Sulfur deposition simulations over China, Japan, and Korea ...20–25°N, 100–150°E, as agreed at...

RESEARCH ARTICLE

Sulfur deposition simulations over China, Japan,and Korea: a model intercomparison study for abatingsulfur emission

Cheol-Hee Kim & Lim-Seok Chang & Fan Meng &

Mizuo Kajino & Hiromasa Ueda & Yuanhang Zhang &

Hye-Young Son & Jong-Jae Lee & Youjiang He & Jun Xu &

Keiichi Sato & Tatsuya Sakurai & Zhiwei Han & Lei Duan &

Jeong-Soo Kim & Suk-Jo Lee & Chang-Keun Song &

Soo-Jin Ban & Shang-Gyoo Shim & Young Sunwoo &

Tae-Young Lee

Received: 5 January 2012 /Accepted: 5 July 2012 /Published online: 7 August 2012# Springer-Verlag 2012

Abstract In response to increasing trends in sulfur depositionin Northeast Asia, three countries in the region (China, Japan,and Korea) agreed to devise abatement strategies. The con-cepts of critical loads and source–receptor (S–R) relationshipsprovide guidance for formulating such strategies. Based on the

Long-range Transboundary Air Pollutants in Northeast Asia(LTP) project, this study analyzes sulfur deposition data inorder to optimize acidic loads over the three countries. Thethree groups involved in this study carried out a full year(2002) of sulfur deposition modeling over the geographic

Responsible editor: Michael Matthies

C.-H. Kim (*) :H.-Y. Son : J.-J. LeeDepartment of Atmospheric Sciences, Pusan National University,Busan 609-735, South Koreae-mail: [email protected]

L.-S. Chang : J.-S. Kim : S.-J. Lee : C.-K. Song : S.-J. BanNational Institute of Environmental Research (NIER),Incheon 404-170, South Korea

F. Meng :Y. He : J. XuChinese Research Academy of Environmental Sciences (CRAES),Beijing 100012, China

M. KajinoMeteorological Research Institute (MRI),Japan Meteorological Agency,Tsukuba 305-0052, Japan

H. UedaToyohashi Institute of Technology,Toyohashi 441-8580, Japan

Y. ZhangCollege of Environmental Sciences, Peking University,Beijing 100871, China

K. SatoAsia Center for Air Pollution Research (ACAP),Niigata 950-2144, Japan

T. SakuraiJapan NUS Co., Ltd.,Tokyo 108-0022, Japan

Z. HanInstitute of Atmospheric Physics, Chinese Academy of Sciences,Beijing 100029, China

L. DuanDepartment of Environmental Science and Engineering,Tsinghua University,Beijing 100084, China

S.-G. ShimKorea Institute of Science and Technology,Seoul 136-791, South Korea

Y. SunwooDepartment of Advanced Technology Fusion, Kunkuk University,Seoul 143-701, South Korea

T.-Y. LeeDepartment of Atmospheric Sciences, Yonsei University,Seoul 120-749, South Korea

Environ Sci Pollut Res (2012) 19:4073–4089DOI 10.1007/s11356-012-1071-1

Author's personal copy

region spanning the three countries, using three air qualitymodels: MM5-CMAQ, MM5-RAQM, and RAMS-CADM,employed by Chinese, Japanese, and Korean modelinggroups, respectively. Each model employed its own meteoro-logical numerical model and model parameters. Only theemission rates for SO2 and NOx obtained from the LTP projectwere the common parameter used in the three models. Threemodels revealed some bias from dry to wet deposition, par-ticularly the latter because of the bias in annual precipitation.This finding points to the need for further sensitivity tests ofthe wet removal rates in association with underlying cloud–precipitation physics and parameterizations. Despite this bias,the annual total (dry plus wet) sulfur deposition predicted bythe models were surprisingly very similar. The ensembleaverage annual total deposition was 7,203.6±370 kt S with aminimal mean fractional error (MFE) of 8.95±5.24 % and apattern correlation (PC) of 0.89–0.93 between the models.This exercise revealed that despite rather poor error scores incomparison with observations, these consistent total deposi-tion values across the three models, based on LTP group'sinput data assumptions, suggest a plausible S–R relationshipthat can be applied to the next task of designing cost-effectiveemission abatement strategies.

Keywords Model intercomparison . Long-rangeTransboundary Air Pollutants in Northeast Asia (LTP) .

Sulfur deposition . Northeast Asia

Introduction

In Northeast Asia, SO2 emissions increased rapidly from the1980s to the early 2000s (Ohara et al. 2007), and previousstudies and projections based on planned development patternsindicate that significant potential increases in emissions mayoccur during the next 20 to 50 years in East Asia (Yu et al. 2008;ESCAP 2000). Consequently, countermeasures are required tokeep the development sustainable because sulfur poses a seriousthreat to terrestrial and aquatic ecosystems for the region and itsneighboring countries (Galloway et al. 1987; Rodhe et al. 2002).

In anticipation of this trend, the trilateral agreementsamong China, Japan, and Korea have launched the Long-range Transboundary Air Pollutants in Northeast Asia (LTP)project, aimed at lowering sulfur and nitrogen emissions bysetting a target percentage level of deposition for each country.To do so, the concept of critical loads was utilized. Criticalload is the maximum allowable depositions without increasingthe probability of damage to the soil ecosystem (Nilsson andGrennfelt 1988; Posch et al. 1993; Fisher et al. 2007). Thisconcept has been extensively used, in Europe particularly, asan aid to developing policy on the control of atmosphericsulfur and nitrogen emissions. For instance, the Conventionof Long-range Transboundary Air Pollution (LRTAP) within

the United Nations Economic Commission for Europe(UNECE) (Bull et al. 2001; Belyazid et al. 2010) and theEuropean Commission's National Emission Ceilings (NEC)directive (Hettelingh et al. 2001, 2007) both used critical loadsas their basis. However, the critical loads approach requires asimulation based on a high resolution acid depositionmodel inorder to diagnose the current acidic loadings for the purpose ofmaximizing cost effectiveness in abating emissions. Towardthis end, the primary focus of the LTP project was agreed tobetter understand the capabilities of regional comprehensiveacid deposition models for quantification of source-receptor(S–R) relationships (Park et al. 2005; Kim et al. 2008).

SO2 emitted from human sources can be removed bothdirectly by dry deposition and wet deposition, and indirectlyby dry and wet deposition of sulfate aerosol after chemicaltransformation. Many works have used chemical transportmodel to study sulfur deposition over East Asia, focusing onthe transport and deposition of sulfur compound to improvemodel performance (Lin et al. 2008; Wang et al. 2008).However, the results for determining long-range transportand the S–R relationship have been inaccurate and unreli-able (Lin et al. 2008; Streets et al. 2006). Various factorsaccount for the discrepancies, such as emissions, horizontalmodel grid size, differences of cloud processing, subgridscale parameterizations, and removal and conversion ratesof chemicals. Therefore, intercountry comparative modelingprovides reasonable estimations of simulated depositions(Carmichael et al. 2002, 2008; Uno et al. 2006), which canbe used for quantifying the robust S–R relationship targetingto abating sulfur emission. In this background, China, Japan,and Korea simulated total (dry plus wet) sulfur depositionsthrough the LTP project and conducted an intercountrycomparative modeling for the year 2002 in order to explorethe total uncertainty of a year-long simulation.

