The impact of global and regional air pollution on climate and the environment is a new focus of atmospheric science. Sulfur dioxide (SO2) and nitrogen oxides (NOx = NO + NO2), mainly emitted from fuel combustion, can be transformed in the atmosphere to aerosol-bound sulfate and nitrate. Aerosols, which have lifetimes of approximately one to two weeks, can be transported over thousands of kilometers resulting in transboundary pollution problems. Nitrogen oxides play a key part in the photochemically induced catalytic production of ozone (O3), which results in summer smog and has increased levels of tropospheric ozone globally. NOx concentrations in many industrialized countries are expected to decrease, but rapid economic development has the potential to increase significantly NOx emissions in parts of Asia. Chemical Transport Models (CTM) are playing increasingly important roles in the studies of long-range transport, chemical transformations and scavenging of trace gases and aerosols in the troposphere. Regional exchange of sulfur compounds in East Asia has been widely studied mainly using simple Lagrangian models during the recent decade. The three-dimensional (3-D) Eulerian models are able to handle more accurately the transport of pollutants in the free troposphere above the boundary layer as contrast to the simple Lagrangian models. However, regional exchange of reactive nitrogen (NOy = NO + NO2 + NO3 + 2N2O5 + HONO + HNO4 + HNO3 + aerosol nitrate + PAN(s) + other organic nitrates) in East Asia have not yet until now, to date, been identified quantitatively by a complex 3-D Eulerian model.

This study has aimed at establishing region-to-grid source-receptor relationships (SRRs) for sulfur and reactive nitrogen deposition in East Asia to support possible integrated assessments of transboundary pollution. A multi-scale long-range transport modeling system has been set up, developed, and evaluated in East Asia for this purpose. The state-of-the-science 3-D Eulerian-type Community Multiscale Air Quality model (CMAQ) is applied in East Asia with horizontal resolutions of 27-km and 81-km. The meteorological fields for the two domains are generated by Mesoscale Model version 5 (MM5) using the NCEP/NCAR global meteorological reanalysis datasets in 2001 for initial and boundary conditions. An emission processing model is newly developed to import and transform various inventory data by seasonal allocation, chemical speciation, vertical assignment (only for large point sources), and horizontal interpolation. To suppress the pronounced inaccuracy of time-invariant lateral-fixed boundary conditions, the seasonal outputs of the global Model for Ozone and Related Chemical Tracers (MOZART) are processed by a newly developed global/regional interface to derive the boundary conditions for the 81-km grid domain of CMAQ. The MOZART-derived boundary conditions are able to account for the global influence to regional air quality in East Asia caused by the long-lived pollutants such as ozone, carbon monoxide, peroxyacetyl nitrate (PAN), and aerosols.

In order to characterize model performance and to build confidence in its use, observational data used in this study include ground-based monitoring from the Acid Deposition Monitoring Network in East Asia (EANET) and satellite measurements from the Global Ozone Monitoring Experiment (GOME) spectrometer. A variety of results and analysis are presented to evaluate how well the current emission estimates and the multi-scale modeling system can represent the observed features of gases and aerosols over a wide range of meteorological conditions and geographical areas.

Firstly, the spatial and temporal variations (from seasonal to diurnal) of ambient concentrations of primary gases SO2 and NOx are analyzed and evaluated. The model successfully reproduces the magnitudes, daily, and diurnal variations of SO2 mixing ratios at most of the EANET sites especially those with 27-km grid spacing. Main uncertainty of SO2 predictions is caused by the representation of model topography and sub-grid variation of emissions in urban areas, which may not be resolved in the coarse 81-km grid resolution. The magnitudes of NOx surface concentrations at the EANET sites in central and southern Japan are slightly underpredicted. Through comparison with the GOME retrieval, the model is shown to be able to capture major spatial and seasonal variations of tropospheric NO2 column abundance in East Asia. Regarding the magnitudes of NO2 columns, however, the model largely (by a factor of 2) underpredicts GOME retrieval over the industrialized area of Central Eastern China in winter, suggesting that anthropogenic NOx emissions over this region, especially domestic heating sources, are likely underestimated in the current emission inventory. In addition, the model also underpredicts GOME retrieval over the remote western China and Mongolia in summer. This is likely caused by the underestimation of soil-biogenic NO emission from the grassland/scrubland/desert over these regions during the wet season.

