, municipal level ⟨M⟩ and grid-cell level ⟨G⟩. The national-level (nat = j) emission E⟨N⟩ src=i,nat=j in emission source category src = i is allocated to the prefectural level to obtain emissions E⟨P⟩ src=i,pref=k at this level (pref ⟨M⟩ = k). These are next resolved into emissions Esrc=i,mun=l at the municipal level (mun = l), which are then allocated further to ⟨G⟩ individual grid cells, yielding emissions Esrc=i,grid=m for source category src = i in grid (cell) = m. Although in this study the number of national levels is one (Japan), that is, nat = 1, the subscript nat was introduced for generality. In Method 2, appropriate statistics and data for estimating the emission from a given source category are unavailable at the municipal level, so the emission in question is allocated directly from the prefectural to the grid-cell level. In Method 3, statistics and data for estimating the emission from a given source category are unavailable at the prefectural level, so the national-level emission is allocated directly to the municipal level, from where it is allocated further to the gridcell level. Method 4, finally, is used in cases where statistics and data for estimating the emission from a given source category are lacking at both the prefectural and municipal level, but where statistics and data are available for estimating the emission in question at the grid-cell level. With this method the nationallevel emission is thus allocated directly to specific grid cells. This stepwise allocation of national emissions was performed using spatial allocation factors to break down the total emission from a given area into emissions from component areas of higher geographical resolution. The spatial allocation factors in this paper take the form of a ratio between the emissions in the ⟨N⟩ former and latter area. For example, in Method 1, Esrc=i,nat=j is multiplied by the spatial allocation factor f⟨N→P⟩ src=i,nat=j,pref=k to allocate national emissions to the geographical attribute pref = k. Superscript ⟨N → P⟩ in f⟨N→P⟩ src=i,nat=j,pref=k signifies allocation from the nation to the geographical attribute prefecture
and gives the proportion of E⟨N⟩ src=i,nat=j to be allocated to ⟨M⟩ ⟨P⟩ prefecture pref = k. Similarly, Esrc=i,mun=l = Esrc=i,pref=k × ⟨P→M⟩ fsrc=i,pref=k,mun=l defines the emissions from a given prefecture ⟨G⟩ to be allocated to municipality mun = l, and Esrc=i,grid=m = ⟨M→G⟩ E⟨M⟩ × f gives the share of these emissions to src=i,mun=l src=i,mun=l,grid=m be allocated to the geographical attribute grid (cell) = m. These spatial allocation factors for each emission source category were prepared according to the allocation method applied for each category. The spatial allocation factor for the emission in source category i for the geographical attribute pref = k, f⟨N→P⟩ src=i,nat=j,pref=k, was calculated by using an activity volume S⟨N→P⟩ src=i,nat=j,pref=k to stand as a proxy for the emission in question: for example, the volume output of products like cement, the volume consumption of fuels like heavy oil and coal, or the volume of final waste disposal. This proxy was normalized to the sum
1300 tonnes annually, or between 46% and 67% of total global emissions. Japan’s emissions were estimated at 143 tonnes in 2000 by Pacyna et al.,6 by applying emission intensities outside Japan to emission sources in Japan. Kida and Takahashi,12 however, argued that the emission intensities of waste incineration and other sources in Japan are substantially lower than those of other countries. Based on actual measurement of emission factors in Japan, they arrived at an estimate of atmospheric emissions in 2002 of 20−24 tonnes. It was on the basis of these latter research results that the Japanese Environment Ministry reported the atmospheric emissions for 2005 that were used in the UNEP report.11 At present, however, Japanese mercury emission inventories12−14 merely provide estimates per emission source category, with no attempt yet having been made to provide any geographical resolution. Because the locations of industrial facilities regarded as emission sources are distributed unevenly in geographic terms, there are presumably also significant regional differences in the mercury emissions themselves. If mercury emission levels and characteristics (type of emission source, point versus diffuse) could be quantified for individual prefectures and municipalities, suitable emission control measures could be designed for each region and regions identified where mercury monitoring should be prioritized. Furthermore, emission inventories of higher geographical resolution could be used as input data for an environmental fate model,15,16 essential for assessing exposure and risk. Particularly for more precise assessments, including those at a local level around specific emission sources, there is a pressing need to identify the distribution of emissions based on the highest geographical resolution possible. The aim of this study, then, is to estimate the distribution of Japanese mercury emissions by prefecture, by municipality and by grid cell and to analyze their characteristics.