This study reports the results of a year-long model intercom-parison study performed on sulfur deposition for the year 2002,carried out as part of the LTP project. Based on a year-longsimulation of sulfur deposition of three Northeast Asiancountries, this study provides current sulfur loads for specifiedareas with discussion on the model uncertainties. In view ofidentifying abatement strategies, this study also deeply consid-ers deposition patterns and seasonal variations.

Material and methods

Comparison of participating models

As participating LTP countries, China, Japan, and Koreautilized the following three-dimensional Eulerian models,respectively: the Community Multi-scale Air Quality(CMAQ, Byun and Ching 1999), the Regional Air QualityModel (RAQM) (An et al. 2002; Han et al. 2006), and the

4074 Environ Sci Pollut Res (2012) 19:4073–4089


Comprehensive Acid Deposition Model (CADM) (Lee et al.1998; Park et al. 2005). The three chemical models calcu-lated concentrations of chemical species in the gas phase,ion concentration in cloud droplets and rainwater, andamounts of wet and dry depositions. The main character-istics of the three models are listed in Table 1, and shortsummaries are described below.

CMAQ, employed by Chinese modeling group, runsin an offline manner. Meteorological fields were allprepared using MM5 (hereafter, MM5-CMAQ), withNCEP data as the initial field and nudging FourDimensional Data Assimilation (FDDA) technique.Global NCEP data has a resolution of 2.5°×2.5°, 16layers, and an interval of 12 h, which is suitable as thegrid nudging field with a resolution of 60 km for thegiven domain. MM5 has the same horizontal resolution,60 km (70×66 grids), as the CMAQ, with 34 verticallayers, and a top at 70 hPa. The dry and wet depositionalgorithms were taken from the method of Wesely(1989) and RADM, respectively.

RAQM, running in an offline manner by the Japanesemodeling group, is a three-dimensional Eulerian modelemploying the meteorological fields of MM5 (hereafter,MM5-RAQM). NCEP data was used as the initial field,and nudging FDDA technique, and a resolution of 45 km(125×95 grids) of MM5 has been employed with 23 verticallayers and a top at 10 hPa. RAQM employed the samehorizontal resolution, 60 km (110×80 grids), as the MM5-CMAQ. Gas phase chemistry of MM5-RAQM can be rep-resented by a condensed mechanism developed on the basisof Atkinson et al. (1982). Dry deposition module for gases isupdated by a modified Wesely's parameterization scheme(Walmsley and Wesely 1996), and cloud information suchas cloud top, cloud base, liquid water content, and cloudfraction are diagnosed using the same method as that inRADM (Chang et al. 1987).

CADM, employed by the Korean modeling group, isan online model directly coupled to a meteorologicalmodel, CSU-RAMS (hereafter, RAMS-CADM). NCEPdata was used as the initial field and nudging FDDAtechnique. RAMS with a resolution of 60 km (110×80grids) has been employed with 29 vertical layers.RAMS-CADM also employed the same horizontal reso-lution, 60 km (110×80 grids), as CSU-RAMS. RAMS-CADM employs a heterogeneous chemical modelRADM (Chang et al. 1987) including gas and aqueousphase atmospheric chemical processes. Dry and wetdepositions are also treated using the same schemeemployed in RADM. Detailed land-use data needed ineach grid is obtained from the EPA/NOAA global eco-system data, and were regrouped for corresponding cat-egories over the model grids. The details of the threemodels are listed in Table 1.

Model intercomparison: experimental design and methods

Each of the three models used the same domains that cover20–25°N, 100–150°E, as agreed at the LTP project meeting,and annual total (dry plus wet) deposition using their ownseparate model were run for the entire year of 2002. As eachcountry employed its own meteorological numerical model,the model parameters, except for emissions, were not nec-essarily the same. Since the meteorological model was spec-ified individually, the map projection also differed (seeTable 1).

As a flowchart of experimental design illustrated inFig. 1, the meteorological fields were designed to respondindividually to the meteorological driver. Each model con-sisted of its own major processes such as advection, diffu-sion, deposition, cloud scavenging, and gas and aqueousphase chemistry.

In order to interpret model uncertainties, the annual totalamount of precipitation and sulfur wet depositions producedby three models are quantitatively compared with observa-tions by employing statistical indexes, such as Root

Mean Square Error (RMSE) ð¼ ½P Si � Oið Þ2 N= �1 2= Þ;Bias ¼ P

Si � Oið Þ½ � N=ð Þ , and pattern correlation PC ¼ðP

Si � S� �

Oi � O� �� P

Si�ð½½= SÞ2 P Oi � O� �2�1 2= �:Þ

where N is the number of pairs of observations/predictions,

Si is the simulation, Oi is the observation, and S and Orepresent the simulated and observed means. In order toexplore the model uncertainties between models, other sta-tistical parameters such as mean fractional errorMFE ¼ P

Ai � Bij jð Þ 0:5�= Ai þ Bið Þ½ � N=ð Þ , i n d e x o f

agreement IOA ¼ 1�PA0 � B0ð Þ2 P

A0j j � Bi0j jð Þ2

.� �,

and absolute gross error AGE ¼ PAi � Bij jð Þ� N=½ð Þ are

also employed where Ai and Bi are variables from different

models, and A′ and B′ represent Ai � A and Bi � B ,respectively.

Emission of SO2 and NOx over Northeast Asia

In cooperation with respective national administrative agen-cies, the three countries compiled individual emission inven-tories of SO2 and NOx over their own countries for the baseyear of 1998. The emission inventories of three were merged,and outer region countries, such as Taiwan and North Korea,were taken from Streets et al. (2003).

Figure 2a and b show the horizontal distributions of theannual SO2 and NOx emissions produced through LTP proj-ect. Emissions in the five subdivided regions are also shown inFig. 2c. Region I, representing northern China, exhibited totalSO2 emissions of 2,540 kt/year. Region II, in central China,showed the largest emission source with 12,239 kt/year.

Environ Sci Pollut Res (2012) 19:4073–4089 4075


Table 1 Description of the three models employed by China, Japan, and Korea for the model intercomparison study

China Japan Korea

Air quality model CMAQa RAQMb,c) CADMd

Model domain 20–50 N, 100–150 E 20–50 N, 100–150 E 20–50 N,100–150 E

Map projection Lambert conformal Lambert conformal Polar stereographic

Model run type Offline Offline Online

Vertical coordinate σ coordinate Terrain following Terrain following

Horizontal resolution 60 km (70×66 grids) 60 km (110×80 grids) 60 km (110×80 grids)

Vertical layers 14 12 12

Gas phase chemistry CB-IV mechanisme Condensed CB-IVb,c,f RADM chemistryg

Chemical species 93 rxns, 36 species 93 rxns, 36 species 157 rxns, 64 species

11 photochemistry 11 photochemistry 27 photochemistry

Aqueous chemistry RADM chemistry RAQM chemistryb,c RADM chemistry

Aqueous phase species 28 rxns, 12 components 21 rxns, 9 components 17 rxns, 9 components

