[U24-P02] COVID-19感染予測数理モデルのアンサンブルデータ同化：感染力と地球環境の関連分析
キーワード：COVID-19、Epidemiological Model、Data Assimilation、Infectivity、Satellite Observations、Atmospheric Reanalysis Data
The worldwide epidemic of COVID-19 has required more understandings of its infectivity and impacts on Earth environments. This study proposes a method enabling such assessments by integrating mathematical approaches and Earth environments. We first developed a modified epidemiological model based on a commonly used compartment model. The model assigns population into five compartments, Susceptible, Exposed, Infectious, Recovered, and Death; hereafter SEIRD model). In this outbreak, there are daily reports of COVID-19-induced deaths, as opposed to influenza epidemics. Hence we updated the commonly used SEIR model to additionally consider deaths. We conducted ensemble data assimilation of the SERID model for individual countries using daily reports of new cases and deaths as observation data. The data assimilation revealed that time series of infectivity of COVID-19 clearly corresponded with political decisions such as the rock-downs and economic activity resume in the U.S., and economic restraint in Japan. Additionally, satellite observations of NO2 also supported the economic restraints. Based on worldwide data assimilation experiments and atmospheric data, we are now investigating the relationships between COVID-19 infectivity and environments, which is one of COVID-19's uncertainty. This presentation will include the most recent progress up to the time of the conference.