On the geometry of aggregate snowflakes

Axel Seifert, Fabian Jakub, Christoph Siewert, Leonie von Terzi, Stefan Kneifel · 2026

Read On the geometry of aggregate snowflakes by Axel Seifert, Fabian Jakub, Christoph Siewert, Leonie von Terzi, Stefan Kneifel (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-10608.txt.

Introduction

Axel Seifert, Fabian Jakub, Christoph Siewert, Leonie von Terzi, and Stefan Kneifel examine the shape of large snowflakes. Snow and ice shape weather and climate in deep ways. Much of the rain that reaches the ground starts as ice high above. The flakes that fall are rarely the neat six-sided crystals of popular images.

Most large flakes are aggregates, clumps formed when small ice crystals collide and stick. The parts include flat plates, columns, and branched dendrites. The authors study the geometry of these clumps, since shape affects how fast flakes fall and melt. Better shape models feed into sharper weather forecasts and climate simulations.

Catalog record

CREATORAxel Seifert, Fabian Jakub, Christoph Siewert, Leonie von Terzi, Stefan Kneifel
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE73 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-10608

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

On the geometry of aggregate snowflakes Axel Seifert, Fabian Jakub, Christoph Siewert, Leonie von Terzi, Stefan Kneifel Abstract Snowflakes play a crucial role in weather and climate. A significant portion of precipitation that reaches the surface originates as ice, even when it ultimately falls as rain. Contrary to the popular image of symmetric, dendritic crystals, most large snowflakes are irregular aggregates formed through the collision of primary ice crystals, such as hexagonal plates, columns, and dendrites. These aggregates exhibit complex, fractal-like structures, particularly at large sizes. Despite this structural complexity, each aggregate snowflake is unique, with properties that vary significantly around the mean — variability that is typically neglected in weather and climate models. Using a physically based aggregation model, we generate millions of synthetic snowflakes to investigate their geometric properties. The resulting dataset reveals that, for a given monomer number (cluster size) and mass, the maximum dimension follows approximately a lognormal distribution. We present a parameterization of aggregate geometry that captures key statistical properties, including maximum dimension, aspect ratio, cross-sectional area, and their joint correlations. This formulation enables a stochastic representation of aggregate snowflakes in Lagrangian particle models. Incorporating this variability improves the realism of simulated fall velocities, enhances growth rates by aggregation, and broadens Doppler radar spectra in closer agreement with observations. 1 Introduction Snowflakes have fascinated scientists for centuries, dating back to the 17th century when German astronomer Johannes Kepler first speculated on the physics behind their hexagonal shapes ( 1 ) . He proposed that the hexagonal geometry of snowflakes might be explained by a close packing of spheres — an idea that led to what is now known as Kepler’s conjecture ( 17 ) . Today, we understand that the intricate shape of dendritic snow crystals arises from the hexagonal molecular structure of ice, combined with the Mullins–Sekerka instability that promotes the growth of filament-like branches ( 24 ) . However, most snowflakes observed in the atmosphere are not pristine dendritic crystals but rather aggregates of primary ice crystals formed through collisions. These primary ice crystals — or monomers — may be hexagonal plates, columns, or dendrites. The resulting aggregate snowflakes are structurally complex and exhibit fractal scaling behavior when composed of many monomers, particularly at large cluster sizes ( 45 ; 44 ) . In weather and climate models, and even in the most detailed cloud-resolving models, snowflake geometry is usually prescribed using empirical relations and simple power-law fits ( 28 ; 29 , e.g.,) . Yet measurements of the mass–size relationships for aggregates consistently show substantial scatter around the mean. This variability is evident in both classic studies ( 25 ; 28 ) and more recent measurements using high-resolution imaging and machine learning ( 22 ) . Importantly, this scatter cannot be fully explained by measurement error alone. To a large extent, it is a reflection of natural variability in aggregate snowflake geometry. For a given mass and monomer number, aggregate snowflakes can vary from compact and nearly spherical to chain-like and highly