Dry deposition Wesely's methodh Modified Wesely's methodi RADM module

Wet deposition RADM module RADM module RADM module

Aerosol thermodynamics ISORROPIAj ISORROPIA UHAEROk

Emission LTP emission (1°×1° resolution) LTP emission (1°×1° resolution) LTP emission + Miyakejima emissionl

Vertical diffusivity Brost's methodm Byun and Dennis' methodn Brost's method

Meteorological model MM5o MM5 CSU-RAMSp

Basic equation Compressible, nonhydrostatic Compressible, nonhydrostatic Compressible, nonhydrostatic

Horizontal resolution 60 km (70×66 grids) 45 km (125×95 grids) 60 km (110×80 grids)

Horizontal grid Arakawa-B Arakawa-B Arakawa-C

Domain structure Nested grid (3 domains) Nested grid (3 domains) Nested grid (3 domains)

Vertical coordinate Terrain following pressure Terrain following pressure Terrain following height

Vertical layers 34 23 29

Data assimilation FDDA with NCEP FDDA with NCEP FNL FDDA with NCEP

Cumulus parameterization Grell schemeso,q Betts-Miller schemer Anthes-Kuo schemeo

TKE closer 1.5 order TKE 1.5 order TKE 1.5 order TKE

PBL scheme MRF PBL schemes MRF PBL scheme MRF PBL scheme

Moisture physics Schultz microphysicst Mixed phase physicsu Mixed phase physicsv

Radiation RRTM longwave radiation schemew RRTM longwave radiation scheme RRTM longwave radiation scheme

Soil layer 7-layer soil model 5-layer soil model 8-layer soil model

Land use type EPA/NOAA (11 categories) EROSx (12 categories) EPA/NOAA (11 categories)

a Byun and Ching (1999)b An et al. (2002)c Han et al. (2006)d Lee et al. (1998)e Gery et al. (1989)f He and Huang (1996)g Chang et al. (1987)h Wesely (1989)i Walmsley and Wesely (1996)j Nenes et al. (1998)k Amundson et al. (2007)l Kajino et al. (2004)m Brost et al. (1988)n Byun and Dennis (1995)o Grell (1993)p Pielke et al. (1992)q Grell et al. (1994)r Betts and Miller (1986)s Hong and Pan (1996)t Schultz (1995)u Reisner et al. (1998)v Dudhia (1989)wMlawer et al. (1997)x Earth Resources Observation System (US Geological Survey), DeFries and Townshend (1994)



Regions III (5,893 kt/year), IV (1,146 kt/year), and V (922 kt/year) corresponded to southern China, South Korea, andJapan, respectively.

The LTP findings show that the emission strength dif-fered by a factor of 0.9–1.2 or less, as compared to thoseused in REAS and INTEX-B (Ohara et al. 2007; Zhang et al.2009). Previously estimated sulfur emissions over Chinawere 20,385, 27,590, and 31,020 kt/year by TRACE/ACE(for the base year of 2000), REAS (for the base year of2000), and INTEX-B (for the base year of 2006) comparedto the LTP estimation with the value of 20,672 kt/year.TRACE/ACE, REAS, and INTEX-B estimated 801–987kt/year (over Japan) and 497–987 kt/year over Japan andKorea, respectively, well compared with LTP estimation of922 kt/year and 1,146 kt/year over Japan (Region V) andKorea (Region IV), respectively, as indicated in Fig. 2c.

Data used

In order to compute the skill score for the simulated precipi-tation, we used GPCP reanalysis data in which regional detailsare considered based on a coarse resolution of 1° x 1°. Inassessing the uncertainties and sensitivities of each of theparticipating three models, the site-specific surface wet depo-sitionmeasurements were compared with those taken from thesimulations. Surface sulfur wet depositions are collected at sixbackground monitoring sites in remote, isolated areas: Dalian(China), Xiamen (China), Ganghwa (Korea), Gosan (Korea),Oki (Japan), and Rishiri (Japan), as illustrated in Fig. 2c.

Results

Precipitation

Figure 3 shows the annual total amount of precipitationproduced by the three models against GPCP. Some

statistical validation scores are presented in Table 2. Thethree models can simulate the pattern of monsoon precipi-tation over the Northeast Asian region (Fig. 3) with thecorresponding PC value of 0.72–0.84 (average00.79) be-tween the models (Table 2). However, compared with thepattern of GPCP precipitation, distinct characteristics arealso apparent due to the different cloud models in the threemodels. As shown in Fig. 3, MM5 simulated reasonablywell over land. MM5-RAQM and MM5-CMAQ overesti-mated at lower latitudes due to the spurious oceanic rainfall.

Fig. 2 Standard emissions of LTP project for a SO2 and b NOx

prepared by China, Japan, and Korea. The figure also shows c the fivesubdivided regions (Regions I–V) for the analysis of annual averagedstatistics and the location of the six wet deposition monitoring sites

Fig. 1 A flowchart for the simulation of meteorological and chemicalvariables



However, RAMS-CADM derived relatively weak convec-tion over the middle latitude ocean (Fig. 3), resulting in anunderestimation against GPCP over the whole domain, witha mean bias of −461.8 mm (Table 2).

Over Region II, the highest emission areas, 935 mm/year,713 mm/year, and 621 mm/year, were produced by MM5-CMAQ, MM5-RAQM, and RAMS-CADM, respectively.MFE (average032.2 %) and RMSE (average0623.3 mm)also show considerably high between the models (Table 2).It has been postulated that these different precipitation char-acteristics, such as big differences between MM5-CMAQand RAMS-CADM over Region II, lower precipitation overRegions IV–V by RAMS-CADM, and overestimation inRegion III by MM5-RAQM and in Region V by MM5-CMAQ, are strongly related to the patterns of the annual wetdepositions (see “Dry and wet sulfur depositions”).

Surface SO2 and sulfate concentrations

The spatial patterns of seasonally averaged SO2 and SO42−

surface concentrations are shown in Fig. 4. All three modelsdemonstrated seasonal variations: the lowest SO2 in summerand the highest in winter. The opposite patterns emerged forSO4

2−. The low SO2 and high SO42− during summer can

undoubtedly be ascribed to both the rapid photochemicalconversion from SO2 to sulfate and the precipitation scav-enging of SO2. Cloud activity has a marked influence on theheterogeneous sink of SO2 in the clouds and rain dropletsand as a result, sulfate concentration is the highest in sum-mer. Moreover, a high sulfate level in summer is closelyrelated to the Pacific high-pressure systems over NortheastAsia, which brings humid northward or northeastward windfields over the southern part of China, resulting in very lowSO2 and an extremely high sulfate concentration. Duringwinter, very few photochemical reactions occur, and therelatively smaller mixed layer height results in a smalldegree of upward transport/diffusion with weak cloud activ-ity; consequently, the sulfate concentration decreases in themixed layer.

