Current Issue

In order to estimate the effect of BVOC emissions from the Kinki region of Japan on urban O₃ concentration in Osaka, this study utilized the Community Multiscale Air Quality model (CMAQ) (Byun and Schere, 2006) version 5.0.1 driven with meteorological fields produced by the Weather Research and Forecasting model (WRF) (Skamarock and Klemp, 2008) version 3.5.1. The baseline WRF/CMAQ simulation in this study is identical to that of Nishimura et al. (2015). In addition, the Hybrid Single Particle Lagrangian Integrated Trajectory model (HYSPLIT) (Draxler and Hess, 1997) version 4 was used with the WRF-simulated wind field for backward trajectory analyses.

The WRF/CMAQ modeling system was run for a period from 6 July to 9 August 2010 with an initial spin-up period of 1 to 5 July 2010. Fig. 1 shows the modeling domains for CMAQ on Lambert conformal conic projection map and the location of Kokusetsu Osaka air pollution monitoring station. The horizontal domains consisted of 3 domains: domain 1 (D1) covering a wide area of East Asia, domain 2 (D2) covering the mainland of Japan, and domain 3 (D3) covering the Kinki region. The Kokusetsu Osaka station is located in Osaka City, the most urbanized area in D3. The horizontal resolutions and the number of grid cells in the domains are 64, 16 and 4 km, and 108×96, 68×68 and 76×76 for D1, D2 and D3, respectively. The vertical layers consisted of 30 sigma-pressure coordinated layers from the surface to 100 hPa with the middle height of the first layer being approximately 28 m.

Fig. 1. 
Modeling domains for air quality simulations and location of Kokusetsu Osaka air pollution monitoring station.

Table 1 shows WRF and CMAQ configurations in this study. The models were configured with the same physics and chemistry options as those used by Shimadera et al. (2014). The WRF simulation was conducted with on-line one-way nesting in the three domains using the mesoscale model grid point value (MSM-GPV) data by the Japan Meteorological Agency (JMA) and the final analysis (FNL) data by the U.S. National Centers for Environmental Prediction (NCEP). Nishimura et al. (2015) showed that WRF well simulated meteorological variables including air temperature, humidity, wind speed and shortwave radiation at the meteorological observatories in D3; and the WRF-simulated-meteorological fields were applicable to O₃ simulations.

For the CMAQ simulations, the hourly WRF results were processed by the Meteorology-Chemistry Interface Processor (MCIP) version 4.1. The MCIP-processed meteorological data were also used for the estimates of biogenic emission and backward trajectory analyses. CMAQ was configured with the Statewide Air Pollution Research Center version 99 (SAPRC99) (Carter, 2000) mechanism and the fifth generation CMAQ aerosol module (AERO5).

Emission data for the CMAQ simulations include NO_x, sulfur oxides, carbon monoxide, ammonia, non-methane VOC (NMVOC) and particulate matter emissions from anthropogenic, biogenic, biomass burning and volcanic sources. The details are described in the previous study (Nishimura et al., 2015). Anthropogenic emissions in Japan from vehicles, ships and other sectors for the simulations in summer 2010 were respectively derived from the Japan Auto-Oil Program (JATOP) Emission Inventory-Data Base in the year 2010 (JEI-DB) developed by the Japan Petroleum Energy Center (JPEC) (2012), the database in the year 2005 developed by the Ocean Policy Research Foundation (OPRF) (2012) and an emissions inventory in the year 2000 developed by Kannari et al. (2007).

Biogenic isoprene and monoterpene emissions in the Kinki region of Japan were estimated on the basis of Bao et al. (2010). They used the experimentally-derived standard emission rates of the dominant tree species and the forest database in the regions by Bao et al. (2008). The three dominant coniferous trees (Cryptomeria japonica, Chamaecyparis obtusa and Pinus densiflora) and the six dominant broadleaf trees (Quercus serrata, Quercus crispula, Fagus crenata, Quercus acutissima Carruthers, Quercus glauca and Quercus myrsinaefolia) respectively occupy 60% and 18% of the forest area in the region. The other biogenic emissions were estimated with the Model of Emissions of Gases and Aerosols from Nature (MEGAN) version 2.04 (Guenther et al., 2006).

The BVOC emission sources contributing to urban O₃ concentration in Osaka were identified by comparing the baseline simulation case of Nishimura et al. (2015) with zero-out emission cases in which emissions from target sources were set to zero. Fig. 2 shows BVOC source regions for zero-out emission cases in D3. The land areas in D3 were divided into 10 source regions: 7 regions in the Kinki region including Mie Prefecture (Reg.1), Shiga Prefecture (Reg.2), Kyoto Prefecture (Reg.3), Osaka Prefecture (Reg.4), Hyogo Prefecture (Reg.5), Nara Prefecture (Reg.6) and Wakayama Prefecture (Reg.7), the Chubu region in D3 (Reg.8), the Chugoku Region in D3 (Reg.9) and the Shikoku Region in D3 (Reg.10).

