Asian Journal of atmospheric environment
Asian Journal of atmospheric environment Asian Journal of atmospheric environment
Asian Journal of atmospheric environment
  Aims and Scope Type of Manuscripts Best Practices Contact Information  
  Editor-in-Chief Associate Editors Editorial Advisory Board  

Journal Archive

Asian Journal of Atmospheric Environment - Vol. 12 , No. 2

[ Research Article ]
Asian Journal of Atmospheric Environment - Vol. 12, No. 2
Abbreviation: Asian J. Atmos. Environ
ISSN: 1976-6912 (Print) 2287-1160 (Online)
Print publication date 30 Jun 2018
Received 21 Oct 2017 Revised 27 Jan 2018 Accepted 21 Feb 2018
DOI: https://doi.org/10.5572/ajae.2018.12.2.139

Urban Air Quality Model Inter-Comparison Study (UMICS) for Improvement of PM2.5 Simulation in Greater Tokyo Area of Japan
Hikari Shimadera* ; Hiroshi Hayami1) ; Satoru Chatani2) ; Tazuko Morikawa3) ; Yu Morino2) ; Yasuaki Mori4) ; Kazuyo Yamaji5) ; Seiji Nakatsuka4) ; Toshimasa Ohara2)
Osaka University, 2-1 Yamada-oka, Suita, Osaka 565-0871, Japan
1)Central Research Institute of Electric Power Industry, 1646 Abiko, Abiko, Chiba 270-1194, Japan
2)National Institute for Environmental Studies, 16-2 Onogawa, Tsukuba, Ibaraki 305-8506, Japan
3)Japan Automobile Research Institute, 2530 Karima, Tsukuba, Ibaraki 305-0822, Japan
4)Japan Weather Association, 3-1-1 Higashi-Ikebukuro, Toshima-ku, Tokyo 170-6055, Japan
5)Kobe University, 5-1-1 Fukae-minami-machi, Kobe, Hyogo 658-0022, Japan

Correspondence to : *Tel: +81-6-6879-7668, E-mail: shimadera@see.eng.osaka-u.ac.jp

Funding Information ▼

Abstract

The urban model inter-comparison study (UMICS) was conducted in order to improve the performance of air quality models (AQMs) for simulating fine particulate matter (PM2.5) in the Greater Tokyo Area of Japan. UMICS consists of three phases: the first phase focusing on elemental carbon (UMICS1), the second phase focusing on sulfate, nitrate and ammonium (UMICS2), and the third phase focusing on organic aerosol (OA) (UMICS 3). In UMICS2/3, all the participating AQMs were the Community Multiscale Air Quality modeling system (CMAQ) with different configurations, and they similarly overestimated PM2.5 nitrate concentration and underestimated PM2.5 OA concentration. Various sensitivity analyses on CMAQ configurations, emissions and boundary concentrations, and meteorological fields were conducted in order to seek pathways for improvement of PM2.5 simulation. The sensitivity analyses revealed that PM2.5 nitrate concentration was highly sensitive to emissions of ammonia (NH3) and dry deposition of nitric acid (HNO3) and NH3, and PM2.5 OA concentration was highly sensitive to emissions of condensable organic compounds (COC). It was found that PM2.5 simulation was substantially improved by using modified monthly profile of NH3 emissions, larger dry deposition velocities of HNO3 and NH3, and additionally estimated COC emissions. Moreover, variability in PM2.5 simulation was estimated from the results of all the sensitivity analyses. The variabilities on CMAQ configurations, chemical inputs (emissions and boundary concentrations), and meteorological fields were 6.1-6.5, 9.7-10.9, and 10.3-12.3%, respectively.


Keywords: Air quality simulation, Sensitivity analysis, Nitrate, Organic aerosol, Variability in simulation

1. INTRODUCTION

Particulate matter (PM) with an aerodynamic diameter of 2.5 μm or less (PM2.5) is an atmospheric pollutant that consists of various chemical species, including the following five major components: sulfate (SO42-), nitrate (NO3-), ammonium (NH4+), elemental carbon (EC) and organic aerosol (OA). Primary OA (POA) and EC are emitted directly through combustions of fossil fuel and biomass (Bond et al., 2004). Secondary OA (SOA) is formed from various volatile organic compounds (VOC) and the corresponding processes are highly complex and varied (Hallquist et al., 2009). PM2.5 SO42- and NO3- are typically secondary pollutants produced through oxidations of sulfur dioxide (SO2) and nitrogen oxides (NOX), respectively. PM2.5 NH4+ is produced from ammonia (NH3) and generally a counter ion of SO42- and NO3-. PM2.5 has been of increasing concern because of its adverse effects on human health (e.g., Goto et al., 2016; Nawahda et al., 2012). While air quality over Japan has been gradually improved, ambient PM2.5 concentration still exceeds the environmental quality standard in Japan (Wakamatsu et al., 2013).

To design effective PM2.5 control strategies, it is essential to use air quality models (AQMs) that represent detailed physical and chemical processes in the atmosphere. However, in the first model inter-comparison study in Japan (Morino et al., 2010a, b), participating AQMs were not able to adequately simulate mass concentrations of PM2.5 and its major components in the Greater Tokyo Area, which includes one of the largest megacities in the world. In order to seek possible ways for improvement of the AQM performance for PM2.5 simulation, the urban air quality model inter-comparison study in Japan (UMICS) was conducted (Chatani et al., 2014; Shimadera et al., 2014a). In UMICS, the major components of PM2.5 in the Greater Tokyo Area of Japan were focused; common datasets, including meteorological, emission and boundary concentration data, were provided to evaluate uncertainties originating from AQMs; and participants conducted sensitivity analyses in their fields of expertise to identify factors influencing the AQM performance for PM2.5 simulation. UMICS consists of three phases: the first phase focusing on EC in summer 2007 (UMICS1), the second phase focusing on SO42-, NO3- and NH4+ in winter 2010 and summer 2011 (UMICS2), and the third phase focusing on OA in winter 2010 and summer 2011 (UMICS3).

In UMICS1 (Chatani et al., 2014), the Community Multiscale Air Quality modeling system (CMAQ) (Byun and Schere, 2006) with seven different configurations and the Regional Air Quality Model 2 (RAQM2) (Kajino et al., 2012) participated in the inter-comparison of simulated PM2.5 EC concentration. All the eight participating AQMs underestimated PM2.5 EC concentrations. Results of sensitivity analyses implied that local emissions, vertical diffusion and long-range transport were critical factors for the AQM performance. According to Shimadera et al. (2016, 2013), CMAQ well simulates the effect of long-range transport from the Asian Continent on PM2.5 concentration in Japan. On the other hand, vertical diffusion process has not been sufficiently verified because of lack of observed vertical concentration profiles.

