Quantifying impacts of crop residue burning in the North China Plain on summertime tropospheric ozone over East Asia

Atmospheric Environment 194 (2018) 14–30

Contents lists available at ScienceDirect

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

Quantifying impacts of crop residue burning in the North China Plain on summertime tropospheric ozone over East Asia

T

Mingliang Maa,b, Kaixu Baia,b,∗, Fengxue Qiaoa,b, Runhe Shia,b,∗∗, Wei Gaoa,b,c,d a

Key Laboratory of Geographic Information Science (Ministry of Education), East China Normal University, Shanghai, 200241, China School of Geographic Sciences, East China Normal University, Shanghai, 200241, China c Natural Resource Ecology Laboratory, USDA UV-B Monitoring and Research Program, Fort Collins, CO, USA d Department of Ecosystem Science and Sustainability, Colorado State University, Fort Collins, CO, USA b

A R T I C LE I N FO

A B S T R A C T

Keywords: Crop residue burning Tropospheric ozone North China Plain (NCP) Statistical modeling Ozone pollution

Crop residue burning has been proved to have negative impacts on regional atmospheric environment. In this study, an evidence-based statistical modeling framework was established to quantify potential impacts of crop residue burning in the North China Plain (NCP) on summertime tropospheric ozone increase over East Asia during 2004–2016. To assess the intensity of crop residue burning, fire occurrence counts derived from the MODerate-resolution Imaging Spectroradiometer onboard the Terra and Aqua satellites were used as a proxy. Additionally, another six factors were employed as potent explanatory variables. Maximum covariance analysis was first applied to decouple spatiotemporal interactions between tropospheric ozone and each explanatory variable. Based on the decoupled modes, multivariate linear regression (MLR) and artificial neural network (ANN) were used to establish statistical relationships between tropospheric ozone and contributing factors, respectively. The results indicate that the ANN-based modeling scheme enables to approximate the observed tropospheric ozone variations better than MLR. Further investigations reveal that the summertime crop residue burning in the NCP is the predominant factor contributing to the observed additive tropospheric ozone increases over East Asia, yielding extra 8% tropospheric ozone elevation on average in June. Moreover, UV radiation and wind also played critical roles in modulating the observed tropospheric ozone variations therein. In general, the critical role of crop residue burning over the NCP in modulating summertime tropospheric ozone increase over East Asia have been well demonstrated based on the proposed evidenced-based modeling framework.

1. Introduction Unlike stratospheric ozone present at high altitudes where it protects Earth from the harmful ultraviolet radiation, tropospheric ozone is commonly deemed “bad ozone” due to its adverse impacts on public health (Ebi and McGregor, 2008; Jerrett et al., 2009; Lim et al., 2013), agriculture (Wittig et al., 2009; Ainsworth et al., 2012), ecosystems (Wittig et al., 2009; Ainsworth et al., 2012; Yue and Unger, 2014), and climate (Skeie et al., 2011; Shindell et al., 2012; Myhre et al., 2013; Stevenson et al., 2013). Scientific investigations revealed that concentrations of tropospheric ozone are controlled primarily by photochemical processes associated with ozone production and destruction, as well as deposition and atmospheric diffusion. Specifically, ozonerelated photochemical processes are influenced primarily by ozone precursors and meteorological factors such as winds, temperature and humidity (e.g., Jacob et al., 1993; Sillman and Samson, 1995; Neftel ∗

