Accumulated phytotoxic ozone dose estimation for deciduous forest in Kanto, Japan in summer

Atmospheric Environment 129 (2016) 176e185

Contents lists available at ScienceDirect

Atmospheric Environment journal homepage: www.elsevier.com/locate/atmosenv

Accumulated phytotoxic ozone dose estimation for deciduous forest in Kanto, Japan in summer* Takanori Watanabe*, Takeki Izumi, Hiroshi Matsuyama Department of Geography, Tokyo Metropolitan University, Tokyo, Japan

h i g h l i g h t s
We investigated the O3 risk for deciduous forest in the Kanto, Japan in summer.
We showed the possibility that plant growth was suppressed in the north of the Kanto.
Air temperature and humidity strongly affected O3 uptake in the dry year.
The O3 concentration and irradiance strongly affected O3 uptake in the humid year.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 23 April 2015 Received in revised form 21 December 2015 Accepted 8 January 2016 Available online 9 January 2016

With ozone concentrations simulated using a regional chemical transport model (ADMER-PRO) and high-spatial resolution meteorological data, we investigated the influence of ozone concentration on deciduous forests in the Kanto region of Japan in summer during 2003, 2004, and 2009: three years for which weather characteristics differed greatly. Ozone risk for plants was assessed by the accumulated phytotoxic ozone dose (POD), a flux-based index. The effects were analyzed by particularly addressing the relation between the stomatal ozone flux and meteorological elements. Results revealed high absorption areas not only where injury to forests had been visually detected in previous studies, but also where injury had not been observed to date. Regarding the relation between the stomatal ozone flux and meteorological elements, air temperature and vapor pressure deficit strongly affected POD in 2004, when high temperature and little rainfall were observed. Additionally, the ozone concentration and irradiance strongly affected POD in 2003 when low temperatures and heavy rainfall were observed. The meteorological elements affecting POD differed from year to year. Results demonstrate the importance of multi-year simulations and analyses in the field of ozone risk assessment. © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Stomatal ozone flux Deciduous forest Chemical transport model Fine mesh Multi-year simulation

1. Introduction In eastern Asia, the tropospheric ozone concentration is increasing year-by-year. Ohara (2010) reported that the annual averaged concentration of ozone in Japan increased rapidly (0.29 ppb/year) during 1985e2007. Moreover, high ozone concentrations were observed not only in urban areas but also in mountainous areas in Japan. Ozone's effects on forests constitute a severe

* This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial-No Derivative Works License, which permits non-commercial use, distribution, and reproduction in any medium, provided the original author and source are credited. * Corresponding author. Department of Geography, Tokyo Metropolitan University, 1-1, Minami-Osawa, Hachioji, Tokyo 192-0397, Japan. E-mail address: [email protected] (T. Watanabe).

and widespread problem. Surface ozone affects plants. Sitch et al. (2007) reported that ozone reduces carbon dioxide absorption by plants. High ozone concentrations injure plants and kill them in the worst cases, underscoring the importance of ozone risk assessment to plants. To date, some ozone risk assessments using concentration-based indexes have been conducted (e.g., Pochanart et al., 2002; Kohno et al., 2005). Concentration-based indexes are easy to use because they are calculable by the ozone concentration alone. However, concentration-based indexes are inadequate for risk assessment for plants growing in harsh climates such as those with high temperature and drought (Matyseek et al., 2007). Conversely, flux-based indexes can assess the risk adequately in harsh climates because the indexes incorporate the effects of meteorological elements (Watanabe and Yamaguchi, 2011). Therefore, risk assessments using flux-based indexes have been increasing during the past decade

http://dx.doi.org/10.1016/j.atmosenv.2016.01.016 1352-2310/© 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

T. Watanabe et al. / Atmospheric Environment 129 (2016) 176e185

(Hoshika et al., 2011). The accumulated phytotoxic ozone dose (POD) is a flux-based index that can evaluate risk more accurately. In Europe, POD has been used to assess risk for agricultural crops and forests. Emberson et al. (2000), after calculating the stomatal ozone flux across Europe, suggested that vapor pressure deficit, soil moisture deficit, and phenology limit the stomatal ozone uptake. Klingberg et al. (2011), after investigating the influence of climate change for POD in Europe, predicted that POD would remain unchanged or would decrease because rising carbon dioxide concentrations decrease the stomatal conductance. Regarding studies using POD in Japan, Hoshika et al. (2011) calculated the horizontal distribution of POD across Japan. They reported that dryness limited the stomatal ozone uptake of

