Japan Geoscience Union Meeting 2022

Presentation information

[E] Oral

M (Multidisciplinary and Interdisciplinary) » M-GI General Geosciences, Information Geosciences & Simulations

[M-GI29] Data assimilation: A fundamental approach in geosciences

Thu. May 26, 2022 3:30 PM - 5:00 PM 104 (International Conference Hall, Makuhari Messe)

convener:Shin ya Nakano(The Institute of Statistical Mathematics), convener:Yosuke Fujii(Meteorological Research Institute, Japan Meteorological Agency), Takemasa Miyoshi(RIKEN), convener:Masayuki Kano(Graduate school of science, Tohoku University), Chairperson:Shin ya Nakano(The Institute of Statistical Mathematics), Yosuke Fujii(Meteorological Research Institute, Japan Meteorological Agency)

4:30 PM - 4:45 PM

[MGI29-11] Snowfall Events and Aerosol Atmospheric Rivers with Changes in Concentrations of Trace Substances in the Atmosphere in the Polar Regions

*Kazue Suzuki1, Kazuyuki Nakamura2, Terumasa Tokunaga5, Keiichiro Hara4, Daisuke Goto3, Naohiko Hirasawa3, Takashi Yamanouchi3 (1.HOSEI University, 2.MEIJI University, 3.NIPR, 4.FUKUOKA University, 5.Kyushu Institute of Technology)

Keywords:time-series modeling, Aerosol transport

In recent years, water vapor transport and heavy rain (snow) have been considered to be associated with atmospheric rivers (ARs), which are narrow (hundreds of kilometers) and long (thousands of kilometers). They have been categorized by the intensity of intergraded water vapor transport (IVT). In the Antarctic region, it is also believed that the AR is related to the poleward flow of water vapor. In the Antarctic, observations of ARs and background fields that enhance water vapor fluxes toward the polar direction have been reported (Gorodetskaya et al., 2014). We have been working on automatically identifying ARs by deep learning using cloud images that can be distinguished from ARs during snowfall at Syowa Station (Suzuki et al., 2021). A method for detecting Aerosol Atmospheric Rivers (AARs) on a global scale using objective analysis data has been developed (Chakraborty et al., 2021). They showed that high concentration aerosols are transported by association with ARs when aerosols are released.
In this study, we aim to construct a prediction model of the atmospheric mass transport process by machine learning by simultaneously capturing not only water vapor transport by AR but also the transport of trace substances in the atmosphere that are considered to be of terrestrial origin. First, the relationship between AR and aerosol transport was investigated for the 2009 event in the Antarctic and Arctic. This paper describes the results of aerosol data (number concentration by particle size, Black carbon (BC) concentration), CO concentration and snow depth, and cloud images from NOAA/AVHRR sensors observed at Syowa Station in 2009. The blizzard followed at the site was used as a snowfall event to study the change in aerosol concentration and the difference in atmospheric transport path by trajectory analysis. The relationship between AR and aerosol transport was discussed. The ERA-Interim data and the National Institute of Polar Research Trajetctory Model (NITRAM) were used to calculate the grain tracks and to produce the figures. Particles were calculated seven days backward from 3000-5000m above sea level.
Figure 1 shows a composite cloud image of the backward trajectory and thermal infrared channel of the B-class blizzard observed from June 6 to 7, 2009. A long line of clouds can be seen ahead of the low-pressure system, and the trajectory suggests that atmospheric transport originated in South America. The number of cloud pixels at high altitudes (white in the image) calculated from the cloud images was small, and the amount of snowfall was not significant. Similarly, the BC concentration peaks during other blizzards showed grain lines of continental origin, and the AR cloud images were also confirmed.
Next, we attempted to extract events from CO concentrations. Since CO concentrations have a large seasonal component and are difficult to extract and contain missing values, we used time series modeling to fit forecast values and extract events. The parameters optimized using AIC are (p,d,q)(P,D,Q)=(1,0,2 )(0,1,0). As a result, we extracted the events in which BC and CO increase simultaneously and the events in which only CO increases and investigated the tracks and background fields for each event. The results show that when BCs are especially transported, they are often transported over long distances from outside Antarctica, while when only CO is transported, it is often originated from Antarctica.
The cloud images show that there are clouds considered ARs during the event, and CO and BC are expected to be good indicators for capturing AARs. In the presentation, we will also show the data analysis of the Arctic and Ny-Alesund.