Despite similarities, each model showed differences inquantity. RAMS-CADM simulated considerably high SO2

concentrations in winter (Fig. 4) with a domain average of3.78 ppb. This is due to the smaller amount of precipitationsimulated by CSU-RAMS (see “Precipitation”), during win-ter particularly, leading to a higher SO2 and a lower SO4

2−

(Fig. 4). Instead, MM5-CMAQ and MM5-RAQM simula-tions yielded considerably lower levels of gas phase SO2

with the domain averages of 1.55 ppb and 2.84 ppb, respec-tively, indicating the important role of wet removal processand/or chemical conversion rate that are more rapid inMM5-CMAQ and MM5-RAQM than RAMS-CADM.This dynamic is more apparent in the seasonal features ofSO4

2−. As shown in Fig. 4, both MM5-CMAQ and MM5-Fig. 3 GPCP precipitation and annual total precipitation patterns assimulated by MM5-CMAQ, MM5-RAQM, and RAMS-CADM



RAQM simulations show higher levels of SO42−. RAMS-

CADM simulated SO42− with a maximum value of only

2.3 μg/m2 during the summer. However, MM5-CMAQ sim-ulated more than 10 μg/m2 with a higher and wider regionof SO4

2− than RAMS-CADM (Fig. 4). This is due to strongcloud activity during summer, reflecting the decisive role ofcloud–precipitation physics in our LTP model frameworksand accounting for the different outcomes. More detaileddiscussions on sulfur concentrations over the LTP domainwere found in Kim et al. (2011).

Dry and wet sulfur depositions

Figure 5 shows the spatial distributions of seasonal sulfurdepositions. All three models simulated lower dry but higherwet deposition in summer than other seasons, showing thehighest total deposition in summer for all three models. Ingeneral, dry deposition velocity was reported to be higherduring summer than winter on vegetative surfaces by afactor of 1.5–5 (Brook et al. 1997), due to dense vegetationand strong turbulent mixing during summer, as described byWesely's parameterization (Wesely 1989; Walmsley andWesely 1996). Our simulations with all three models simu-lated the same tendencies of dry deposition velocity (notshown). In the current study, however, the results obtainedwith the three models indicate that the degree of dry depo-sition shows no significance, with opposite seasonal trends

at times; a lower rate of dry deposition in summer and ahigher rate in winter. This is due to lower gas phase SO2 insummer (see “Surface SO2 and sulfate concentrations”),which was due mainly to a greater degree of turbulentmixing and partly to the rapid photochemical conversionto sulfate. Horizontal distributions of the year-long drydepositions show that, for every season, significant drydeposition was simulated over eastern China and Korea,almost identical to those of areas with large emission sour-ces, basically reflecting the emission strength, as shown inFig. 1.

Wet depositions were found to be larger in summer(Fig. 5), with a maximum of 980, 813, and 796 mg/m2,indicated by MM5-CMAQ, MM5-RAQM, and RAMS-CADM, respectively, in the decreasing order of the precip-itation amount (Table 2), again indicating the importance ofclouds and precipitation. A high level of cloud activity insummer was found to strongly influence the heterogeneousproduction of sulfate because most of the conversion of SO2

to sulfate occurs in clouds and rain droplets. In addition, thehigh temperature and greater amount of solar radiation thatlead to a faster photochemical conversion of SO2 to sulfateduring summer (“Surface SO2 and sulfate concentrations”)is also of great importance for high wet deposition process-es. However, total deposition shows almost the same pat-terns with detectable but small seasonal variation in all threemodels (Fig. 5). This is because the lower wet deposition

Table 2 Statistical summary of the simulated precipitation simulated by MM5-CMAQ, MM5-RAQM, and RAMS-CADM in comparison withGPCP precipitation

GPCP CMAQ (MM5) RAQM (MM5) CADM (CSU-RAMS)

Precipitation (mm)

Region I 378.9 509.1 338.9 288.2

Region II 696.8 935.2 713.1 620.6

Region III 1,617.7 1,359.1 1,983.5 1,317.0

Region IV 1,481.1 1,829.2 1,229.1 969.8

Region V 1,618.0 2,242.5 1,870.9 1,257.6

Model versus GPCP

BIAS (mm) – 279.6 −0.13 −461.8

Root mean square error (RMSE) (mm) – 539.5 532.3 463.2

Pattern correlation (PC) – 0.68 0.81 0.72

Mean fractional error (MFE) (%) – 22.1 6.4 26.8

Absolute gross error (AGE) (mm) – 399.9 373.2 345.0

Index of agreement (IOA) – 0.82 0.87 0.86

Model versus model

BIAS (mm) −284.3–387.6

RMSE (mm) 434.8–739.6 (average0623.3 mm)

Pattern correlation (PC) 0.72–0.84 (average00.79)

Mean fractional error (MFE) (%) 15.0–48.5 (average032.2 %)

Absolute gross error(AGE) (mm) 324.5–543.4 (average0469.2 mm)

Index of agreement (IOA) 0.74–0.91 (average00.81)



Fig. 4 Monthly averaged surface SO2 and SO42− concentrations simulated by MM5-CMAQ (left panel), MM5-RAQM (middle panel) and RAMS-

CADM (right panel) for 2002



Fig. 5 Spatial distributionpatterns of seasonally averageddry, wet, and total (0dry + wet)sulfur depositions (S mg/m2) assimulated by MM5-CMAQ,MM5-RAQM, and RAMS-CADM during 2002



Tab

le3

Sum

maryof

theseason

alvariations

inthewetanddrydepo

sitio

nsof

sulfur

(ktS

)foreach

ofthefive

subd

ivided

region

s,as

simulated

byMM5-CMAQ,M

M5-RAQM,and

RAMS-CADM

during

2002

Region

III

III

IVV

Season

MM5-

CMAQ

MM5-

RAQM

RAMS-

CADM

MM5-

CMAQ

MM5-

RAQM

RAMS-

CADM

MM5-

CMAQ

MM5-

RAQM

RAMS-

CADM

MM5-

CMAQ

MM5-

RAQM

RAMS-

CADM

MM5-

CMAQ

MM5-

RAQM

RAMS-

CADM

Spring(M

arch,April,

May)

Total

248.2

116.7

170.8

882.1

751.9

846.8

583.5

708.0

429.3

41.1

66.7

46.2

109.7

299.8

148.3

Wet

175.0

60.1

73.7

533.1

308.7

332.7

433.2

444.2

234.0

9.7

24.6

10.7

71.3

205.9

71.4

Dry

73.2

56.6

97.1

349.0

443.2

514.1

150.3

263.8

195.3

31.4

42.1

35.5

38.4

93.9

76.9

Sum

mer

(Jun

e,July,

Aug

ust)

Total

303.0

200.3

257.2

1,12

2.1

886.2

922.4

560.4

823.6

550.7

49.8

72.9

36.3

119.3

262.3

107.9

Wet

196.6

133.8

159.4

769.6

453.0

423.6

423.9

601.0

431.8

25.7

38.0

15.5

71.0

190.6

78.4

Dry

106.4

66.5

97.9

352.6

433.2

398.8

136.5

222.6

119.0

24.2

34.9

20.7

22.9

71.7

29.4

Autum

n(September,

Octob

er,Nov

ember)

Total

187.5

112.2

158.9

803.9

672.4

815.7

492.1

661.6

499.8

43.5

62.0

40.2

94.1

329.5

151.9

Wet

93.9

57.5

85.1

359.5

240.5

293.8

295.7

376.5

254.1

14.0

23.6

8.4

57.0

219.3

67.4

Dry

93.6

54.7

73.8

444.4

431.9

521.9

196.4

285.1

245.8

29.6

38.4

31.7

37.2

110.2

84.5

Winter(January,F

ebruary,

Decem

ber)