177

deciduous trees in western Japan. Saito et al. (2013) calculated POD for the Tanzawa Mountains (Fig. 1). Their results demonstrated that POD in the region greatly exceeded the critical level. Izuta (2013), after evaluating ozone effects on photosynthesis of deciduous trees in Japan, proposed locations where observation stations should be built. Nevertheless, risk assessment studies using POD are far fewer in Japan than in Europe. Actually, POD depends not only on the ozone concentration; it also depends to a great degree on meteorological elements (air temperature, humidity, irradiance, and wind speed), soil moisture, and phenology. Numerous studies have been conducted of the effects of meteorological elements (e.g., Emberson et al., 2000). However, most of these studies have targeted the effect in a single year despite that fact that the weather characteristics vary from

139˚

140˚

141˚

37˚

37˚

36˚

36˚

35˚

35˚

139˚

140˚

141˚ (m)

0

500

1000

1500

2000

2500

3000

Fig. 1. Topography of the Kanto region. Solid circles represent monitoring stations. The open circle and the open rectangle respectively denote the Tanzawa Mountains and Tokyo Bay.

178

T. Watanabe et al. / Atmospheric Environment 129 (2016) 176e185

Domain 2

Domain 1

Fig. 2. Computational domain.

year to year. Therefore, it is necessary to study the effects of meteorological elements in multiple years including high temperature and drought. In addition, simulations should be conducted with high spatial resolution to calculate POD accurately because the ozone concentration and meteorological elements change greatly in space. However, the horizontal resolution of calculating POD in previous studies is coarse, i.e., the resolutions used by Hoshika et al. (2011) and Simpson et al. (2007) were 40e50 km because these studies were conducted to evaluate widespread POD. It is necessary to simulate POD with high spatial resolution to assess the effects of meteorological elements accurately. Based on the high-spatial simulations, the objectives of this study are to ascertain the areas where forests might be harmed by the ozone, and to evaluate the effects of meteorological elements on ozone absorption by the forest with emphasis on year-to-year variations. For this study, we simulated ozone concentration and meteorological fields in the summers of 2003, 2004, and 2009, years known to have different weather characteristics. Then we calculated POD to assess the risk to deciduous trees. We also evaluated high-risk areas of POD in relation to the analyses of the meteorological elements. The study area was the Kanto region, including the Tokyo metropolitan area in Japan (Fig. 1), where ozone concentrations show an increasing trend (Ohara, 2010). 2. Study methods 2.1. Model description We used ADMER-PRO Ver. 1.0, developed by the National Institute of Advanced Industrial Science and Technology, Japan. The

software comprises Regional Atmospheric Modeling System (RAMS; Pielke et al., 1992) Ver. 4.4, incorporating anthropogenic heat processes and several chemical processes including reactions, emissions, and deposition of chemical substances. A salient benefit of the system is that it can calculate meteorological fields and transports of chemical substances simultaneously, which is certainly different from other chemical transport models. The configurations of ADMER-PRO used for this study included radiation (Chen and Cotton, 1983), microphysics (Walko et al., 1995), and the planetary boundary layer (Mellor and Yamada, 1982). The reaction process of Carbon Bond Mechanism e IV (CB-IV_99; Adelman, 1999) was used in this model. The deposition process described by Zhang et al. (2003) was also adopted. Two domains were set for calculations (Fig. 2). The coarse domain (Domain 1) was 800 km  800 km in the horizontal direction: its resolution was 20 km. The fine domain (Domain 2) was 250 km  270 km, with resolution of 5 km the resolution used for this study was higher than that used in previous studies. For that reason, it is possible that this study calculated the influence of meteorological elements for POD more accurately. In addition, the top of the model was configured at 20 km with the vertical level of 29 layers. The models were run respectively from 28 June through 31 August in 2003, 2004, and 2009. Regarding the weather characteristics of the Kanto region in each year, the air temperature was below normal; the rainfall was above normal in 2003 (JMA, 2003). For 2004, the air temperature was above normal; the rainfall was below normal (JMA, 2004). The air temperature and rainfall were almost equal to the long-term means in 2009 (JMA, 2009). Clearly, the selected years had different weather characteristics. We used NCEP final analysis as boundary data of the