Fig. 2. 
BVOC source regions for zero-out emission cases in D3.

Fig. 3 shows spatial distributions of mean NMVOC emission rates and mean fractions of BVOC to the NMVOC emission rates for the simulation period. Major emission sources of NMVOC are stationary evaporative sources and biogenic sources. While the large emissions in the coastal lowland areas are dominated by AVOC from urban areas, large emissions in the mountainous areas are dominated by BVOC from forest vegetation. The temporal mean value of the total NMVOC emission rates for the simulation period in D3 was 3703 molC s^-1, 57% of which was accounted for by biogenic isoprene and monoterpenes. In respect of NO_x, there were large emissions from the coastal urban areas and from ships, and the temporal mean total emission rate in D3 was 360 mol s^-1 (Nishimura et al., 2015). Fig. 4 shows temporal variations of the total AVOC and BVOC emissions and mean regional AVOC and BVOC emissions in D3 in the simulation period. Although diurnal, weekday-weekend (vehicles only) and monthly variations were considered in AVOC emissions, the temporal variation of AVOC was much smaller than that of BVOC that was strongly affected by temperature and light intensity. As described by Nishimura et al. (2015), the meteorological conditions in the first 10 days of the simulation period were generally influenced by a seasonal rain front over Japan. Then, the study region tended to be covered with high pressure systems, which resulted in clear sky and high temperature conditions, except on 29 July, when low pressure systems brought precipitation. These variations of the meteorological conditions were reflected in the temporal variation of BVOC emission rate. In respect of the mean regional emissions, the ratios of Reg.4 was the highest for AVOC emissions and the lowest for BVOC emissions because Reg.4 has the most urbanized area and the smallest forest area in D3. However, Reg.4 is surrounded by regions with large BVOC emissions, such as Reg.3, Reg.5 and Reg.7.

Fig. 3. 
Horizontal distributions of mean NMVOC emission rates (a) and mean fractions of BVOC to the NMVOC emission rates (b) from 6 July to 9 August 2010 in D3.

Fig. 4. 
Temporal variations of total AVOC and BVOC emissions (a) and mean regional AVOC and BVOC emissions (b) from 6 July to 9 August 2010 in D3.

Zero-out cases for BVOC emissions were conducted for each source region in D3 with boundary concentrations identical to those of the baseline simulation case for the entire simulation period. The contribution of emissions from a source region to O₃ concentration was defined as the difference of O₃ concentration in the baseline case and a zero-out case for the region. In addition, a zero-out case for BVOC emissions in the entire D3 was conducted in order to estimate uncertainty associated with nonlinearity in atmospheric reactions. This study conducted much more detailed analyses than Bao et al. (2010) because the zero-out emission case in the previous study was only a single case for BVOC emissions in the entire Kinki region.

Fig. 5 shows hourly time series of observed and simulated wind speed and direction at Osaka meteorological observatory, and nitrogen dioxide (NO₂) and O₃ concentrations (the baseline case) at the Kokusetsu Osaka air pollution monitoring station from 6 July to 9 August 2010. The Osaka meteorological observatory is located at about 2 km west of the Kokusetsu Osaka station. Under clear sky conditions in summer in Osaka, wind speed exhibits a clear diurnal cycle with a peak in the daytime because of prevailing southwesterly sea breeze as shown in Fig. 5a and b. For NO₂, the baseline CMAQ simulation sometimes overestimated diurnal peaks in the morning but well captured day-by-day variation. Consequently, the Pearson’s correlation coefficient between the observation and the baseline simulation was 0.57 for hourly NO₂ concentrations and 0.80 for daily mean values. For O₃, high concentration peaks exceeding 100 ppb were observed on 8, 23 and 25 July, 2 and 3 August under clear sky and high temperature conditions. The baseline CMAQ simulation captured well the temporal variations including the occurrence of the O₃ concentration peaks. Consequently, the Pearson’s correlation coefficient between the observation and the baseline simulation was 0.87 for hourly O₃ concentrations, and the mean simulated daily maximum 1-h concentration (64.5 ppb) was close to the observed value (60.5 ppb). The performance evaluation of the baseline CMAQ simulation was described in greater detail in the previous study (Nishimura et al., 2015). The good CMAQ performance for urban O₃ simulation indicate that the model is applicable to estimates of the contribution of BVOC emissions to urban O₃ concentration.