et al., 2002; Menut, 2003; Bärtsch-Ritter et al., 2004; McMillan et al., 2005; Dawson et al., 2007; Lou et al., 2015). Atmospheric diffusion alters the distribution of ozone and its precursors at a scale ranging from regional to inter-continental and even hemispheric-scale through long-range transport (e.g., Lin et al., 2014; Derwent et al., 2015), stratosphere–troposphere exchange (Cristofanelli et al., 2010), seasonal transport patterns (Safieddine et al., 2015), and climate variability (Lin et al., 2014). With respect to ozone-related photochemical processes, precursors such as nitrogen oxides (NOX), volatile organic compounds (VOCs), and carbon monoxide (CO) have been proven to play a critical role as their relative amounts in the atmosphere largely control the production of tropospheric ozone (Sillman, 1999; Sillman and He, 2002). In general, surface anthropogenic emissions from industry and automobiles, emissions from fires, and natural emissions are deemed as three primary sources of these ozone precursors (Turquety et al., 2007; Guenther et al., 2012; Huo et al., 2012; Jaffe and Wigder, 2012; Monks

Corresponding author. Key Laboratory of Geographic Information Science (Ministry of Education), East China Normal University, Shanghai, 200241, China. Corresponding author. Key Laboratory of Geographic Information Science (Ministry of Education), East China Normal University, Shanghai, 200241, China. E-mail addresses: [email protected] (K. Bai), [email protected] (R. Shi).

∗∗

https://doi.org/10.1016/j.atmosenv.2018.09.018 Received 20 January 2018; Received in revised form 8 September 2018; Accepted 11 September 2018 Available online 15 September 2018 1352-2310/ © 2018 Elsevier Ltd. All rights reserved.

Atmospheric Environment 194 (2018) 14–30

M. Ma et al.

should be quantitatively evaluated. Toward such a goal, evidencedbased statistical modeling frameworks were established between the observed tropospheric ozone and a set of potent explanatory variables, including UV radiation, wind speed, air temperature, boundary layer height and relative humidity, in addition to satellite detected fire occurrence counts (as a proxy for the intensity of crop residue burning). Moreover, estimated NOX emission inventories over China were also employed to examine the associations between crop residue burning and tropospheric ozone variations since NOX is one important precursor for ozone production besides others like VOCs. The following science questions could be answered by this study: (1) is there any linkage between crop residue burning in the NCP and the summertime tropospheric ozone increase over East Asia? and (2) if so, to what extent can the observed tropospheric ozone increase be explained by crop residue burning? Our hypothesis is that before prominent crop residue burning, the ozone photochemical reaction process over East Asia is primarily limited by NOX concentrations, and thus the elevated NOX concentrations resulting from crop residue burning would accelerate ozone photochemical processes and in turn yield regional tropospheric ozone increase.

