IceDetectNet: a rotated object detection algorithm for classifying components of aggregated ice crystals with a multi-label classification scheme

• Published in: Atmospheric Measurement Techniques
• Main Authors: H. Zhang, X. Li, F. Ramelli, R. O. David, J. Pasquier, J. Henneberger
• Format: Article in Journal/Newspaper
• Language: English
• Published: Copernicus Publications 2024
• Subjects: Environmental engineering; TA170-171; Earthwork. Foundations; TA715-787; Ny-Ålesund; Svalbard; Ny Ålesund
• Online Access: https://doi.org/10.5194/amt-17-7109-2024; https://doaj.org/article/1dafafd8ac344e54b6514c2a8a984af3

The shape of ice crystals affects their radiative properties, growth rate, fall speed, and collision efficiency; thus, it plays a significant role in cloud optical properties and precipitation formation. Ambient conditions, like temperature and humidity, determine the basic habit of ice crystals, while microphysical processes, such as riming and aggregation, further shape them, resulting in a diverse set of ice crystal shapes and effective densities. Current classification algorithms face two major challenges: (1) ice crystals are often classified as a whole (at the image scale), necessitating identification of the dominant component of aggregated ice crystals, and (2) single-label classifications lead to information loss because of the compromise between basic habit and microphysical process information. To address these limitations, we present a two-pronged solution here: (1) a rotated object detection algorithm (IceDetectNet) that classifies each component of an aggregated ice crystal individually and (2) a multi-label classification scheme that considers both basic habits and physical processes simultaneously. IceDetectNet was trained and tested on two independent datasets obtained by a holographic imager during the NASCENT campaign in Ny-Ålesund, Svalbard, in November 2019 and April 2020. The algorithm correctly classified 92 % of the ice crystals as either aggregate or non-aggregate and achieved an overall accuracy of 86 % for basic habits and 82 % for microphysical process classification. At the component scale, IceDetectNet demonstrated high detection and classification accuracy across all sizes, indicating its ability to effectively classify individual components of aggregated ice crystals. Furthermore, the algorithm demonstrated a good generalization ability by classifying ice crystals from an independent generalization dataset with overall accuracies above 70 %. IceDetectNet can provide a deeper understanding of ice crystal shapes, leading to better estimates of ice crystal mass, fall velocity, and ...