Uncertainty in biogenic isoprene emissions and its impacts on ... - Core

Science of the Total Environment 463–464 (2013) 754–771

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Review

Uncertainty in biogenic isoprene emissions and its impacts on tropospheric chemistry in East Asia K.M. Han a,b, R.S. Park a,b, H.K. Kim c, J.H. Woo c, J. Kim d, C.H. Song a,b,⁎ a

School of Environmental Science and Engineering, Gwangju Institute of Science and Technology (GIST), Gwangju, 500-712, Republic of Korea Advanced Environmental Monitoring Research Center (ADEMRC), Gwangju Institute of Science and Technology (GIST), Gwangju, 500-712, Republic of Korea Department of Advanced Technology Fusion, Konkuk University, 1 Hwayang dong, Gwangjin-gu, Seoul, 143-701, Republic of Korea d Department of Atmospheric Sciences, Yonsei University, 134 Sinchon-dong, Seodaemoon-gu, Seoul, 120-749, Republic of Korea b c

H I G H L I G H T S • GEIA isoprene emissions were possibly overestimated over East Asia. • Using MEGAN or MOHYCAN emissions in CMAQ well captured HCHO columns from SCIAMAHCY. • Changes in HOy cycle can affect tropospheric chemistry during summer over East Asia.

a r t i c l e

i n f o

Article history: Received 15 January 2013 Received in revised form 31 May 2013 Accepted 2 June 2013 Available online 15 July 2013 Editor: Pavlos Kassomenos Keywords: Tropospheric HCHO columns CMAQ model SCIAMACHY Biogenic isoprene emission

a b s t r a c t In this study, the accuracy of biogenic isoprene emission fluxes over East Asia during two summer months (July and August) was examined by comparing two tropospheric HCHO columns (ΩHCHO) obtained from the SCIAMACHY sensor and the Community Multi-scale Air Quality (CMAQ v4.7.1) model simulations, using three available biogenic isoprene emission inventories over East Asia: i) GEIA, ii) MEGAN and iii) MOHYCAN. From this comparative analysis, the tropospheric HCHO columns from the CMAQ model simulations, using the MEGAN and MOHYCAN emission inventories (ΩCMAQ, MEGAN and ΩCMAQ, MOHYCAN), were found to agree well with the tropospheric HCHO columns from the SCIAMACHY observations (ΩSCIA). Secondly, the propagation of such uncertainties in the biogenic isoprene emission fluxes to the levels of atmospheric oxidants (e.g., OH and HO2) and other atmospheric gaseous/particulate species over East Asia during the two summer months was also investigated. As the biogenic isoprene emission fluxes decreased from the GEIA to the MEGAN emission inventories, the levels of OH radicals increased by factors of 1.39 and 1.75 over Central East China (CEC) and South China, respectively. Such increases in the OH radical mixing ratios subsequently influence the partitioning of HOy species. For example, the HO2/OH ratios from the CMAQ model simulations with GEIA isoprene emissions were 2.7 times larger than those from the CMAQ model simulations based on MEGAN isoprene emissions. The large HO2/OH ratios from the CMAQ model simulations with the GEIA biogenic emission were possibly due to the overestimation of GEIA biogenic isoprene emissions over East Asia. It was also shown that such large changes in HOx radicals created large differences on other tropospheric compounds (e.g., NOy chemistry) over East Asia during the summer months. © 2013 The Authors. Published by Elsevier B.V. Open access under CC BY-NC-SA license.

Contents 1. 2.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental methods . . . . . . . . . . . . . . . . . . . . . . 2.1. US EPA CMAQ model simulations . . . . . . . . . . . . . . 2.2. Emission inventories . . . . . . . . . . . . . . . . . . . . 2.3. Tropospheric HCHO columns from the SCIAMAHCY instrument

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

755 756 758 759 761

⁎ Corresponding author at: School of Environmental Science and Engineering, Gwangju Institute of Science and Technology (GIST), Gwangju, Republic of Korea. Tel.: +82 62 715 3276; fax: +82 62 715 3404. E-mail address: [email protected] (C.H. Song). 0048-9697 © 2013 The Authors. Published by Elsevier B.V. Open access under CC BY-NC-SA license. http://dx.doi.org/10.1016/j.scitotenv.2013.06.003

K.M. Han et al. / Science of the Total Environment 463–464 (2013) 754–771

3.

Results and discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Modeling performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Evaluation of the accuracy of the biogenic isoprene emissions over East Asia . . . . . . 3.2.1. Comparison between two tropospheric HCHO columns obtained from the CMAQ 3.2.2. Statistical analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Impacts of biogenic isoprene emission fluxes on tropospheric chemistry . . . . . . . . 3.3.1. Isoprene, monoterpen, and formaldehyde (HCHO) . . . . . . . . . . . . . . 3.3.2. Atmospheric oxidants: ozone and HOy . . . . . . . . . . . . . . . . . . . . 3.3.3. Other atmospheric particulate species . . . . . . . . . . . . . . . . . . . . 4. Summary and conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Introduction Volatile organic compounds (VOCs) are of primary importance in the formation of tropospheric ozone (Wang and Shallcross, 2000) and secondary organic aerosols (SOAs) (Claeys et al., 2004a,b), and also play important roles in HOx and NOy cycles. In particular, isoprene (C5H8) and monoterpenes (C10H16) emissions during summer can influence the concentrations of atmospheric species by changing the atmospheric oxidant levels (e.g., OH and HO2). Several studies have attempted to estimate the isoprene and mono-terpene emission fluxes on both regional and global scales, using bottom-up approaches (e.g., Zimmerman, 1979; Lamb et al., 1987; Mueller, 1992; Guenther et al., 1995, 2006). For example, Guenther et al. (1995, 2006) reported the annual global isoprene emissions as a function of the driving variables, such as temperature, solar radiation, leaf area index (LAI) and plant functional type. Their studies estimated that the biogenic VOC (BVOC) emissions account for about 40% of the total global VOCs emissions of > 1000 Tg yr−1. However, despite the many efforts to estimate the BVOC emission fluxes, large uncertainties still exist in the seasonal variations and magnitudes of the BVOC emissions. For example, the total isoprene emission fluxes of 7.1 Tg C over USA during summer estimated from the GEIA (Global Emissions Inventory Activity) inventory differed greatly from the 2.6 Tg C calculated from the BEIS2 (second generation of the Biogenic Emissions Inventory System) modeling during the same summer period (Guenther et al., 1995; Pierce et al., 1998; Palmer et al., 2003).

755

. . . . . . . . . . . . . . . . . . . . . . . . . . . model simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . and . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . the SCIAMAHCY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . sensor . . . . . . . . . . . . . . . . . . . . . . . .

761 761 762 762 764 765 765 765 768 768 770 770

On the other hand, the formaldehyde (HCHO) columns (ΩHCHO) retrieved from satellite observations have also been used to quantify the fluxes of VOC emissions, because HCHO is a short-lived intermediate compound formed from the rapid oxidations of anthropogenic VOCs (AVOCs) and BVOCs. Recently, several attempts to quantify and constrain the global isoprene emission fluxes (Shim et al., 2005; Stavrakou et al., 2009) and regional fluxes over North America (Abbot et al., 2003; Palmer et al., 2003, 2006; Millet et al., 2006), over Europe (Dufour et al., 2009), and over Asia (Fu et al., 2007), were made using the tropospheric HCHO columns from satellite observations and 3D CTM (three dimensional Chemistry–Transport Model) simulations. For example, Abbot et al. (2003) showed that the tropospheric HCHO columns from the GEOS-CHEM (Goddard Earth Observing System– Chemistry) model simulations with the GEIA isoprene emission inventory have seasonal and inter-annual consistency with those from the GOME (Global Ozone Monitoring Experiment) observations over North America. Dufour et al. (2009) also reported a bias of less than 20% between two tropospheric HCHO columns from 3D CTM (CHIMERE) model simulations and SCIAMACHY (SCaning Imaging Absorption spectroMeter for Atmospheric CHartographY) measurements over Europe. Tropospheric HCHO columns from satellite observations have also been used over East and South Asia to improve the regional emission fluxes of reactive AVOCs and BVOCs (e.g., Fu et al., 2007). East Asia is a region, where large amounts of NMVOCs are emitted from multiple emission source types (i.e. anthropogenic NMVOCs from urban and industrial areas, biogenic VOCs from forest area, and pyrogenic

Table 1 Brief description of biogenic emission models and/or inventories considered in this study. Inventory

GEIA (POET)a

MEGANb

MOHYCANa

Emission model

Canopy radiative transfer model

MEGAN model

Base year in this study Developing institution

1990 IGAC-GEIA natural VOCs working Group

Input data used in this study

Ecosystem type, global vegetation indices (GVI), precipitation, temperature, and cloudiness

Reference Output species

Guenther et al. (1995) Isoprene, mono-terpenes, and other biogenic NMVOCs 20.01 Tg yr−1

2006 National Center for Atmospheric Research (NCAR) Temperature and solar radiation above the canopy, Plant Functional Type (PFTc), Leaf Area Index (LAIc), Area average Emission Factor (EF v2.1d), fractional values of PFT (PFTFc), and MET fields from NCEP reanalysis Guenther et al. (2006) Isoprene, mono-terpenes, and other biogenic NMVOCs 14.99f Tg yr−1

MEGAN + multi-layer canopy environmental model (MOHYCAN) 2006 Belgian Institute for Space Aeronomy (BIRA-IASB) Leaf temperature and radiative fluxes as functions of height inside the canopy, Plant Functional Type (PFTd), Leaf Area Index (LAIe), Area average Emission Factor (EF v2.0d), fractional values of PFT (PFTFd), and MET fields from ECMWF Müller et al. (2008) Isoprene

5.07 Tg month−1 (July) 4.98 Tg month−1 (August)

2.56 Tg month−1 (July) 2.13 Tg month−1 (August)

Annual isoprene flux for 2000 (20°–50°N, 100°–150°N) Isoprene fluxes Over entire study domain a b c d e f g

Biogenic emission fluxes obtained directly from the emission inventories. Biogenic emission fluxes obtained from the MEGAN model simulations for July and August, 2006. Data obtained from MODIS sensor. Data obtained from the Community Data Portal of the NCAR (http://cdp.ucar.edu). Data obtained from the MODIS sensor based on the study of Zhang et al., 2004. The annual emissions were obtained from ‘http://accent.aero.jussieu.fr/database_table_inventories.php’. The annual emissions were obtained from ‘http://tropo.aeronomie.be/models/mohycan.htm’.