T. Watanabe et al. / Atmospheric Environment 129 (2016) 176e185

meteorological field. Surface anthropogenic emissions data for primary chemical species and non-methane volatile organic compounds (NMVOCs) were used as emission data (Kannari et al., 2007). These emission data are monthly mean hourly data in 2005, for which the horizontal resolution is approximately 1 km. We used these data as the lower boundary condition for each year simulation because of the data availability of NMVOCs. Although the amount of NMVOCs decreased monotonically from 2000 onward (Ministry of the Environment, 2015), the effect was not incorporated into the simulation, as discussed in Section 3.1. Regarding biogenic VOC (BVOC) emission data, this model used upwardly revised BVOC data created by Kannari et al. (2007). In the model, BVOC emissions were modified by air temperature and solar radiation. Therefore, the application of these BVOC data to 2003, 2004, and 2009 affected the results only slightly. Inoue et al. (2010a, b) reported that modification of the BVOC data improved the reproducibility of the model. Therefore, we adopted the same method.

179

ftemp is a function of temperature. It is given by Equations (3) And (4),

n    ftemp ¼ max fmin ; ðT  Tmin Þ= Topt  Tmin *   bt o ; ðTmax  TÞ= Tmax  Topt     bt ¼ Tmax  Topt = Topt  Tmin ;

(3)

(4)

where Tmin, Tmax, and Topt respectively represent the minimum, maximum, and optimum temperature [ C] for stomatal conductance. Stomatal closure occurs at Tmin and Tmax. fVPD, the function of vapor pressure deficit, is given by Equation (5).

fVPD ¼ minf1; maxffmin ; ðð1  fmin Þ*ðVPDmin  VPDÞ=ðVPDmin  VPDmax ÞÞ þ fmin gg: (5)

2.2. Estimation of stomatal ozone flux We calculated the accumulated phytotoxic ozone dose (POD) based on UNECE (2010). To estimate POD, it is necessary to calculate the stomatal conductance, which greatly affects POD. The stomatal conductance (gsto [mmol/m2PLA/s]) was calculated using the following multiplicative algorithm:

i o h  n  gsto ¼ gmax * min fphen ;fO3 *flight *max fmin ; ftemp *fVPD *fSWP : (1) In that equation, gmax denotes the maximum stomatal conductance [mmol/m2PLA/s]. The projected leaf area, PLA, is the total area of sunlit leaves. The functions (fphen, fO3, flight, fmin, ftemp, fVPD, fSWP) are limiting factors of the stomatal conductance, which take values between 0 and 1. The decrease of the limiting factor engenders the decrease of stomatal conductance. We set fmin, which determines the minimum stomatal conductance, as 0.1 based on UNECE (2010). Here, fphen is the function of phenology, which expresses the difference of stomatal flux by leaf age. We set fphen to 1 because the limitation of stomatal ozone flux by leaf age in summer is extremely small. We set fO3, the function of ozone concentration, to 1 based on UNECE (2010). The function of soil water potential, fSWP, was set to 1 following Hoshika et al. (2011) and Saito et al. (2013) because the influence of soil moisture on temperate deciduous trees was assumed to be small by reflecting the fairly large precipitation in summer in the Kanto region. flight is the function of photosynthetically photon flux density, which is given as

flight ¼ 1  expððlighta ÞPFDÞ;

(2)

where lighta is the parameter that determines the function shape. PFD is the photosynthetic photon flux density [mmol/m2/s]. Table 1 Required parameters for calculating POD. Variable

Unit

Value

gmax flight Tmin Tmax Topt VPDmin VPDmax

mmol O3/m2 PLA/s Constant Degrees Celsius Degrees Celsius Degrees Celsius kPa kPa

150 0.006 0 35 21 3.25 1.00

Therein, VPD stands for the vapor pressure deficit [kPa]. When VPD is less than VPDmin [kPa], fVPD is set to 1. Conversely, when VPD is greater than VPDmax [kPa], fVPD is set to fmin. The stomatal ozone flux (Fst [nmol/m2/s]) was calculated using Equation (6), which assumes that ozone concentration at the top of the canopy represents the concentration at the sunlit upper canopy leaves:

rc Fst ¼ Cðz1 Þ*gsto * rb þ rc :

(6)

In Equation (6), C(z1) stands for the ozone concentration [nmol/ m3] at the canopy top, gsto represents the converted unit from [mmol/m2/s] to [m/s], rb denotes the quasi-laminar resistance [s/ m], and rc represents the leaf surface resistance [s/m]. Also, rb is given by Equation (7) as

sffiffiffiffiffiffiffiffiffiffiffi L ; rb ¼ 1:3*150* uðz1 Þ

(7)

where 1.3 represents the difference in diffusivity between heat and ozone. L [m] is the mean leaf width and u(z1) [m/s] is the wind speed at height z1. Also, rc is given by Equation (8) as

rc ¼ 1=ðgsto þ gext Þ;