2. MATERIALS AND METHODS 2.1. Stepwise Spatial Allocation of National-Level Mercury Emissions. In this study, a stepwise procedure was used to allocate annual Japanese mercury emissions by source category with increasing spatial resolution.17 A general tendency of Japanese statistics by prefecture, by municipality and by grid cell is that the greater the regional resolution, the fewer types of statistics are available and the less detailed the information is on attributes of interest. For example, at the national level statistics are available on the production output of numerous industrial categories, but as resolution increases down to prefectural, municipal and grid-cell levels, these categories are aggregated, with definitions becoming ever more general. To estimate the spatial distribution of mercury emissions in Japan with greater accuracy and at higher resolution, we therefore used a geographic information system (GIS) and a stepwise allocation procedure using the statistics compiled by prefectural and municipal governments and statistics available at grid-cell level. Depending on data availability, one of the four allocation methods described in the next section was applied to estimate emissions per source category for 47 prefectures, 1840 municipalities and grid cells measuring approximately 1 × 1 km and covering the whole of Japan. 2.2. Four Methods of Spatial Allocation. To improve the geographical resolution of national-level emissions in stepwise fashion, one of the four following methods was 4934
dx.doi.org/10.1021/es203687r | Environ. Sci. Technol. 2012, 46, 4933−4940
Environmental Science & Technology
Article
total of the activity volumes of all the prefectures making up nat = j to yield f⟨N→P⟩ src=i,nat=j,pref=k. The details and benefits of stepwise spatial allocation of emissions and of the four allocation methods are described in the Supporting Information (SI). 2.3. National Mercury Emissions Inventory. This study used the existing national mercury emissions inventory for the year 200513 for national-level emissions by source category, ⟨N⟩ Esrc=i,nat=j . This inventory is an update of the estimate for the year 2002 in Kida and Takahashi12 obtained by updating the activity volume of each source category to the year 2005, using the same emission factors as those for 2002 in Kida and Takahashi.12 The emission in each source category was calculated by multiplying the activity volume of that category by an emission factor in which the mercury emission removal rate by exhaust gas treatment had been factored in and based, further, on current research papers, calculations based on site surveys and spot measurements, and estimates of mercury levels in products and raw materials. In this way it was endeavored to provide an as realistic picture as possible of the actual situation. Although these estimates are based on simple assumptions or assume that the mercury emission is directly proportional to the activity volume, the national-level mercury emissions inventory by source category is currently the most up-to-date available for use in Japan. To enhance the quality of spatial allocation, in this study the emission source categories in the nonferrous metal and transportation sectors defined in the national inventory13 were further subdivided into four subcategories (zinc, copper, lead, and nickel production) and five subcategories (transportation burning gasoline, jet fuel, kerosene, diesel and heavy oil). Meanwhile, emissions attributable to natural sources (volcanoes) (1400 kg) and the chlor-alkali industry (with zero emissions in the national inventory13) were excluded from the estimate of spatial distribution. The study also excluded the category of collection and crushing of fluorescent lights, mercury emissions from which were estimated to be extremely low (between 0.00723 and 0.00903 kg). Transportation using heavy oil (237 kg; likely to be mainly for shipping) was also excluded because of the difficulties inherent in developing spatial allocation factors for shipping on varying routes. In the national inventory, emissions from certain source categories are estimated as a range, and in such cases spatial allocation was therefore performed using both the maximum and the minimum value reported there. For source categories reported without a range, the maximum and minimum were assumed to be the same. Table 1 specifies the emissions in each of the source categories considered in this study and which spatial allocation method was used in each case. 2.4. Definition of Average Mercury Emissions. As stated, the national-level emissions inventory13 reports maximum and minimum estimates of annual emissions for some source categories. We therefore made two calculations of grid-level emissions by source category, one on the assumption that all sources take the maximum value of the annual emission, the other on the assumption that all sources take the minimum. Because the aggregate emission in a grid cell is given by the sum of emissions from all the sources in that cell, the maximum emission in a cell naturally occurs when all the sources it comprises emit the maximum amount. Conversely, the minimum emission occurs when all the sources emit the minimum. In other words, estimated grid-cell emissions have a
Table 1. Annual Japanese Mercury Emissions in 2005 Per Source Category Used to Estimate Geographical Distribution (Source: ref 13) annual mercury emission in 2005 source code S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 S13 S14 S15 S16 S17 S18 S19 S20 S21 S22 S23 S24 S25 S26 S27 S28 S29 S30 S31 S32 S33 S34 S35 S36 S37 Total Total
name of emission source thermal power generation (coal) industrial boilers (coal) thermal power generation (petroleum) industrial boilers (petroleum) incineration of general waste incineration of medical waste incineration and dissolution of sewage sludge industrial waste (waste plastics) industrial waste (wastepaper) industrial waste (waste wood) industrial waste (waste textile) industrial waste (waste rubber) industrial waste (other sludge) industrial waste (shredder dust) steel and iron manufacturing nonferrous metal (zinc) nonferrous metal (copper) nonferrous metal (lead) nonferrous metal (nickel) cement manufacturing quicklime and slaked lime manufacturing carbon black manufacturing coke manufacturing pulp and paper manufacturing chlor-alkali industry battery manufacturing electric switch manufacturing fluorescent light manufacturing cremation collection and crushing of fluorescent lights dental service (amalgam) transportation (gasoline) transportation (jet fuel) transportation (kerosene) transportation (diesel) transportation (heavy oil) volcanoes excluding S30, S36 and S37
minimum [kg/y]
maximum [kg/y]
allocation method
1229
1229
method 4
569 299
569 299
method 2 method 4
1050
1050
method 2
98 570
236 1680
method 3 method 1
258
1480
method 1
17
657
method 1
5.5
5.5
method 1
13
116
method 1
3.3
11
method 1
0.021
1.9
method 1
661
661
method 1
49
793
method 1
3260 505 12 2.5 0.70 8940 1060
3260 3362 79 1164 4.7 8940 1060
method method method method method method method
121 886 427
121 886 652
method 4 method 4 method 1
0 1.8 4.3
0 1.8 4.3
excluded method 2 method 2
18
18
method 2
56 0.00723
56 0.00903
method 3 excluded
3 175 0 0 361 237 1400 21063 19426
3 175 0 0 361 237 1400 29344 27707
method 2 method 2 excluded excluded method 2 excluded excluded
2 4 4 4 4 4 1
range bounded by the sum total of the maxima and the sum total of the minima from the sources within the cell in question. 4935
dx.doi.org/10.1021/es203687r | Environ. Sci. Technol. 2012, 46, 4933−4940
Environmental Science & Technology
Article
Figure 1. Estimated annual Japanese mercury emissions in 2005 in 47 prefectures. Label represents PrefCode (pref = 1...47).