Secondly, the model's ability to reproduce the photochemical production of O3 is evaluated by the analysis of photochemistry sensitivity with the CBIV and SAPRC99 mechanism. Analysis of seasonal cycle of surface O3 suggests that the use of MOZART-derived boundary conditions improves the O3 prediction during the winter season when the inflow from Europe across Eurasia is significant. The exchange between continental polluted air mass and clean marine air mass results in the summer minimum of ozone over the south of 30°N in the study region. The summer-time O3 production in central eastern China and central Japan is highly sensitive to the chemical mechanisms applied in CMAQ. Both 27-km and 81-km simulations of the CBIV mechanism well reproduce the observed seasonal cycle of surface ozone at most sites. Although the SAPRC99 mechanism provides a more-detailed representation of hydrocarbon classification, the 81-km simulation largely overpredicts the observed summer-time O3 especially on low O3 days near central Japan.

Thirdly, precipitation and the chemical components in rainwater are analyzed. The model generally reproduces ionic concentrations in rainwater and annual wet deposition loads of sulfate and ammonium. Except the uncertainty in NOx emissions, CMAQ has a tendency to underestimate the production of nitrate aerosol. More integrated observation of reactive nitrogen gases and aerosols are needed to evaluate the thermodynamic aerosol module of CMAQ.

Finally, the complex 3-D Eulerian model CMAQ and related components newly developed in this study are applied, for the first time, to quantify region-to-grid source-receptor relationships (SRRs) both for sulfur and reactive nitrogen in East Asia. Investigation of model responses to emission changes suggest that reductions of ammonia emissions have significant nonlinear effects on the deposition of sulfur and reactive nitrogen as a result of non-linearity in the atmospheric chemistries of sulfate-nitrate-ammonia system and the deposition process. A source region attribution methodology is proposed based on the analysis of non-linear effects. Sensitivity simulations were conducted where emissions of SO2, NOx, and its primary particles from a specific source region were reduced by 25%. The difference between the base and sensitivity simulations was multiplied by a factor of four, and then defined as the contribution from that source region. Source attributions are carried out for eighteen designated source regions as well as international shipping and volcanoes within the study domain.

Region-to-region quantitative SRRs provide an important connection between emissions and fates of pollutants over different time and spatial scales. The transboundary influences exhibit strong seasonal variation and are generally maximum during the dry seasons reflecting the Asian monsoon circulation. Long-range transport from central eastern China contributes a significant percentage (>20%) of anthropogenic sulfur deposition as well as reactive nitrogen deposition throughout Asia. At the same time, northwestern China receives approximately 35% of sulfur load and 45% of nitrogen load due to foreign emissions mainly from the Indian subcontinent. Volcanic sources including emissions from Miyakejima volcano in 2001 contribute approximately 50% of total sulfur deposition in Japan. Sulfur inflows from regions outside the study domain, which is attributed from the MOZART-derived boundary conditions, are pronounced (10~40%) over different parts of Asia. Compared with previous studies using simple Lagrangian models, the results derived in this study show more effects from long-range transport. The relationships are more realistic including the global influences and internal interactions among different parts of China. Region-to-region source-receptor relationships of acid deposition provide quantitative information for environmental negotiation, cost allocation, and possible international initiatives to mitigate global air pollution and climate change.

主に燃料の燃焼によって生じる二酸化硫黄や窒素酸化物は、大気中に放出されてエアロゾルとなり、約1~2週間の浮遊期間中に数千kmも輸送され、越境大気汚染の現況となっている。窒素酸化物は対流圏オゾン生成の触媒となり、対流圏オゾン濃度を増加させている。工業化が進んだ国々では窒素酸化物濃度は減少する傾向にあるが、今後発展が期待されるアジアの途上国では窒素酸化物の放出量が増加する恐れがある。

「Long-range Transport and Source-receptor Relationships of Acidifying Substances in East Asia」(東アジアにおける酸性化物質の長距離輸送と排出─受容関係)と題する本研究は、越境大気汚染の総合的なアセスメントを可能とするために、東アジア、東南アジアの各地域から排出される二酸化硫黄、反応性の窒素酸化物が、どの場所に乾性沈着、湿性沈着として受容されるのかを推定する手法を確立したものである。