8.85g Tg yr−1 7.28g Tg yr−1 (for 2006) 2.33 Tg month−1 (July) 2.35 Tg month−1 (August)

756


NMVOCs from biomass burning). Particularly, isoprene emissions are highly uncertain. For example, isoprene emission fluxes of ~20 Tg yr−1 were reported from the GEIA and POET (Precursors of Ozone and their Effects in the Troposphere) inventories over East Asia, whereas, isoprene emission fluxes of approximately ~15 and ~9 Tg yr−1 were estimated from the MEGAN (Model of Emissions of Gases and Aerosols from Nature) and MOHYCAN (MOdel for HYdrocarbon emissions by the CANopy) inventories, respectively, as shown in Table 1. Based on the above discussions, this study aimed to evaluate the fluxes of biogenic isoprene emissions over East Asia by comparing two tropospheric HCHO columns from SCIAMACHY observations and 3D-CTM simulations with three available biogenic emission inventories. Also, the influences (or propagation) of uncertainty in the biogenic isoprene emission fluxes on the tropospheric HOx cycle over East Asia were then investigated. In order to perform such investigations, the outline of

a)

this manuscript is as follows: the description of the CTM model and the emission inventories used in this study are explained in Sections 2.1 and 2.2. The HCHO retrieval method from the SCIAMACHY observations follows in Section 2.3. In Section 3, the CMAQ-simulated tropospheric HCHO columns were compared with SCIAMACHY-retrieved tropospheric HCHO columns, and the influences of uncertainty in the isoprene emission fluxes on the concentrations of atmospheric species (i.e. oxidants, gases, and particulates) are discussed. The conclusion and summary are presented in Section 4. 2. Experimental methods Three crucial components for this study were: (i) three biogenic isoprene emission inventories, (ii) 3-D Eulerian CTM simulations with anthropogenic emissions, and (iii) tropospheric HCHO columns

d)

A

D C

B

b)

e)

c)

f)

Fig. 1. Isoprene fluxes obtained from the GEIA, MEGAN and MOHYCAN emission inventories for July–August, 2006: a) GEIA for July; b) MEGAN for July; c) MOHYCAN for July; d) GEIA for August; e) MEGAN for August; and f) MOHYCAN for August (Unit: ×102 Ton month−1).


1.0

a1) July NO conc. (ppb)

O3 conc. (ppb)

60

40

20

b1) July

757

Modeling (MOHYCAN) Observations (EANET) Detection Limit (EANET)

0.8 0.6 0.4 0.2 0.0 1.0

0

NO conc. (ppb)

O3 conc. (ppb)

60

a2) August

40

20

b2) August

0.8 0.6 0.4 0.2 0.0

0

HD

3 2 1 0 4

SO2 conc. (ppb)

c1) July

c2) August

3 2 1 0

Ammonium conc. (µg m-3) Ammonium conc. (µg m-3)

16

YS

OK

IJ

HP

SD

TP

RS

d1) July

14 12 10 8 6 4 2 0 16

d2) August

14 12 10 8 6 4 2 0

TR CJ IS HD YS OK PM IJ HP SD TP RS

5

Sulfate conc. (µg m-3)

SO2 conc. (ppb)

4

Sulfate conc. (µg m-3)

CJ KH IS HD YS OK IJ HP SD TP RS

e1) July

CJ IS PM HD YS OK IJ HP SD TP RS

f)

4 3 2 1 0 5

e2) August

4

TR (Terelj); CJ (Cheju); KH (Kanghwa); IS (Imsil); HD (Hedo); YS (Yusuhara); OK (Oki); PM (Primorskaya); IJ (Ijira); HP (Happo); SD (Sado-seki); TP (Tappi); RS (Rishiri);

3 2 1 0 TR CJ IS HD YS OK PM IJ HP SD TP RS

Fig. 2. Comparison between EANET-observed concentrations (red circle) and CMAQ-predicted concentrations (bars) of O3 (a), NO (b), SO2 (c), sulfate (d) and ammonium (e) at 13 EANT monitoring sites (f) in July and August, 2006. Here, detection limits for NO and SO2 with blue lines.

758


retrieved from the SCIAMACHY sensor. In the following sections, the three crucial components will be described in order. 2.1. US EPA CMAQ model simulations The regional-scale US EPA models-3/Community Multi-scale Air Quality (CMAQ) model v4.7.1 was used in this study (Byun and Ching, 1999; Byun and Schere, 2006), in conjunction with the Pennsylvania State University/National Center for Atmospheric Research Meso-scale Model 5 (PSU/NCAR MM5) for meteorological fields. The CMAQ model was driven by the MET fields generated using the PSU/NCAR MM5 model. Two month simulations were carried out over East Asia for July and August, 2006. MM5 modeling was conducted using 2.5° × 2.5° resolved re-analyzed NCEP (National Centers for Environmental Prediction) data, coupled with automated data processing (ADP) of the global surface and upper air observations employing four-dimensional data assimilation (FDDA) techniques (Stauffer and

a)

d)

A

D Beijing

B

Xian

Seaman, 1990). In the MET modeling, the MRF (Medium Range Forecast) and Grell schemes were selected for the planetary boundary layer (PBL) and cumulus parameterization, respectively (Grell et al., 1994; Hong and Pan, 1996). The output data from the MM5 modeling were processed by the Meteorology-Chemistry Interface Processor (MCIP) to generate model-ready MET inputs for the CMAQ model simulations. Here, it should be emphasized that the model-ready MET input data were also utilized in the MEGAN modeling to generate isoprene emission fluxes (this will be discussed further in Section 2.2). For the CMAQ model simulations, the domain covers the areas of East Asia, including Korea, Japan and China, as well as parts of Mongolia and Russia (approximately 95°E to 150°E and 20°N to 50°N), with a grid resolution of 30 × 30 km2 on a Lambert map projection. Fourteen σ-coordinated layers in the vertical direction were considered, with a model-top at 118 hPa. The total number of grid points was 216,734 in the CMAQ model simulations. The schemes selected in the CMAQ simulations were the global massconserving method (YAMO) for advection (Yamartino, 1993), and the

A1

Tianjin

C

Qingdao

Chengdu Chongqing Guangzhou

b)

e)

c)

f)

Fig. 3. Spatial distributions of tropospheric HCHO columns from the CMAQ model simulations (a, d) and SCIAMACHY observations (b, e) in January and December, 2006. Differences between two HCHO columns for January (c) and December (f) were also shown (Unit: ×1015 molecules cm−2).


was applied for China, which was previously developed for the INTEX-B (Intercontinental Chemical Transport Experiment-Phase B) field campaign. The REAS (Regional Emission inventory in ASia) and CAPSS (Clean Air Policy Support System, http://airemiss.nier.go.kr/ main.jsp) emission inventories for the year of 2006 were used for the consideration of the emissions from Japan and Korea, respectively (Ohara et al., 2007; NIER, 2008). Also, in order to take ship emissions into account, the REAS ship emissions were used in this study. The anthropogenic emission inventories have been used for several 3D-CTM simulations for East Asia (e.g., Wang et al., 2010; Han et al., 2011; Park et al., 2011; He et al., 2012). In addition to the AVOC emissions, three biogenic emission inventories for isoprene and monoterpene (the GEIA, MEGAN, and MOHYCAN emission inventories) were used for the CMAQ model simulations in order to assess the magnitudes of biogenic emission fluxes over East Asia, which are summarized in Table 1. First of all, the GEIA inventory estimated global BVOC emissions, with a resolution of 1° × 1°, for the base year of 1990. The GEIA biogenic emissions were obtained from the official GEIA website (http://www.geiacenter.org/ presentData/nvoc.html; Guenther et al., 1995). Although this GEIA inventory is old, it has previously been applied to many 3D-CTM simulations over East Asia, even very recently (e.g., Zhang et al., 2004; Han et al., 2005, 2009; Liu et al., 2007; Chen et al., 2008; Li et al., 2008; Lin et al., 2008, 2009; Song et al., 2008; Yamaji et al., 2008; van Donkelaar et al., 2008; Matsui et al., 2009). Another dataset for the BVOC emissions was prepared from the MEGAN model simulations (Guenther et al., 2006). The MEGAN model requires two types of input information (i.e. meteorological fields and land-cover data). As meteorological information, the temperature and solar radiation data were obtained directly from MCIP processing, using the output data generated from the MM5 modeling (refer to Section 2.1). As the land-cover data, the Leaf Area Index (LAI) from the MODIS (Moderate Resolution Imaging Spectroradiometer) observations was used to determine the vegetation growth. In this study, the vegetation coverage for the MEGAN modeling was derived from the MODIS and then classified into four major categories: i) broadleaf tree, ii) needle leaf tree, iii) shrub lands and iv) herbaceous lands. Lastly, the MOHYCAN emission inventory, with a resolution of 0.5° × 0.5°, was obtained from the official website of the MOHYCAN inventory (http://tropo.aeronomie.