calculated for each cell. For example, G80 grid cells comprise 6400 parcels of approximately 1 × 1 km (G1), whereas G10 cells comprise 100 G1 parcels, and G5 25 parcels. The average value of the emission intensity of a G80 grid cell was termed ⟨G80⟩ ⟨G80⟩ Igrid=m⟨G80⟩ and the standard deviation SDgrid=m⟨G80⟩ = 2 6400 ⟨G80⟩ ⟨G1⟩ (∑m⟨G80⟩ ∈ m⟨G1⟩(Igrid=m⟨G80⟩ − Igrid=m⟨G1⟩) /6400)1/2 for the emission intensity I⟨G1⟩ grid=m⟨G1⟩ of a G1 grid cell was obtained. This was divided by the emission intensity to calculate the variation ⟨G80⟩ ⟨G80⟩ coefficient vc⟨G80⟩ grid=m⟨G80⟩ = SDgrid=m⟨G80⟩/Igrid=m⟨G80⟩ × 100. In similar fashion, the variation coefficient vc⟨G10⟩ grid=m⟨G10⟩ of G10 cells and the variation coefficient vc⟨G5⟩ grid=m⟨G5⟩ of G5 cells were obtained. 2.6. Specifications of the Emissions Inventory Database. The Japanese grid-based mercury emissions inventory for 2005 developed in this study is attached in the SI in electronic form. These data, in CSV (comma separated value) format, report annual Japanese mercury emissions in 2005 per grid cell (approximately 1 × 1 km) and per source category and allow users to aggregate emissions by prefecture or by municipality as they see fit. Detailed specifications of the data format are provided in the SI.
To obtain a straightforward definition of the annual average emission per grid cell, this study used Monte Carlo simulation with pseudorandom numbers uniformly distributed between the maximum and minimum emissions in source category i within the cell. To this end, one hundred thousand samples of emissions in source category i within the cell were generated, and the average emission E̅s⟨G⟩ rc=i,grid=m in each category and its ⟨G⟩ were then determined from the standard deviation σsrc=i,grid=m simple average of the sample obtained. These average emissions in each category were then added to arrive at the average ⟨G⟩ emission for the grid cell E̅⟨G⟩ grid=m = ∑srcE̅ src,grid=m. The standard of the cell was calculated from σ⟨G⟩ deviation σ⟨G⟩ grid=m grid=m = ⟨G⟩ ⟨G⟩ ⟨G⟩ 2 1/2 (∑src((E̅src,grid=m/∑srcE̅src,grid=m)σsrc,grid=m) ) by combining the emissions in the respective source categories on the assumption of noncorrelation among them. 2.5. Effect of Grid Cell Size on Emission Intensity of the Cell. In this study, emissions were estimated for grid cells measuring approximately 1 × 1 km. Emissions in grid cells of 5 × 5 km, say, can therefore be derived by consolidating 25 of these smaller cells. However, use of such larger cells for estimating the distribution of mercury emissions means losing the detailed characteristics available in the smaller cells, leading to marked uncertainty in the spatial distribution of the emission in question, which is now assumed to be uniformly distributed throughout the larger cell. To assess the benefits of the higher-resolution emission inventory, the emissions from grid cells of approximately 1 × 1 km were consolidated to cells of approximately 80 × 80 km (represented by G80), approximately 10 × 10 km (represented by G10) and approximately 5 × 5 km (represented by G5) according to the codes given in the Standard Regional Grid and Mesh Codes for Statistical Use in Japan.18 The emission intensity (g/km2), that is, the annual emission divided by area, was then
3. RESULTS AND DISCUSSION 3.1. Prefectural Distribution of Mercury Emissions in Japan. Figure 1 presents the mercury emissions of each Japanese prefecture calculated by aggregating estimated emissions in 1 × 1 km grid cells. The highest emission, found in Fukuoka (pref = 40), was estimated as 2255 kg/y, representing 9.1% of total Japanese emissions. The second largest was for Yamaguchi (pref = 35) with 2201 kg/y (8.9%), followed by Hyogo (pref = 28) with 1578 kg/y (6.4%), Oita (pref = 44) with 1372 kg/y (5.5%) and Hokkaido (pref = 1) with 1329 kg/y (5.4%). Together, these top five prefectures 4936
dx.doi.org/10.1021/es203687r | Environ. Sci. Technol. 2012, 46, 4933−4940
Environmental Science & Technology
Article
Figure 2. Estimated annual mercury Japanese emissions in 2005 in 1840 municipalities.