東アジアにおける硫黄酸化物の越境輸送に関してはこれまでにもラグランジェ型の輸送モデルを用いて研究されてきているが、より正確に汚染物質の輸送を扱うことができるオイラー型の3次元輸送モデルを用いて、反応性の窒素酸化物の輸送を取り扱った研究はこれまでなされていなかった。本研究では最新のオイラー型の3次元大気化学輸送モデルであるCMAQを東アジア、東南アジアの領域に空間解像度27km、81kmで適用している。大気場に関してはNCEP/NCARのグローバル再解析データを初期値、ならびに境界条件としてメソスケール領域大気モデルMM5を用いて2001年の1年分が計算されている。

二酸化硫黄、窒素酸化物の大気への排出量については様々なインベントリーデータに基づいて季節変化や大きな点源については鉛直分布も含めて新たに推定され、格子点化して用いられた。大気化学反応輸送モデルの側方境界条件は、global Model for Ozone and Related Chemical Tracers (MOZART)の算定値の空間解像度、化学物質分類などをCMAQ用に変換して与えられている。これにより、オゾンや一酸化炭素、peroxyacetyl nitrate (PAN)など大気中での寿命が長く長距離輸送される物質の対象領域外からの影響を考慮することが可能となっている。このように境界条件を与えることにより、特に冬季においてモデルのオゾン濃度の推定精度を向上させることが確認されている。また、夏のオゾン濃度に関しては、光化学反応過程に鋭敏で、CBIVとSAPRC99の2つの光化学反応過程モデルが試され、SAPRC99の方が揮発性有機化合物の分類は多いもののCBIVの方が概ね観測をよく再現することが検証されている。

さらに、こうして構築された東アジア、東南アジアの越境大気化学物質輸送モデルの推定結果は、Acid Deposition Monitoring Network in East Asia (EANET)ならびにGlobal Ozone Monitoring Experiment (GOME)センサによる観測データと比較されている。それによると、モデルの推定結果は、EANETの地点観測結果と比べて、季節変動、日々変動、日周変動をよく捕らえているが、モデルの空間解像度が81kmの場合、地形が十分に表現されないため、推定精度が落ちることが示されている。窒素酸化物に関しては日本中南部でモデルがやや過小評価しているものの、衛星観測と比較すると、対象領域内の鉛直積分された窒素酸化物濃度の空間分布はよく捕らえられている。窒素酸化物濃度の過小評価はGOMEと比較して冬の中国東部の工業化地帯でも観察され、インベントリーデータにおいて、家庭からの暖房に伴う窒素酸化物の排出量がやや低めなのではないか、ということが示唆されている。夏の中国西部やモンゴルにおける過小評価に関しては、土壌からの生化学的な窒素酸化物の排出を過小評価しているためではないかと推察されている。

降水中の濃度に関しては、硫酸ならびにアンモニアの濃度は概ね適切に再現されているものの、硝酸濃度に関しては排出量の不確実性を除いてもやや過小評価する傾向にあり、CMAQの化学反応過程を評価し改良するためには、より包括的な反応性窒素酸化物とエアロゾルの観測が必要であることが指摘されている。

このように精度が検証された越境大気化学物質輸送モデルを用いて東アジアにおける汚染物質の排出─受容関係が推定された。様々な感度分析の結果、アンモニア排出量を減らすと大気化学過程の非線形効果によって硫黄酸化物や反応性窒素酸化物の沈着量も大きく変化してしまうことがわかった。こうした知見に基づき、二酸化硫黄や反応性窒素酸化物の排出量を特定地域についてのみ25%削減するといった摂動を与えてコントロール実験との沈着量の差をとり、排出─受容関係が月ごとに推定された。

中国東部からの排出量はアジア全体における人為起源硫黄沈着のかなりの量を占める一方で、中国北部では硫黄沈着、窒素沈着のかなりの部分をインド亜大陸など国外の排出から受容していること、三宅島の火山の影響が日本における硫黄沈着の半分を占めること、アジアモンスーンの季節変化に従い、乾季に長距離輸送が顕著であることなどが明快に示されている。

このように、本研究は、最新のオイラー型3次元大気化学輸送モデルと新たに構築した排出量データセットを用いて東アジア地域の排出─受容関係を推定する枠組みを構築したもので、地上観測、衛星観測を用いた丁寧な精度の検証もなされており、定量的な信頼性も高く、有用性に富む研究成果と評価できる。よって本論文は博士(工学)の学位請求論文として合格と認められる。