elongated. This geometric diversity directly affects particle properties such as terminal fall velocity: more compact aggregates fall faster than elongated ones of the same mass. Since snowflake growth via aggregation is largely driven by differential sedimentation, variations in fall velocity can enhance the growth rate. This effect is similar to atmospheric turbulence leading to an increase in the collision rate by introducing variability in the particle velocity ( 37 ; 16 ; 33 ; 10 ) . Given these implications, it is important to characterize not only the mean behavior of snowflakes but also the statistical variability of their geometry. Specifically, we investigate the probability distribution governing deviations in maximum dimension for a given mass and monomer number. This leads to the development of a stochastic model for aggregate snowflake geometry, suitable for use in Lagrangian Monte Carlo particle simulations of cloud microphysics. Decades after the pioneering work of 40 and 15 , Lagrangian particle-based models have gained renewed interest and offer new opportunities to address longstanding challenges in cloud physics ( 38 ; 7 ; 32 ; 39 ; 9 ; 43 ; 31 ) . Previous studies of aggregate snowflake geometry, including those by 14 , 13 , and 35 , have examined snowflake shape, density, and fall speed using observational datasets and Monte Carlo aggregation models. 14 employed bivariate statistical methods to show how monomer habit and aggregation history influence the observed geometric variability. 35 expanded on this by using the Ice Particle and Aggregate Simulator (IPAS) to generate a large database of computer-simulated aggregates, aimed at informing bulk microphysical parameterizations. 13 focused on how variability in snowflake geometry affects terminal fall velocity, a process that is revisited in this study. While these studies offer important insights, their statistical frameworks are not readily applicable in Lagrangian particle models (LPMs). To address this, we develop a stochastic parameterization tailored for use in LPMs. In this approach, key geometric properties such as maximum dimension, aspect ratio, and cross-sectional area are probabilistically sampled based on three inputs: ice mass, monomer number, and monomer habit. This allows for efficient and physically consistent generation of snowflake geometries, supporting both LPMs and remote sensing forward operators. The paper is organized as follows: Section 2 introduces the aggregation model and the dataset that underpins our parameterization. Section 3 describes the statistical model of snowflake geometry, focusing on maximum dimension, aspect ratio, and projected area, derived as functions of mass, monomer number, and habit. In Section 4, we demonstrate the implementation of this parameterization within the Lagrangian particle model McSnow. In Section 5 we investigate the influence of the new parameterization on common radar variables, also introducing a new scattering database developed for this study. Section 6 concludes with a summary and outlook. 2 Aggregation modeling The aggregation model described by 23 and 21 constructs three-dimensional representations of aggregate snowflakes by combining primary crystals of predefined shapes such as hexagonal columns, plates, or dendrites. This approach is conceptually similar to the model presented by 45 ; 44 . Our analysis focuses exclusively on unrimed snowflakes, although the geometry of rimed aggregates has been addressed in previous work ( 36 ) . The same aggregation framework has also been used to study the physical properties of snowflake aggregates ( 18 ) , their scattering characteristics ( 34 ) , and to develop retrieval algorithms based on in-situ observations ( 22 ) . In addition, the model has been applied to generate 3D-printed snowflake analogs for laboratory experiments ( 20 ) . Building on this framework, we refine the constant penetration depth concept introduced by 21 to allow the formation of slightly more compact aggregates. The basic implementation for attaching monomers and aggregates follows a procedure similar to the game Tetris . Two particles are randomly aligned in the vertical direction such that their horizontal, axis-aligned bounding boxes overlap. The lower one is then moved upward until it intersects the upper one at the point of minimum vertical distance. If no intersection occurs, the orientation is randomized again until contact is achieved. This method, however, tends to produce very fragile connections, comparable to two needles just touching at their tips, resembling the near-touching fingers in Michelangelo’s The

Read the complete text