(8)

where gext [m/s] stands for the external leaf, or cuticular, conductance. We set gext 0.0004 based on UNECE (2010). POD [mmol/m2] was calculated using Equation (9), which shows the accumulated stomatal ozone flux as

Z PODy ¼

maxðFst  y; 0Þdt:

(9)

Therein, y denotes a threshold of the stomatal ozone flux [nmol/ m2/s]. We set y 0.0 based on a description by Hoshika et al. (2011). Similarly, we chose the required parameters of calculating POD following UNECE (2010) (Table 1).

3. Results 3.1. Validation of simulated values To validate the simulated ozone concentrations, we compared the simulated data and observed data reported by the Ministry of the Environment and local governments in Japan (Fig. 1). Fig. 3

180

T. Watanabe et al. / Atmospheric Environment 129 (2016) 176e185

(b)

(a) (ppb)

(c)

(e)

(ppb)

(ppb)

(ppb)

(d)

(f)

(ppb)

(ppb)

Fig. 3. Observed and simulated distributions of ozone concentration (ppb) in summer (average of July and August) in the study area. (a) Observation in 2003, (b) simulation in 2003, (c) same as (a) but for 2004, (d) same as (b) but for 2004, (e) same as (a) but for 2009, and (f) same as (b) but for 2009.

displays observed and simulated distributions of ozone concentrations in summer of 2003, 2004, and 2009, respectively averaged for July and August. In each year, low concentrations are found around Tokyo Bay. High concentrations are distributed in the suburbs of the study area both in observations and simulations. Although the absolute values are systematically different between them, the spatial pattern of the ozone concentration is reproduced well.

Fig. 4 displays the spatially averaged time series of ozone concentration at monitoring stations in Domain 2 (Fig. 1) and the model's grids corresponding to these stations. This study calculated the ozone concentration at a bottom layer (50 m) of the model, whereas the actual concentration was observed between 1 m and 30.2 m from the surface. In comparison with the observations, simulations overall overestimated them in summer. However, the diurnal variations of ozone concentration were well reproduced in

T. Watanabe et al. / Atmospheric Environment 129 (2016) 176e185

(a)

120

181

(a)

100

(mmol/m2) 80 60 40 20 0

(b)

7/1

7/11

7/21

7/31

8/10

8/20

8/30

120

100

(b) 80

(mmol/m2)

60 40 20 0

(c)

7/1

7/11

7/21

7/31

8/10

8/20

8/30

120

100 80

(c) 60

(mmol/m2)

40 20 0

7/1

7/11

7/21

7/31

8/10

8/20

8/30

Fig. 4. Time series of ozone concentration (ppb) in summer of (a) 2003, (b) 2004, and (c) 2009. Solid and dashed lines respectively represent calculated and observed values.

Table 2 Performance statistics on predictions of one hour ozone concentration.

Benchmark 2003 2004 2009

MNB

MNGE

N

 ±0.15 0.06 0.15 0.15

0.30 0.26 0.26 0.28

21,948 44,147 23,817

each year. The difference in height between observations and simulations might contribute to the overestimation of the latter because the ozone concentration was observed under the tree canopy or the urban canopy. Grant and Wong (1999) reported that diurnal variation of the ozone concentration differed above and below canopies over suburban areas. Although differences between observations and simulations are large, especially at night, the

Fig. 5. Spatial distributions of POD0 in summer of (a) 2003, (b) 2004, and (c) 2009. This study did not calculate POD0 at non-forest meshes which were not colored.

differences do not affect the stomatal ozone flux because the calculation method based on UNECE (2010) does not incorporate the absorption of ozone by plants at night. Moreover, studies of species-specific absorption at night are still inadequate and are in progress. Matyseek et al. (2004) reported that absorption at night was approximately 20% of that in the daytime, so that improvement of the model reproducibility at night remains as an issue for future work. Another candidate of overestimation is the difference of the period of NMVOCs between observation (Kannnari et al., 2007) and

182

T. Watanabe et al. / Atmospheric Environment 129 (2016) 176e185

(a)

MNB ¼

N 1 X Cali  Obsi : N i¼1 Obsi

(10)

(m/s)

MNGE ¼

(b)

(m/s)

N 1 X jCali  Obsi j : N i¼1 Obsi

(11)