Total

138.4

107.6

114.2

634.3

688.0

837.2

502.5

599.7

461.3

47.3

70.6

46.4

128.5

302.3

209.1

Wet

74.2

45.7

75.8

243.4

184.3

249.9

262.1

277.7

149.5

13.4

26.1

9.2

75.2

199.3

84.5

Dry

64.2

61.9

38.6

390.9

503.7

587.3

240.3

322.0

311.8

33.9

44.5

36.9

53.3

103.0

124.6

Ann

ual

Total

877.1

536.8

701.1

3,44

2.4

2,99

8.5

3,32

2.1

2,13

8.5

2,79

2.9

1,94

1.1

181.7

272.2

169.1

426.2

1,19

3.9

617.2

Wet

539.7

297.1

394.0

1,90

5.6

1,18

6.5

1,39

0.0

1,41

4.9

1,69

9.4

1,06

9.4

62.8

112.3

43.8

274.5

815.1

301.7

Dry

337.4

239.7

307.4

1,53

6.9

1,81

2.0

2,02

2.1

723.5

1,09

3.5

871.9

119.1

159.9

124.8

151.8

378.8

315.4



due to smaller precipitation is compensated by higher drydeposition due to higher SO2 concentrations and vice versa.

Table 3 summarizes the wet and dry deposition over thefive subregions, as indicated in Fig. 2c. As expected, astrong seasonal variation of wet depositions was found inevery region, but there was a little fluctuation of drydeposition. In every season, wet deposition prevailed overRegion III, while dry deposition was more significant overthe Korea (Region IV). Three models simulated the largestannual total deposition in Region II, which has pro-nounced SO2 emission sources, with deposition amountsup to 3.0–3.4×103 (kt S) (Table 3). The total depositioncomprises of nearly equal levels of dry and wet depositionin Region II. The next largest total deposition was foundin Region III, the annual total up to 1.9–2.8×103 kt S).Wet deposition was accounting for about 55–66 % of thetotal, mainly due to the high precipitation characteristics,by MM5-RAQM particularly (Table 2), as described in“Precipitation”. In Region IV, where the total annualdeposition was only 0.17–0.27×103 kt S), a much smallerfraction (26–41 %) of wet deposition was shown than inother regions. RAMS-CADM, in particular, showed thehighest percentage (74 %) of dry deposition in RegionIV, due to its underestimation of the amount of precipita-tion (Table 2). Overall, most of the wet depositions arewell characterized by the features of precipitation pro-duced by the models.

The ensemble average of three annual accumulated totalsulfur depositions over China, Japan, and Korea totaled 7,066kt S), 7,794 kt S, and 6,851 kt S), respectively, by MM5-CMAQ, MM5-RAQM, and RAMS-CADM, with only asmall deviation (5–7 %) among the three models. The ensem-ble calculation of the mean deposition versus emission frac-tions over three countries reached only 62.1, 68.6, and 60.3 %in the three models, which, in turn, indicates a transportedfraction of 31–40 % outside of Northeast Asia. MM5-CMAQand RAMS-CADM showed small (~10%) differences in theirtotal sulfur deposition, except in Region V, because higherbiases were simulated proportional to the downstream dis-tance from the highest emission area.

The features of annual total sulfur deposition are shown inFig. 6, and statistical skills between the three models aresummarized in Table 4. Here, the statistical method recommen-ded by Willmott (1981), Fox (1981) and Klung et al. (1992)was employed. The results showed that the 1-year accumulatedtotal (dry plus wet) depositions simulated in three modelsyielded nearly identical sulfur deposition patterns (Fig. 6), withthe MFR of 8.95±5.24(%) (Table 4). Mean depositions werealso similar with the magnitudes of 1,256–1,769 Smg/m2,which was revealed to be in reasonable agreement within afactor of 1.3. PC of 0.92±0.02 shows much closer agreementthan expected from other skill scores. This similarity is becauselower wet depositionmore or less canceled out the effects of the

higher dry deposition to a certain degree. The index of agree-ment (IOA00.91) also suggests that the results are mostly thesame in all simulations.

In Table 4, overall similarity betweenmodels is found, withno significant biases, low RMSE, and high IOA of more than0.9. These low errors and high IOAs indicates the strength of

Fig. 6 Spatial distribution patterns of annual total sulfur deposition(S mg/m2) as simulated by a MM5-CMAQ, b MM5-RAQM, and cRAMS-CADM during 2002



sulfur abatement strategy by the LTP group, which took intoconsideration the assumptions and uncertainties in the inputdata for calculating the S–R relationship over China, Japan,and Korea. Only the systematic mean square error (MSES) isthe most sensitive parameter used in the current study.

Comparison with observation

Figure 7 shows the three-model simulations of monthlyaccumulated wet depositions in six background monitoringstations indicated in Fig. 2c. Model simulations showed wellthe observed seasonal trends; with the summer level of~8 kg/ha at Ganghwa, the maximum value among total sixmonitoring sites over the current study period. In general, allthree models tend to underestimate with higher measure-ment values and overestimate with lower measurement(Fig. 7). This is believed to be responsible for the inappro-priate time variation of the emission rate. In most of the sixbackground sites, RAMS-CADM tends to be relatively low-er values over four seasons. The study also found that thediscrepancies in the simulations from measurements are notsignificant—by a factor of around 2.5 or lesser—the soleexception being the month of January at Oki with an ob-served level of ~7.5 kg/ha and minimum 0.26 kg/ha in July,which deviated from simulations by a factor of more than10, the largest factor in this study (Fig. 7). This difference isdue to the climatology of Oki, a northern Japanese coastalregion influenced by the prevailing northerly monsoon,which causes frequent heavy precipitation in winter, induc-ing opposite monthly variations in wet deposition.

Figure 8 shows scatter plot comparisons of the simulatedsulfur wet depositions against measurements. The modelreproduces the observed reasonable variability with R2 of0.57, 0.31, and 0.43 for MM5-CMAQ, MM5-RAQM, andRAMS-CADM, respectively. All three models underesti-mate the values by approximately 40–60 % with the regres-sion slopes of 0.61, 0.49, and 0.40, respectively, again indecreasing order of the precipitation amount as indicated inMM5-CMAQ, MM5-RAQM, and RAMS-CADM, respec-tively (Table 2). CADM, in particular, showed the lowestregression slope due to the same reasons described in“Surface SO2 and sulfate concentrations” to “Dry and wetsulfur depositions”.