In UMICS2 (Shimadera et al., 2014a), CMAQ with five different configurations participated in the intercomparison of simulated PM2.5 SO42-, NO3- and NH4+ concentrations. The different configurations approximately reproduced PM2.5 SO42- concentration, but consistently overestimated PM2.5 NO3- concentration. Sensitivity analyses revealed that simulated PM2.5 NO3- concentration was highly sensitive to NH3 emissions and dry deposition of nitric acid (HNO3) and NH3. The importance of NH3 emissions for PM2.5 simulation has been also emphasized in other studies (e.g., Gonzalez-Abraham et al., 2015; Kim et al., 2014). The results indicate that larger dry deposition velocities of the precursor gases may improve the overestimation of PM2.5 NO3- concentration, and the finding has been applied to some air quality modeling studies in Japan (Itahashi et al., 2017; Morino et al., 2015).

This paper describes additional results of UMICS2 on PM2.5 NO3- simulation, and results of UMICS3 on PM2.5 OA simulation. In addition, variability in PM2.5 simulation was estimated from results of all the sensitivity analyses on CMAQ configurations, chemical inputs (emissions and boundary concentrations), and meteorological fields. Findings obtained in UMICS will be useful for future air quality modeling studies including a new model inter-comparison project called Japan’s study for reference air quality modeling (J-STREAM) (Chatani et al., 2018).


2. METHODOLOGY
2. 1 Common Dataset

Common datasets, including meteorological, emission and boundary concentration data, were prepared for UMICS2/3. Fig. 1 shows common modeling domains and the target area for UMICS2/3. The computational domain consists of three nested horizontal domains from East Asia domain with 64-km grid cells (D1), East Japan domain with 16-km grid cells (D2), Kanto domain with 4-km grid cells (D3), and 30 vertical layers from the surface to 100 hPa with the middle height of the first layer being approximately 28 m. For comparisons between model results, the target area is defined as land areas with elevations lower than 200 m above sea level in D3. The target area includes the Greater Tokyo Area that is the largest urban areas in Japan, and therefore is characterized by huge population and air pollutant emissions. Field campaigns were conducted at Kisai, Komae, Maebashi, Saitama and Tsukuba in the Greater Tokyo Area (Fig. 1) to provide observation data for model validation (Ministry of the Environment, 2013).


Fig. 1. 
Common modeling domains and the target area for UMICS2/3 with location of observation sites.

Meteorological fields were produced using the Weather Research and Forecasting model (WRF) (Skamarock and Klemp, 2008) v3.2.1 for periods from November 15 to December 5, 2010 and July 11 to 31, 2011. Topography height and land use were obtained from the United States Geological Survey (USGS) data with 30-sec resolution. Initial and boundary conditions were derived from the final analysis data of the U.S. National Centers for Environmental Prediction (NCEP FNL) and the high-resolution, real-time, global sea surface temperature analysis (RTG_SST_HR) data of NCEP. WRF was configured with the cumulus parameterization of Kain (2004) (D1 and D2), the Asymmetrical Convective Model v2 (ACM2) (Pleim, 2007), the 5-class microphysics scheme of Hong et al. (2004), the Pleim-Xiu land surface model (PX LSM) (Xiu and Pleim, 2001), the longwave radiation scheme of Mlawer et al. (1997) and the shortwave radiation scheme of Dudhia (1989). Shimadera et al. (2014a) showed that the WRF simulations well agreed with the observed ground-level meteorology in D3.

Anthropogenic emissions in D1 were derived from an emission inventory by Zhang et al. (2009) except NH3 and from an emission inventory by Ohara et al. (2007) for NH3. Anthropogenic emissions in D2 and D3 were estimated with the Japan Auto-Oil Program (JATOP) vehicle emission estimate model (Chatani et al., 2011) and the Georeference-Based Emission Activity Modeling System (Nansai et al., 2004). Biogenic VOC (BVOC) emissions were estimated by using the Model of Emissions of Gases and Aerosols from Nature (MEGAN) (Guenther et al., 2006) v2.04 with the common meteorological fields processed by the Meteorology Chemistry Interface Processor (MCIP). Common initial and boundary concentrations for D1 were obtained from the Model for Ozone and Related Chemical Tracers v4 (Emmons et al., 2010).

2. 2 Participating AQM

Table 1 summarizes configurations of AQMs participating in UMICS2/3. All the six participating AQMs (M0 to M4 in Shimadera et al. (2014a) and M5) are CMAQ with different configurations. M0 is the baseline simulation of UMICS2/3. The Statewide Air Pollution Research Center v99 (SAPRC99) gas-phase chemistry (Carter, 2000) was used with the fourth or fifth generation CMAQ aerosol module (AERO4 or 5). Although emissions of sea salt particles were computed with the aerosol modules, the emission enhancement in the coastal surf zone (Kelly et al., 2010) was not considered in M0-M5 runs. The aqueous chemistry and wet deposition process were based on the Regional Acid Deposition Model (RADM) (Chang et al., 1987) in all the configurations. The sub-grid cloud process was based on RADM in M2 and ACM (Pleim and Chang, 1992) in the others. All the models used ACM2 for the vertical diffusion and the Models-3 routine (Pleim et al., 2001) for the dry deposition process.

Table 1. 
Configurations of AQMs participating in UMICS2/3.
M0 M1 M2 M3 M4 M5
AQM CMAQ v4.7.1 CMAQ v4.7.1 CMAQ v4.6 CMAQ v4.7.1 CMAQ v5.0 CMAQ v5.0.1
Met. MCIP v3.6 MCIP v3.6 MCIP v3.6 MCIP v3.6 MCIP v3.6 MCIP v3.6
Domain D1, D2, D3 D3a D3a D1, D2, D3 D3a D2b, D3
Advectionc Yamartino/
Yamartino
Yamartino/
Yamartino
Yamartino/
Yamartino
PPM/
PPMd
Yamartino/
WRF
Yamartino/
WRF
Diffusionc Multiscale/
ACM2
Multiscale/
ACM2
Multiscale/
ACM2
Multiscale/
ACM2
Multiscale/
ACM2
Multiscale/
ACM2
Photolysis Lookup table In-line calce Lookup table In-line calce In-line calce In-line calce
Chem. SAPRC99
(EBIf)
SAPRC99
(EBIf)
SAPRC99
(ROS3g)
SAPRC99
(EBIf)
SAPRC99
(EBIf)
SAPRC99
(EBIf)
Aerosol AERO5 AERO5 AERO4 AERO5 AERO5 AERO5
Cloud RADM/ACM RADM/ACM RADM RADM/ACM RADM/ACM RADM/ACM
Dry Dep. Models-3 Models-3 Models-3 Models-3 Models-3 Models-3
aBoundary concentrations were derived from D2 of M0.
bBoundary concentrations were derived from D1 of M0.
cHorizontal/Vertical.
dPiecewise-Parabolic Method (Colella and Woodward, 1984).
fEuler Backward Iterative solver (Hertel et al., 1993).
gRosenbrock solver (Sandu et al., 1997).