In these equations, N represents the number of data, Cal stands for the calculated ozone concentration [ppb], and Obs is the observed ozone concentration [ppb]. We calculated MNB and MNGE only if the observed value was greater than or equal to 60 ppb, which is the air quality standard of ozone concentration in Japan (Ministry of the Environment, 1993). For validation, this study used the benchmark which is a criterion used to judge the reproducibility of the model (U.S. EPA, 2007). Table 2 presents performance statistics for predictions of hourly ozone concentration. Results show that all of MNB are positive in 2003, 2004, and 2009, i.e., the calculated values overestimated the observations, as already pointed out by Figs. 3 and 4. Still, both MNB and MNGE are smaller than their respective benchmarks in all years (±0.15 for NMB, and 0.30 for MNGE), along with the reproducibility of spatial pattern and diurnal variation (Figs. 3 and 4). Probably, the ozone concentration was fairly well reproduced, although systematic biases appear in the simulations. We can discuss the simulated results carefully if we incorporate consideration of the characteristics of this overestimation. 3.2. Stomatal ozone flux Fig. 5 presents spatial distributions of POD0 in summer (JulyeAugust) of 2003, 2004 and 2009. In these figures, POD0 was calculated only for forest areas. POD0 in 2003 and 2009 presented

(c)

(m/s)

Fig. 6. Same as Fig. 5 but for averaged stomatal conductance at daytime (6:00e18:00).

simulation. Although the amount of NMVOCs decreased monotonically from 2000 onward (Ministry of the Environment, 2015), the simulated ozone concentration still overestimated the observation in all three years (Fig. 4), so that we cannot attribute it to the cause of the overestimation. Next, we evaluated the calculated values using Mean Normalized Bias (MNB) and Mean Normalized Gross Error (MNGE) based on U.S. EPA data (2007). MNB and MNGE are calculated, respectively, using Equations (10) And (11).

Fig. 7. Averaged diurnal variations of (a) the stomatal ozone flux (nmol/m2/s) and (b) the stomatal conductance (m/s). Solid, dashed, and dotted lines respectively present values from 2003, 2004, and 2009.

T. Watanabe et al. / Atmospheric Environment 129 (2016) 176e185

Fig. 8. Same as Fig. 7 but for (a) flight, (b) ftemp, and (c) fVPD, all of which are unitless (ratio).

approximately 8e14 mmol/m2 in Domain 2 (Fig. 5a and c). In particular, the highest POD0 areas are located at the border of the plains and mountains. Low temperature and heavy rainfall were observed in 2003. The temperature and rainfall in 2009 were almost equal to their long-term mean values. In contrast, high temperature and little rainfall were observed in 2004 when POD0 was larger than those in 2003 and 2009. In 2004, the areas larger than 16 mmol/m2 were found (Fig. 5b), i.e., POD0 is very high at mountainous areas and high at coastal areas in that year. Regarding high POD0 areas, the Tanzawa Mountains (Fig. 1) exhibited higher POD0 every year, where forest injury was observed by Saito et al. (2013). POD0 in this area is 12e16 mmol/m2, which is slightly larger than the result of Saito et al. (2013), who reported POD1 in this area as approximately 12 mmol/m2 in 2010. Considering the characteristics of the overestimation of the model (Figs. 3 and 4), this study and that of Saito et al. (2013) quantitatively agree. Fig. 5 shows that plant growth might be suppressed at the west and north of the Kanto plain, but no visible injury caused by ozone has

