This study aims to update the emission inventories and to evaluate the air quality model (AQM) performance for atmospheric ammonia (NH₃) and secondary inorganic aerosols. The update of the ammonia emissions was conducted using EAGrid2010-Japan. The update methodology was divided into the emission sources from agriculture and human & pets. Regarding the update for the agricultural emissions, the volatilization from soil after compost application was separated from the ammonia emissions from the livestock industry, and the monthly variation considering the domestic fertilization pattern was applied to the fertilization of the compost and chemical fertilizer. Regarding the update for the emissions from human & pets, the activities such as population density were updated. In addition, the emission factors for human sweat, human breath and pet dogs were modified based on the latest literature. The performance of the AQM was evaluated by a sensitivity analysis based on the multiple emission cases for the Tokyo metropolitan area in 2015. It was found that the overestimation of the simulated NH₃ concentration was improved at the monitoring sites in the agricultural areas and the urban suburbs in the warm season by updating the NH₃ emissions from agriculture and human & pets, respectively. In addition, the overestimation of the simulated NO₃^- in PM_2.5 was improved over a wide area of the Kanto region by updating the NH₃ emission from human & pets.

High PM_2.5 concentrations (daily mean exceeding 50 µg/m³) were observed from July 16 to 21, 2018, over a wide area of Japan. The surface level aerosol observation data by the Aerosol Chemical Speciation Analyzer (ACSA-14) were used for the analysis of the PM_2.5 compositions, and the Community Multi-scale Air Quality (CMAQ) were used for the detailed analysis of the observation data. Our findings from this study can be summarized as follows:

1) The PM_2.5 was mainly composed of SO₄²⁻. SO₂ emitted from the Sakurajima volcano is converted to SO₄²⁻ over the East China Sea, then transported to the Japan Sea region along the marginal flow of the Pacific High. SO₄²⁻ over the Japan Sea, then penetrates to the San-in and the Hokuriku area by the sea breeze.

2) A sensitivity analysis determined that the volcanic SO₂ contribution was 80% at the sea near Sakurajima, 70% at the Hokuriku and off the coast of the San-in area.

3) The Process analysis of CMAQ showed that SO₄²⁻ near Sakurajima was formed by a gas phase reaction in the daytime and the liquid phase reaction in the clouds, then transported out by horizontal advection and diffusion. SO₄²⁻ at the Japan Sea increased by horizontal advection and diffusion. It indicated that SO₄²⁻ formed near Sakurajima was transported to the Japan Sea.

The appropriateness of the method to estimate the annual appearance frequency of the stability classes from the meteorological observations executed in a week during each of the four seasons was examined. This method is often employed for the environmental impact assessment of NO₂ for rather small scale constructions. Meteorological data were obtained from 12 observatories throughout Japan for 11 years. Two methods were compared; Method 1: to calculate the variance of the annual appearance frequency for stability D for 10 years obtained from all the data of each year except for the year used in Method 2, and Method 2: to calculate the variance of the estimated annual appearance frequency of stability D from the summation of oneweek of observations in each of the four seasons. The F-test was applied to the hypothesis that the variance of Method 1 and Method 2 is equal. The hypothesis was rejected for more than half of the years and observatories, which suggests that it is generally not acceptable to estimate the annual appearance frequency of the stability class from four weeks of observations. Conservative results were obtained for the Gaussian dispersion calculation by considering the stability D only for the distance farther than 40 m from the source point for construction in the daytime.