M0 was driven by the common datasets in order to produce common boundary concentrations for the inner domains, which were used for M1, M2, M4, M5, and sensitivity runs. CMAQ runs were conducted for periods from November 15 to December 5, 2010 and July 11 to 31, 2011. The first seven days of each period are initial spin-up periods, and the next 14 days of each period are the target periods for comparisons between model results.

2. 3 Sensitivity Analysis

Table 2 summarizes all the sensitivity analysis cases in this study. The cases are categorized into CMAQ configuration (CC), emission and boundary concentration (EB) as chemical inputs, and meteorological model (MM). There are 8 cases for CC (CC01-08), 11 cases for EB (EB01-11) and 4 cases for MM (MM01-04). Each individual case focuses on a specific target and has 1-5 runs with different settings. There are 27, 24 and 11 runs in total for CC, EB and MM cases, respectively. The sensitivity was evaluated by the difference of mean PM2.5 concentrations between the baseline and sensitivity runs in the target area (Fig. 1) during the target periods in November 22 to December 5, 2010 (winter 2010) and July 18 to 31, 2011 (summer 2011).

Table 2. 
Sensitivity analysis cases in UMICS2/3.
ID Target Baseline Modified setting
CC01 CMAQ participating in
UMICS2/3
M0 M1, M2, M3, M4, M5
CC02 CMAQ v4 and v5 M0 M1 with minimum eddy diffusivity of CMAQ v5 (M1Kmod); M4 with
Yamartino (M4yamo); M5 with MCIP v4.1 (M5MC41) in D3
CC03 Gas/particle equilibrium M0 Uniform change of temperature by ±2 K (T+2; T-2); relative humidity by
±10% (RH+10; RH-10) in aerosol module in D3
CC04 Dry deposition M1 Uniform multiplication of HNO3 and NH3 dry deposition velocities by 5
(Vd5); 2 (Vd2); 0.5 (Vd0.5); 0.2 (Vd0.2) in D3
CC05 Heterogeneous reaction M0 N2O5 reaction probability set to 0 (Γ0); 0.1 (Γ0.1); and calculated by the
method in AERO3 (Γae3); AERO4 (Γae4) in D3
CC06 Photolysis M1 Reduction of photolysis rate of HNO3 (jHNO3); NO22 (jNO22); O3 → O (1D)
(jO3O1D); O3 → O (3P) (jO3O3P) by 20% in D3
CC07 Advection scheme M3 Eddy (M3eddy); Yamartino (M3yamo) in D1/D2/D3
CC08 Aerosol scheme M5 AERO6 (M5ae6) in D2/D3
EB01 NOX emission M1 Uniform change by from - 40% to +40% (NE-40; NE-20; NE+20;
NE+40) in D3
EB02 VOC emission M1 Uniform change by from - 40% to +40% (VE-40; VE-20; VE+20;
VE+40) in D3
EB03 NH3 emission M0 Monthly profile (larger in winter and smaller in summer) (NH3Emod) in
D2/D3
EB04 AVOC emission M0 Evaporative AVOC with maximum (EVEmax); minimum (EVEmin) in
95% confidence interval of emission factor in D3
EB05 BVOC emission M5 MEGAN v2.04 with J-IBIS (JMG20); MEGAN v2.1 with J-IBIS (JMG21)
in D2/D3
EB06 SVOC and COC emissions M5 Additional emission of SVOC (as POA) with boiling point 240 to 400°C
(SVE400); <240°C (SVE240); and additional emission of COC (as POA)
(COCE) in D2/D3
EB07 HONO emission M5 HONO from diesel-powered vehicles with HONO/NO22 ratio of 1%
(HONO1); 5% (HONO5) in D2/D3
EB08 Sea salt emission M5 Considering surf zone (SSEsurf); uniform 30% increase (SSE+30) in D2/D3
EB09 Emission inventory in
Japan
M0 JATOP emission inventory in D1/D2/D3 (EJ_JEI); EAGrid2000-JAPAN in
D2/D3 (EJ_EAG)
EB10 Boundary concentration M0 D1 boundary concentration derived from CMAQ default profile (D1BDP)
EB11 Boundary concentration M5 D3 boundary concentration derived from M0 (D3BM0)
MM01 WRF version M0 WRF v3.4.1 (W341) in D1/D2/D3
MM02 PBL scheme M0 YSU; MYJ; MYNN in D2/D3
MM03 Analysis data M0 ACM2 (MACM2); MYJ (MMYJ) with JMA MANAL in D2/D3
MM04 Land surface M0 ACM2/Noah with USGS (UAN); ACM2/PX with USGS (twofold
roughness length) (UAPz0x2); ACM2/PX with JIBIS (JAP); ACM2/Noah
with J-IBIS (JAN); MYJ/Noah with J-IBIS (JMN) in D2/D3

For CC cases, there are a comparison of participating CMAQ in UMICS2/3 (CC01) and vertical transport processes associated with the difference between CMAQ v4 and v5 (CC02), sensitivity analyses on factors influencing PM2.5 NO3- simulation in UMICS2 (CC03/04/05/06), sensitivity analyses using different advection schemes (CC07) and the sixth generation CMAQ aerosol module (AERO6) (CC08). In CC02, the impact of reduction in the minimum eddy diffusivity introduced in CMAQ v5.0 (from 2.0 to 1.0 m2 s-1 for urban areas and from 0.5 to 0.01 m2 s-1 for the other areas) (Maker et al., 2014), changes in the vertical advection scheme or the treatment of non-precipitating cloud convection associated with the CMAQ and MCIP versions were investigated to complement the analysis in CC01. For UMICS2, CC03/04/05 were described in Shimadera et al. (2014a), and photolysis rates relevant to HNO3 formation were reduced by 20% in CC06.