Pleim scheme for dry depositions of both gas and particulate species (Pleim et al., 1996). For gas-phase chemistry, the SAPRC-99 (Statewide Air Pollution Research Center-99) mechanism (Carter, 2000) was partly modified as shown in Reaction 1 (R1) based on the study of Butler et al. (2008), in order to consider missing OH radical processes due to unknown chemical mechanism(s). ISOPO2 þ HO2 →ISOPOOH þ 2OH

759

ðR1Þ

Here, ISOPO2 and ISOPOOH represent isoprene peroxy radical and isoprene peroxide, respectively. Currently, this OH recycling in isoprene chemistry is a controversial issue. Butler et al. (2008) reported that an OH recycling of about 40–50% (compared with 5-10% in current isoprene mechanisms) is necessary for their modeled OH to approach the lower error bounds of OH observed during the GABRIEL (Guyanas Atmosphere-Biosphere exchange and Radicals Intensive Experiment with a Learjet) field campaign. With respect to such uncertainty of the chemical mechanism, detailed analyses were made and then discussed in Section 3.3.2 in terms of the ratio of HO2 to OH. In addition to the gas-phase chemistry, the AERO4 module was selected for aerosol dynamic and thermodynamic calculations (Binkowski and Roselle, 2003). Cloud physics and chemistry were also considered using the sub-grid convective cloud scheme derived from the diagnostic cloud model within RADM (Walcek and Taylor, 1986; Chang et al., 1987, 1990; Dennis et al., 1993). In the sub-grid convective cloud scheme, aqueous chemistry was also turned on, which included the condensation of chemical compounds into cloud droplets. In addition, the CMAQsimulated tropospheric HCHO columns in this study were excluded for the cloudy regions, and were then averaged between 10:00 and 12:00 local time (LT) for synchronization with the tropospheric HCHO columns from the SCIAMACHY sensor, since the local scanning time of the SCIAMACHY sensor is approximately 10:30–11:30 LT over East Asia. 2.2. Emission inventories In the CMAQ model simulations, accurate emissions are an indispensable component for the precise prediction of concentrations of atmospheric species. For anthropogenic species (such as SO2, NOx, CO, AVOCs, BC and OC), the emission inventory from Zhang et al. (2009)

Table 2 Statistical analyses between the CMAQ-modeled and SCIAMACHY-retrieved HCHO columns and average HCHO columns for January and December. Regions

Season

Ω(1) CMAQ

Ω(1) SCIA

R(3)

RMSE(1),(4)

MNGE(2),(5)

PE(2),(6)

MB(1),(7)

MNB(2),(8)

Central East China (CEC)

Jan. n(9) = 315 Dec. n = 325 Jan. n = 192 Dec. n = 189 Jan. n = 648 Dec. n = 665 Jan. n = 83 Dec. n = 75 Jan. n = 193 Dec. n = 230

2.91

3.86

0.75

2.96

215.29

−24.56

−0.95

167.11

2.68

3.27

0.82

2.37

165.47

−17.92

−0.59

119.93

4.68

4.52

1.04

0.40

3.06

3.53

−0.01

1.52

4.56

4.21

1.08

0.49

2.07

8.36

0.03

0.28

5.72

6.75

0.85

4.21

167.55

−15.28

−1.04

129.30

5.40

6.59

0.82

3.78

143.50

−18.15

−1.20

103.27

2.90

3.91

0.74

2.55

122.78

−25.80

−1.01

74.86

3.25

2.51

1.29

2.62

305.10

29.27

7.36

284.92

2.06

2.81

0.73

1.90

293.24

−26.63

−0.75

243.23

2.23

3.73

0.60

2.81

105.87

−40.06

−1.49

40.33

East China (EC)

South China (SC)

South Korea (SK)

Japan (JPN)

Ω

Unit, ×1015 molecules cm−2; (2) Unit, %; (3) R ¼ ΩCMAQ ; SCIA vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u N N X X u Ω 2 CMAQ −ΩSCIA (4) RMSE ¼ tN1 ΩCMAQ −ΩSCIA ; (5) MNGE ¼ N1 100; Ω SCIA 1 1 (1)

100; (9) n: the number of grid data used in the statistical analysis.

N X (6)

PE ¼

1

N X ΩCMAQ − ΩSCIA N X 1

1

ΩSCIA

100;

(7)

MB ¼ N1

N X ΩCMAQ −ΩSCIA ; 1

(8)

MNB ¼ N1

N X ΩCMAQ −ΩSCIA Ω SCIA 1

760


be/models/mohycan.htm, refer to Müller et al., 2008). Müller et al. (2008) estimated the global isoprene emission fluxes based on the MEGAN and MOHYCAN model calculations, which were driven by the meteorological fields (i.e. air temperature, cloud cover, downward solar irradiance, wind speed and volumetric soil moisture) generated from the ECMWF (European Centre for Medium-Range Weather

a)

Forecasts). The MOHYCAN model was used for the calculations of critical parameters: the leaf temperature and visible radiation fluxes as a function of height inside the canopy. The two parameters were then incorporated into the MEGAN model simulations. In this sense, the “MOHYCAN emissions” should be labeled the “MEGAN + MOHYCAN emissions”, because the MOHYCAN model was coupled to the

e) A

D C

B

b)

f)

c)

g)

d)

h)

Fig. 4. Spatial distributions of the tropospheric HCHO columns from the CMAQ model simulations, using the GEIA, MEGAN and MOHYCAN emission inventories, and SCIAMACHY observations for July–August 2006. The tropospheric HCHO columns from the CMAQ simulations: a) and e) using the GEIA emission inventory for July and August; b) and f) using the MEGAN emission inventory for July and August; c) and g) using the MOHYCAN emission inventory for July and August; and d) and h) from the SCIAMACHY observations for July and August (Unit: ×1015 molecules cm−2).


MEGAN model. However, for the sake of convenience, this emission was simply referred to as the MOHYCAN emissions inventory. As shown in Table 1, the isoprene emission fluxes from the GEIA, MEGAN and MOHYCAN inventories were estimated to be 5.07, 2.56 and 2.33 Tg month−1 for July and 4.98, 2.13 and 2.35 Tg month−1 for August, respectively. Fig. 1 shows the spatial distributions of isoprene emissions over East Asia for July and August. As presented in Fig. 1, isoprene emissions were emitted mainly from South China, i.e. southern parts of the Yangtze river (region B in Fig. 1). In order to reduce possible discrepancy between the CMAQcalculated and SCIAMAHCY-derived HCHO levels, the pyrogenic VOC emissions from biomass burning (BB) activities, which were obtained from the GFED (Global Fire Emission Database) v3.0, were also considered in our simulations (van der Werf et al., 2010). 2.3. Tropospheric HCHO columns from the SCIAMAHCY instrument Tropospheric HCHO columns retrieved from the SCIAMACHY observations were used in this study. The HCHO columns were based on slant column retrievals with the DOAS (Differential Optical Absorption Spectroscopy) method. The SCIAMACHY is one of the instruments on the European Space Agency's ENVISAT satellite, which was launched in March 2002. The SCIAMACHY is an imaging spectrometer, which covers the spectral region from 240 to 1700 nm and the region of the short wave infrared between 1900 and 2400 nm, with nadir and limb view capacities and a relatively high spectral resolution of 0.2 to 0.5 nm. In the descending node, the equator was passed at 10:00 local time (LT). Observations at northern mid-latitudes occurred between 10:30 and 11:30 LT over East Asia. The tropospheric HCHO columns data retrieved from the SCIAMACHY sensor have a typical surface spatial resolution of 30 km along track by 60 km across track. Global coverage at the equator is achieved within 6 days. In this study, monthly level products of the tropospheric HCHO columns were obtained for July and August, 2006, from the official website of the TEMIS (Tropospheric Emission Monitoring Internet Service, http://www.temis.nl/airpollution/ch2o.html), hosted by KNMI (Royal Netherlands Meteorological Institute). Tropospheric HCHO columns from the SCIAMACHY observations were retrieved for the products in a three step procedure: (i) retrieval of the slant column density of HCHO, (ii) FRESCO cloud detection and (iii) retrieval of the tropospheric

761

HCHO vertical column density. A complete description of the tropospheric HCHO retrieval has previously been reported by De Smedt et al. (2008, 2010). The uncertainty associated with each SCIAMACHY scene for the HCHO columns retrieved included random errors of 1 × 1016 molecules cm−2 for the slant columns. The AMF error was also a large contributor to the total uncertainty in the HCHO columns retrieved. At low and mid-latitudes, the contributions from the slant column random error and air mass factor error were almost equivalent. The total errors associated with the monthly tropospheric HCHO columns have been reported to range from 20 to 40% (De Smedt et al., 2008). 3. Results and discussions The aim of this study was to evaluate the accuracy of biogenic isoprene emissions during summer by comparing two tropospheric HCHO columns (ΩHCHO) obtained from CMAQ model simulations using three biogenic emission inventories and SCIAMACHY observations (Section 3.2). The propagation of uncertainty in the isoprene emission fluxes to the levels of atmospheric oxidants and the tropospheric gas-aerosol chemistry over East Asia were then investigated (Section 3.3). 3.1. Modeling performance To evaluate the CMAQ modeling performance, surface-level measurement data in East Asia were obtained from the EANET (Acid Deposition Monitoring Network in East Asia). Fig. 2 shows the comparisons of monthly average concentrations of O3, NO, SO2, sulfate (SO2− 4 ) and ammonium (NH+ 4 ) from the CMAQ model simulations and the observations at 13 rural and remote EANET monitoring sites (EANET, 2007). Detection limits of O3, NO, SO2, sulfate (SO2− 4 ) and ammonium are 1 ppb, 0.1 ppb, 0.1 ppb, 0.01 μg m−3 and 0.01 μg m−3, respectively. In general, the CMAQ model reasonably well simulated the concentrations of gaseous and particulate species at the 13 EANET monitoring sites. Although some over-predictions of the O3 concentrations by the CMAQ model simulations were found at several sites, the spatial trend of the O3 concentrations agrees well with those from the EANET data. The ozone titration due to relatively high NO concentrations at the IJ (Ijira) site, was also found from the CMAQ model simulations. As

Table 3 Average tropospheric HCHO columns, their standard deviations and the ratios of the tropospheric HCHO columns (ΩCMAQ) obtained from the CMAQ model simulations to those (ΩSCIA) from the SCIAMACHY sensor for July and August. Regions

Season

Average value of tropospheric HCHO columns(1) (Standard deviation of tropospheric HCHO columns(1)) ΩCMAQ,GEIA

ΩCMAQ,MEGAN

ΩCMAQ,MOHYCAN

ΩSCIA

RGEIA

RMEGAN

RMOHYCAN

A

Jul.

12.15 (5.50) 14.49 (5.55) 22.43 (5.65) 22.81 (5.30) 11.65 (1.44) 13.91 (2.46) 7.94 (1.88) 9.92 (2.41) 10.35 (6.80) 11.36 (6.85)

8.71 (4.02) 9.64 (3.98) 14.14 (3.22) 12.50 (2.65) 8.86 (0.80) 10.75 (1.58) 6.76 (1.43) 8.29 (1.58) 7.76 (4.02) 8.02 (3.40)

8.23 (3.66) 9.47 (3.83) 13.95 (3.06) 13.88 (3.10) 8.94 (0.83) 13.25 (2.77) 6.83 (1.43) 9.10 (2.06) 7.67 (4.01) 8.48 (3.96)

8.85 (4.93) 7.83 (4.09) 9.89 (4.27) 11.42 (3.54) 6.26 (3.96) 6.47 (3.29) 6.48 (3.43) 6.01 (2.69) 6.73 (4.01) 6.68 (3.95)

1.37

0.98

0.93

1.85

1.23

1.21

2.27

1.43

1.41

2.00

1.09

1.22

1.86

1.41

1.43

2.15

1.66

2.05

1.23

1.04

1.05

1.65

1.38

1.51

1.54

1.15

1.14

1.70

1.20

1.27

(CEC, Central East China) Aug. B

Jul. (SC, South China) Aug.