et al., 2015). A variety of studies found that biomass burning could significantly affect tropospheric ozone variations nearby the burning regions and their potential associations have been well investigated around the world, e.g. over Southeast Asia (Liu et al., 1999; Deng et al., 2008; Lin et al., 2009a; Toh et al., 2013; Yadav et al., 2017), especially in Indonesia (Thompson A.M., 2001; Duncan et al., 2003; Chandra et al., 2009; Zhang et al., 2011), tropical Africa (Mauzerall et al., 1998; Martin et al., 2002; Sauvage et al., 2005; Jourdain et al., 2007; Real et al., 2010), India (Kumar et al., 2011; Sarangi et al., 2014; Sinha et al., 2014), Siberia (Jaffe et al., 2004; Verma et al., 2009; Konovalov et al., 2011), South America (Watson et al., 1990; Jonquières et al., 1998; Schultz et al., 1999), North America (Val Martín et al., 2006; Parrington et al., 2012, 2013; Alvarado et al., 2015), and Europe (Amiridis et al., 2012; Cristofanelli et al., 2013). In general, these studies focused primarily on forest fires (Kaufman et al., 1992; Real et al., 2007; Yokelson et al., 2009; Alvarado et al., 2010; Akagi et al., 2012) and savanna fires (Kaufman et al., 1992; Sanhueza et al., 1999; Chandra et al., 2002; Trentmann et al., 2005). As one distinct type of biomass burning, crop residue burning can also yield large amounts of ozone precursors, which may result in significant tropospheric ozone variations in turn. Nevertheless, the impacts of crop residue burning on regional tropospheric ozone variations have been seldom evaluated relative to forest and savanna fires (Stohl et al., 2007; Li et al., 2008; Yamaji et al., 2010; Ding et al., 2013; Kanaya et al., 2013; Tang et al., 2013; Pan et al., 2015; Stavrakou et al., 2016; Lu et al., 2017), especially in those agricultural dominated economies like China where crop residue burning is salient. As a major wheat production region, the North China Plain (NCP) plays a critical role in wheat yield in China (Huang et al., 2012). However, due to the lack of cost-efficient crop residue recycling technologies, the wheat residue and straw are oftentimes burnt directly in farmland after the wheat harvest from late May to June (Huang et al., 2012; Xue et al., 2014). By taking advantage of atmospheric chemistry transport models and ground-measured air pollutant emissions from field campaigns, Li et al. (2008) and Yamaji et al. (2010) revealed that crop residue burning over the southern part of the NCP could result in about 37.9% and 26% surface ozone increases at the summit of Mount Tai in June 2006, respectively. Similar results were also evidenced in Kanaya et al. (2013). Likewise, Stavrakou et al. (2016) and Lu et al. (2017) found that crop residue burning could result in 7% increases of the maximum surface ozone in June 2010 over the NCP and accounted for 18% surface ozone formation in the autumn of 2013 in Wuhan, respectively. In addition to model simulated results, associations between crop residue burning and tropospheric ozone variations were also examined by making use of satellite observations (Dufour et al., 2010) and in situ measurements (e.g., Li et al., 2007; Ding et al., 2008; He et al., 2008; Lin et al., 2008, 2009b; Meng et al., 2009; Wang et al., 2010; Tang et al., 2013; Pan et al., 2015). By referring to spatial and temporal variations of tropospheric ozone nearby the NCP and the number of fire occurrence counts (derived from the MODerate-resolution Imaging Spectroradiometer, MODIS) detected in the NCP (Fig. 1), a very tight spatiotemporal coincidence was revealed between the regional tropospheric ozone peaks and the maximum number of fire occurrence counts detected therein. Given the fact that such a vast area of biomass burning could release ample amount of air pollutants like NOX, VOCs and CO that are often deemed as ozone production precursors (Guo et al., 2004; Yan et al., 2006; Zhang et al., 2008; Suthawaree et al., 2010; Huang et al., 2012; Li et al., 2016; Wu et al., 2016), in conjunction with the observed spatial and temporal coincidence patterns, we may deduce that there is a potential linkage between the observed summertime tropospheric ozone increase and crop residue burning in the NCP. To demonstrate this hypothesis, impacts of crop residue burning in the NCP on regional tropospheric ozone variations observed nearby