Table 5 summarizes the statistical skill scores of wet depo-sition simulations in comparison with measurements. All of thesimulations showed different ranges, with a variety of similar-ities and differences. The mean ratio between the valueobtained from the model the observed value (S/O) ranges from0.48 to 0.76, which indicates an underestimation (in S/O, Sdenotes simulation and O denotes observation). Nevertheless,all three models score IOAs of 0.76, 0.61, and 0.68, respec-tively, which are apparently > 0.5, the level of good gradesdiscussed in previous studies (Hurley et al. 2002). RMSE forthe three models are 1.37, 1.70, and 1.85 kg/ha for MM5-CMAQ, MM5-RAQM, and RAMS-CADM, respectively,indicating MM5-CMAQ gives much better simulation thanthe other two models. However, MFE (21.5–59.7 %)together with difference of standard deviation (1.27–1.52 kg/ha) suggest some discrepancies between the gridpoint-based results and the observed results. Unsystematic

Table 4 Summary of statisticalcomparisons of modeled total(0dry + wet) sulfur depositionsbetween models

MM5-CMAQ MM5-RAQM RAMS-CADM

Sample size 5,400 5,400 5,400

Range (S mg/m2) ~10,245 ~9,314 ~7,239

Mean (S mg/m2) 1,760 1,869 1,634

Standard deviation(SD) (S mg/m2)

2,215 2,095 1,538

SD of ratio between models 0.89 (±0.41)

Absolute gross error(AGE) (S mg/m2)

599 (±80.8)

Mean difference (S mg/m2) 551 (±55)

Difference of standard deviation(S mg/m2)

925 (±51)

Root mean square error (RMSE)(S mg/m2)

1,031 (±121)

Unsystematic mean square error(MSEN) (S mg/m2)

1,604 (±424)

Systematic mean square error (MSES) (S mg/m2) 2,482 (±754)

Index of agreement (IOA) 0.91 (±0.13)

Mean fractional error (MFE) (%) 8.95 (±5.24)

Pattern correlation (PC) 0.92 (±0.02)



error (MSEN) were again smaller than the systematic error(MSES) (Table 4), suggesting that the model performancewould be more reliable if we could obtain more preciseseasonal emission and other input data, rather than resolvingany intrinsic uncertainty in the model itself.

Discussion

Deposition processes for removal of sulfur from the atmo-sphere to the surface occur in two ways: dry deposition andwet deposition. Clouds and precipitation play a critical role inthe removal of air pollutants via wet deposition. Dry depositionvelocity, which is governed by atmospheric turbulence and thenature of the surface, is also considered to be an important

parameter in the removal process. In our study, the results fordry deposition showed that the distribution of SO2 concentra-tionwas simulated to be more predominant than the type of dryremoval parameterization scheme. Our analysis showed thatdry deposition velocity varies with a factor of only ~1.2 amongmodels (not shown here). In long-term simulation, a ratherrelatively larger variation in regions such as southern Chinaand Japan did not indicate well the detectable variation in drydeposition for all three models. In our analysis, the seasonaltrend of dry deposition velocity was completely overwhelmedby the variation of gas phase SO2 concentrations. As a result,only the higher emission area showed a significantly predom-inant amount of dry deposition in the current study.

However, we discovered some bias in the cloud physicsand dynamics due to different adapted microphysics and/or

Fig. 7 Monthly variations ofobserved and simulated sulfurwet depositions (SO4

2− kg/ha)at the six monitoring stations



cumulus parameterization, which mainly caused model un-certainty in the estimation of the amounts and patterns ofwet deposition in our study. For example, the cumulusparameterization schemes, Anthes-Kuo (Grell et al. 1994)

employed by RAMS-CADM, and Grell (1993) and Bettsand Miller (1986) by MM5-CMAQ and MM5-RAQM,respectively, produce much different total annual amountsof precipitation by up to a factor of almost 2 in some regions(Fig. 3). General overestimation by MM5 is centered overthe ocean regions near the south of both Japan and Taiwan.By and large, the excessive convection in the ocean at lowerlatitudes accounts for the deviations in simulated precipita-tion values from those of the GPCP. These different precip-itation characteristics, such as lower precipitation overRegions IV–V particularly by RAMS-CADM, overestima-tion of precipitation in Region III by MM5-RAQM, and inRegion V by MM5-CMAQ, are clearly related to the pat-terns of both sulfur concentrations and annual total wetdeposition, as discussed in“Surface SO2 and sulfate concen-trations” and “Dry and wet sulfur depositions”.

Through this project, we have attempted to investigate thesensitivity to model variability arising from different modeltypes, assumptions, and meteorological parameterizations in-cluding microphysics, cloud schemes, and other surface bound-ary forcing parameterizations. Among them, amodel sensitivitytest of the wet removal rate in association with the cloud–precipitation physics should at least be a priority for furthertests and applications to long-term sulfur depositions simula-tions. Despite smaller bias of dry deposition, we also plan tointroduce a new parameterization for dry deposition in order toassess the velocity of dry deposition on Asian vegetation indifferent seasons; for this purpose, we plan to categorize Asianvegetation by using the realistic observation data obtained froma satellite.

Given the critical load map over Northeast Asia (Tao andFeng 2000; Park and Lee 2001; Shindo et al. 1995), under-standing of S–R relationships at regional scale is importantin establishing effective emission management strategies.As S–R relationships are estimated using comprehensivemodels, it is also important to evaluate the uncertainty insimulated S–R relationship for formulating the strategies.Inevitably such analysis contains considerable uncertaintydepending on the model and various meteorological condi-tions and assumptions in parameterization. Evaluations per-formed in this study pointed to an importance of cloudprocess for reducing the model uncertainty, but further mod-el development and more detailed testing of parameteriza-tion including cloud–precipitation process, differentinitiations, and realistic boundary conditions would be need-ed over Northeast Asian.

Summary and conclusion

In this study, comprehensive acid deposition models designedby the three countries participating in the LTP project wereemployed, and the results of the simulation of sulfur

Fig. 8 Scatter plots of the daily mean modeled versus observed wetdepositions with optimal regression lines over six background sites



deposition in the year 2002 were compared. The comprehen-sive MM5-CMAQ (China), MM5-RAQM (Japan), andRAMS-CADM (Korea) were used, and a year-long intercom-parison study was performed. The emission rates for SO2 andNOx obtained from the report of the LTP project were the onlycommon parameters used in these three models.

The results of meteorological models employed by threecountries revealed that annual precipitation from three mod-els biased high. MM5-CMAQ and MM5-RAQM overesti-mated the precipitation, while RAMS-CADM seriouslyunderestimated against GPCP. In turn, precipitation biasbetween the three models revealed some bias from dry towet deposition. In particular, wet deposition bias arisesdirectly from the bias in annual precipitation.

In comparison with observations, all three modelsshowed underestimation with rather poor MFR and meandifference. Nevertheless, the simulated pattern and amountof annual total (dry plus wet) sulfur deposition determinedusing these three models give satisfying results; the ensem-ble average annual total (dry plus wet) sulfur deposition was7,203.6±370 kt S with a minimal MFR of 8.95±5.24 %. Agood PC (0.92±0.02) between the models also suggestedthe plausible results of the S–R relationship and abatementstrategy as a next step for abating sulfur emission, based onLTP group's input data assumptions.

Finally, intercountry alliance and cooperation throughcollaborative modeling studies are critical to makingimprovements on identifying and assessing other transboun-dary air pollutants in the region. For this purpose, intensivemonitoring and comprehensive modeling that are beingcarried out by the LTP group in collaboration with theNortheast Asian countries, with the aim of making improve-ments to the existing models, will be required. Therefore,through the LTP project, further detailed model sensitivity

tests including the wet removal rate in association with thecloud physics will be carried out.