C

Jul. (SK, South Korea) Aug.

D

Jul. (JPN, Japan) Aug.

Entire domain

Jul. Aug.

(1)

Unit, ×1015 molecules cm−2;

(2)

R¼

ΩCMAQ ΩSCIA .

Ratio of ΩCMAQ to Ω(2) SCIA

762


shown in Fig. 2(c), the relatively high over-predictions of SO2 at the IJ (Ijira) site were also reported in the previous study of the MICS-Asia Phase I (Model Intercomparison Study Asia Phase I; Carmichael et al., 2008). 3.2. Evaluation of the accuracy of the biogenic isoprene emissions over East Asia 3.2.1. Comparison between two tropospheric HCHO columns obtained from the CMAQ model simulations and the SCIAMAHCY sensor Prior to evaluation of the accuracy of the biogenic isoprene emissions, we wish to provide information on the accuracy of the AVOC emission fluxes over East Asia through the comparisons between the CMAQ-calculated and SCIAMACHY-derived HCHO columns during the winter episode. For this analysis, additional CMAQ model simulations using the same AVOC emission fluxes used in the summerepisode simulations were carried out over the same modeling domain for January and December, 2006. During these winter months, the effects of biogenic isoprene on the levels of the HCHO columns

can be almost negligible, because of inactive emissions of biogenic isoprene in winter over East Asia. For example, in the MOHYCAN emission inventory used in this study, less than 1.5% of annual isoprene emissions were only emitted in January and December, 2006, whereas 64% of annual isoprene emissions were emitted from June to August, 2006. Fig. 3 shows the spatial distributions of tropospheric HCHO columns obtained from the CMAQ model simulation (ΩCMAQ) and from the SCIAMACHY observations (ΩSCIA), together with the differences between the two HCHO columns. For this analysis, five regions were defined: (i) Region A for Central East China (CEC), (ii) Region A1 for East China (EC), (iii) Region B for South China (SC), (iv) Region C for South Korea (SK), and (v) Region D for Japan (JPN), as shown in Fig. 3. The average tropospheric HCHO columns, the ratio (R) of the average ΩCMAQ to the average ΩSCIA, and other statistical analyses over the five regions were summarized in Table 2. As shown in Fig. 3 and Table 2, although ΩCMAQ were slightly smaller than ΩSCIA, particularly around the Xian, Chengdu, and Chongqing areas due to possible underestimations of the AVOCs emissions, the CMAQ-estimated tropospheric HCHO columns successfully captured

a)

d)

b)

e)

c)

f)

Fig. 5. Differences between the CMAQ-predicted HCHO columns (ΩCMAQ) and SCIAMACHY-retrieved HCHO columns (ΩSCIA) over East Asia: a) and d) ΩCMAQ,GEIA–ΩSCIA for July and August; b) and e) ΩCMAQ,MEGAN–ΩSCIA for July and August; c) and f) ΩCMAQ,MOHYCAN–ΩSCIAE for July and August, respectively (Unit: ×1015 molecules cm−2).


the magnitudes of tropospheric HCHO columns obtained from the SCIAMACHY observations, particularly over East China (EC, Region A1), in which the large amounts of AVOCs were mainly emitted. Also, some negative biases over South China (SC) were possibly caused by the combined uncertainties due to the AVOC emissions and biomass burning events. It is therefore thought that from this evaluation study, the BVOC emission analysis during the summer episode would not be hampered greatly by the uncertainty in the AVOC emissions. The spatial distributions of tropospheric HCHO columns obtained from the CMAQ model simulations using three biogenic emission inventories and the SCIAMAHCY observations for July and August, 2006, are shown in Fig. 4. The tropospheric HCHO columns obtained from the CMAQ model simulations appeared to be spatially consistent with the tropospheric HCHO columns retrieved from the SCIAMACHY sensor, as shown in Fig. 4. The ΩCMAQ and ΩSCIA exhibited their largest values over South China during the two summer months. In particular, the maximum tropospheric HCHO columns (ΩCMAQ,GEIA) from the CMAQ model simulation using the GEIA isoprene emission reached 3.68 × 1016 molecules cm−2 over Southern China. The average ΩCMAQ,GEIA, ΩCMAQ,MEGAN, and ΩCMAQ,MOHYCAN for July and August ranged between 2.24 × 1016 and 2.28 × 1016, 1.25 × 1016 and 1.41 × 1016, and 1.39 × 1016 and 1.40 × 1016 molecules cm−2, respectively, over South China. As expected, the tropospheric HCHO columns (Ω) decrease with reducing isoprene emissions fluxes from the GEIA to MEGAN and/or MOHYCAN inventories. Similar trends were also observed for the tropospheric HCHO columns over CEC, South Korea, and Japan. The average tropospheric HCHO columns over the four regions and for the entire domain are summarized in Table 3. As indicated in Fig. 4 and Table 3, the ΩCMAQ,MEGAN and ΩCMAQ,MOHYCAN showed more consistent values with the ΩSCIA over East Asia. The ratios of the average ΩCMAQ,MEGAN and ΩCMAQ, MOHYCAN to the average ΩSCIA over the entire domain were roughly close to 1. Despite the ideal value over the entire domain with the MEGAN and MOHYCAN emission inventories, the ratios were still larger over South China (approximately 1.4 for July), over Japan (1.38–1.51 for August), and particularly over

763

South Korea (approximately 1.4 for July and 1.66–2.05 for August). On the other hand, when the average ΩSCIA was compared with the ΩCMAQ,GEIA, the ratio increased to 1.54 for July and 1.70 for August over the entire domain, and was even larger than ~2.90 over some parts of South China. Overall, our analysis suggested that the CMAQ model simulation using the GEIA isoprene emission, could produce over-predicted values of the tropospheric HCHO columns over East Asia. The differences between the ΩCMAQ and ΩSCIA were analyzed further over the four regions in Fig. 5. The difference between ΩCMAQ, GEIA and ΩSCIA exhibited large positive values over East Asia, particularly over Southern China, where the differences increase up to 2.89 × 1016 molecules cm−2. In contrast, for the cases using the MEGAN and/or MOHYCAN inventories in the CMAQ model simulations, the differences between ΩCMAQ, MEGAN (or ΩCMAQ, MOHYCAN) and ΩSCIA became smaller and showed negative values in some regions. The results were in line with the ratio analyses as shown in Table 3. In Fig. 5, these negative values around the industrial areas such as Xian, Chengdu, Chongqing and Guangzhou were also found in the comparisons between the CMAQ-calculated and SCIAMACHY-retrieved HCHO columns during the winter episode, as shown in Fig. 3. This spatial consistency could be caused by the possible underestimations of the AVOCs emissions around these regions. Collectively, Figs. 4, 5, and Table 3 suggest that the use of the MEGAN and MOHYCAN emission in the CMAQ model simulations leads to a better agreement with the SCIAMACHY observations over East Asia during the two summer episodes, although two HCHO columns using the MEGAN and MOHYCAN emissions produce overestimated values over some regions, particularly over South Korea. In addition, this study again confirmed that the isoprene emissions obtained from the GEIA emission inventory were actually overestimated over East Asia, as found by Steiner et al. (2002), Han et al. (2009), and Stavrakou et al. (2009). Although uncertainty in the biogenic isoprene emissions could be the major cause of the discrepancies between ΩCMAQ and ΩSCIA over East Asia, they could also have resulted from several other

Table 4 Statistical analyses between the CMAQ-modeled and SCIAMACHY-retrieved tropospheric HCHO columns over East Asia for July and August. Regions

Season

Inventory

RMSE(1),(6)

MNGE(2),(7)

PE(3),(7)

MB(4),(6)

MNB(5),(7)


Jul. n(8) = 360

GEIA MEGAN MOHYCAN GEIA MEGAN MOHYCAN GEIA MEGAN MOHYCAN GEIA MEGAN MOHYCAN GEIA MEGAN MOHYCAN GEIA MEGAN MOHYCAN GEIA MEGAN MOHYCAN GEIA MEGAN MOHYCAN

5.64 4.12 4.08 8.06 4.10 3.94 14.13 6.49 6.26 12.64 3.97 4.59 6.66 4.70 4.71 8.15 5.08 7.69 3.70 3.29 3.31 4.91 3.51 4.22

135.02 78.66 75.24 206.92 112.41 109.53 236.30 123.45 120.67 119.62 34.19 41.40 183.14 123.74 124.81 170.59 108.87 160.61 96.11 77.72 78.63 158.80 123.89 142.11

37.21 -1.65 -7.05 85.15 23.13 21.03 126.73 42.96 41.04 99.71 9.40 21.56 85.97 41.39 42.65 114.85 65.99 104.65 23.46 5.08 6.11 65.06 37.87 51.42

3.29 -0.15 -0.62 6.66 1.81 1.65 12.53 4.24 4.05 11.39 1.07 2.46 5.39 2.59 2.67 7.44 4.27 6.78 1.46 0.28 0.35 3.91 2.28 3.09

125.73 51.89 44.90 205.12 97.91 94.85 235.44 116.77 113.79 118.73 20.39 33.38 180.58 114.39 116.31 170.59 107.73 160.37 82.69 56.10 57.57 156.15 118.00 137.80

Aug. n = 365 South China (SC)

Jul. n = 692 Aug. n = 740

South Korea (SK)

Jul. n = 92 Aug. n = 100

Japan (JPN)

Jul. n = 264 Aug. n = 276

(1)

N N X X vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ΩCMAQ − ΩSCIA u N N N N X X X u X Ω −Ω ΩCMAQ −ΩSCIA 2 CMAQ SCIA (2) (3) 1 RMSE ¼ tN1 ΩCMAQ −ΩSCIA ; MNGE ¼ N1 PE ¼ 1 100; (4) MB ¼ N1 ΩCMAQ −ΩSCIA ; (5) MNB ¼ N1 100; N X Ω Ω SCIA SCIA 1 1 1 1 ΩSCIA

100;

(6)

Unit, ×1015 molecules cm−2; (7) Unit, %;

1 (8)

n: the number of grid data used in the statistical analysis.