2. Data sources 2.1. Tropospheric ozone column data As one conventional algorithm proposed by Fishman et al., in 1990, the tropospheric ozone residual method (TOR) has been widely used to derive tropospheric column ozone (TCO) by subtracting the stratospheric column ozone (SCO) from the total column ozone (Fishman et al., 1990, 2003; Chandra et al., 2003). Although there are several other techniques available for the derivation of TCO from satellite observations, e.g., the modified residual method (Hudson and Thompson, 1998), the UV radiance scan angle method (Kim et al., 2001), the topography differencing method (Newchurch et al., 2001) and the Convective Cloud Differential (CCD) method (Ziemke et al., 2009), TOR provides tropospheric ozone products with better spatial and temporal coverage and is thus widely used (Ziemke et al., 2014). Compared with total column ozone products, tropospheric column ozone are rarely provided due to instrumental and algorithmic constraints. At present, there are few tropospheric column ozone products available to the public. One data set is a multi-sensor integrated record by combining the CCD method derived tropospheric column ozone from the first Global Ozone Monitoring Experiment (GOME-1) onboard ERS2 (1995–2011), the Scanning Imaging Absorption Spectrometer for Atmospheric Chartography (SCIAMACHY) onboard ENVISAT (2002–2012) and the second Global Ozone Monitoring Experiment (GOME-2) onboard MetOp-A/MetOp-B (2007-present) over tropics (20°S-20°N) (Leventidou et al., 2016). Another one is derived from the Ozone Monitoring Instrument (OMI) and the Microwave Limb Sounder (MLS) through the TOR method, providing continuous monthly tropospheric ozone over 60°S–60°N from the late August 2004 to present. In this study, the monthly OMI/MLS tropospheric column ozone product was employed to resemble spatiotemporal variations of tropospheric ozone over East Asia from October 2004 to September 2016 (Table 1). The product is derived by subtracting the SCO (derived from MLS measurements) from the total column ozone (derived from OMI measurements) after adjusting the inter calibration differences between these two instruments through the convective-cloud differential method (Ziemke et al., 2006). OMI and MLS are two critical instruments onboard the Aura satellite. As a nadir-scanning instrument, OMI provides near global coverage total column ozone data at a nadir resolution of 13 km × 24 km, by detecting backscattered solar radiance from visible (350–500 nm) and ultraviolet (UV) wavelength channels (UV-1: 270–314 nm; UV-2: 306–380 nm) (McPeters et al., 2008). The MLS instrument is a thermal-emission microwave limb sounder that measures vertical profiles of mesospheric, stratospheric, and upper tropospheric 15

Atmospheric Environment 194 (2018) 14–30

M. Ma et al.

Fig. 1. Spatial and temporal distribution of Ozone Monitoring Instrument (OMI) and Microwave Limb Sounder (MLS) derived tropospheric ozone columns and Moderate-resolution Imaging Spectroradiometer (MODIS) detected fire occurrence counts over the North China Plain (NCP) and nearby regions. (a) Spatial distribution of the averaged OMI/MLS tropospheric ozone column in June during 2004–2016 (data has been centered by removing the averaged tropospheric ozone across the whole area), (b) spatial distribution of averaged MODIS detected fire occurrence counts (a natural logarithmic transformation was performed to reduce data amplitude) in June during 2004–2016, (c) temporal variability of the averaged OMI/MLS tropospheric ozone shown in the selected region in Fig. 1a (dotted green rectangle) during October 2004 to September 2016 (data has been centered by removing the averaged tropospheric ozone across the whole time period); (d) same as (c) but for the number of MODIS detected fire occurrence counts in the NCP (dotted blue rectangle shown in Fig. 1b). Locations of two UV radiation sites were also depicted in Fig. 1a by open blue circles. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

et al., 2008; Miyazaki et al., 2012), many researchers have successfully examined the biomass burning related tropospheric ozone increases around the world by using this product (e.g., Chandra et al., 2009; David et al., 2011; Lin et al., 2009b; Ziemke et al., 2009; Sonkaew and Macatangay, 2015). Therefore, this product was also employed in this study to evaluate tropospheric ozone variations over East Asia.

temperature, ozone and other constituents from limb scans (Waters et al., 2006). This unique capacity renders MLS the capability of measuring SCO that cannot be derived from a nadir mapper. The quality of OMI/MLS tropospheric column ozone product has been well documented in the literature after comparing with both ground-based measurements (Thompson, 2003) and simulated TCO data derived from Global Modeling Initiative's Chemical Transport Model (Ziemke et al., 2006). In general, OMI/MLS TCO data agree well with ground-measured (with a bias less than 5 DU) and model simulated TCO data (bias within 2 DU) (Ziemke et al., 2006, 2009, 2011, 2014; Liu et al., 2010a; Tang and Prather, 2012; Mielonen et al., 2015; Huang et al., 2017a, 2018). Although some researchers raised concerns that OMI/MLS TCO products are more sensitive to ozone at the middle and upper troposphere rather than at the low troposphere (Bethan et al., 1996; Stajner