Figure 3. Estimated annual mercury Japanese emissions in 2005 in approximately 1 × 1 km grid cells.
account for 35.2% of all national emissions. These are followed by Gifu (pref = 21) with 1261 kg/y (5.1%), Hiroshima (pref =
34) with 1241 kg/y (5.0%), Chiba (pref = 12) with 1043 kg/y (4.2%), Saitama (pref = 11) with 964 kg/y (3.9%), and Aichi 4937
dx.doi.org/10.1021/es203687r | Environ. Sci. Technol. 2012, 46, 4933−4940
Environmental Science & Technology
Article
In Figure S3 in the SI, the horizontal axis shows the cumulative number of grid cells with emissions, arranged in decreasing order of emission magnitude, while the vertical axis shows the share of the cumulative emissions of these grid cells in total national emissions. As can be seen, 50% of all emissions derive from a mere 32 of the 255 954 cells to which emissions were allocated in this study, while the accumulated emissions from the 2000 grid cells with the highest emissions accounts for 92.6% of all emissions. Given this geographical distribution of mercury emissions, to improve the accuracy of data on Japan’s emissions inventory with high geographical resolution, it makes more sense to examine individual facilities with high emissions than to collect data from across a vast geographical area. This can be demonstrated as follows. In Figure 4, the vertical axis depicts the variation coefficient (%) of G80 (a), G10 (b) and G5 (c), while the horizontal axis represents the annual emission intensity per grid cell (g/km2). In (a), the variation coefficient can be seen to have a maximum
(pref = 23) with 855 kg/y (3.4%), which together make up the top 10 prefectures, accounting for 56.9% of national emissions. Figure S1 in the SI provides a breakdown of the source categories (src = S1−S35 in Table 1) contributing most to emissions by considering the five prefectures with the highest emissions. Only sources contributing 2% or more to total emissions are shown. The contribution of cement manufacturing (S20) is large in all prefectures, accounting for 74% of emissions in Fukuoka (pref = 40) and 76% in Yamaguchi (pref = 35), in particular. Although Oita (pref = 44) and Hokkaido (pref = 1) show similar relative contributions by cement manufacturingapproximately half the total: 56% and 48%, respectivelythe shares of the remaining source categories vary depending on the region. Major emission sources in Oita (pref = 44) are technologies associated with iron and steel production, including steel and iron manufacturing (S15), with 26%, and coke manufacturing (S23), with 7%. In contrast, Hokkaido (pref = 1) includes pulp and paper manufacturing (S24), with 11%, steel and iron manufacturing, with 7%, coalfired power generation (S1), with 5%, medical waste incineration (S6), with 5%, and many other sources with contributions of 2% or more. Hyogo (pref = 28), with the third highest emissions, appears to be characterized by a wide range of emission sources. This quantitative expression of differences in the causes of emissions among prefectures facilitates identification of priority emission sources for mercury control policies in the respective regions. 3.2. Municipal Distribution of Mercury Emissions in Japan. Figure 2 presents the emissions of each Japanese municipality calculated by aggregating the estimated grid-cell emissions. Comparison with Figure 1 reveals that even in prefectures characterized by relatively high emissions, these are by no means evenly spread across the region, but are found to be concentrated in certain municipalities. Figure S2 in the SI provides a breakdown of emissions by municipality in each of the five prefectures with the highest emissions (with the municipality code attached to those accounting for 2% or more). The codes follow the Standard Codes for Areas of Prefectures and Municipalities for Statistical Use in Japan.19 In Fukuoka (pref = 40), with the highest emissions, emissions are concentrated above all in three municipalities: Kanda, Miyako District (mun = 40 621) with 62%, Kitakyushu (mun = 40 100) with 22% and Tagawa (mun = 40 206) with 9% of total emissions. In Yamaguchi (pref = 35), emissions are concentrated in four municipalities: Shunan (mun = 35 215) with 41%, Mine (mun = 35 213) with 29%, Shimonoseki (mun = 35 213) with 12% and Ube (mun = 35 202) with 11%. Similar tendencies are also evident in Hyogo (pref = 28), Oita (pref = 44) and Hokkaido (pref = 1). 3.3. Grid-Cell Distribution of Mercury Emissions in Japan. Figure 3 presents the estimated emissions on an approximately 1 km ×1 km grid, defined according to the Standard Regional Grid and Mesh Codes for Statistical Use in Japan.18 Comparison with the emissions by municipality in Figure 2 reveals further local concentrations of mercury emissions. Although grid cells with high emissions are observed near highly populated areas like Tokyo (pref = 13), Aichi (pref = 23), and Osaka (pref = 27), their contributions to nationwide emissions are relatively small: respectively 445 kg/y (1.8%), 855 kg/y (3.4%) and 316 kg/y (1.3%). This fact suggests that the scattering of grid cells with locally high emissions determines the regions with the highest emissions shown in Figures 1 and 2.