183

been reported in these areas. To clarify the effects of meteorological elements that caused different spatial distributions of POD0 in each year, we showed the spatial distribution of averaged stomatal conductance at daytime (6:00e18:00) (Fig. 6) because equation (1) means that the stomatal conductance is affected by meteorological elements. Fig. 6 displays the averaged stomatal conductance during daytime. It is 0.0014e0.0034 m/s, which is of the same order as the figure reported by Kadaira and Yoshida (2006), showing that averaged stomatal conductance at daytime of deciduous trees (Zelkova serrata) was 0.0047 m/s (converted from mol/m2/s units). Fig. 6 also shows that the stomatal conductance was large at high elevation areas of the west and north of the Kanto region in all three years. In particular, those areas where the stomatal conductance was greater than 0.0032 m/s are found in 2004 (Fig. 6b). Therefore, the deciduous forest in mountainous areas absorbed the ozone most easily in the Kanto region. From the viewpoint of year-to-year variation, stomatal conductance in 2004 was the highest, whereas that in 2004 at low elevation areas was less than those in 2003 and 2009. The stomatal conductance in 2003 was distributed between 0.0016 and 0.0028 m/s (Fig. 6a), i.e., the difference between the maximum and minimum was the smallest in these three years. For further analysis of factors affecting the stomatal ozone flux, we compared diurnal variations of the stomatal ozone flux, the stomatal conductance, flight, ftemp, and fVPD. Fig. 7 shows averaged diurnal variations of the stomatal ozone flux and stomatal conductance, whereas Fig. 8 shows the data in Fig. 7 but for flight, ftemp, and fVPD. These figures were calculated only for forest areas. Fig. 7a shows that the stomatal ozone flux in 2004 was the largest (18,614 nmol/m2/s) in the morning among all three years. Stomatal ozone flux in 2004 started decreasing at 10:00, but those in 2003 and 2009 continued increasing until 13:00 and 14:00, respectively. They showed similar diurnal variations (Fig. 7a). Fig. 7b depicts the diurnal variations of the stomatal conductance. In each year, the maximum values appeared in the morning. The maximum (0.003 m/s) appeared at 8:00 in 2004, decreasing rapidly thereafter. In contrast, the decreases of the stomatal conductance in 2003 and 2009 were gentle. As for diurnal variations of flight, the maximum appeared at 12:00 in all three years (Fig. 8a), among which that in 2004 (0.95) was the highest. In the three years, ftemp and fVPD presented similar diurnal variations in that the minimum appeared at 13:00 or 14:00 (Fig. 8b and c). In comparison with other years, fVPD in 2004 showed noticeable diurnal variation in that the minimum of fVPD (0.70) appeared at 14:00. 4. Discussion In this study, we showed that the spatial distributions of POD0 in summer had different characteristics in 2003, 2004, and 2009 (Fig. 5). In particular, POD0 in 2004 was very high in mountainous areas and low at the border of the plains and mountains (Fig. 5b). The distribution caused the stomatal conductance there (Fig. 6b) because POD0 and the stomatal conductance showed strong positive correlation (r ¼ 0.80) in 2004. Additionally, it seems that the stomatal conductance in 2004 was affected by weather characteristics in 2004, when high temperature and little rainfall were observed. The stomatal conductance in 2004 was low at the border of the plains and the mountains (Fig. 6b) because the stomatal conductance was limited by high temperature and dryness there in 2004. In 2003, POD0 was high at the border of the plains and the mountains (Fig. 5a). In comparison with the stomatal conductance in 2003 (Fig. 6a), POD0 was affected only slightly by the stomatal conductance because neither showed strong correlation

184

T. Watanabe et al. / Atmospheric Environment 129 (2016) 176e185

(r ¼ 0.05) in 2003. The results also suggest that the stomatal conductance did not affect POD0 because they did not have strong correlation (r ¼ 0.09) either in 2009. In other words, the spatial distributions of POD0 were affected by those of the ozone concentration in 2003 and 2009 because POD was proportional to the product of the ozone concentration and the stomatal conductance (Eq. (6)). The respective correlations between POD0 and the ozone concentration in 2003 and 2009 were 0.78 and 0.75, although the correlation was low (r ¼ 0.10) in 2004. One purpose of this study was to evaluate the effect of meteorological elements on POD from the viewpoint of year-to-year variations. Emberson et al. (2000) similarly compared the stomatal conductance and the effect of meteorological elements every 6 h in the Czech Republic. In their study, VPD and air temperature were the most important factors limiting stomatal conductance during the daytime; VPD remained the predominant limiting factor until the early evening. As a result, the stomatal ozone uptake decreased greatly at noon, although the ozone concentration at noon was the highest of the day. In contrast, irradiance limited stomatal conductance strongly in the early morning and the early evening. In this study, diurnal variations of the limiting factors (flight, ftemp, and fVPD) generally agreed with the results reported by Emberson et al. (2000) (Figs. 7 and 8). However, the effect of the limiting factor varied greatly from year to year. In 2004, VPD and air temperature caused the rapid decline of the stomatal conductance, which caused the decrease of the stomatal ozone flux in the afternoon (Fig. 7b). However, the stomatal conductance in 2004 was not less than those in 2003 and 2009 because flight in 2004 was the highest (Fig. 8a). In contrast, irradiance limited the stomatal conductance in 2003, and air temperature and VPD had little effect on the stomatal conductance. Results of diurnal variations of the stomatal conductance and the limiting factors suggest that air temperature and VPD strongly affected POD in 2004 when high temperature and little rainfall were observed. High temperatures greatly decreased POD because high temperatures tended to cause high VPD. Not only the ozone concentration, but also irradiance strongly affected POD in 2003 when low temperature and heavy rainfall were observed. These results underscore the importance of evaluating the effect of meteorological elements on POD from the viewpoint of year-toyear variations because the effect varied greatly from year to year. 5. Summary and conclusion For this study, we evaluated POD using multi-year numerical simulations in the Kanto region of Japan in summer. We analyzed the relations between POD and meteorological elements that affect POD. New findings obtained from this study are summarized as follows.
Our results show a high absorption area in the Tanzawa Mountains, which agrees well with the results reported by Saito et al. (2013). However, damage caused by ozone has not been reported yet, even in the inland area of Kanto plain. Our results show the possibility that plant growth is suppressed in northern areas of the Kanto plain.
Our results suggest that air temperature and VPD strongly affected POD in 2004 when high temperature and little rainfall were observed. In contrast, the ozone concentration and irradiance strongly affected POD in 2003 when low temperature and heavy rainfall were observed. We demonstrated that the meteorological elements affecting POD changed from year to year, depending on the weather characteristics of the year, and demonstrated the importance of multi-year simulations and analyses.