2.2. MODIS fire occurrence counts To evaluate the intensity of crop residue burning, the monthly MODIS fire location information product (MCD14ML) generated by combining worldwide fire occurrence counts detected by MODIS onboard Terra and Aqua satellites, was employed. This product has been widely used to detect biomass burning or fire disasters on Earth

Table 1 A brief summary of data sets used in this study. Name

Type

Unit

Spatial resolution

Temporal resolution

Data set

Tropospheric Ozone Fire occurrence counts Tropospheric NO2 Boundary Layer Height Relative Humidity Temperature Zonal Wind Meridional Wind UV radiation NOX

Integrated Surface Integrated Integrated 850 hpa 850 hpa 850 hpa 850 hpa Surface Surface

DU / molec/cm2 m % K m/s m/s kJ/m2 ton/month

1° × 1.25° 1 km × 1 km 0.25° × 0.25° 0.5° × 0.5° 0.5° × 0.5° 0.5° × 0.5° 0.5° × 0.5° 0.5° × 0.5° 0.25° × 0.25° 0.1° × 0.1°

monthly monthly daily monthly monthly monthly monthly monthly daily monthly

OMI/MLS MCD14ML OMNO2d ERA-Interim ERA-Interim ERA-Interim ERA-Interim ERA-Interim TEMIS NOX-PKU

16

Atmospheric Environment 194 (2018) 14–30

M. Ma et al.

2.4. UV radiation

(Chuvieco et al., 2008; Vadrevu et al., 2012; Chen et al., 2013; Peterson et al., 2014). A brief introduction to this product was summarized as follows: brightness temperatures measured at the 4 μm (denoted as BT4) MODIS channel are used for active fire detection, while brightness temperatures at 11 μm and 12 μm (denoted as BT11 and BT12) are used for cloud masking and forest clearing rejection (Giglio et al., 2016). Meanwhile, top-of-atmosphere reflectance at 0.65 μm, 0.86 μm, and 2.1 μm (denoted as ρ0.65 , ρ0.86 and ρ2.1, respectively) aggregated to 1-km spatial resolution are utilized to mask clouds, bright surface, sun glint, and coastal false alarm rejection cloud masking. The fire pixel identification thresholds BT4* and ΔBT* are automatically determined for each MODIS scan by averaging the values of BT4 and ΔBT for all cloudfree and glint-free land pixels within a 301-sample by 30-line large moving window (Giglio et al., 2016). A daytime pixel will be marked as a potential fire occurrence counts once it satisfies BT4 > BT4* and ΔBT > ΔBT* (where ΔBT≡BT4−BT11), and ρ0.86 < 0.35 (does not apply for nighttime pixels). The MODIS daily fire product is a 1-km gridded product compositing fire occurrence counts detected over each grid cell on a daily basis (24-h) across the globe, and the monthly MCD14ML product is a composition of daily fire products. In terms of the MCD14ML product, the persistent fire occurrence counts can be detected every time during the overpass, and the number of fire occurrence counts over the specific region depicts the burning area and duration of active fires. More details related to the MCD14ML product can be found in Giglio (2015 & 2016) and Schroeder et al. (2008). By comparing with more than 2500 high resolution images observed by the Advanced Spaceborne Thermal Emission and Reflection Radiometer, the MODIS collection 6 active fire products showed excellent accuracy, with a global commission error of 1.2% (Giglio et al., 2016). As an essential quality control, in this study, only those fire occurrence counts with a confidence level greater than 70% were applied in the further modeling framework.