EB cases include sensitivity analyses on factors influencing PM2.5 NO3- simulation in UMICS2 (EB01/ 03/07/08), factors influencing PM2.5 OA simulation in UMICS3 (EB02/04/05/06), different emission inventories in Japan (EB09) and different boundary concentrations (EB10/11). For UMICS2, EB01/03 were described in Shimadera et al. (2014a). Emissions of nitrous acid (HONO), which is an important source of hydroxyl radical, from diesel-powered vehicles were estimated based on the concentration ratio of HONO to nitrogen dioxide (NO22) measured by chassis dynamometer experiments (Ministry of the Environment, 2013) in EB07. The impact of sea salt emissions to PM2.5 NO3- was investigated by increasing the emissions uniformly and by taking into account enhanced emissions from the coastal surf zone (Kelly et al., 2010) in EB08. For UMICS3, VOC emissions were uniformly changed in EB02. Anthropogenic VOC (AVOC) emissions from evaporative sources were respectively increased and decreased based on upper and lower limits of 95% confidence interval of emission factors (European Environment Agency, 2009) in EB04. The upper and lower limits of AVOC emissions were respectively larger and smaller by about 80% and 50% than those in the common emission data. BVOC emissions were estimated by MEGAN v2.04 and v2.1 (Guenther et al., 2012) with land cover data obtained from the Japan Integrated Biodiversity Information System (J-IBIS) in EB05 (Chatani et al., 2015). Emissions of semi-volatile organic compounds (SVOC) were estimated based on the Pollutant Release and Transfer Register (PRTR) system of the Ministry of Economy, Trade and Industry in Japan (2011) and emissions of condensable organic compounds (COC) from large point sources were estimated based on a survey on emission sources by Tokyo Metropolitan Government (2011) in EB06. SVOC with boiling point of 240 to 400°C (SVE400), SVOC with boiling point less than 240°C and not included as VOC in the common emission data (SVE240), or COC (COCE) were assumed to be emitted as non-volatile POA in the case. The total POA emissions in SVE400, SVE200 and COCE were respectively larger by 3, 277, and 597% than that of the common emission data. In EB09, anthropogenic emissions in Japan were replaced by the JATOP emission inventory database (JEI-DB) (Japan Petroleum Energy Center, 2012a, b) (EJ_JEI) or an emission inventory called EAGrid2000-JAPAN (Kannari et al., 2007) (EJ_EAG). The amounts of anthropogenic NOX, SO2 and PM2.5 emissions changed from the common emission data by -1, +2 and -7% in EJ_JEI, and -8, -39 and +6% in EJ_EAG in the target area, respectively.

For MM cases, WRF v3.4.1 was utilized with the identical configurations described in subsection 2.1 in MM01. Three different PBL parameterizations, namely the Yonsei University (YSU) scheme (Hong et al., 2006), the Mellor-Yamada-Janjic (MYJ) scheme (Janjic, 1994) and the Mellor-Yamada Level-2.5 model proposed by Nakanishi and Niino (2006) (MYNN), with the Noah LSM (Chen and Dudhia, 2001) were compared with the ACM2 scheme with PX LSM in MM02. Initial and boundary conditions for WRF were obtained from the mesoscale objective analysis data of the Japan Meteorological Agency (JMA MANAL) with higher spatial and temporal resolutions than those of NCEP FNL in MM03. The impacts of land surface data and model on PM2.5 simulations were investigated in MM04. The land use data obtained from J-IBIS were characterized by more widely distributed urban areas than the USGS data used in the WRF simulation to produce the common meteorological fields.


3. RESULTS AND DISCUSSION
3. 1 Comparison with Observed PM2.5

Fig. 2 shows comparisons of observed and M0-simulated mean concentrations of PM2.5 and the five major components in winter 2010 and summer 2011. The M0-simulated mean concentrations of PM2.5, and PM2.5 EC and SO42- approximately agreed with the observations. However, the model overestimated PM2.5 NO3- concentration and substantially underestimated PM2.5 OA concentration at the observation sites. The model also overestimated total (gaseous+PM2.5) nitrate and ammonium, and the overestimation of PM2.5 NO3- likely attributed to overestimation of particulate NH4NO3 (Shimadera et al., 2014a). The agreement of mean PM2.5 concentrations between the observation and simulation does not indicate a good model performance because it is due to compensation of the overestimations and underestimations of PM2.5 components. Sensitivity analyses to identify influencing factors on the overestimation of PM2.5 NO3- and the underestimation of PM2.5 OA were conducted in UMICS2 and UMICS3, respectively.


Fig. 2. 
Comparisons of observed and M0-simulated mean concentrations of PM2.5 and its major components in (a) winter 2010 and (b) summer 2011. Error bars represent standard deviation.

3. 2 Comparison of Participating AQM

Fig. 3 shows percentage differences between M0 and the other participating AQMs, all of which are CMAQ with different configurations, in UMICS2/3 for mean concentrations of PM2.5 and its major components in the target area during the target periods in winter 2010 and summer 2011. While the differences between the results of M0 and those of M1 and M3, which are CMAQ v4.7.1, were relatively small, M2 showed relatively large differences compared to the others because of various different configurations, such as CMAQ version, gas-phase chemistry solver, aerosol and aqueous modules. M4 and M5, which were CMAQ v5.0, tended to predict higher ground-level concentrations for primary pollutants such as PM2.5 EC that was mainly emitted from ground-level emission sources.


Fig. 3. 
Percentage differences between M0 and the other participating AQMs in UMICS2/3 for mean concentrations of PM2.5 and its major components in the target area during the target periods in (a) winter 2010 and (b) summer 2011.

The differences in simulated PM2.5 concentrations among the participating CMAQ with different configurations typically ranged within 20%, which was smaller than the magnitudes of the overestimation of PM2.5 NO3- and the underestimation of PM2.5 OA by M0 (Fig. 2). Therefore, all the participating CMAQ shared common problems in PM2.5 simulation.

3. 3 Sensitivity Analysis on PM2.5 NO3-

Fig. 4 shows results of UMICS2 by percentage differences between the baseline and sensitivity runs in CC03/04/05/06 and EB01/03/07/08 for mean PM2.5 NO3- concentration in the target area during the target periods in winter 2010 and summer 2011. As described in Shimadera et al. (2014a), there were large sensitivities of PM2.5 NO3- concentration to dry deposition velocities of precursor gases (CC04) and NH3 emissions (EB03). Compared to these factors, the reduction in photolysis rates (CC06) and the estimated HONO emissions (EB07) showed quite small impacts on PM2.5 NO3- simulation. In EB08, the increase in sea salt emissions slightly decreased PM2.5 NO3- concentration, which was due to enhancement of the reaction between HNO3 and sea salt aerosols to produce coarse particulate NO3-. Although the sensitivity of PM2.5 NO3- was smaller in EB08 than CC04 and EB03, the impacts of ocean-related phenomena including sea salt emissions should be further investigated given the regional characteristics of Japan surrounded by ocean.


Fig. 4. 
Percentage differences between the baseline and sensitivity runs in CC03/04/05/06 and EB01/03/07/08 for mean PM2.5 NO3- concentration in the target area during the target periods in winter 2010 and summer 2011.

The finding that larger dry deposition velocities of precursor gases may improve the overestimation of PM2.5 NO3- has been applied to some air quality modeling studies in Japan (Itahashi et al., 2017; Morino et al., 2015). However, corresponding field measurements have to be performed in order to verify the finding. While several factors were investigated in UMICS, there are still other unconsidered factors relevant to the AQM performance for PM2.5 NO3- simulation (e.g., a heterogeneous reaction of nitric acid to NOx (Li et al., 2015)). In addition, although the CMAQ performance for PM2.5 SO42- simulation was better than that for PM2.5 NO3-, some studies have reported that AQMs tend to underestimate PM2.5 SO42- concentration particularly in winter (e.g., Zheng et al., 2015; Shimadera et al., 2014b). Some of the remaining challenges in UMICS will be tackled in J-STREAM (Chatani et al., 2018).