As remaining problems, it is necessary to incorporate nocturnal absorption of ozone into POD calculations (see Section 3.1). Additionally, it is important to calculate POD including the effects of soil moisture to assess risk more adequately (see Section 2.2). Both are issues to be resolved in the near future. Acknowledgments ADMER-PRO Ver. 1.0, developed by National Institute of Advanced Industrial Science and Technology, was used for these calculations. The observed ozone concentration data were provided by the National Institute for Environmental Studies, Japan. References Adelman, Z.E., 1999. A Reevaluation of the Carbon Bond-iv Photochemical Mechanism. School of Public Health, University of North Carolina, Chapel Hill, North Carolina. Chen, C., Cotton, W.R., 1983. A one-dimensional simulation of the stratocumuluscapped mixed layer. Bound.-Layer Meteor 25, 289e321. Emberson, L.D., Ashmore, M.R., Cambridge, H.M., Simpson, D., Tuovinen, J.-P., 2000. Modelling stomatal ozone flux across Europe. Environ. Pollut. 109, 403e413. Grant, R.H., Wong, K.-L., 1999. Ozone profiles over a suburban neighborhood. Atmos. Environ. 33, 51e63. Hoshika, Y., Shimizu, Y., Omasa, K., 2011. A comparison between stomatal ozone uptake and AOT40 of deciduous trees in Japan. iForest 4, 128e135. Inoue, K., Yasuda, R., Yoshikado, H., Higashino, H., 2010a. Spatial distribution of summer-time surface ozone sensitivity to NOx and VOC emissions for the Kanto area Part 1: estimation by numerical simulations with two kinds of (larger and smaller) biogenic emission estimates. J. Jpn. Soc. Atmos. Environ. 45, 183e194 (in Japanese with English summary). Inoue, K., Yoshikado, H., Higashino, H., 2010b. Spatial distribution of summer-time surface ozone sensitivity to NOx and VOC emissions for the Kanto area Part 2: estimation based on the measurement of a photochemical indicator. J. Jpn. Soc. Atmos. Environ. 45, 195e204 (in Japanese with English summary). Izuta, T., 2013. Evaluation of Effects of Ozone on Japanese Forest Trees Based on Leaf Ozone Uptake. Final Report of the Environment Research and Technology Development Fund, pp. 1e115 (in Japanese with English summary). JMA, 2003. Weather in Summer in 2003 (From June to August). Available from: http://www.data.jma.go.jp/obd/stats/data/stat/tenko030608.pdf (accessed November 29, 2015) (in Japanese). JMA, 2004. Weather in Summer in 2004 (From June to August). Available from: http://www.data.jma.go.jp/obd/stats/data/stat/tenko040608.pdf (accessed November 29, 2015) (in Japanese). JMA, 2009. Weather in Summer in 2009 (From June to August). Available from: http://www.data.jma.go.jp/obd/stats/data/stat/tenko090608.pdf (accessed November 29, 2015) (in Japanese). Kadaira, A., Yoshida, H., 2006. Study on a model of stomatal conductance on an evergreen broad leaves tree and a deciduous broad leaves tree. J. Environ. Eng. AIJ 608, 105e111 (in Japanese with English summary). Kannari, A., Tonooka, Y., Baba, T., Murano, K., 2007. Development of multiplespecies 1 km  1 km resolution hourly basis emissions inventory for Japan. Atmos. Environ. 41, 3428e3439. Klingberg, J., Engardt, M., Uddling, J., Karlsson, P.E., Pleijel, H., 2011. Ozone risk for vegetation in the future climate of Europe based on stomatal ozone uptake calculations. Tellus 63A, 174e187. Kohno, Y., Matsumura, H., Ishii, T., Izuta, T., 2005. Establishing Critical Levels of Air Pollutants for Protecting East Asian Vegetation e a Challenge. Plant Responses to Air Pollution and Global Change. Springer, Japan, pp. 243e250. Matyssek, R., Bytnerowicz, A., Karlsson, P.E., Paoletti, E., Sanz, M., Schaub, M., Wieser, G., 2007. Promoting the O3 flux concept for European forest trees. Environ. Pollut. 146, 587e607. Matyssek, R., Wieser, G., Nunn, A.J., Kozovits, A.R., Reiter, I.M., Heerdt, C., Winkler, J.B., Baumgarten, M., Haberle, K.-H., Grams, T.E.E., Werner, H., Fabian, P., Havranek, W.M., 2004. Comparison between AOT40 and ozone uptake in forest trees of different species, age and site conditions. Atmos. Environ. 38, 2271e2281. Mellor, G.L., Yamada, T., 1982. Development of a turbulence closure model for geophysical fluid problems. Rev. Geophys. Space Phys. 20, 851e875. Ministry of the Environment, 1993. The Basic Environmental Low: Environmental Quality Standards in Japan e Air Quality. Available from: http://www.env.go.jp/ en/air/aq/aq.html (accessed 04.04.15.). Ministry of the Environment, 2015. Report on Volatile Organic Compound (VOC) Emission Inventory. Available from: http://www.env.go.jp/air/osen/voc/H26main.pdf (accessed 29.11.15.) (in Japanese). Ohara, T., 2010. Research Report from the National Institute for Environmental Studies. Study on Characteristics of Photochemical Oxidants and Particulate Matter, vol. 203, pp. 1e219 (in Japanese). Pielke, R.A., Cotton, W.R., Walko, R.L., Tremback, C.J., Lyons, W.A., Grasso, L.D., Nicholls, M.E., Moran, M.D., Wesley, D.A., Lee, T.J., Copeland, J.H., 1992.

