| Summary: | Low level clouds are ubiquitous over the Southern Ocean. However, climate and weather models fail to accurately simulate their radiative impact. This has been attributed in part to the inadequate representation of cloud phase distributions. Using in situ airborne observations acquired during the Southern Ocean Clouds, Radiation, Aerosol Transport Experimental Study (SOCRATES) campaign, this dissertation classifies cloud samples with horizontal spatial resolutions ranging from 120‒150 m as either liquid, ice or mixed phase (i.e., liquid and ice particles in the same volume). Cloud phase is determined using a combination of data from the in situ cloud probes and a supervised machine learning algorithm, which determines phase based on particle imagery. An abundance of liquid phase samples is observed over the region (70%) at temperatures from -20° to 0°C. The prevalence of supercooled liquid abruptly decreases to single digit percentages at temperatures less than -20°C. There is also a notable ice phase presence (10%) at relatively high temperatures (> -5°C). Ice nucleating particle (INP) and cloud condensation nuclei (CCN) concentrations are compared with relative cloud phase frequencies within and above the boundary layer. A positive correlation is found between INP concentrations and ice-containing cloud phase (i.e., ice and mixed phase) frequencies in select cases. However, many cases do not exhibit significant correlation, suggesting a prevalence of alternative ice initiation/growth processes, such as secondary ice production. CCN concentrations are negatively correlated with ice-containing frequencies above the boundary layer, which may be related to longer lifetimes of supercooled liquid clouds in high CCN environments. A strong negative correlation is also found between CCN and large cloud drop (> 25 μm) number concentrations, suggesting secondary ice production may be inhibited in the presence of high CCN concentrations. A novel cloud layer classification method is introduced to classify cloud layers ...
|
|---|