3. 4 Sensitivity Analysis on PM2.5 OA

Fig. 5 shows results of UMICS3 by simulated mean concentrations of mean POA, anthropogenic and biogenic SOA (ASOA and BSOA) in CC01 and EB02/ 04/05/06 in the target area during the target periods in winter 2010 and summer 2011. Although the fraction of SOA was higher in summer than in winter, POA accounted for most of PM2.5 OA in the participating CMAQ in UMICS2/3 in the both seasons. Therefore, there were quite small sensitivities of PM2.5 OA concentrations in EB02/04/05, in which VOC emissions were modified. Meanwhile, PM2.5 OA concentrations substantially increased in SVE240 and COCE of EB06, in which POA emissions were increased based on the estimated SVOC and COC emissions, respectively. These results indicate the importance of estimates of additional emission sources that are not considered in the existing emission data for improvement of the AQM performance for PM2.5 OA simulation.


Fig. 5. 
Simulated mean PM2.5 OA concentrations in CC01 and EB02/04/05/06 in the target area during the target periods in (a) winter 2010 and (b) summer 2011.

It must be noted that the contributions of SVOC and COC were likely overestimated in this study because these semi-volatile compounds were assumed to be non-volatile POA in EB06. AQMs incorporating the volatility basis set (VBS) approach (Robinson et al., 2007; Donahue et al., 2006), which can deal with direct emissions and gas-particle partitioning of these semivolatile compounds as well as corresponding SOA formations, may provide more realistic results. According to Morino et al. (2015), the VBS approach improves OA simulation in warm season in Japan.

The fraction of oxygenated OA, which is a proxy of SOA, to the total OA concentration was estimated to be 44-55% at Kisai, Maebashi and Saitama (Fig. 1) according to the measurement and receptor modeling based approach by Hagino et al. (2012) in summer 2011. Therefore, the participating CMAQ in UMICS2/3 probably underestimated the formation of SOA and the sensitivity of PM2.5 OA concentration to VOC emissions. Updated SOA formation processes, such as epoxide pathways of isoprene (Pye et al., 2013), have to be employed for further discussions.

3. 5 Improvement of PM2.5 Simulation

Another sensitivity analysis (Mmod), in which modifications of multiple factors that could improve the PM2.5 NO3- overestimation (Shimadera et al., 2014a) and the PM2.5 OA underestimation were applied to M0, was conducted in D2 to D3. Mmod is different from M0 in using in-line photolysis rate calculation, modified monthly profile of NH3 emission, fivefold dry deposition velocities of HNO3 and NH3, N2O5 parameterization in the third generation CMAQ aerosol module (AERO3), and COC emissions as POA.

Fig. 6 shows mean observed, M0-and Mmod-simulated concentrations of major PM2.5 components at Kisai and Maebashi in summer 2011. While there were only slight changes in PM2.5 EC and SO42- concentrations between M0 and Mmod, PM2.5 NO3- and OA concentrations substantially decreased and increased in Mmod compared to M0, respectively. As a result, the ratio of the five major components in the Mmod-simulation was much closer to that in the observation. This result indicates a possibility of improvement of the AQM performance for PM2.5 simulation.


Fig. 6. 
Mean observed, M0-and Mmod-simulated concentrations of major PM2.5 components at (a) Kisai and (b) Maebashi in summer 2011.

3. 6 Variability in PM2.5 Simulation

Fig. 7 shows results of all the sensitivity analyses by percentage differences between the baseline and sensitivity runs in CC, EB and MM cases for mean PM2.5 concentrations in the target area during the target periods in winter 2010 and summer 2011. Secondary PM2.5 concentration was assumed to be the total concentration of PM2.5 SO42-, NO3-, NH4+ and SOA.


Fig. 7. 
Percentage differences between the baseline and sensitivity runs in (a) CC, (b) EB and (c) MM cases for mean concentrations of PM2.5, primary PM2.5 (PM2.5-secondary PM2.5) and secondary PM2.5 (SO42-+NO3-+NH4++SOA) in the target area during the target periods in winter 2010 and summer 2011.

In CC01/02, the difference between M1Kmod and M1 indicates that the smaller minimum diffusivity introduced in CMAQ v5.0 increased primary PM2.5 concentration by about 10%. The difference between M5MC41 and M5 indicates that the difference between M4 and M5 in summer was attributed to the treatment of non-precipitating cloud convection associated with the MCIP versions. In CC07, the sensitivities of PM2.5 concentration to the advection schemes were ranged within 5%. In CC08, the difference was mainly due to the treatment of aging of OA introduced in AERO6. Although UMICS2/3 compared results of PM2.5 simulations by CMAQ with various different configurations, there was no comparison between AQMs with different model frameworks. Therefore, UMICS2/3 was an intracomparison of CMAQ rather than an inter-comparison of AQMs. In addition, the effect of chemical scheme on PM2.5 simulation was not sufficiently evaluated in UMI CS2/3 because all the CMAQ runs used the SAPRC99 gas-phase chemistry mechanism. In J-STREAM, intercomparisons between different AQMs with different chemical schemes will be conducted for further comprehensive study (Chatani et al., 2018).

In EB09, the usage of different emission inventories in Japan caused larger differences in PM2.5 simulation than in the sensitivity runs in EB01-08 except NH3 Emod, SVOC240 and COCE runs. EB10 showed that the impacts of the change in boundary concentrations of D1 on air quality in the Greater Tokyo Area was comparable to those of the change in CMAQ configurations although air quality in the area was dominantly controlled by pollutants emitted within D1.

For MM cases, the change in WRF version had negligible impacts on PM2.5 simulation in MM01. Meanwhile, the impacts of the changes in PBL scheme, analysis data and LSM in MM02-04 were larger than those in most sensitivity runs in CC and EB cases. The ACM2 scheme tended to predict higher ground-level PM2.5 concentration because of lower PBL height compared to the other PBL schemes, particularly in summer 2011. The increase in urban areas in J-IBIS typically decreased ground-level PM2.5 concentration because of enhanced vertical diffusion associated with higher ground-level air temperature and larger minimum eddy diffusivity in urban areas. The PX LSM, in which land use parameters in a grid cell are derived from fractional land use coverages in the grid cell, was less sensitive to the change in land use data than the Noah LSM, in which land use parameters in a grid cell are determined by a dominant land use category in the grid cell.