764


factors, such as: (i) uncertainties in the AVOC and other BVOC (e.g., monoterpenes and sesquiterpenes) emission fluxes; (ii) possible seasonal variations in the AVOC emission fluxes; (iii) BB emissions; (iv) retrieval algorithm errors for the tropospheric HCHO columns from the SCIAMACHY sensor. Regarding the last factor, De Smedt et al. (2008) reported that the tropospheric HCHO columns retrieved through the KNMI algorithm were about ~ 20% lower than the dataset from Harvard University (Fu et al., 2007) over South China during summer. As for the first factor, several studies have reported that the impacts of the other BVOC species are not large. For example, Palmer et al. (2006) showed that monoterpene and methylbutenol (MBO) tend to make only small contributions to the tropospheric HCHO columns obtained from the GOME sensor over North America. 3.2.2. Statistical analysis In order to assess the accuracy of the three biogenic isoprene emission inventories through tropospheric HCHO columns, five statistical parameters were introduced: i) the Root Mean Square Error (RMSE, absolute error), ii) the Mean Normalized Gross Error (MNGE, relative error), iii) the Percent Error (PE), iv) the Mean Bias (MB, absolute bias), and v) the Mean Normalized Bias (MNB, relative

bias). The definitions of these five parameters used in the statistical analyses are shown in the footnote to Tables 2 and 4. As shown in Table 4, the use of the MEGAN and MOHYCAN inventories in the CMAQ model simulations decreased the RMSE, MNGE, PE, MB, and MNB, compared to the use of the GEIA inventory. For example, it was clearly seen that the MBs over South China for August decreased by factors of ~ 0.09 and ~ 0.22 from GIEA to MEGAN and from GEIA to MOHYCAN, respectively. Also, when the isoprene emission fluxes of the GEIA, MEGAN, and MOHYCAN emission inventories were used for July, the percentage error (PEs) were evaluated to be 127, 43, and 41% over South China, respectively. In addition, statistical parameters for the MEGAN and MOHYCAN cases showed almost similar values except for August over South Korea and Japan. These exceptions were also consistent with the ratio analyses, which are shown in Table 3. For these exceptions, the use of the MEGAN inventory in the CMAQ model simulation showed better agreement with the SCIAMACHY observations than the use of the MOHYCAN inventory. The statistical analyses again confirmed that the GEIA isoprene emissions are overestimated over East Asia, particularly over South China, whereas there appear to be relatively small errors and weak biases in the MEGAN and MOHYCAN inventories.

a1)

a2)

a3)

b1)

b2)

b3)

c1)

c2)

c3)

Fig. 6. Spatial distributions of isoprene, mono-terpene, and HCHO at the surface obtained from the CMAQ model simulations: a1), b1) and c1) using GEIA; a2), b2) and c2) using MEGAN; a3), b3) and c3) using MOHYCN emission inventories, respectively.


3.3. Impacts of biogenic isoprene emission fluxes on tropospheric chemistry

also an important species in tropospheric chemistry, particularly for the formation of OH from the photolysis of ozone (i.e. Reactions 3 and 4).

3.3.1. Isoprene, monoterpen, and formaldehyde (HCHO) Isoprene and mono-terpene (TRP1) are the main precursors of tropospheric HCHO, particularly during summer. Fig. 6 shows the spatial distributions of the isoprene, monoterpene, and HCHO mixing ratios at the surface obtained from the CMAQ model simulations using the GEIA, MEGAN and MOHYCAN isoprene emissions. As expected, the mixing ratios of two biogenic VOC species, isoprene and monoterpene at the surface were spatially consistent with the isoprene emission fluxes shown in Fig. 1, as well as the HCHO mixing ratios shown in Fig. 6. The HCHO mixing ratios decrease with reducing isoprene emissions (Table 5). Also, the average HCHO mixing ratio over South China in the case of the GEIA isoprene emissions were larger than those in the cases of the MEGAN and MOHYCAN by factors of 2.06 and 1.92, respectively. 3.3.2. Atmospheric oxidants: ozone and HOy Ozone is formed by photochemical reactions involving the oxidations of AVOCs and BVOCs in the presence of NOx and sunlight: NOx þ ðA=BÞVOCs þ hv→O3

765

ðR2Þ

Therefore, high ozone concentrations would be expected over CEC and South Korea (Over Southern China (B1) defined in Fig. 7(a) as a blue dashed line, the BVOC emissions were large, but the NOx emissions were small). As shown in Fig. 7(a), the ozone mixing ratios were relatively high over CEC and South Korea, where large amounts of NOx were also emitted. However, the average ozone mixing ratios decrease by ~ 4 ppbv and ~ 3 ppbv as the biogenic isoprene emissions from the GEIA to the MEGAN and/or MOHYCAN emission inventories decreased over CEC and South Korea, respectively. There were only slight differences in the ozone mixing ratios over Southern China (B1), where a NOx-limited environment would be created. These results indicate that the formation of ozone over Southern China would be controlled by NOx levels, not by those of isoprene during summer (this will be confirmed partly in Fig. 7(d) again). Ozone is

1D

O3 þ hv→O 1D

O

þ O2

ðR3Þ

þ H2 O→2OH

ðR4Þ

Fig. 7(b,c,d) shows the spatial distribution of HOy (HOy ≡ OH + HO2 + H2O2) over East Asia. Two molecules of hydroxyl radicals (OH) are produced from the photolysis of ozone at wavelengths between 305 and 320 nm. Therefore, high levels of OH radicals would be expected over areas where the levels of ozone are high. However, as shown in the second row of Fig. 7, the levels of OH radicals did not follow those of ozone shown in Fig. 7(a). In other words, the OH levels over CEC were increased from the GEIA case to the MEGAN and MOHYCAN cases, because the OH radicals formed via the reaction of O1D + H2O would be consumed due to isoprene oxidation. The levels of OH radicals increased with decreasing strength of the isoprene emissions. Interestingly, some high levels of OH radicals were found over the Northwestern Pacific ocean. These OH distributions are caused by the ship NOx emissions over the ocean areas, which would create a NOx-abundant environment in the ship-going marine boundary layer (MBL). The ship-emitted NOx can create a shift in the HOx cycle, which is in favor of OH generation via the active HO2 + NO reaction under the NOx-abundant environment. Such elevations of the OH levels affected by the ship NOx emissions were also analyzed by Lawrence and Crutzen (1999) and Kim et al. (2009). Also, some high levels of OH radicals were found within the marine and planetary boundary layers (i.e. northwestern Pacific marine boundary layer and planetary boundary layer (PBL) on Tibetan Plateau). These high levels of OH are possibly caused by strong solar radiation (in cases of Pacific ocean MBL and Tibetan Plateau PBL) and/or by high relative humidity (in case of Pacific ocean MBL). OH radicals are converted to perhydroxyl radicals (HO2) or organic peroxyl radicals (RO2) through their reactions with isoprene. As shown in Fig. 7, the levels of HO2 radicals were inversely related to

Table 5 Average concentrations of atmospheric species at the surface obtained from the CMAQ model simulations using three biogenic emission inventories. Species

Unit

Isoprene

ppbv

TRP1

ppbv

HCHO

ppbv

O3

ppbv

OH

pptv

HO2

pptv

H2O2

ppbv

Nitrates

μg m−3

Sulfates

μg m−3

Biogenic SOA

μg m−3

PAN

ppbv

NO2

ppbv

a


South China (SC)

GEIA

MEGAN

MOHYCAN

GEIA

MEGAN

MOHYCAN

GEIA

MEGAN

MOHYCAN

GEIA

MEGAN

MOHYCAN

1.08 (0.47)a 0.07 (0.04) 3.28 (1.27) 51.66 (7.61) 0.18 (0.05) 18.39 (4.15) 1.24 (0.31) 9.46 (6.91) 14.26 (5.42) 0.67 (0.32) 1.00 (0.47) 3.16 (3.25)

0.34 (0.29) 0.02 (0.02) 1.96 (0.91) 47.63 (5.08) 0.25 (0.05) 15.44 (3.14) 1.06 (0.16) 12.77 (8.77) 14.59 (5.50) 0.24 (0.16) 0.56 (0.24) 3.47 (3.38)

0.24 (0.23) 0.02 (0.02) 1.81 (0.80) 47.07 (4.68) 0.26 (0.05) 15.03 (3.37) 1.05 (0.15) 13.14 (9.14) 14.56 (5.48) 0.22 (0.13) 0.51 (0.22) 3.52 (3.39)

5.23 (2.69) 0.37 (0.22) 7.00 (1.79) 48.36 (6.77) 0.12 (0.06) 25.21 (5.30) 2.08 (0.44) 7.93 (4.12) 15.33 (4.72) 2.13 (0.90) 1.17 (0.35) 2.95 (2.99)

1.14 (0.57) 0.15 (0.08) 3.40 (0.86) 45.39 (5.57) 0.21 (0.05) 18.83 (4.94) 1.66 (0.30) 12.12 (5.53) 16.08 (5.15) 0.99 (0.38) 0.71 (0.18) 3.55 (3.00)

1.41 (0.85) 0.11 (0.07) 3.64 (1.06) 45.61 (5.55) 0.21 (0.07) 19.29 (5.10) 1.69 (0.30) 11.93 (5.71) 16.00 (5.07) 0.71 (0.42) 0.74 (0.21) 3.53 (3.00)

1.01 (0.98) 0.09 (0.09) 2.53 (1.25) 49.82 (6.84) 0.18 (0.05) 14.12 (3.96) 1.24 (0.13) 11.04 (3.89) 8.07 (2.47) 0.32 (0.20) 0.83 (0.33) 5.85 (7.83)

0.41 (0.33) 0.05 (0.05) 1.70 (1.11) 44.59 (6.28) 0.21 (0.05) 11.93 (3.87) 1.11 (0.14) 12.48 (4.19) 8.10 (2.45) 0.16 (0.08) 0.50 (0.25) 5.94 (7.26)

0.72 (0.71) 0.06 (0.07) 2.16 (1.21) 46.55 (6.45) 0.19 (0.05) 13.23 (3.93) 1.18 (0.12) 11.93 (4.04) 8.11 (2.46) 0.20 (0.13) 0.64 (0.30) 5.85 (7.46)

0.92 (0.63) 0.13 (0.09) 1.91 (0.83) 42.48 (3.80) 0.19 (0.06) 14.74 (2.46) 1.18 (0.11) 4.74 (2.05) 4.62 (0.97) 0.34 (0.23) 0.53 (0.23) 2.20 (2.18)

0.47 (0.27) 0.08 (0.05) 1.39 (0.48) 39.90 (2.75) 0.22 (0.07) 12.97 (2.32) 1.09 (0.09) 5.64 (2.38) 4.65 (0.97) 0.17 (0.07) 0.36 (0.13) 2.36 (2.28)

0.58 (0.43) 0.10 (0.12) 1.56 (0.60) 40.62 (2.96) 0.21 (0.07) 13.71 (2.32) 1.13 (0.10) 5.30 (2.27) 4.64 (0.97) 0.27 (0.23) 0.42 (0.16) 2.31 (2.26)

Average (±standard deviation).