Fig. 5. 
Hourly time series of observed and simulated wind speed (a) and direction (b) at Osaka meteorological observatory, and NO₂ (c) and O₃ (d) concentrations (baseline case) at Kokusetsu Osaka air pollution monitoring station from 6 July to 9 August 2010.

Fig. 6 shows estimated contributions of BVOC emissions from the 10 source regions in D3 and from the entire D3 to simulated daily maximum 1-h O₃ concentration at the Kokusetsu Osaka station in the simulation period. The BVOC contributions to daily maximum 1-h O₃ concentration at the Kokusetsu Osaka station were large for emissions from Reg.4 including the station and Reg.3, 5, 6 and 7 surrounding Reg.4 with dayby-day variations of contributing source regions. The results indicate that the BVOC emissions in the Kinki region substantially contribute to urban O₃ concentration in Osaka. Considering the finding of Nishimura et al. (2015) that monoterpenes tended to play a role of reducing O₃ concentration in the baseline simulation, the BVOC contribution is dominated by O₃ produced through isoprene reactions.

Fig. 6. 
Estimated contributions of BVOC emissions from each source region in D3 and from the entire D3 to daily maximum 1-h O₃ concentration at Kokusetsu Osaka air pollution monitoring station from 6 July to 9 August 2010.

Although the temporal variation of the sum of estimated BVOC contributions from Reg.1-10 was similar to that of the BVOC contribution from the entire D3, there were some discrepancies between the estimated BVOC contributions. The mean BVOC contribution to daily maximum 1-h O₃ concentration at the Kokusetsu Osaka station was 10.3 ppb (15.9%) for the sum of contributions from Reg.1-10 and 11.8 ppb (18.3%) for the contribution from the entire D3. This difference can be considered as an uncertainty of the estimated BVOC contribution by the zero-out emission method in this study, and is associated with nonlinearity in photochemical O₃ production through NO_x and VOC reactions, which has been discussed in a number of earlier literatures (e.g., Zhang et al., 2008; Sillman, 1999).

Fig. 7 shows horizontal distributions of O₃ concentration with wind fields and backward trajectories arriving at the Kokusetsu Osaka station on the days with relatively high concentration peaks (Fig. 5b; 14 local time (UTC+9) on 8 July, 15 local time on 23 July, at 14 local time on 2 August and 15 local time on 7 August). The backward trajectories make it easy to understand the transport pathway of an air mass arriving at the Kokusetsu Osaka station. Fig. 8 shows horizontal distributions of estimated contribution of BVOC emissions from the entire D3 to O₃ concentration in the same way as Fig. 7. BVOC emissions in D3 substantially contributed to high O₃ concentrations over Reg.4. The day-by-day variations of BVOC source regions contributing to O₃ concentration at the Kokusetsu Osaka station (Fig. 6) can be qualitatively explained by the backward trajectories. On 8 July, the air mass passed from Reg.5 to near the border of Reg.4, 6 and 7 before arriving at the Kokusetsu Osaka station (Fig. 7a), and the contributions of BVOC emitted from the regions were relatively large. On 23 July and 2 August, the trajectories indicated that air mass arriving at the Kokusetsu Osaka station was transported from the northwesterly direction (Fig. 7b and c), and the BVOC emissions from Reg.3 (the north from Reg.4) and Reg.5 (the west from Reg.4) showed relatively large contributions to O₃ concentrations at the station. On 7 August, the air mass passed from Reg.6 to Reg.4 (Fig. 7d), and the BVOC emissions from the regions contributed to O₃ concentration at the Kokusetsu Osaka station. In addition, on 16-19 July with prevailing southerly wind in D3 (not shown), although O₃ concentrations at the Kokusetsu Osaka were relatively low (Fig. 5d), the BVOC emissions from Reg.7 (the south from Reg.4) dominantly contributed to O₃ at the station among those from the 10 source regions.

Fig. 7. 
Horizontal distributions of O₃ concentration with wind field at the first layer and backward trajectory arriving at 300 m above Kokusetsu Osaka air pollution monitoring station (circles represent 3-hour transport distances) at 14 local time on 8 July (a), 15 local time on 23 July (b), 14 local time on 2 August (c) and 15 local time on 7 August (d).

Fig. 8. 
Horizontal distributions of estimated contribution of BVOC emissions from the entire D3 to O₃ concentration with wind field at the first layer and backward trajectory arriving at 300 m above Kokusetsu Osaka air pollution monitoring station (circles represent 3-hour transport distances) at 14 local time on 8 July (a), 15 local time on 23 July (b), 14 local time on 2 August (c) and 15 local time on 7 August (d).