Variability in PM2.5 simulation was estimated from the results of 27 runs for CC cases, 24 runs for EB cases and 11 runs for MM cases. Standard deviations of the percentage differences in CC, EB and MM cases were 6.5, 9.7 and 10.3% in winter 2010 and 6.1, 10.9 and 12.3% in summer 2011, respectively. Therefore, AQM input data are as important as, or more important than AQM configurations for improvement of the AQM performance for PM2.5 simulation.


4. CONCLUSIONS

The model inter-comparison study called UMICS was conducted in order to improve the AQM performance for PM2.5 simulation in the Greater Tokyo Area of Japan. In UMICS, there were three phases: UMICS1 focusing on EC (Chatani et al., 2014), UMICS2 focusing on SO42-, NO3- and NH4+ (Shimadera et al., 2014a), and UMICS3 focusing on OA. This paper described the results of UMICS3 with the additional results of UM ICS2. In UMICS2/3, all the participating AQMs, which were six CMAQ with different configurations, similarly overestimated PM2.5 NO3- and underestimated PM2.5 OA. According to the additional sensitivity analyses in UMICS2, simulated PM2.5 NO3- concentration was less sensitive to photolysis rates, HONO emissions and sea salt emissions than the two important factors for improvement of PM2.5 NO3- simulation, namely NH3 emissions and dry deposition velocities of NNO3 and NH3 that were identified in the previous study (Shimadera et al., 2014a). The sensitivity analyses in UMI CS3 focused on emissions of AVOC, BVOC, SVOC and COC. Because POA accounted for most of PM2.5 OA in all the participating CMAQ, there were quite small sensitivities of PM2.5 OA concentration to AVOC and BVOC emissions. Meanwhile, the increase in POA emissions based on the estimated SVOC and COC emissions substantially increased PM2.5 OA concentration. Therefore, to estimate additional emission sources that are not considered in the existing emission data is one of the most important factors for improvement of the PM2.5 OA underestimation. Based on the results of UMICS2/3, multiple factors, such as the modified NH3 emissions, the increased HNO3 and NH3 dry deposition velocities and the estimated COC emissions, were applied to the baseline simulation case to improve the PM2.5 NO3- overestimation and the PM2.5 OA underestimation. As a result, the ratio of the five major components in the simulated PM2.5 concentration approximately agreed with the observation. Variability in PM2.5 simulation was estimated from all the sensitivity runs for CMAQ configurations, emission and boundary concentrations, and meteorological fields. The variability on input data was comparable to, or larger than that on CMAQ configurations, indicating the importance of the input data for improvement of the AQM performance. Although several important findings were obtained in UMICS, there are a lot of remaining challenges (e.g., corresponding measurements to verify the findings, inter-comparisons between different AQMs with different chemical schemes; improvement of SO42- simulation, SOA formation from directly emitted semi-volatile organic species), some of which will be tackled in a new model inter-comparison project called J-STREAM (Chatani et al., 2018).


Acknowledgments

This research was supported by the Environment Research and Technology Development Fund (C-1001 and 5-1601).


