the bin and bulk microphysics schemes used in weather and climate models (the 'can we do better?' question of Grabowski et al. 2019). A related explicit approach is the super-droplet method of Shima et al. (2009), in which each computational particle statistically represents many identical droplets.

Two largely separate strands of prior work exist. Periodic-cubic-box DNS studies (from the foundational collision-in-turbulence analysis of Saffman and Turner 1956 onward; reviewed by Grabowski and Wang 2013; Onishi et al. 2015; Chen et al. 2018; Michel et al. 2023) resolve homogeneous isotropic turbulence (HIT) and quantify how turbulent shear and preferential concentration (the clustering of inertial particles out of turbulent eddies) enhance droplet collision rates, including aerodynamic droplet–droplet interactions — but a periodic box has no vertical cloud structure, no cloud base or top, and cannot represent the altitude dependence of supersaturation (the excess of water vapor above its equilibrium value over a droplet surface) and droplet size distributions. Kinematic and Lagrangian frameworks with vertical development, such as the Kinematic Driver (KiD) framework of Shipway and Hill (2012) used in the warm-rain benchmark intercomparison of Hill et al. (2023), capture a prescribed updraft and the full column from ground to cloud top but contain no resolved turbulence. Kunishima and Onishi (2018) combined a vertically elongated quasi-1D domain (ground to cloud top) with direct Lagrangian tracking of all warm-cloud processes, still without turbulence. A third strand — eddy hopping (Grabowski and Abade 2017; Abade et al. 2018; Saito et al. 2021) — models how supersaturation fluctuations along differing turbulent trajectories broaden the droplet size spectrum in parcel/Lagrangian frameworks without resolving the turbulence.

Iwashima and Onishi (2026) bridged the first two strands: they inserted a frozen snapshot of HIT into the vertically elongated quasi-1D Lagrangian domain (the snapshot is repeated in the vertical periodic direction to fill the column) and compared simulations with (TURB) and without (LAM) turbulent wind fluctuations. Turbulence roughly tripled the total collision count over the first 600 s (from 1.72×1051.72\times10^{5} to 5.47×1055.47\times10^{5}), promoted mid-layer autoconversion during the updraft stage and accretion afterward, advanced surface precipitation onset by about 270 s, and enlarged the first surface-reaching raindrops by a factor of 1.5. Analyzing the collision histories of surface-reaching raindrops, they found the average volume of constituent droplets (the originally activated single droplets that later merged into each raindrop) was larger in TURB by a factor 1.1, while the mean volume per droplet inferred from total droplet numbers and the (essentially equal) total condensed water was larger by a factor 1.37 — a discrepancy they attribute qualitatively to turbulence-modified droplet size distributions altering subsequent condensation growth histories, since condensation growth rate depends on droplet size. The study used the KiD warm-1 configuration (a prescribed sinusoidal updraft lasting 600 s over a shallow convective cloud), maritime CCN activation, coalescence efficiency assumed equal to 1, two-sphere Stokes-flow hydrodynamic interaction, and no droplet breakup (assumed negligible because the Weber number — the ratio of disruptive aerodynamic stress to restoring surface-tension stress — remains ≲1\lesssim1 at the simulated droplet sizes).