Acknowledgments The authors acknowledge members of modelingsubworking group of LTP Project organized by Korea, China, andJapan for LTP discussions. Also authors thank the researchers fromthe National Institute of Environmental Research (NIER, SouthKorea), Chinese Research Academy of Environmental Sciences(CRAES, China), and Asia Center for Air Pollution Research (ACAP,Japan) for both carrying out the models and providing the measure-ments for model verification. The authors thank Joon K. Kim atColorado State University for assistance with the manuscript.

References

Amundson NR, Caboussat A, He JW, Martynenko AV, Savarin VB,Seinfeld JH, Yoo KY (2007) A new atmospheric aerosol phaseequilibrium model (UHAERO): organic systems. Atmos ChemPhys 7:4675–4698

An J, Ueda H, Wang Z, Matsuda K, Kajino M, Cheng X (2002)Simulations of monthly mean nitrate concentrations in precipita-tion over East Asia. Atmos Environ 36:4159–4171

Atkinson R, Lloyd AC, Winges D (1982) An updated chemical mech-anism for hydrocarbon/NOx/SO2 photooxidations suitable forinclusion in atmospheric simulation models. Atmos Environ16:1341–1355

Belyazid S, Kurz D, Braun S, Sverdrup H, Rihm B, Hettelingh JP(2010) A dynamic modeling approach for estimating critical loadsof nitrogen based on plant community changes under a changingclimate. Environ Pollut 159:789–801

Betts AK, Miller MJ (1986) A new convective adjustment scheme. PartII: single column tests using GATE wave, BOMEX, ATEX, andArctic air-mass data sets. Q J R Meteorol Soc 112:693–709

Brook JR, Di-Giovanni F, Cakmak S, Meyers TP (1997) Estimation ofdry deposition velocity using inferential models and site-specificmeteorology: uncertainty due to sitting of meteorological towers.Atmos Environ 31(23):3911–3919

Table 5 Summary of pairedstatistical comparisons of mod-eled sulfur wet depositions withobservations

Observation MM5-CMAQ MM5-RAQM RMAS-CADM

Sample size 72 72 72 72

Range (kg/ha) 0.0006–15.13 ~5.00 ~3.60 ~4.60

Mean (kg/ha) 2.26 1.72 1.48 1.08

Standard deviation (kg/ha) 1.81 0.91 0.79 0.79

Mean of ratio model/observation (S/O) 0.76 0.66 0.48

Standard deviation of ratio model/observation (S/O) 1.27 1.70 0.90

Absolute gross error (AGE) (kg/ha) 0.95 1.18 1.31

Mean difference (kg/ha) 0.54 0.78 1.18

Difference of standard deviation (kg/ha) 1.27 1.52 1.42

Root mean square error (RMSE) (kg/ha) 1.37 1.70 1.85

Unsystematic mean square error (MSEN) (kg/ha) 0.35 0.43 0.36

Systematic mean square error (MSES) (kg/ha) 1.56 2.50 3.09

Index of agreement (IOA) 0.76 0.61 0.68

Mean fractional error (MFE) (%) 21.5 36.7 59.7

Pattern correlation (PC) 0.76 0.57 0.65



Brost RA, Haagenson PL, Kuo YH (1988) The effects of diffusion ontracer puffs simulated by a regional scale Eulerianmodel. J GeophysRes 93(D3):2389–2404

Bull KR, Achermann B, Bashkin V, Chrast R, Fenech G, Forsius M,Gregor HD, Guardans R, Haussmann T, Hayes F, Hettelingh JP,Johannessen T, Krzyzanowski M, Kucera V, Kvaeven B, Lorenz M,Lundin L, Mills G, Posch M, Skjelkvale BL, Ulstein MJ (2001)Coordinated effects monitoring and modeling for developing andsupporting international air pollution control agreements. Water AirSoil Pollut 130:119–130

Byun DW, Ching JKS (eds) (1999) Science algorithms of the EPAModels-3 Community Multiscale Air Quality(CMAQ) modelingsystem, EPA-600/R-99/030. Office of Research and Develop-ment, US Environmental Protection Agency, Washington, DC

Byun DW, Dennis R (1995) Design artifacts in Eulerian air qualitymodels: evaluation of the effects of layer thickness and verticalprofile correction on surface ozone concentrations. Atmos Envi-ron 29:105–126

Carmichael GR, Calori G, Hayami H, Uno I, Cho SY, Engardt M, KimS-B, Ichikawa Y, Ikeda Y, Woo J-H, Ueda H, Amann M (2002)The MICS-Asia study: model intercomparison of long-rangetransport and sulfur deposition in East Asia. Atmos Environ26:175–199

Carmichael GR, Sakurai T, Streets D, Hozumi Y, Ueda H, Park S-U,Fung C, Han Z, Kajino M, Engardt M, Bennet C, Hayami H,Sartelet K, Holloway T, Wang Z, Kannari A, Fu J, Matsuda K,Thongboonchoo N, Amann M (2008) MICS-Asia II: the modelintercomparison study for Asia Phase II methodology and over-view of findings. Atmos Environ 42:3468–3490

Chang JS, Brost RA, Isaksen ISA, Madronich S, Middleton P, StockwellWR, Walcek CJ (1987) A three-dimensional Eulerian acid deposi-tion model: physical concepts and formulation. J Geophys Res92:14681–14700

DeFries RS, Townshend JRG (1994) NDVI-derived land coverclassification at global scales. Int J Remote Sens 15:3567–3586

Dudhia J (1989) Numerical study of convection observed during theWinter Monsoon Experiment using a mesoscale two-dimensionalmodel. J Atmos Sci 46:3077–3107

ESCAP (2000) State of the environment in Asia and Pacific. UnitedNations

Fisher LS, Mays PA, Wylie CL (2007) An overview of nitrogen criticalloads for policy makers, stakeholders, and industries in the UnitedStates. Water Air Soil Pollut 179:3–18

Fox DG (1981) Judging air quality model performance. Bull AmMeteorol Soc 62:599–609

Galloway JN, Dianwu Z, Jiling X, Likens GE (1987) Acid rain:China, United States, and a remote area. Science 236(4808):1559–1562

Gery MW, Whitten GZ, Killus JP, Dodge MC (1989) A photochemicalkinetics mechanism for urban and regional scale computer mod-eling. J Geophys Res 94:12925–12956

Grell G (1993) Prognostic evaluation of assumptions used by cumulusparameterizations. Mon Weather Rev 121:764–787

Grell GA, Dudhia J, Stauffer DR (1994) A description of the fifthgeneration Penn State/NCAR mesoscale model (MM5). NCARtechnical note NCAR/TN-398+STR, 138 pp

Han Z, Ueda H, Sakurai T (2006) Model study on acidifying wetdeposition in East Asia during wintertime. Atmos Environ40:2360–2373

He D, Huang M (1996) A numerical study on tropospheric ozone overChina. Atmos Environ 30:2449–2451

Hettelingh JP, Posch M, De Smet PAM (2001) Multi-effect criticalloads used in multi-pollutant reduction agreements in Europe.Water Air Soil Pollut 130:1133–1138

Hettelingh JP, Posch M, Slootweg J, Reinds GJ, Spranger T, Tarrason L(2007) Critical loads and dynamic modeling to assess Europeanareas at risk of acidification and eutrophication. Water Air SoilPollut Focus 7:379–384

Hong SY, Pan HL (1996) Nonlocal boundary layer vertical diffusion ina medium-range forecast model. Mon Weather Rev 124(10):2322–2339