In this study, the UV radiation provided by the Tropospheric Emission Monitoring Internet Service (TEMIS) under the Royal Netherlands Meteorological Institute (KNMI) was also employed to model tropospheric ozone variations because of its critical role played in the ozone production process. The TEMIS UV data products are derived from the empirically based parametrization method of Allaart et al. (2004), which means that the amount of UV radiation incident at the surface is a function of total ozone column, solar zenith angle, cloud mask and an appropriate action spectrum (Derrien and Le Gléau, 2005; Zempila et al., 2017). The ground-based validations indicated that the TEMIS UV data products agreed well with in situ measurements, showing an error budget less than 1% for erythemal doses data in cloud free days (Zempila et al., 2017). In this study, the TEMIS UV erythemal dose time series recorded at Baoding and Tianjin (two stations in the region of the NCP and shown as blue circles in Fig. 1a) were averaged as the proxy for the UV radiation variability over the NCP. 2.5. Meteorological factors Besides ozone production related factors such as ozone-precursors (NOX and VOCs) and UV radiation, tropospheric ozone variations are also influenced by atmospheric conditions. Therefore, a set of meteorological factors were also employed to better characterize the observed tropospheric ozone variations in the statistical modeling framework. Previous studies have revealed that biomass burning related tropospheric ozone variations mainly occur in the lower troposphere (Chan et al., 2003, 2006; Logan et al., 2008; Ziemke et al., 2009; Liu et al., 2010b). Hence, five meteorological parameters in the lower troposphere (at the 850 hpa level), i.e., relative humidity, temperature, boundary layer height, U (Zonal wind) and V (Meridional wind), were applied to represent atmospheric conditions. As the largest global atmospheric reanalysis data set provided by the European Center for Medium–Range Weather Forecasts, ERA-Interim reanalysis provides a description of the Earth surface (including land and ocean) and the atmospheric conditions in the troposphere and stratosphere (Dee et al., 2011). The products cover a time period from 1979 to present, and are updated routinely. This reanalysis has been extensively used in various studies, and the quality of this product has been well documented in the literature (Bormann and Thépaut, 2004; Andersson et al., 2005; Palm et al., 2005; Tompkins et al., 2007; Rotach and Zardi, 2007; Von Engeln and Teixeira, 2013; Kalmus et al., 2015; Chen et al., 2016; Bai et al., 2016; Guo et al., 2016). Therefore, the monthly reanalysis of the aforementioned five meteorological parameters was extracted from the ERA-Interim and used in this study for further statistical modeling practices.

2.3. NOX emission inventory To justify the hypothesized association between the observed tropospheric ozone increase over East Asia and the mass crop residue burning in the NCP, the NOX emission inventory was thereby applied to reflect the elevated NOX concentrations resulting from crop residue burning in June in the NCP. The emission inventory provided by the Emissions Database for Global Atmospheric Research (EDGAR), a global bottom-up emission inventory which compiles the gaseous and particulate air pollutant emissions using a consistent methodology based on international statistics such as those of IEA (2014) and FAOSTAT (2014) rather than regional offices (Crippa et al., 2016, 2018; JanssensMaenhout et al., 2017), has been widely applied toward such a goal. However, this inventory is incapable of capturing the crop residue burning emitted NOX peaks in June over the NCP due to the lack of monthly averaged emissions. Therefore, the estimated monthly gridded NOX emission inventory over China provided by the group in Peking University was employed. This product is a bottom-up global emission inventory, which is based on an updated fuel consumption data set from Peking University Fuel Inventories (PKU-FUEL) in conjunction with detailed source information of residential sector and over 1000 emission factors data (Huang et al., 2017b). The updated PKU-FUEL includes subnational disaggregation energy data specific to China and the detailed residential energy consumption were updated for rural China based on numerous surveys and field campaigns. The emission factor data from various sources which were collected from a large number of studies were used to generate country and time specific emission factors for different sources. This NOX emission inventory provides a monthly global emission inventory from 1960 to 2014 with a spatial resolution of 0.1° × 0.1°. A detailed description and validation of this product can be found in Huang et al. (2017b). In this study, the monthly NOX emission inventory over China from 2005 to 2014 was applied to examine the crop residue burning related NOX emissions.

