Quantifying Ice Aggregation in Stratus Clouds through Targeted Cloud Seeding Experiments

Ice aggregation in clouds plays a crucial role in cloud development and precipitation processes, both of which are important for weather forecasting and climate studies. Initially, ice crystals grow by water vapor diffusion inside a cloud, and as the clouds mature, aggregation can become more important. Field studies have shown that aggregated ice crystals can make up a large proportion (38%-54%) of the total cloud ice, predominantly in precipitating clouds. Despite its importance, aggregation is poorly understood due to several challenges: (1) mimicking natural cloud conditions in laboratory studies, and (2) difficulties in tracking and quantifying aggregation in field measurements.

To bridge this knowledge gap, the CLOUDLAB project conducted cloud seeding experiments by injecting silver iodide into low stratus clouds to induce ice crystal formation and growth. After seeding, within a residence time of 5-15 minutes, the resulting microphysical changes in the cloud downwind of seeding were observed using an in situ balloon-borne platform equipped with a holographic imager and ground-based cloud radars. The number and habits of aggregated ice crystal components were retrieved using IceDetectNet (Zhang et al. 2024) - a newly developed algorithm capable of classifying ice crystals down to the aggregate ice component level. Ice aggregation was quantified by deriving the number of collisions per aggregated ice crystal. The analysis included ice crystal images from approximately 30 seeding experiments conducted at temperatures ranging between -10°C and -5°C.

We quantified how the aggregation rate varies with ice residence times by conducting experiments at different distances between the seeding and measurement sites for similar ambient (e.g., temperature, wind speed) and cloud microphysical conditions (e.g., cloud droplet number concentration). The dependence of aggregation on factors such as temperature and turbulence has been investigated by comparing experiments at varying ambient conditions and microphysics but with comparable residence times. In our seeding experiments, we observed that within just 5-10 minutes, over 50 % of the freshly formed ice crystals were already aggregated, likely due to the high ice crystal number concentration produced in the seeding plume. This study presents the first successful field measurements and quantification of the ice aggregation rate by applying targeted cloud seeding and IceDetectNet. It provides valuable empirical data that will enhance our understanding of the ice aggregation mechanisms and hence will improve climate projections and precipitation forecasts.

Reference:
Zhang H, Li X, Ramelli F, et al. IceDetectNet: A rotated object detection algorithm for classifying components of aggregated ice crystals with a multi-label classification scheme[J]. EGUsphere, 2024, 2024: 1-27.