Problem Statement

Within the direct numerical simulation (DNS) + Lagrangian particle tracking framework for a vertically developing warm cloud (vertically elongated quasi-1D domain from ground to cloud top, with explicit CCN activation, condensation/evaporation, collision–coalescence, and sedimentation), establish — quantitatively and mechanistically — how turbulence modifies droplet growth and the initiation of surface precipitation when the turbulence is physically realistic and the microphysical parameters are not idealized. This requires all three of the following components, each of which must be addressed: (1) replacing the frozen, vertically repeated snapshot of homogeneous isotropic turbulence with a time-evolving homogeneous isotropic turbulence field whose turbulent statistics (intensity and integral scale) are matched to those of the frozen-snapshot baseline, so that the comparison isolates turbulence time evolution rather than a change in turbulent intensity (and assessing which of the paper's conclusions — enhanced autoconversion then accretion, earlier precipitation onset, larger first raindrops — survive); (2) resolving the two-way coupling between turbulence-enhanced collision-coalescence and subsequent condensation growth, of which the paper's unexplained quantitative mismatch between the constituent-droplet volume ratio (vc,TURB/vc,LAM=1.1v_{c,\mathrm{TURB}}/v_{c,\mathrm{LAM}}=1.1) and the total-number-inferred mean-volume ratio (1.371.37) is the direct signature — i.e., quantifying how turbulence-altered droplet size distributions feed back on condensation histories and on the sizes of the droplets that become rain; and (3) testing the robustness of the turbulence-induced acceleration of surface precipitation onset across the microphysical dimensions that the source and its referee report identify as unexplored: (a) thermodynamic environments and cloud types, (b) aerosol/CCN populations, (c) collision efficiencies and hydrodynamic-interaction models, (d) coalescence efficiencies below unity, (e) droplet breakup regimes, and (f) stronger turbulence. Component (3) contrasts each of these six dimensions with the KiD warm-1 baseline at one representative non-baseline departure per dimension. With quantitative support, establish: (1) which of the paper's conclusions — enhanced autoconversion then accretion, earlier precipitation onset, larger first raindrops — survive when the frozen, vertically repeated snapshot is replaced by a time-evolving homogeneous isotropic turbulence field matched in statistics to the baseline, in terms of the autoconversion-to-accretion transition time, the surface precipitation-onset time, and the first-raindrop size; (2) the mechanism and magnitude of the two-way turbulence–collision–condensation coupling in vertically developing clouds, quantitatively accounting for the reported discrepancy between the constituent-droplet volume ratio (vc,TURB/vc,LAM=1.1v_{c,\mathrm{TURB}}/v_{c,\mathrm{LAM}}=1.1) and the total-number-inferred mean-volume ratio (1.371.37) by reproducing both ratios from the size dependence of condensational growth; and (3) whether, from a single representative departure per dimension, the turbulence-induced acceleration of surface precipitation onset persists, weakens, or reverses across the six dimensions (a)-(f).