Hurley PJ, William LP, Ashok KL (2002) The Air Pollution Model(TAPM) version 2: summary of some verification studies. CSIRO,Clayton

Kajino M, Ueda H, Satsumabayashi H, An J (2004) Impacts of theeruption of Miyakejima Volcano on air quality over far east Asia.J Geophys Res 109(D21204) doi:10.1029/2004JD004762

Kim C-H, Park I-S, Kim S-K, Son H-Y, Lee J-J, Lee J-B, Song C-K,Shim J-M (2008) Estimation and mapping of nitrogen uptake byforest in South Korea. Water Air Soil Pollut 187:315–325

Kim C-H, Chang L-S, Kim J-S, Meng F, Kajino M, Ueda H, Zhang Y,Son H-Y, He Y, Xu J, Sato K, Song C-K, Ban S-J, Sakurai T, HanZ, Duan L, Lee S-J, Shim S-G, Sunwoo Y, Lee T-Y (2011) Long-term simulations of the sulfur concentrations over the China,Japan and Korea: a model comparison study. Asia Pac J AtmosSci 47(4):399–411

Klung W, Graziani G, Grippa G, Pierce D, Tassone C (1992) TheATMES report. Elsevier, London

Lee T-Y, Kim S-B, Lee S-M, Park S-U, Kim D-S, Shin H-C (1998)Numerical simulation of air quality and acid deposition for epi-sodic cases in eastern Asia. Korean J Atmos Sci 1(2):126–144

Lin M, Oki T, Holloway T, Street DG, Bengtsson M, Kanae S (2008)Long-range transport of acidifying substances in East Asia—partI: model evaluation and sensitivity studies. Atmos Environ 42(24):5939–5955

Mlawer EJ, Taubman SJ, Brown PD, Iacono MJ, Clough SA (1997)Radiative transfer for inhomogeneous atmospheres: RRTM, avalidated correlated-k model for the longwave. J Geophys Res102:16663–16682

Nenes A, Pandis SN, Pilinis C (1998) ISORROPIA: a new thermody-namic equilibrium model for multiphase multicomponent inor-ganic aerosols. Aquat Geochem 4:123–152

Nilsson J, Grennfelt P (1988) Critical loads for sulfur and nitrogen.Proceedings of workshop on critical loads for sulfur and nitrogen.United Nations Economic Council for the Environment andNordic Council of Ministers. Skokloster, Sweden

Ohara T, Akimoto H, Kurokawa J, Horii N, Yamaji K, Yan X, HayasakaT (2007) An Asian emission inventory of anthropogenic emissionsources for the period 1980–2020. Atmos Chem Phys 7:4419–4444

Park S-U, Lee Y-H (2001) Estimation of the maximum critical load forsulfur in South Korea. Water Air Soil Pollut 130:1145–1150

Park I-S, ChoiW-J, Lee T-Y, Lee S-J, Han J-S, KimC-H (2005) Simulationof long-range transport of air pollutants over Northeast Asia using acomprehensive acid deposition model. Atmos Environ 39:4075–4085

Pielke RA, Cotton WR, Walko RL, Tremback CJ, Nicholls ME, MoranMD, Wesley DA, Lee TJ, Copeland JH (1992) A comprehensivemeteorological modeling system-RAMS. Meteorog Atmos Phys49:69–91

Posch M, Forsius M, Kamari J (1993) Critical loads of sulfur andnitrogen for lakes. 1. Model description and estimation of uncer-tainty. Water Air Soil Pollut 66:173–192

Reisner J, Rasmussen RM, Bruintjes RT (1998) Explicit forecasting ofsupercooled liquid water in winter storms using a mesoscalemodel. Q J Royal Meteorol Soc 124:1071–1089

Rodhe H, Dentener F, Schulz M (2002) The global distribution ofacidifying wet deposition. Environ Sci Technol 36:4382–4388

Schultz P (1995) An explicit cloud physics parameterization for oper-ational numerical weather prediction. Mon Weather Rev 123(11):3331–3343



http://dx.doi.org/10.1029/2004JD004762

Shindo J, Bregt AK, Hakamata T (1995) Evaluation of estimationmethods and base data uncertainties for critical loads of aciddeposition in Japan. Water Air Soil Pollut 85:2571–2576

Streets DG, Bond TC, Carmichael GR, Fernandes SD, Fu Q, He D,Klimont Z, Nelson SM, Tsai NY, Wang MQ, Woo J-H, YarberKF (2003) An inventory of gaseous and primary aerosolemissions in Asia in the year 2000. J Geophys Res 108(D21):8809

Streets DG, Zhang Q, Wang L, He K, Hao J, Wu Y, Tang Y,Carmichael GR (2006) Revisiting China's CO emissions afterthe Transport and Chemical Evolution over the Pacific (TRACE-P) mission: synthesis of inventories, atmospheric modeling, andobservations. J Geophys Res 111(D14306), doi:10.1029/2006JD007118

Tao F, Feng Z (2000) Critical loads of SO2 dry deposition and theirexceedance in South China. Water Air Soil Pollut 124:429–438

Uno I, He Y, Ohara T, Yamaji K, Kurokawa J-I, Katayama M, Wang Z,Noguchi K, Hayashida S, Richter A, Burrows JP (2006) System-atic analysis of interannual and seasonal variations of model-simulated tropospheric NO2 in Asia and comparison withGOME-satellite data. Atmos Chem Phys Discuss 6:11181–11207

Walmsley JL, Wesely ML (1996) Modification of coded parameter-izations of surface resistances to gaseous dry deposition. AtmosEnviron 30:1181–1188

Wang Z, Xie F, Sakurai T, Ueda H, Han Z, Carmichael GR, Streets D,Engardt M, Holloway T, Hayami H, Kajino M, ThongboonchooN, Bennet C, Park SU, Fung C, Chang A, Sartelet K, Amann M etal (2008) MICS-Asia II: model inter-comparison and evaluationof acid deposition. Atmos Environ 42:3528–3542

Wesely ML (1989) Parameterization of surface resistances to gaseousdry deposition in regional-scale numerical model. Atmos Environ23:1293–1304

Willmott CJ (1981) On the validation of models. Phys Geogr 2:184–194Yu H, Remer LA, Chin M, Bian H, Kleidman RG, Diehl T (2008) A

satellite-based assessment of transpacific transport of pollutionaerosol. Journal of Geophysical Research 113(D14S12),doi:10.1029/2007JD009349

Zhang Q, Street DG, Carmichael GR, He KB, Huo H, KannariA, Klimont Z, Park IS, Reddy S, Fu JS, Chen D, Duan L,Lei Y, Wang LT, Yao ZL (2009) Asian emissions in 2006for the NASA INTEX-B mission. Atmos Chem Phys9:5131–5153



http://dx.doi.org/10.1029/ 2006JD007118

http://dx.doi.org/10.1029/ 2006JD007118

http://dx.doi.org/10.1029/2007JD009349

Sulfur deposition simulations over China, Japan, and Korea ...20–25°N, 100–150°E, as agreed at...

Documents

Transcript of Sulfur deposition simulations over China, Japan, and Korea ...20–25°N, 100–150°E, as agreed at...