REFERENCES
1. Binkowski, F.S., Arunachalam, S., Adelman, Z., Pinto, J., (2007), Examining photolysis rates with a prototype online photolysis module in CMAQ, Journal of Applied Meteorology and Climatology, 46, p1252-1256.
2. Bond, T.C., Streets, D.G., Yarber, K.F., Nelson, S.M., Woo, J.-H., Klimont, Z., (2004), A technology-based global inventory of black and organic carbon emissions from combustion, Journal of Geophysical Research D: Atmospheres, 109(D14203).
3. Byun, D.W., Schere, K.L., (2006), Review of the governing equations, computational algorithms, and other components of the Models-3 Community Multiscale Air Quality (CMAQ) modeling system, Applied Mechanics Reviews, 59, p51-77.
4. Carter, W.P.L., (2000), Implementation of the SAPRC-99 Chemical Mechanism into the Models-3 Framework, Report to the United States Environmental Protection Agency.
5. Chang, J.S., Brost, R.A., Isaksen, I.S.A., Madronich, S., Middleton, P., Stockwell, W.R., Walcek, C.J., (1987), A Three-Dimensional Eulerian Acid Deposition Model: Physical Concepts and Formulation, Journal of Geophysical Research, 92, p14681-14700.
6. Chatani, S., Morikawa, T., Nakatsuka, S., Matsunaga, S., Minoura, H., (2011), Development of a framework for a high-resolution, three-dimensional regional air quality simulation and its application to predicting future air quality over Japan, Atmospheric Environment, 45, p1383-1393.
7. Chatani, S., Morino, Y., Shimadera, H., Hayami, H., Mori, Y., Sasaki, K., Kajino, M., Yokoi, T., Morikawa, T., Ohara, T., (2014), Multi-model analyses of dominant factors influencing elemental carbon in Tokyo Metropolitan Area of Japan, Aerosol and Air Quality Research, 14, p396-405.
8. Chatani, S., Matsunaga, S.N., Nakatsuka, S., (2015), Estimate of biogenic VOC emissions in Japan and their effects on photochemical formation of ambient ozone and secondary organic aerosol, Atmospheric Environment, 120, p38-50.
9. Chatani, S., Yamaji, K., Sakurai, T., Itahashi, S., Shimadera, H., Kitayama, K., Hayami, H., (2018), Overview of model inter-comparison in Japan’s study for reference air quality modeling (J-STREAM), Atmosphere, 9, p18.
10. Chen, F., Dudhia, J., (2001), Coupling an advanced landsurface/hydrology model with the Penn State/NCAR MM5 modeling system. Part I: Model implementation and sensitivity, Monthly Weather Review, 129, p569-585.
11. Colella, P., Woodward, P.L., (1984), The piecewise parabolic method (PPM) for gasdynamical simulations, Journal of Computational Physics, 54, p174-201.
12. Donahue, N.M., Robinson, A.L., Stanier, C.O., Pandis, S.N., (2006), Coupled partitioning, dilution, and chemical aging of semivolatile organics, Environmental Science and Technology, 40, p2635-2643.
13. Dudhia, J., (1989), Numerical study of convection observed during the Winter Monsoon Experiment using a mesoscale two-dimensional model, Journal of the Atmo-spheric Sciences, 46, p3077-3107, 2.
14. Emmons, L.K., Walters, S., Hess, P.G., Lamarque, J.-F., Pfister, G.G., Fillmore, D., Granier, C., Guenther, A., Kinnison, D., Laepple, T., Orlando, J., Tie, X., Tyndall, G., Wiedinmyer, C., Baughcum, S.L., Kloster, S., (2010), Description and evaluation of the Model for Ozone and Related chemical Tracers, version 4 (MOZART-4), Geoscientific Model Development, 3, p43-67.
15. European Environment Agency, (2009), EMEP/EEA air pollutant emission inventory guidebook-2009, http://www.eea.europa.eu/publications/emep-eea-emissioninventory-guidebook-2009.
16. Gonzalez-Abraham, R., Chung, S.H., Avise, J., Lamb, B., Salathé, E.P. Jr., Nolte, C.G., Loughlin, D., Guenther, A., Wiedinmyer, C., Duhl, T., Zhang, Y., Streets, D.G., (2015), The effects of global change upon United States air quality, Atmospheric Chemistry and Physics, 15, p12645-12665.
17. Goto, D., Ueda, K., Ng, C.F.S., Takami, A., Ariga, T., Matsuhashi, K., Nakajima, T., (2016), Estimation of excess mortality due to long-term exposure to PM2.5 in Japan using a high-resolution model for present and future scenarios, Atmospheric Environment, 140, p320-332.
18. Guenther, A., Karl, T., Harley, P., Wiedinmyer, C., Palmer, P.I., Geron, C., (2006), Estimates of global terrestrial isoprene emissions using MEGAN (Model of Emissions of Gases and Aerosols from Nature), Atmospheric Chemistry and Physics, 6, p3181-3210.
19. Guenther, A.B., Jiang, X., Heald, C.L., Sakulyanontvittaya, T., Duhl, T., Emmons, L.K., Wang, X., (2012), The Model of Emissions of Gases and Aerosols from Nature version 2.1 (MEGAN2.1): an extended and updated framework for modeling biogenic emissions, Geoscientific Model Development, 5, p1471-1492.
20. Hagino, H., Morikawa, T., Hasegawa, S., Yonemochi, S., Sekiguchi, K., Kumagai, K., Yamaguchi, N., Iijima, A., Shimadera, H., Hayami, H., (2012), Measurement of Atmospheric Fine Particles using Aerosol Mass Spectrometer -Characterization of Organic Aerosols at Urban and Rural Sites in the Kanto Area during Summer of 2011-, JARI Research Journal, 20120803.
21. Hallquist, M., Wenger, J.C., Baltensperger, U., Rudich, Y., Simpson, D., Claeys, M., Dommen, J., Donahue, N.M., George, C., Goldstein, A.H., Hamilton, J.F., Herrmann, H., Hoffmann, T., Iinuma, Y., Jang, M., Jenkin, M.E., Jimenez, J.L., Kiendler-Scharr, A., Maenhaut, W., McFiggans, G., Mentel, T.F., Monod, A., Prévôt, A.S.H., Seinfeld, J.H., Surratt, J.D., Szmigielski, R., Wildt, J., (2009), The formation, properties and impact of secondary organic aerosol: Current and emerging issues, Atmospheric Chemistry and Physics, 9, p5155-5236.
22. Hertel, O., Berkowicz, R., Christensen, J., Hov, O., (1993), Test of two numerical schemes for use in atmospheric transport-chemistry models, Atmospheric Environment, 27A, p2591-2611.
23. Hong, S.-Y., Dudhia, J., Chen, S.-H., (2004), A Revised Approach to Ice Microphysical Processes for the Bulk Parameterization of Clouds and Precipitation, Monthly Weather Review, 132, p103-120.
24. Hong, S.-Y., Noh, Y., Dudhia, J., (2006), A new vertical diffusion package with an explicit treatment of entrainment processes, Monthly Weather Review, 134, p2318-2341.
25. Itahashi, S., Uno, I., Osada, K., Kamiguchi, Y., Yamamoto, S., Tamura, K., Wang, Z., Kurosaki, Y., Kanaya, Y., (2017), Nitrate transboundary heavy pollution over East Asia in winter, Atmospheric Chemistry and Physics, 17, p3823-3843.
26. Janjic, Z.I., (1994), The step-mountain eta coordinate model: further developments of the convection, viscous sublayer and turbulence closure schemes, Monthly Weather Review, 122, p927-945.
27. Japan Petroleum Energy Center, (2012a), Technical report of the Japan Auto-Oil Program: Emission inventory of road transport in Japan, Rep. JPEC-2011AQ-02-06, 2012, (in Japanese).
28. Japan Petroleum Energy Center, (2012b), Technical report of the Japan Auto-Oil Program: Emission inventory of sources other than road transport in Japan, Rep. JPEC-2011AQ-02-07, (in Japanese).
29. Kain, J.S., (2004), The Kain-Fritsch convective parameterization: An update, Journal of Applied Meteorology, 43, p170-181.
30. Kajino, M., Inomata, Y., Sato, K., Ueda, H., Han, Z., An, J., Katata, G., Deushi, M., Maki, T., Oshima, N., Kurokawa, J., Ohara, T., Takami, A., Hatakeyama, S., (2012), Development of the RAQM2 aerosol chemical transport model and predictions of the Northeast Asian aerosol mass, size, chemistry, and mixing type, Atmospheric Chemistry and Physics, 12, p11833-11856.
31. Kannari, A., Tonooka, Y., Baba, T., Murano, K., (2007), Development of multiple-species 1 km × 1 km resolution hourly basis emissions inventory for Japan, Atmospheric Environment, 41, p3428-3439.
