**1. Introduction**

Particulate matter (PM) consists of a complex mixture of solid and liquid particles of organic and inorganic substances suspended in the atmosphere. The major components of PM are sulfates (SO4 <sup>2</sup>−), nitrates (NO3 −), ammonium (NH4 +), sodium chloride (NaCl), black carbon (BC) or elemental carbon (EC), organic carbon (OC), mineral dust, and water. The PMs with the greatest negative health e ffects are those with a diameter of 2.5 μm or less (PM2.5), which can penetrate and lodge deep inside the lungs [1]. Some PMs are also climate-dependent, known as short-lived climate pollutants (SCLPs) [2]. Warming due to sunlight absorption (e.g., BC) and cooling due to sunlight scattering (e.g., SO4 <sup>2</sup>−) directly a ffect radiative forcing in the earth's climate system. Additionally, water-soluble PMs a ffect the regional climate system interacting with cloud microphysics. These radiative and microphysical interactions can induce changes in regional precipitation and atmospheric circulation patterns. Some PM2.5 particles are directly emitted from natural sources and human activities, while others are formed through complex oxidation reactions and particle agglomeration. Combining the regional three-dimensional chemical transport model (CTM) with comprehensive particulate formations may be a useful tool for understanding the detailed behavior of short-lived PM2.5 components in the atmosphere.

Recently, PM2.5 air quality has been improved in East Asian countries, e.g., China [3] and Japan [4]; however, PM2.5 concentrations at Japanese air pollution monitoring stations (APMSs) have not met ye<sup>t</sup> the environmental quality standard, defined as 15 μg/m<sup>3</sup> for the annual PM2.5 mean and 35 μg/m<sup>3</sup> for 24-h PM2.5 mean, or the World Health Organization (WHO) air quality guidelines, with corresponding values of 10 and 35 μg/m3. An established reference regional CTM system should be applied to design effective PM2.5 control strategies [5]. However, accurately reproducing or predicting the concentrations and distributions of PM2.5 and its relevant substances remains challenging, due to inaccurate emission inventories, poorly represented initial and boundary conditions, imperfect physical, dynamical, and chemical parameterizations, and limited observations for verification, as noted for previous Asian scale model inter-comparisons, i.e., the model inter-comparison study for Asia (MICS-Asia) series [6–9] and the urban air quality model inter-comparison study in Japan (UMICS) series [10–12].

A model inter-comparison framework, Japan's study for reference air quality modeling (J-STREAM), was designed to establish a reference regional CTM system to consider strategies for reducing PM2.5 and its relevant substances [5]. In this paper, the capacities of participant models for J-STREAM to simulate PM2.5 and its components were evaluated for two urban areas in Japan in each season. The model improvements will be discussed based on the inter-model di fferences.