References

  1. Iwashima, M., and Onishi, R.: Direct Lagrangian tracking simulation of droplet growth in vertically-developing turbulent cloud, arXiv:2604.10564, 2026 (submitted 12 Apr 2026). https://arxiv.org/abs/2604.10564 (also available as ACP/EGUsphere preprint doi:10.5194/egusphere-2026-2143)
  2. Anonymous Referee #1: RC1 comment on 'Direct Lagrangian tracking simulation of droplet growth in vertically-developing turbulent cloud' (egusphere-2026-2143), 19 May 2026, doi:10.5194/egusphere-2026-2143-RC1. https://doi.org/10.5194/egusphere-2026-2143-RC1
  3. Saffman, P. G., and Turner, J. S.: On the collision of drops in turbulent clouds, J. Fluid Mech., 1, 16–30, 1956, doi:10.1017/S0022112056000020. https://doi.org/10.1017/S0022112056000020
  4. Kunishima, Y., and Onishi, R.: Direct Lagrangian tracking simulation of droplet growth in vertically developing cloud, Atmos. Chem. Phys., 18, 16619–16630, 2018, doi:10.5194/acp-18-16619-2018. https://doi.org/10.5194/acp-18-16619-2018
  5. Shipway, B. J., and Hill, A. A.: Diagnosis of systematic differences between multiple parametrizations of warm rain microphysics using a kinematic framework, Q. J. Roy. Meteor. Soc., 138, 2196–2211, 2012, doi:10.1002/qj.1913. https://doi.org/10.1002/qj.1913
  6. Hill, A. A., Lebo, Z. J., Andrejczuk, M., Arabas, S., Dziekan, P., Field, P., Gettelman, A., Hoffmann, F., Pawlowska, H., Onishi, R., and Vié, B.: Toward a Numerical Benchmark for Warm Rain Processes, J. Atmos. Sci., 80, 1329–1359, 2023, doi:10.1175/JAS-D-21-0275.1. https://doi.org/10.1175/JAS-D-21-0275.1
  7. Onishi, R., Matsuda, K., and Takahashi, K.: Lagrangian Tracking Simulation of Droplet Growth in Turbulence–Turbulence Enhancement of Autoconversion Rate, J. Atmos. Sci., 72, 2591–2607, 2015, doi:10.1175/JAS-D-14-0292.1. https://doi.org/10.1175/JAS-D-14-0292.1
  8. Chen, S., Yau, M. K., and Bartello, P.: Turbulence Effects of Collision Efficiency and Broadening of Droplet Size Distribution in Cumulus Clouds, J. Atmos. Sci., 75, 203–217, 2018, doi:10.1175/JAS-D-17-0123.1. https://doi.org/10.1175/JAS-D-17-0123.1
  9. Saito, I., and Gotoh, T.: Turbulence and cloud droplets in cumulus clouds, New J. Phys., 20, 023001, 2018, doi:10.1088/1367-2630/aaa229. https://doi.org/10.1088/1367-2630/aaa229
  10. Michel, A., Ababaei, A., and Rosa, B.: The Influence of Gravity and Mass Loading on the Coalescence of Aerodynamically Interacting Droplets in Homogeneous Isotropic Turbulence, J. Atmos. Sci., 80, 2207–2220, 2023, doi:10.1175/JAS-D-22-0267.1. https://doi.org/10.1175/JAS-D-22-0267.1
  11. Grabowski, W. W., and Wang, L.-P.: Growth of Cloud Droplets in a Turbulent Environment, Annu. Rev. Fluid Mech., 45, 293–324, 2013, doi:10.1146/annurev-fluid-011212-140750. https://doi.org/10.1146/annurev-fluid-011212-140750
  12. Grabowski, W. W., Morrison, H., Shima, S.-I., Abade, G. C., Dziekan, P., and Pawlowska, H.: Modeling of Cloud Microphysics: Can We Do Better?, Bull. Amer. Meteor. Soc., 100, 655–672, 2019, doi:10.1175/BAMS-D-18-0005.1. https://doi.org/10.1175/BAMS-D-18-0005.1
  13. Grabowski, W. W., and Abade, G. C.: Broadening of Cloud Droplet Spectra through Eddy Hopping: Turbulent Adiabatic Parcel Simulations, J. Atmos. Sci., 74, 1485–1493, 2017, doi:10.1175/JAS-D-17-0043.1. https://doi.org/10.1175/JAS-D-17-0043.1
  14. Abade, G. C., Grabowski, W. W., and Pawlowska, H.: Broadening of Cloud Droplet Spectra through Eddy Hopping: Turbulent Entraining Parcel Simulations, J. Atmos. Sci., 75, 3365–3379, 2018, doi:10.1175/JAS-D-18-0078.1. https://doi.org/10.1175/JAS-D-18-0078.1
  15. Saito, I., Watanabe, T., and Gotoh, T.: Statistical properties of a stochastic model of eddy hopping, Atmos. Chem. Phys., 21, 13119–13130, 2021, doi:10.5194/acp-21-13119-2021. https://doi.org/10.5194/acp-21-13119-2021
  16. Chandrakar, K., Morrison, H., Grabowski, W., and Lawson, R.: Are turbulence effects on droplet collision–coalescence a key to understanding observed rain formation in clouds?, Proc. Natl. Acad. Sci. USA, 121, e2319664121, 2024, doi:10.1073/pnas.2319664121. https://doi.org/10.1073/pnas.2319664121
  17. Kuntzleman, J., Krueger, S., and Shaw, R.: Cloud Droplet Size Distributions Governed by Condensation, Collision–Coalescence, and Sedimentation. Part I: Monte Carlo Simulations and Scaling Relations for Growth in a Convection–Cloud Chamber, J. Atmos. Sci., 81, 2115–2128, 2024, doi:10.1175/JAS-D-23-0233.1. https://doi.org/10.1175/JAS-D-23-0233.1
  18. Shima, S., Kusano, K., Kawano, A., Sugiyama, T., and Kawahara, S.: The super-droplet method for the numerical simulation of clouds and precipitation: a particle-based and probabilistic microphysics model coupled with a non-hydrostatic model, Q. J. Roy. Meteor. Soc., 135, 1307–1320, 2009, doi:10.1002/qj.441. https://doi.org/10.1002/qj.441
  19. Yin, C., Shima, S.-I., and Lu, C.: A Lagrangian Particle Tracking Framework for the Super-Droplet Method: Development, Implementation, and Application of Backward and Forward Algorithms in SCALE-SDM 5.2.6-2.3.1, EGUsphere preprint, 2025, doi:10.5194/egusphere-2025-6221. https://doi.org/10.5194/egusphere-2025-6221