Abstract. Atmospheric new particle formation (NPF) is an important source of climate-relevant aerosol particles which has been observed at many locations globally. To study this phenomenon, the first step is to identify whether an NPF event occurs or not on a given day. In practice, NPF event identification is performed visually by classifying the NPF event or non-event days from the particle number size distribution surface plots. Unfortunately, this day-by-day visual classification is time-consuming, labor-intensive, and the identification process renders subjective results. To detect NPF events automatically, we regard the visual signature (banana shape) which has been observed all over the world in NPF surface plots as a special kind of object, and a deep learning model called Mask R-CNN is applied to localize the spatial layouts of NPF events in their surface plots. Utilizing only 358 human-annotated masks on data from the Station for Measuring Ecosystem and Atmospheric Relations (SMEAR) II station (Hyytiälä, Finland), the Mask R-CNN model was successfully generalized for three SMEAR stations in Finland and the San Pietro Capofiume (SPC) station in Italy. In addition to the detection of NPF events (especially the strongest events), the presented method can determine the growth rates, start times, and end times for NPF events automatically. The automatically determined growth rates agree with the growth rates determined by the maximum concentration and mode fitting methods. The statistical results valid the potential of applying the proposed method on different sites, which will improve the automatic level for NPF events detection and analysis. Furthermore, the proposed automatic NPF event analysis method provides more consistent results compared with human-made analysis, especially when long-term data series are analyzed and statistically comparisons between different sites are needed for event characteristics such as the start and end times, thereby saving time and effort of scientists studying NPF events.

中文翻译：

---

使用深度学习的新粒子形成事件检测

摘要。大气新粒子形成 (NPF) 是与气候相关的气溶胶粒子的重要来源，已在全球许多地方观察到。要研究这种现象，第一步是确定某一天是否发生了 NPF 事件。在实践中，NPF 事件识别是通过从粒子数大小分布表面图中对 NPF 事件或非事件天数进行分类来直观地进行的。不幸的是，这种日复一日的视觉分类既费时又费力，而且识别过程呈现主观结果。为了自动检测 NPF 事件，我们将 NPF 曲面图中在世界各地观察到的视觉特征（香蕉形状）视为一种特殊的物体，并且应用名为 Mask R-CNN 的深度学习模型来定位 NPF 事件在其表面图中的空间布局。仅利用来自测量生态系统和大气关系站 (SMEAR) II 站（芬兰海蒂拉）的数据的 358 个人工标注的掩码，Mask R-CNN 模型成功地推广到芬兰的三个 SMEAR 站和圣彼得罗卡菲乌姆（ SPC) 驻意大利。除了检测 NPF 事件（尤其是最强事件）外，所提出的方法还可以自动确定 NPF 事件的增长率、开始时间和结束时间。自动确定的增长率与通过最大浓度和模式拟合方法确定的增长率一致。统计结果验证了将所提出的方法应用于不同站点的潜力，这将提高 NPF 事件检测和分析的自动化水平。此外，与人工分析相比，所提出的自动NPF事件分析方法提供了更一致的结果，特别是在分析长期数据系列并且需要在不同站点之间对开始和结束时间等事件特征进行统计比较时，从而节省科学家们研究 NPF 事件的时间和精力。