Figure 4. Coefficient of variation of annual mercury emission intensity of grid cells larger than in the G1 grid (approximately 1 × 1 km): (a) G80 grid (approximately 80 × 80 km), (b) G10 grid (approximately 10 × 10 km) and (c) G5 grid (approximately 5 × 5 km). 4938
dx.doi.org/10.1021/es203687r | Environ. Sci. Technol. 2012, 46, 4933−4940
Environmental Science & Technology
Article
with respect to sewage sludge treatment by outsourcing in the former municipalities and take disposal capacities as the “activity volume” of the incineration facilities in the latter, this would be extremely labor, time and cost-intensive. The same approach was also adopted for the combustion of medical waste (src = 6) and industrial wastes (src = 8−14). While the accuracy of the spatial allocation factors used in this paper for these emission sources could certainly be improved by considering the specific activity volume of each facility, such data collection is likely to require substantial efforts. Of the emissions attributable to transportation, those due to fuel oil probably derive mainly from shipping. To determine spatial allocation factors for shipping emissions would require identification of all shipping routes and their activity volumes as a proxy for emissions. Given the limited availability of shipping statistics in Japan, however, this would be extremely laborious. In the present study this emission source was therefore excluded from the estimate of mercury distribution. 3.4.3. Cases in Which Distinction of Activity Volumes Is Difficult. In Japan, national consumption of amalgam has been declining every year since use of the substance began to be questioned, and today it is rarely used at dental clinics. Because of unavailability of information to help identify which dental clinics still use amalgam, this study allocated emissions based on the number of employees at dental clinics published in the Establishment and Enterprise Census (grid-cell statistics) irrespective of their actual use of amalgam. As a result, allocation may have been made to grid cells with no actual mercury emissions from this source. Except for those from transportation, the bulk of Japanese mercury emissions are caused by a relatively small number of specific industries and processes, with actual spatial distribution in all likelihood being limited to a small number of particular locations. Nonetheless, the industry classification of those gridcell statistics that are regularly updated and most accessible for estimating spatial allocation factors to cells are less detailed than in the case of national and prefectural statistics. For this reason, the study fails to reflect the reality of emissions generated only by a particular industry within more roughly classified industries when allocating emissions to individual grid cells. This is because the higher emissions are distributed to the grid cells with the larger activity volumes even if the activity volume of the particular industry in a grid cell is in reality very small and that of other industries dominates the cell instead. There is thus a tendency for estimated emissions to be distributed more widely geographically than they are in reality. To further improve the precision of the results in the future, geographical information (coordinates of manufacturing and incineration facilities, etc.), detailed information related to activity (manufacturing outputs and incineration volumes) and other aspects will need to be thoroughly investigated to identify inadequate categories in the grid-cell statistics.