South Korea (SK)

Japan (JPN)

766


a1)

a2)

a3)

b1)

b2)

b3)

c1)

c2)

c3)

d1)

d2)

d3)

B1

Fig. 7. Spatial distributions of ozone and HOy (OH + HO2 + H2O2) at the surface obtained from the CMAQ model simulations: a1), b1), c1) and, d1) using GEIA; a2), b2), c2), and d2) using MEGAN; a3), b3), c3), and d3) using MOHYCN emission inventories, respectively.

those of the OH radical mixing ratios. Also, HO2 and RO2 are removed via the formation reactions of hydrogen peroxide (H2O2) and organic hydroperoxide (ROOH). This pathway is also illustrated as thick arrows in Fig. 8. Therefore, the H2O2 mixing ratios, shown in the last row of Fig. 7, decreased with reducing isoprene emission fluxes. Also, it should be noted that the active H2O2 formation over South China (Region B) would be further evidence that this region is likely to have a NOx-limited environment, as illustrated in Fig. 8. Since isoprene can create a shift in the HOx cycle during summer, the ratios of

HO2 to OH can be determined by the following Eq. (1), under pseudo steady-state conditions (Stevens et al., 1997): ½HO2 k1 ½CO þ k2 ½O3 þ k3 ½CH4 þ k4 ½HCHO þ k5 ½ISOP þ k6 ½VOCs ð1Þ ≡ ½OH k7 ½NO þ k8 ½O3 where k1, k2, k3, k4, k5, k6, k7, and k8 (cm3 molecules−1 s−1) are the reaction constants for OH + CO, OH + O3, OH + CH4, OH + HCHO, OH + ISOP (isoprene), OH + VOCs, HO2 + NO and HO2 + O3,


Primary production

O3 H 2O

O*

HONO hv

OH

HCHO

HO2 RO2

NO

OH hv

O3 NO 2

respectively. Fig. 9 represents how HOx responds to variations in the levels of NO and isoprene. The ratios decrease with increasing the levels of NO due to the conversion of HO2 to OH via the reaction of HO2 with NO (i.e. HO2 + NO → OH + NO2). The results show that under an isoprene-abundant environment (e.g., GEIA case), the maximum [HO2]/[OH] ratios reach ~740, particularly over South China (a low NOx region), whereas the maximum [HO2]/[OH] ratios decrease to ~320 over South China under less isoprene-rich conditions (e.g., in the MEGAN case). The average [HO2]/[OH] ratios over CEC, South China, Japan and South Korea were 142.55 ± 34.08, 370.32 ± 137.80, 117.25 ± 30.46 and 100.28 ± 28.83 for the GEIA case, 79.34 ± 21.05, 138.30 ± 59.78, 82.88 ± 27.35 and 67.16 ± 20.46 for the MEGAN case and 73.20 ± 19.87, 154.18 ± 70.72, 93.20 ± 26.20 and 84.74 ± 25.71 for the MOHYCAN case, respectively. It was clearly shown that strong biogenic isoprene emissions tended to enhance the [HO2]/[OH] ratios by a factor of 2.7 from the GEIA to the MEGAN cases over South China. Such enhancements of the [HO2]/[OH] ratios due to increased biogenic isoprene mixing ratios have previously been reported in several other studies conducted for other forest regions (Stevens et al., 1997; Carslaw et al., 2001; Kanaya et al., 2001; Kubistin et al., 2010).

HCHO

Isoprene

SO2 SO2

HNO3 H2SO4

SO2

H2O2, ROOH Anth-/Bio-NMVOCs

NO3- SO42-

SO42-

SOAs

Aerosol & cloud droplet Fig. 8. An illustration of the simplified HOx/RO2–NOx–BVOCs–aerosol photochemistry.

[HO2]/[OH] vs. [ISOP]

[HO2]/[OH] vs. [NO]

[HO2]/[OH]

1000

[HO2]/[OH]

b)

GEIA

d)

MEGAN

SC CEC JPN SK

100

10 1000

c)

MEGAN

e)

MOHYCAN

100

10 1000

[HO2]/[OH]

GEIA

a)

767

MOHYCAN

f)

100

10

0

5

10

NO (ppbv)

15

20 0

5

10

15

20

Isoprene (ppbv)

Fig. 9. [HO2]/[OH] vs. [NO] in the first column, and [HO2]/[OH] vs. [Isoprene] in the second from the CMAQ model simulations over SC (red open circle), CEC (blue cross), JPN (sky-blue open triangle), and SK (green open square), using three different biogenic emission inventories: a) and b) GEIA; c) and d) MEGAN; and e) and f) MOHYCAN emission inventories.

768


Observation Modeling (w/o modification) Modleing (w/ modification)

[HO2]/[OH]

Previous studies

This study

1000

100

10 AEROBIC97

ORION99

PROPHET

PRD

GEIA

MEGAN

MOHYCAN

1)

2)

3)

4)

5)

6)

7)

Fig. 10. Maximum and minimum ratios of [HO2]/[OH] from several studies conducted over forest regions during summer episodes. These works were conducted: (1) over Northwestern Greece during the AEROBIC 97 campaign for July and August 1997 (Carslaw et al., 2001); (2) over Okinawa, Japan, during the ORION99 campaign for August, 1999 (Kanaya et al., 2001); (3) over North Michigan, USA, during the PROPHET campaign for August, 1998 (Tan et al., 2001); and (4) over Pearl River Delta (PRD), China, for July, 2006 (Hofzumahaus et al., 2009). For comparative purposes, the [HO2]/[OH] ratios from this study were obtained only over South China (large forest regions) from the CMAQ model simulations without (“w/o modification” case) and with (“w/modification” case) modified isoprene chemistry in SAPRC99 mechanism, using: (5) GEIA; (6) MEGAN; and (7) MOHYCAN emission inventories. Here, the data in the top and bottom 5th percentiles from our results, were excluded in this analysis.

In order to evaluate approximate magnitudes of the [HO2]/[OH] ratios influenced by such enhancements of BVOC emissions, we compared the modeled [HO2]/[OH] ratios over South China (Regions B) with those from previous studies conducted for other forest regions as shown in Fig. 10. In the previous studies, the measured [HO2]/[OH] ratios ranged from several tens to ~100, whereas, the model-simulated ratios were up to several hundreds in Fig. 10. Regarding the persistent over-predictions in the [HO2]/[OH] ratios by the model simulations, several studies have suggested that such discrepancies would possibly be caused by missing OH radicals in unknown chemical mechanism (i.e. OH recycling or OH regeneration) (Carslaw et al., 2001; Kanaya et al., 2001; Tan et al., 2001; Thornton et al., 2002; Hasson et al., 2004; Butler et al., 2008; Hofzumahaus et al., 2009; Peeters et al., 2009; Kubistin et al., 2010; Peeters and Müller, 2010). In order to investigate the impacts of OH recycling, two separate CMAQ model simulations were carried out in Fig. 10 with and without the modification of the OH recycling, which were discussed in Section 2.1 (check the cases of “w/o modification” and “w/modification” in Fig. 10). As shown in Fig. 10, the [HO2]/[OH] without OH recycling in the CMAQ model simulations (“w/o modification”), calculated over South China for the GEIA case, were far larger than those reported from Pearl River Delta (PRD) in the southern part of China, and were also large compared to those measured and modeled for other forest regions (Carslaw et al., 2001; Kanaya et al., 2001; Tan et al., 2001; Hofzumahaus et al., 2009). From the large over-predictions in the [HO2]/[OH] ratios shown in Fig. 10, the possible overestimation of the GEIA isoprene emissions can be inferred again. On the other hand, the ratios of HO2 to OH for the MEGAN and MOHYCAN cases were relatively close to the measured ratios from the previous studies. Additionally, considering OH recycling (i.e. “w/modification”) based on the approach of Butler et al. (2008) improved the model performances in terms of the ratios of HO2 to OH, as indicated in Fig. 10.

MOHYCAN emission cases (also, refer to Table 5). This is undoubtedly because the OH radical mixing ratios in HOy increase by ~ 40% over CEC and ~ 75% over South China (shown in Table 5) with reducing isoprene emissions from the GEIA to the MEGAN and MOHYCAN emission cases. In addition to OH radicals, NO2 has a contribution to the increase of the nitrate levels. In this study, the levels of NO2 increased by ~ 10% over CEC and ~ 20% over South China, as the biogenic isoprene emissions decreased from the GEIA to the MEGAN and/or MOHYCAN emissions, as shown in Table 5. High levels of HNO3 were formed via the reaction of high levels of OH radicals with NO2, where HNO3 is then condensed into the particulate phase. These results suggested that the rates of NOx chemical loss could be highly influenced by biogenic isoprene emissions (Han et al., 2009). In the second row of Fig. 11, the sulfate concentrations over the four regions showed similar trends with those for nitrates. Here, OH radicals also have important role to produce sulfuric acid and sequentially sulfate. In the third and fourth rows of Fig. 11, BSOA and PAN decreased rapidly from the GEIA to MEGAN and MOHYCAN cases, particularly over South China due to the decrease of biogenic emission fluxes. Overall, these examples clearly indicate that changes in the partitioning of HOy (and/or in the BVOC emissions) could greatly influence (and perturb) the tropospheric chemistry over East Asia during summer. Therefore, the use of the correct/accurate BVOC emissions in the CTM model simulations is of critical importance. In addition to the air quality issue in East Asia, isoprene emissions can also greatly influence climate forcing over East Asia, since different isoprene emissions can affect the concentrations of ozone, nitrate, SOAs and other particulate species, as indicated in this section. The different levels of OH radicals can also affect the oxidation rate of CH4. All the gaseous and particulate species mentioned here are important climate forcers in the atmosphere (IPCC, 2007). 4. Summary and conclusions

