Laboratory Measurements and Modelling of the Scattering Phase Functions of Hollow and Solid Ice Crystals

In 2013, the International Panel on Climate Change (IPCC) concluded that the coupling of clouds with the Earth's atmosphere is currently the largest uncertainty faced in predicting climate change.  Cirrus clouds are of particular interest due to their large global coverage and diverse microphysical properties.  The large variety in size, shape and complexity poses many problems in accurately representing ice particles in scattering models.  As such, simple geometric models are commonly used, leading to errors in predicted phase function and asymmetry parameter. In recent years, surface roughness and internal structure have gained recognition as important contributors to the scattering properties of cirrus. Although many models have begun to take these properties into account, further laboratory studies are necessary in order to parameterize them correctly.

This work focuses on experimentally measuring the scattering properties of ice crystals, with particular emphasis on particles with hollow cavities.  Experiments were conducted in the Manchester Ice Cloud Chamber (MICC).  The chamber consists of a 10 m tall fall tube and is capable of maintaining temperatures down to -50^oC.  A wide range of particle habits are achievable by altering the water inputs and temperature.  The clouds were sampled using a Cloud Particle Imager (CPI), which gave information on the particle size distribution, concentration and habit.  In addition to this, ice crystal replicas were taken using formvar resin.  These were used to gain additional insight into the crystal habit including internal structure.  This information was used to create realistic particle geometries for use in scattering models.  Measurements were carried out in a smaller ‘scattering chamber' which was attached to the bottom of the cloud chamber.  The scattering set up utilizes a 405nm and a 635nm laser.  These are directed horizontally through the cloud as it falls out of the chamber, giving a scattering path length of 0.3m.  The detector optics samples the portion of light scattered in a horizontal plane at a range of angular bins from 0-150^o.  The measured data was combined with theoretical results in order to normalise them.  From this, the phase function, P₁₁, and asymmetry parameter, g, were found.

Observations from the formvar replicas revealed two distinct internal structures.  The hollow cavities found at -7^o C were simple hexagonal pyramids.  At -30^oC, the hollow cavity was more complex, consisting of a series of stepped intrusions.  Particle models were created to represent the -7^oC (‘warm') intrusions and the -30^oC (‘cold') intrusions.  These were tested using Ray Tracing (RT) and Ray Tracing with Diffraction on Facets (RTDF).  For an example column of aspect ratio of 2 (length/diameter), RT showed a reduction in g for both of the cavities, compared with a solid particle.  In comparison to this, RTDF showed an increase in g for the ‘warm' column, and a decrease in g for the ‘cold' column.

In order to investigate the effect of the two types of cavity experimentally, several ice clouds were created. The vapour input was varied in order to produce a cloud of either solid or hollow particles, and temperature was altered to produce particular habits. Each cloud was found to consist of only one habit, although there were variations in particle size and aspect ratio.    As the cloud conditions were changed in order to produce either solid or hollow columns, the average aspect ratio and size distribution varied between different clouds.  Consequently, the measurements from a cloud of solid particles cannot be directly compared with measurements from a cloud of hollow particles.   The measured data is therefore compared to theoretical results from RT and RTDF.  For solid columns, both the RT and RTDF code overestimated g by 1.4% and 1.2% respectively, compared with measured results.  For the ‘warm' columns cloud, RT under predicts g by 11.2%, and RTDF over predicts it by 5%.  For the ‘cold' column cloud, RT under predicts asymmetry parameter by 21.5%, and RTDF over predicts it by 3.6%. 

Discrepancies in the measured and modelled results for the ‘warm column' make it difficult to determine the effect of the hollow cavity.  Due to the conditions needed to grow hollow particles, the ice crystals may have rough surfaces which may mask the effects of the cavity.  Both measured and modelled results agree that the more complex hollow cavity in the ‘cold column' model acts to decrease the asymmetry parameter.  A comparison of measured and modelled results show both RT and RTDF to be suitable models for solid particles.  However, RT largely under predicts g for both hollow columns, and large differences are evident in the phase function.  Therefore RT is not a suitable model for hollow particles.

In addition to P₁₁ and g, knowledge of the polarization properties of cloud particles is necessary for remote sensing applications.  The backscattering linear depolarization ratio has long been used by LIDAR instruments/analysis to determine the thermodynamic phase of cloud particles, and to estimate ice particle habit.  Observed linear depolarization ratios in cirrus clouds are frequently lower than those predicted by theory which may be caused by complexities such as surface roughness and hollowness.  The experimental set up in the MICC has been expanded in order to measure linear depolarization ratio over a range of angles including backscattering.