of 7966%, showing that there is significant heterogeneity in the emission distribution in the grid. The average value of the variation coefficient for all grid cells is 3180%, while the minimum value is 568%. Observations (b) and (c) reveal that the maximum value of the variation coefficient is 996% for G10 and 490% for G5. These maxima occur when the G5 or G10 cell contains only one G1grid cell with emission greater than zero, with no emission in any other grid cell. The maximum thus decreases as the number of G1 grid cells of included is reduced. The average values are 379% and 213%, respectively, whereas the minimum values fall to 74% and 30%. It was thus confirmed that improving the geographical resolution of the grid provides a far better reflection of the characteristics of emission distribution across the grid. 3.4. Limitations Attributable to Data Availability, And Future Work on Inventory Improvement. Using data on mercury emissions per source category and statistics involving geographical information, this study has developed data for a mercury emissions inventory with high geographical resolution. This was accompanied by problems, however, in collecting the data required to establish spatial allocation factors. This section identifies these problems and proposes solutions, suggesting means by which accuracy can be improved in future studies. To establish spatial allocation factors, this study confronted the following three main issues in selecting suitable activity data to act as proxy variables for emission quantities. 3.4.1. Cases in Which Activity Volumes Are Unavailable. For the spatial allocation of mercury emissions from cremation (src = 29), the locations of individual crematoria were identified using the crematorium database of the Ministry of Health, Labor and Welfare (MHLW). However, these statistics do not include the number of people cremated, the number of furnaces, the operating ratio or other information eligible for use as the “activity volume” of individual facilities. This study therefore simply allocated the emissions of municipalities to grid cells based on the number of facilities located in each cell. As the activity volume of each facility was not taken into account, these figures do not accurately reflect the actual emissions in each grid cell. A similar difficulty arises in the spatial allocation of emissions associated with nonferrous metal (nickel) (src = 19), quicklime and slaked lime manufacturing (src = 21) and pulp manufacturing (src = 24). 3.4.2. Cases in Which Collecting Activity Volumes Would Be Very Labor and Time-Consuming. Major Japanese municipalities (“government-ordinance-designated cities”) and particular wards of Tokyo and surrounding areas have established their own incineration facilities to dispose of sewage sludge. In such municipalities, then, treatment of sewage sludge consists of incineration in these facilities as municipal waste or outsourcing to a waste disposal company as industrial waste. However, we assumed that all the sewage sludge in these municipalities was treated in their “own” incineration facilities and we obtained the locations of facilities and their treatment capacities, for use as the “activity volume”, as far as possible from the Web sites of the Bureaus of Sewerage operating the incineration plants. We then calculated spatial allocation factors for the emissions from the incineration and dissolution of sewage sludge (src = 7) from municipalities to grid cells on the basis of plant capacity. For all other municipalities, this study simply used the number of incineration facilities for “sludge” as industrial waste for the spatial allocation factors from municipalities to grid cells. Although it would be possible to investigate the actual situation
■
ASSOCIATED CONTENT
S Supporting Information *
Additional descriptions of methodologies, data and results are included. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*Phone: +81 29-850-2889; fax: +81 29-850-2917; e-mail: [email protected]. 4939
dx.doi.org/10.1021/es203687r | Environ. Sci. Technol. 2012, 46, 4933−4940
Environmental Science & Technology
Article
Notes
(19) MIAC Standard Codes for Areas of Prefectures and Municipalities for Statistical Use in Japan. http://www.stat.go.jp/index/seido/95.htm (accessed October 1, 2011).
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We thank three anonymous reviewers for the helpful comments received, which contributed to improve the quality of this paper. We are also grateful to Nigel Harle of Gronsveld, The Netherlands, for his conscientious revision of our English.