3.3.3. Other atmospheric particulate species In Section 3.3.2, the partitioning of HOy species varies greatly, and is influenced by the strength of the isoprene emissions over East Asia. Therefore, in this section, the impacts of partitioning of HOy on the levels of several atmospheric particulate species are examined briefly. Fig. 11 shows some examples: the spatial distributions of nitrate 2− (NO− 3 ), sulfate (SO4 ), biogenic secondary organic aerosols (BSOAs) and PAN at the surface level from the CMAQ model simulations using the GEIA, MEGAN, and MOHYCAN emissions. The nitrate concentrations in the first row of Fig. 11 increased with reducing isoprene emissions from the GEIA to the MEGAN and

The isoprene emission fluxes over East Asia are highly uncertain. For example, the highly variable isoprene emission fluxes of 5.07, 2.56 and 2.33 Tg month−1 for July, 2006, were estimated using GEIA, MEGAN and MOHYCAN emission inventories, respectively. Therefore, this study attempted to evaluate the accuracy of biogenic emission fluxes during summer over East Asia by comparing two tropospheric HCHO columns obtained from the CMAQ v4.7.1 model simulations, using three BVOC emission inventories and SCIAMACHY observations. From the comparative study between the two tropospheric HCHO columns, the tropospheric HCHO columns obtained from the CMAQ


a1)

a2)

a3)

b1)

b2)

b3)

c1)

c2)

c3)

d1)

d2)

d3)

769

Fig. 11. Spatial distributions of nitrate, sulfate, BSOA, and PAN at the surface, obtained from the CMAQ model simulations: a1), b1), c1), and d1) using GEIA; a2), b2), c2), and d2) using MEGAN; a3), b3), c3) and d3) using MOHYCN emission inventories, respectively.

model simulation using the MEGAN and MOHYCAN emission inventories was revealed to best agree with those from the SCIAMACHY observations. Statistical analyses also showed that the magnitudes of the RMSE, MNGE, PE, MB and MNB became smaller with decreases in BVOC emission fluxes obtained from the GEIA to the MEGAN and MOHYCAN cases over CEC, South China, South Korea and Japan. The ratios of the average ΩCMAQ,MEGAN and ΩCMAQ,MOHYCAN to the average ΩSCIA were close to 1 over the entire domain. However, the ratios of ΩCMAQ to ΩSCIA increased by between 1.54–1.70 over East Asia when

the GEIA isoprene emission was used in the CMAQ model simulations. This indicates that isoprene emission fluxes of the GEIA inventories may be overestimated. The possible impacts of the uncertainty in the isoprene emissions on the levels of atmospheric oxidants and the cycle of HOy during summer over East Asia were also investigated. As the biogenic isoprene emission fluxes decrease from the GEIA to the MEGAN and MOHYCAN emissions, the levels of OH radicals were increased by factors of 1.75 over South China, and accordingly, the productions of

770


both HO2 (or RO2) and H2O2 became active. In the analyses of ratios of [HO2]/[OH], high ratios were generally found when the levels of NO and isoprene were low and high, respectively. Also, strong BVOC emissions tended to enhance the [HO2]/[OH] ratios by a factor of 2.7, when the GEIA case was compared to the MEGAN case. Such changes in the cycle of HOy can also affect tropospheric chemistry during summer over East Asia. The decreasing isoprene emission fluxes between the GEIA to the MEGAN and MOHYCAN emissions 2− enhanced the levels of nitrate (NO− 3 ) and sulfate (SO4 ), particularly over CEC, but decreased the mixing ratios of BSOAs and PAN over South China. It is expected that this study will enhance our understanding of uncertainty in the isoprene emissions fluxes over East Asia, and their impacts on the tropospheric chemistry. In the future, it would be desirable to conduct one-year-long inverse model simulations (i.e. top-down modeling) to confirm and/or provide an alternative estimate of biogenic isoprene emission fluxes over East Asia. More accurate BVOC emissions would undoubtedly provide more powerful tools for the analyses and forecasting of air quality and climate forcing over East Asia.

Acknowledgments This research was supported by the GEMS program of the Ministry of Environment, Korea and the Eco Innovation Program of KEITI (2012000160004). Also, this work was funded by the Eco-Innovation project 411-113-013 from the Korean Ministry of Environment and partly supported by the Korea Meteorological Administration Research and Development Program under Grant CATER 2012-7110.

References Abbot D, Palmer PI, Martin RV, Chance KV, Jacob DJ, Guenther A. Seasonal and interannual variability of North American isoprene emissions as determined by formaldehyde column measurements from space. Geophys Res Lett 2003;30: 1886. http://dx.doi.org/10.1029/2003GL017336. Binkowski FS, Roselle SJ. Models-3 community multi-scale air quality (CMAQ) model aerosol components: 1. model description. J Geophys Res 2003;108(D6):4183. http://dx.doi.org/10.1029/2001JD001409. Butler TM, Taraborrelli D, Brühl C, Fischer H, Harder H, Martinez M, et al. Improved simulation of isoprene oxidation chemistry with the ECHAM5/MESSy chemistryclimate model: lessons from the GABRIEL airborne field campaign. Atmos Chem Phys 2008;8:4529–46. http://dx.doi.org/10.5194/acp-8-4529-2008. Byun DW, Ching JKS. Science algorithms of the EPA models-3 Community Multiscale Air Quality (CMAQ) modeling system. EPA/600/R-99/030. Research Triangle Park, USA: US EPA; 1999. Byun DW, Schere KL. Review of the governing equations, computational algorithm, and other components of the Models-3 Community Multi-scale Air Quality (CMAQ) Modeling system. Appl Mech Rev 2006;59(2):51–77. Carmichael GR, Sakurai T, Streets D, Hozumi Y, Ueda H, Park SU, et al. MICS-Asia II: The model intercomparison study for Asia Phase II methodology and overview of findings. Atmos Environ 2008;42:3468–90. Carslaw NC, Creasey DJ, Harrison D, Heard DE, Hunter MC, Jacobs PJ, et al. OH and HO2 radical chemistry in a forested region of north-western Greece. Atmos Environ 2001;35:4725–37. Carter WPL. Implementation of the SAPRC-99 Chemical Mechanism into the Models-3 Framework. Report to the United States Environmental Protection Agency; 2000 [Online Available At. http://www.engr.ucr.edu/_carter/pubs/s99mod3.pdf]. Chang JS, Brost RA, Isaksen ISA, Madronich S, Middleton P, Stockwell WR, et al. A three-dimensional Eulerian acid deposition model: physical concepts and formation. J Geophys Res 1987;92:14681–700. Chang JS, Middleton PB, Stockwell WR, Walcek CJ, Pleim JE, Lansford HH, et al. The regional acid deposition model and engineering model. Acidic Deposition: State of Science and Technology, Report 4, National Acid Precipitation Assessment Program; 1990. Chen DS, Cheng SY, Liu L, Lei T, Guo XR, Zhao XY. Assessment of the integrated ARPS-CMAQ modeling system through simulating PM10 concentration in Beijing, China. Environ Eng Sci 2008;25(2):191–206. Claeys M, Graham B, Vas G, Wang W, Vermeylen R, Pashynska V, et al. Formation of secondary organic aerosols through photooxidation of isoprene. Science 2004a;303: 1173–6. Claeys M, Wang W, Ion AC, Kourtchev I, Gelencser A, Maenhaut W. Formation of secondary organic aerosols from isoprene and its gas-phase oxidation products through reaction with hydrogen peroxide. Atmos Environ 2004b;38:4093–8.