32. Kelly, J.T., Bhave, P.V., Nolte, C.G., Shankar, U., Foley, K.M., (2010), Simulating emission and chemical evolution of coarse sea-salt particles in the Community Multiscale Air Quality (CMAQ) model, Geoscientific Model Development, 3, p257-273.
33. Kim, Y.J., Spak, S.N., Carmichael, G.R., Riemer, N., Stanier, C.O., (2014), Modeled aerosol nitrate formation pathways during wintertime in the great lakes region of North America, Journal of Geophysical Research, 119, p12,420-12,445.
34. Li, J., Dong, H., Zeng, L., Zhang, Y., Shao, M., Wang, Z., Sun, Y., Fu, P., (2015), Exploring Possible Missing Sinks of Nitrate and Its Precursors in Current Air Quality Models-A Case Simulation in the Pearl River Delta, China, Using an Observation-Based Box Model, Scientific Online Letters on the Atmosphere, 11, p124-128.
35. Makar, P.A., Nissen, R., Teakles, A., Zhang, J., Zheng, Q., Moran, M.D., Yau, H., Dicenzo, C., (2014), Turbulent transport, emissions and the role of compensating errors in chemical transport models, Geoscientific Model Development, 7, p1001-1024.
36. Ministry of Economy, Trade and Industry, (2011), Summary of PRTR Data in Japanese Fiscal Year 2009, http://www.meti.go.jp/policy/chemical_management/law/prtr/h21kohyo/gaiyou.html. (in Japanese).
37. Ministry of the Environment, (2013), Final Report of the Environment Research and Technology Development Fund (C-1001: Evaluation of Validity and Predictability of Air Quality Modeling for Urban PM2.5 in Japan), (in Japanese).
38. Mlawer, E.J., Taubman, S.J., Brown, P.D., Iacono, M.J., Clough, S.A., (1997), Radiative transfer for inhomogeneous atmospheres: RRTM, a validated correlated-k model for the longwave, Journal of Geophysical Research, 102, p16663-16682.
39. Morino, Y., Chatani, S., Hayami, H., Sasaki, K., Mori, Y., Morikawa, T., Ohara, T., Hasegawa, S., Kobayashi, S., (2010a), Inter-comparison of Chemical Transport Models and Evaluation of Model Performance for O3 and PM2.5 Prediction-Case Study in the Kanto Area in Summer 2007, Journal of Japan Society for Atmospheric Environment, 45, p212-226.
40. Morino, Y., Chatani, S., Hayami, H., Sasaki, K., Mori, Y., Morikawa, T., Ohara, T., Hasegawa, S., Kobayashi, S., (2010b), Evaluation of Ensemble Approach for O3 and PM2.5 Simulation, Asian Journal of Atmospheric Environment, 4, p150-156.
41. Morino, Y., Nagashima, T., Sugata, S., Sato, K., Tanabe, K., Noguchi, T., Takami, A., Tanimoto, H., Ohara, T., (2015), Verification of chemical transport models for PM2.5 chemical composition using simultaneous measurement data over Japan, Aerosol and Air Quality Research, 15, p2009-2023.
42. Nakanishi, M., Niino, H., (2006), An improved MellorYamada Level-3 model: Its numerical stability and application to a regional prediction of advection fog, Boundary-Layer Meteorology, 119, p397-407.
43. Nansai, K., Suzuki, N., Tanabe, K., Kobayashi, S., Moriguchi, Y., (2004), Design of Georeference-Based Emission Activity Modeling System (G-BEAMS) for Japanese emission inventory management, Proc.: the 13th International Emission Inventory Conference in Clearwater, Florida, United States of America.
44. Nawahda, A., Yamashita, K., Ohara, T., Kurokawa, J., Yamaji, K., (2012), Evaluation of premature mortality caused by exposure to PM2.5 and ozone in East Asia: 2000, 2005, 2020. Water, Air, and Soil Pollution, 223, p3445-3459.
45. Ohara, T., Akimoto, H., Kurokawa, J., Horii, N., Yamaji, K., Yan, X., Hayasaka, T., (2007), An Asian emission inventory of anthropogenic emission sources for the period 1980-2020, Atmospheric Chemistry and Physics, 7, p4419-4444.
46. Pleim, J.E., Chang, J.S., (1992), A non-local closure model in the convective boundary layer, Atmospheric Environment, 26A, p965-981.
47. Pleim, J.E., Xiu, A., Finkelstein, P.L., Otte, T.L., (2001), A coupled land-surface and dry deposition model and comparison to field measurements of surface heat, moisture, and ozone fluxes, Water, Air and Soil Pollution: Focus, 1, p243-252.
48. Pleim, J.E., (2007), A combined local and non-local closure model for the atmospheric boundary layer. Part 1: Model description and testing, Journal of Applied Meteorology and Climatology, 46, p1383-1395.
49. Pye, H.O.T., Pinder, R.W., Piletic, I.R., Xie, Y., Capps, S.L., Lin, Y.-H., Surratt, J.D., Zhang, Z., Gold, A., Luecken, D.J., Hutzell, W.T., Jaoui, M., Offenberg, J.H., Kleindienst, T.E., Lewandowski, M., Edney, E.O., (2013), Epoxide pathways improve model predictions of isoprene markers and reveal key role of acidity in aerosol formation, Environmental Science and Technology, 47, p11056-11064.
50. Robinson, A.L., Donahue, N.M., Shrivastava, M.K., Weitkamp, E.A., Sage, A.M., Grieshop, A.P., Lane, T.E., Pierce, J.R., Pandis, S.N., (2007), Rethinking Organic Aerosols: Semivolatile Emissions and Photochemical Aging, Science, 315(1259).
51. Sandu, A., Verwer, J.G., Blom, J.G., Spee, E.J., Carmichael, G.R., Potra, F.A., (1997), Benchmarking stiff ode solvers for atmospheric chemistry problems. II: Rosenbrock solvers, Atmospheric Environment, 31, p3459-3472.
52. Shimadera, H., Hayami, H., Morino, Y., Ohara, T., Chatani, S., Hasegawa, S., Kaneyasu, N., (2013), Analysis of summertime atmospheric transport of fine particulate matter in Northeast Asia, Asia-Pacific Journal of Atmospheric Sciences, 49, p347-360.
53. Shimadera, H., Hayami, H., Chatani, S., Morino, Y., Mori, Y., Morikawa, T., Yamaji, K., Ohara, T., (2014a), Sensitivity analyses of factors influencing CMAQ performance for fine particulate nitrate, Journal of Air Waste Management Association, 64, p374-387.
54. Shimadera, H., Hayami, H., Ohara, T., Morino, Y., Takami, A., Irei, S., (2014b), Numerical simulation of extreme air pollution by fine particulate matter in China in winter 2013, Asian Journal of Atmospheric Environment, 8, p25-34.
55. Shimadera, H., Kojima, T., Kondo, A., (2016), Evaluation of Air Quality Model Performance for Simulating Long-Range Transport and Local Pollution of PM2.5 in Japan, Advances in Meteorology, No. 5694251.
56. Skamarock, W.C., Klemp, J.B., (2008), A time-split nonhydrostatic atmospheric model for weather research and forecasting applications, Journal of Computational Physics, 227, p3465-3485.
57. Tokyo Metropolitan Government, (2011), Report for survey on emission sources of fine particulate matter, https://www.kankyo.metro.tokyo.jp/air/attachement/02_hassei.pdf. (in Japanese).
58. Wakamatsu, S., Morikawa, T., Ito, A., (2013), Air pollution trends in Japan between 1970 and 2012 and impact of urban air pollution countermeasures, Asian Journal of Atmospheric Environment, 7, p177-190.
59. Xiu, A., Pleim, J.E., (2001), Development of a land surface model part I: Application in a mesoscale meteorology model, Journal of Applied Meteorology, 40, p192-209.
60. Zhang, Q., Streets, D.G., Carmichael, G.R., He, K., Huo, H., Kannari, A., Klimont, Z., Park, I., Reddy, S., Fu, J.S., Chen, D., Duan, L., Lei, Y., Wang, L., Yao, Z., (2009), Asian emissions in 2006 for the NASA INTEXB mission, Atmospheric Chemistry and Physics, 9, p5131-5153.
61. Zheng, B., Zhang, Q., Zhang, Y., He, K.B., Wang, K., Zheng, G.J., Duan, F.K., Ma, Y.L., Kimoto, T., (2015), Heterogeneous chemistry: A mechanism missing in current models to explain secondary inorganic aerosol formation during the January 2013 haze episode in North China, Atmospheric Chemistry and Physics, 15, p2031-2049.