■
REFERENCES
(1) UNEP. The Global Atmospheric Mercury Assessment: Sources, Emissions and Transport; UNEP Chemicals Branch: Geneva, 2008. (2) UNEP. Report of the Governing Council Twenty-Fifth Session (16− 20 February 2009); United Nations: New York, NY, 2009. (3) Nriagu, J. O.; Pacyna, J. M. Quantitative assessment of worldwide contamination of air, water and soils by trace-metals. Nature 1988, 333 (6169), 134−139. (4) Pirrone, N.; Keeler, G. J.; Nriagu, J. O. Regional differences in worldwide emissions of mercury to the atmosphere. Atmos. Environ. 1996, 30 (17), 2981−2987. (5) Pacyna, J. M.; Pacyna, E. G. Global emission of mercury from anthropogenic sources in 1995. Water, Air, Soil, Pollut. 2002, 137 (1− 4), 149−165. (6) Pacyna, J. M.; Pacyna, E. G.; Steenhuisen, F.; Wilson, S. Global anthropogenic mercury emission inventory for 2000. Atmos. Environ. 2006, 40 (22), 4048−4063. (7) Wilson, S. J.; Steenhuisen, F.; Pacyna, J. M.; Pacyna, E. G. Mapping the spatial distribution of global anthropogenic mercury atmospheric emission inventories. Atmos. Environ. 2006, 40 (24), 4621−4632. (8) Pacyna, J. M.; Pacyna, E. G.; Sundseth, K.; Munthe, J.; Kindbom, K.; Wilson, S.; Steenhuisen, F.; Maxson, P. Global emission of mercury to the atmosphere from anthropogenic sources in 2005 and projections to 2020. Atmos. Environ. 2010, 44 (20), 2487−2499. (9) Streets, D. G.; Zhang, Q.; Wu, Y. Projections of Global Mercury Emissions in 2050. Environ. Sci. Technol. 2009, 43 (8), 2983−2988. (10) Streets, D. G.; Devane, M. K.; Lu, Z.; Bond, T. C.; Sunderland, E. M.; Jacob, D. J. All-Time Releases of Mercury to the Atmosphere from Human Activities. Environ. Sci. Technol. 2011, 45 (24), 10485− 10491. (11) AMAP/UNEP. Technical Background Report to the Global Atmospheric Mercury Assessment; Arctic Monitoring and Assessment Programme; UNEP Chemicals Branch, 2008. (12) Kida, A.; Takahashi, F. Japanese and global mercury emission inventories. Global Environ. Res. 2008, 13 (2), 167−180. (13) MOE. Report of the Ministry of Environment Committee on Measures for Hazardous Metals; The Ministry of Environment, Japan: Tokyo, Japan, 2008. (14) Kida, A.; Sakai, S. Preliminary estimation of mercury emission inventories for Japan’s air. Waste Manage. Res. 2005, 16, 191−203. (15) Suzuki, N.; Murasawa, K.; Sakurai, T.; Nansai, K.; Matsuhashi, K.; Moriguchi, Y.; Tanabe, K.; Nakasugi, O.; Morita, M. Georeferenced multimedia environmental fate model (G-CIEMS): Model formulation and comparison to the generic model and monitoring approaches. Environ. Sci. Technol. 2004, 38 (21), 5682−5693. (16) Scheringer, M.; Hollander, A.; Shatalov, V.; Mantseva, E.; Sweetman, A.; Roemer, M.; Baart, A.; Suzuki, N.; Wegmann, F.; van de Meent, D. Estimating overall persistence and long-range transport potential of persistent organic pollutants: a comparison of seven multimedia mass balance models and atmospheric transport models. J Environ. Monit. 2008, 10 (10), 1139−1147. (17) Nansai, K.; Suzuki, N.; Tanabe, K.; Kobayashi, S.; Moriguchi, Y. In Design of Georeference-Based Emission Activity Modeling System (G-BEAMS) for Japanese Emission Inventory Management, 13th International Emission Inventory Conference, Florida, 2004. (18) MIAC Standard Regional Grid and Mesh Codes for Statistical Use in Japan. http://www.stat.go.jp/data/mesh/gaiyou.htm (accessed October 1, 2011). 4940
dx.doi.org/10.1021/es203687r | Environ. Sci. Technol. 2012, 46, 4933−4940