De Smedt I, Müller J-F, Stavrakou T, van der AR, Eskes H, Van Roozendael M. Twelve years of global observations of formaldehyde in the troposphere using GOME and SCIAMACHY sensors. Atmos Chem Phys 2008;8:4947–63. De Smedt I, Stavrakou T, Müller J-F, van der AR, Van Roozendael M. Trend detection in satellite observations of formaldehyde tropospheric columns. Geophys Res Lett 2010;37:L18808. http://dx.doi.org/10.1029/2010GL044245. Dennis RL, McHenry JN, Barchet WR, Binkowski FS, Byun DW. Correcting RADM's sulfate under prediction: discovery and correction of model errors and testing the corrections through comparisons against field data. Atmos Environ 1993;26A(6): 975–97. Dufour G, Wittrock F, Camredon M, Beekmann M, Richter A, Aumont B, et al. SCIAMACHY formaldehyde observations: constraint for isoprene emission estimates over Europe? Atmos Chem Phys 2009;9:1647–64. EANET. Data Report on the Acid Deposition in the East Asia Region 2006. Available at (http://www.eanet.asia/product/index.html) Network Center for EANET; 2007. Fu T, Jacob DJ, Palmer PI, Chance K, Wang YX, Barletta B, et al. Space-based formaldehyde measurements as constrains on volatile organic compound emissions in east and south Asia and implications for ozone. J Geophys Res 2007;112:D06312. http://dx.doi.org/10.1029/2006JD007853. Grell G, Dudhia J, Stauffer D. A description of the Fifth-Generation Penn State/NCAR Mesoscale Model (MM5). NCAR Technical Note, NCAR/TN-398 + STR, NCAR, Boulder, CO; 1994. Guenther A, Hewitt CN, Erickson D, Fall R, Geron C, Graedel T, et al. A global model of natural volatile organic compound emissions. J Geophys Res 1995;100:8873–92. Guenther A, Karl T, Harley P, Wiedinmyer C, Palmer P, Geron C. Estimates of global terrestrial isoprene emissions using MEGAN (Model of Emissions of Gases and Aerosols from Nature). Atmos Chem Phys 2006;6:3181–210. Han Z, Ueda H, Matsuda K. Model study of the impact of biogenic emission on regional ozone and the effectiveness of emission reduction scenarios over eastern China. Tellus B 2005;57:12–27. Han KM, Song CH, Ahn HJ, Park RS, Woo JH, Lee CK, et al. Investigation of NOx emissions and NOx related chemistry in East Asia using CMAQ-predicted and GOME-derived NO2 columns. Atmos Chem Phys 2009;9:1017–36. Han KM, Lee CK, Lee J, Kim J, Song CH. A comparison study between model-predicted and OMI-retrieved tropospheric NO2 columns over the Korean peninsula. Atmos Environ 2011;45(17):2962–71. Hasson AS, Tyndall GS, Orlando JJ. A product yield study of the reaction of HO2 radicals with ethyl peroxy (C2H5O2), acetyl peroxy (CH3C(O)O2), and acetonyl peroxy (CH3C(O)CH2O2) radicals. J Phys Chem A 2004;108:5979–89. He H, Li C, Loughner CP, Li Z, Krotkov NA, Yang K, et al. SO2 over central China: measurements, numerical simulations and the tropospheric sulfur budget. J Geophys Res 2012;117:D00K37. http://dx.doi.org/10.1029/2011JD016473. Hofzumahaus A, Rohrer F, Lu K, Bohn B, Brauers T, Chang C, et al. Amplified trace gas removal in the troposphere. Science 2009;324:1702–4. Hong S-Y, Pan HL. Nonlocal boundary layer vertical diffusion in a medium-range forecast model. Mon Weather Rev 1996;124:2322–39. IPCC. Climate change 2007: the physical science basis. Contribution of Working Group I. Cambridge University Press; 2007. Kanaya Y, Sadanaga Y, Nakamura K, Akimoto H. Behavior of OH and HO2 radicals during the observations at a remote island of Okinawa (ORION99) field campaign. J Geophys Res 2001;106(D20):24197–208. Kim HS, Song CH, Park RS, Huey G, Ryu JY. Investigation of ship-plume chemistry using a newly-developed photochemical/dynamic ship-plume model. Atmos Chem Phys 2009;9:7531–50. Kubistin D, Harder H, Martinez M, Rudolf M, Sander R, Bozem H, et al. Hydroxyl radicals in the tropical troposphere over the Suriname rainforest: comparison of measurements with the box model MECCA. Atmos Chem Phys 2010;10:9705–28. Lamb B, Guenther A, Gay D, Westburg H. A national inventory of biogenic hydrocarbon emissions. Atmos Environ 1987;21:1695–705. Lawrence MG, Crutzen PJ. Influence of NOx emissions from ships on tropospheric photochemistry and climate. Nature 1999;402:167–70. Li J, Wang Z, Akimoto H, Yamaji K, Takigawa M, Pochanart P, et al. Near-ground ozone source attributions and outflow in central eastern China during MTX2006. Atmos Chem Phys 2008;8:7335–51. Lin M, Oki T, Holloway T, Streets DG, Bengtsson M, Kanae S. Long-range transport of acidifying substances in East Asia-Part I Model evaluation and sensitivity studies. Atmos Environ 2008;42:5939–55. Lin M, Holloway T, Oki T, Streets DG, Richter A. Multi-scale model analysis of boundary layer ozone over East Asia. Atmos Chem Phys 2009;9:3277–301. Liu L, Sundet JK, Liu Y, Berntsen TK, Isaksen ISA. A study of tropospheric ozone over China with a 3-D global CTM model. Terr Atmos Ocean Sci 2007;18(3):515–45. Matsui H, Koike M, Kondo Y, Takegawa N, Kita K, Miyazaki Y, et al. Spatial and temporal variations of aerosols around Beijing in summer 2006: Model evaluation and source apportionment. J Geophys Res 2009;114:D00G13. http://dx.doi.org/10.1029/ 2008JD010906. Millet DB, Jacob DJ, Turquety S, Hudman RC, Wu S, Fried A, et al. Formaldehyde distribution over North America: implications for satellite retrievals of formaldehyde columns and isoprene emission. J Geophys Res 2006;111:D24S02. http://dx.doi.org/10.1029/ 2005JD006853. Mueller J-F. Geographical distribution and seasonal variation of surface emissions and deposition velocities of atmospheric trace gases. J Geophys Res 1992;97:3787–804. Müller J-F, Stavrakou T, Wallens S, De Smedt I, Van Roozendael M, Potosnak M, et al. Global isoprene emissions estimated using MEGAN, ECMWF analyses and a detailed canopy environment model. Atmos Chem Phys 2008;8:1329–41. NIER. 2006 Greenhouse gas and air pollutants emissions in Korea. National Institute of Environmental Research (NIER); 2008.

K.M. Han et al. / Science of the Total Environment 463–464 (2013) 754–771 Ohara T, Akimoto H, Kurokawa J, Horii N, Yamaji K, Yan X, et al. An Asian emission inventory of anthropogenic emission sources for the period 1980–2020. Atmos Chem Phys 2007;7:4419–44. Palmer PI, Jacob DJ, Fiore AM, Martin RV, Chance K, Kurosu TP. Mapping isoprene emissions over North America using formaldehyde column observations from space. J Geophys Res 2003;108:4180. http://dx.doi.org/10.1029/2002JD002153. Palmer P, Abbot DS, Fu T-M, Jacob DJ, Chance K, Kurosu TP, et al. Quantifying the seasonal and interannual variability of North American isoprene emissions using satellite observations of the formaldehyde column. J Geophys Res 2006;111: D12315. http://dx.doi.org/10.1029/2005JD006689. Park RS, Song CH, Han KM, Park ME, Lee S-S, Kim S-B, et al. A study on the aerosol optical properties over East Asia using a combination of CMAQ-simulated aerosol optical properties and remote-sensing data via a data assimilation technique. Atmos Chem Phys 2011;11:12275–96. http://dx.doi.org/10.5194/acp-11-12275-2011. Peeters J, Müller J-F. HOx radical regeneration in isoprene oxidation via peroxy radical isomerisation. II: experimental evidence and global impact. Phys Chem Chem Phys 2010;12:14227–35. Peeters J, Nguyen TL, Vereecken L. HOx radical regeneration in the oxidation of isoprene. Phys Chem Chem Phys 2009;11:5935–9. Pierce T, Geron C, Bender L, Dennis R, Tonnesen G, Guenther A. Influence of increased isoprene emissions on regional ozone modeling. J Geophys Res 1998;103: 25611–29. Pleim JE, Clarke JF, Finkelstein PL, Cooter EJ, Ellestad TG, Xiu A, et al. Comparison of measured and modeled surface fluxes of heat, moisture and chemical dry deposition. In: Gryning, Schiermeier, editors. Air Pollution Modeling and Its Application XI. New York.: Plenum Press; 1996 Shim C, Wang Y, Choi Y, Palmer PI, Abbot DS, Chance K. Constraining global isoprene emissions with Global Ozone Monitoring Experiment (GOME) formaldehyde column measurements. J Geophys Res 2005;110:D24301. http://dx.doi.org/10.1029/2004JD005629. Song CH, Park ME, Lee KH, Ahn HJ, Lee YJ, Kim JY, et al. An investigation into seasonal and regional aerosol characteristics in East Asia using model-predicted and remotely-sensed aerosol properties. Atmos Chem Phys 2008;8:6627–54. Stauffer DR, Seaman NL. Use of four-dimensional data assimilation in a limited-area mesoscale model. Part I: experiments with synoptic-scale data. Mon Weather Rev 1990;118(6):1250–77. Stavrakou T, Müller J-F, De Smedt I, Van Roozendael M, van der Werf GR, Giglio L, et al. Evaluating the performance of pyrogenic and biogenic emission inventories against one decade of space-based formaldehyde columns. Atmos Chem Phys 2009;9: 1037–60. Steiner A, Luo C, Huang Y, Chameides WL. Past and present-day biogenic volatile organic compound emissions in East Asia. Atmos Environ 2002;36:4895–905.

771

Stevens PS, Mather JH, Brune WH, Eisele F, Tanner D, Jefferson A, et al. HO2/OH and RO2/HO2 ratios during the tropospheric OH photochemistry experiment: measurement and theory. J Geophys Res 1997;102:6379–91. Tan D, Faloona I, Simpas JB, Brune W, Shepson PB, Couch TL, et al. HOx budgets in a deciduous forest: results from the PROPHET summer 1998 campaign. J Geophys Res 2001;106:24407–27. Thornton JA, Wooldridge PJ, Cohen RC, Martinez M, Harder H, Brune WH, et al. Ozone production rates as a function of NOx abundances and HOx production rates in the Nashville urban plume. J Geophys Res 2002;107(D12):4146. http://dx.doi.org/ 10.1029/2001JD000932. van der Werf GR, Randerson JT, Giglio L, Collatz GJ, Mu M, Kasibhatla PS, et al. Global fire emissions and the contribution of deforestation, savanna, forest, agricultural, and peat fires (1997–2009). Atmos Chem Phys 2010;10:11707–35. http://dx.doi.org/ 10.5194/acp-10-11707-2010. van Donkelaar A, Martin RV, Leaitch WR, Macdonald AM, Walker TW, Streets DG, et al. Analysis of aircraft and satellite measurements from the Intercontinental Chemical Transport Experiment (INTEX-B) to quantify long-range transport of East Asian sulfur to Canada. Atmos Chem Phys 2008;8:2999–3014. Walcek CJ, Taylor GR. A theoretical method for computing vertical distributions of acidity and sulfate production within cumulus clouds. J Atmos Sci 1986;43:339–55. Wang KY, Shallcross DE. Modelling terrestrial biogenic isoprene fluxes and their potential impact on global chemical species using a coupled LSM-CTM model. Atmos Environ 2000;34:2909–25. Wang S, Streets DG, Zhang Q, He K, Chen D, Kang S, et al. Satellite detection and model verification of NOx emissions from power plants in Northern China. Environ Res Lett 2010;5:044007. http://dx.doi.org/10.1088/1748-9326/5/4/044007. Yamaji K, Ohara T, Uno I, Kurokawa J, Pochanart P, Akimoto H. Future prediction of surface ozone over East Asia using Models-3 Community Multiscale Air Quality Modeling System and Regional Emission Inventory in Asia. J Geophys Res 2008;113: D08306. http://dx.doi.org/10.1029/2007JD008663. Yamartino RJ. Nonnegative, conserved scalar transport using grid-cell-centered, spectrally constrained Blackman cubics for applications on a variable-thickness mesh. Mon Weather Rev 1993;131:753–63. Zhang P, Anderson B, Barlow M, Tan B, Myneni R. Climate related vegetation characteristics derived from MODIS LAI and NDVI. J Geophys Res 2004;109:D20105. http://dx.doi.org/ 10.1029/2004JD004720. Zhang Q, Streets DG, Carmichael GR, He KB, Huo H, Kannari A, et al. Asian emissions in 2006 for NASA INTEX-B mission. Atmos Chem Phys 2009;9:5131–53. Zimmerman P. Testing of hydrocarbon emissions from vegetation, leaf, litter and aquatic surfaces, and development of a method for compiling biogenic emission inventories. Rep. EPA-450-4-70-004. Research Triangle Park, N.C: US EPA; 1979.