Ice crystal images from optical array probes: classification with convolutional neural networks

Abstract

Abstract. Although airborne optical array probes (OAPs) have existed for decades, our ability to maximize extraction of meaningful morphological information from the images produced by these probes has been limited by the lack of automatic, unbiased, and reliable classification tools. The present study describes a methodology for automatic ice crystal recognition using innovative machine learning. Convolutional neural networks (CNNs) have recently been perfected for computer vision and have been chosen as the method to achieve the best results together with the use of finely tuned dropout layers. For the purposes of this study, The CNN has been adapted for the Precipitation Imaging Probe (PIP) and the 2DS Stereo Probe (2DS), two commonly used probes that differ in pixel resolution and measurable maximum size range for hydrometeors. Six morphological crystal classes have been defined for the PIP and eight crystal classes and an artifact class for the 2DS. The PIP and 2DS classifications have five common classes. In total more than 8000 images from both instruments have been manually labeled, thus allowing for the initial training. For each probe the classification design tries to account for the three primary ice crystal growth processes: vapor deposition, riming, and aggregation. We included classes such as fragile aggregates and rimed aggregates with high intra-class shape variability that are commonly found in convective clouds. The trained network is finally tested through human random inspections of actual data to show its real performance in comparison to what humans can achieve.

Distinct secondary ice production processes observed in radar Doppler spectra: insights from a case study

“…1c), a few large heavily rimed or graupel particles can be seen, as well as numerous columnar particles and aggregates of needles or columns. The independent PIP-based morphological classification (Jaffeux et al, 2022, for particles with a maximum dimension greater than 2 mm,) shown in Fig. 1d also confirms this partitioning, with 18% of rimed particles (graupel and rimed aggregates), 36% of columnar crystals and 42% of aggregates which are either distinctly classified as made of columns and needles, or simply labeled as fragile, which denote weakly bounded crystals).…”

“…100 µm to 6.4 mm, 10 µm to 1.28 mm). An automatic classification algorithm (Jaffeux et al, 2022) allows identifying particle habits from 2D-S and PIP images. In addition, cloud liquid water content (LWC) is estimated with a cloud droplet probe (CDP-2, Faber et al, 2018), which samples droplets up to 50 µm.…”

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

“…When considering only non-aggregated ice crystals, IceDetectNet has an accuracy of 82 % for all data, 90 % for basic habits, and 85 % for microphysical processes (Table 2). Previous studies using single-label classification (Xiao et al, 2019;Jaffeux et al, 2022;Zhang, 2021) have reported overall accuracies above 90 %, which is higher compared to IceDetectNet for all data (82 %). However, for the multi-label classification, IceDetectNet classifies both basic habits and microphysical processes.…”

Classification of Ice Crystal Habits by Deep Transfer Learning

“…Wu et al (2020) employed nine distinct categories for classifying ice crystal images captured by the Cloud Imaging Probe (CIP), including tiny, sphere, column, needle, irregular, dendrite, graupel, plate, and aggregate. Jaffeux et al (2022) classified ice crystals from 2D-S probe into nine categories: graupel, fragile aggregate, column, rosette, plate, sphere, capped column, artifact, and aggregate, and particle habits for the Precipitation Imaging Probe (PIP) were classified into six categories: graupel, fragile aggregate, needle, plate, rosette, and rimed aggregate. Chen et al (2022) classified images of ice crystal sampled by 2D-S and CPI probes into ten categories: plate, plate aggregate, irregular, rosette, aggregate, sphere, graupel, dendrite, column, and column aggregate.…”

“…Liao et al (2021) proposed a CNN embedded with hypergraphic convolutional module, Hy-INet, which achieved a high accuracy of 98.08% in a ten-category classification task for ice crystals. Jaffeux et al (2022) devised two models to classify ice crystal images acquired by the Precipitation Imaging Probe (PIP) and the Two-dimensional Stereo (2D-S) Probe into 6 and 9 classes using CNN models, respectively, and obtained 85.44% and 91.02% accuracy. Zhang et al (2023) introduced a CNN-based classification method for 2D-S cloud particle images, achieving an average accuracy of 97% in eight cloud particle habits classification.…”