T. Watanabe et al. / Atmospheric Environment 129 (2016) 176e185 A comprehensive meteorological modeling system RAMS. Meteor. Atmos. Phys. 49, 69e91. Pochanart, P., Akimoto, H., Kinjo, Y., Tanimoto, H., 2002. Surface ozone at four remote island sites and the preliminary assessment of the exceedances of its critical level in Japan. Atmos. Environ. 36, 4235e4250. Saito, M., Wakamatsu, S., Aihara, K., 2013. Estimation of accumulated phytotoxic ozone dose in the Tanzawa Mountains. J. Jpn. Soc. Atmos. Environ. 48, 252e259 (in Japanese with English summary). Simpson, D., Ashmore, M.R., Emberson, L., Tuovinen, J.-P., 2007. A comparison of two different approaches for mapping potential ozone damage to vegetation. A model study. Environ. Pollut. 146, 715e725. Sitch, S., Cox, P.M., Collins, W.J., Huntingford, C., 2007. Indirect radiative forcing of climate change through ozone effects on the land-carbon sink. Nature 448, 791e794.

185

UNECE, 2010. Manual on Methodologies and Criteria for Modelling and Mapping Critical Loads & Levels and Air Pollution Effects, Risks and Trends, Chapter 3: Mapping Critical Levels for Vegetation, pp. 1e115. U. S. EPA, 2007. Guidance on the Use of Models and Other Analyses for Demonstrating Attainment of Air Quality Goals for Ozone, PM2.5, and Regional Haze. EPA-454/B-07e002. Walko, R.L., Cotton, W.R., Meyers, M.P., Harrington, J.Y., 1995. New RAMS cloud microphysics parameterization. Part I: the single-moment scheme. Atmos. Res. 38, 29e62. Watanabe, M., Yamaguchi, M., 2011. Risk assessment of ozone impact on 6 Japanese forest tree species with consideration of nitrogen deposition. Jpn. J. Ecol. 61, 89e96 (in Japanese). Zhang, L., Brook, J.R., Vet, R., 2003. A revised parameterization for gaseous dry deposition in air-quality models. Atmos. Chem. Phys. 3, 2067e2082.