3. Methods To evaluate the impacts of crop residue burning in the NCP on tropospheric ozone variations over East Asia, a statistical modeling framework is proposed here by establishing relationships between the observed tropospheric ozone variations and a set of potent explanatory variables (i.e., fire occurrence counts, UV radiation, relative humidity, temperature, boundary layer height, U and V). Since each factor (except UV radiation) is provided in a three-dimensional data set representing the observed variations in both spatial and temporal domains on a large scale, reducing the dimensionality of each data set is thus of critical importance. In this study, maximum covariance analysis (MCA), a popular dimensionality reduction method frequently used in geophysical research studies, was applied to detect and extract the dominant coupled patterns between the response variable and the explanatory factors. By maximizing covariance rather than correlation between two spatial-temporal fields, the coupled interactions between tropospheric ozone and the relevant explanatory factors were extracted in both 17

Atmospheric Environment 194 (2018) 14–30

M. Ma et al.

where k = 1, 2, …, r. There is no doubt that the first coupled mode explains the most covariance between two data sets, and the explained variance can be evaluated by the fraction of squared covariance (SCF), and the SCF of the ith coupled mode can be calculated as:

spatial and temporal domains. The derived time series were then used as inputs for further modeling analyses. Subsequently, two distinct modeling methods, multivariate linear regression (MLR) and artificial neural networks (ANN) were used to link seven potential explanatory variables to tropospheric ozone. The reason to apply ANN as another modeling tool lies in the fact that the statistical relationships between tropospheric ozone and the selected contributing factors could be highly nonlinear, whereas MLR may be inadequate to model complex nonlinear interactions. Time series of tropospheric ozone were used as the response variable while the others were applied as explanatory factors. Before modeling, time series of all explanatory variables were normalized by subtracting their means and then dividing by their standard deviations. Finally, tropospheric ozone responses to each potential explanatory variable were evaluated through a sensitivity analysis scheme. The ultimate goal is to quantify tropospheric ozone changes associated with the dynamics of each explanatory variable. More detail of the sensitivity analysis will be discussed in subsection 3.3.

SCF =

MCA was developed mainly to capture the dominant coupled modes between two geophysical spatial-temporal fields by maximizing their covariance rather than correlation (Bretherton et al., 1992; Wallace et al., 1992). Due to the distinct advantage in dimensionality reduction by decomposing the cross-covariance matrix of two fields and then mapping the leading modes (i.e., modes with large singular values) onto each field to depict the possible covariance between them in either spatial or temporal domain, MCA has been widely used in the literature to detect possible linkages or interactions between two geophysical fields (e.g., Bretherton et al., 1992; Wallace et al., 1992; Ding et al., 2011; Bai et al., 2016). Mathematically, assuming two geophysical fields that can be represented by X and Y, where X is a N × p matrix (e.g., N samplings over p grids) and Y is a N × q matrix (e.g., N samplings over q grids), the cross-covariance matrix between X and Y can be calculated as follows:

1 XTY N−1

3.2. Statistical modeling framework 3.2.1. Multivariate linear regression MLR has been widely used to model the dependence of a response variable on several explanatory variables (Thompson M.L. et al., 2001; Sousa et al., 2007; Rajab et al., 2013; Bai et al., 2017). In most MLR modeling schemes, the explanatory variables are oftentimes used solely, without considering the internal interactions among them. However, this assumption does not always hold, particularly in modeling complex natural systems like the atmosphere in which parameters are oftentimes highly interactive with each other and serve to modulate the atmospheric state collectively. Bearing this in mind, the quadratic interaction terms among seven explanatory variables were added into the MLR modeling framework (Eq. (8)). The interactions include not only the external connections between two different variables, but also the internal interactions (self–correlation and denoted as the quadratic form) of each explanatory variable. The MLR-based statistical modeling framework can be simplified and formulated (only statistically significant terms were included here) as follows:

(1)

It is clear that the cross-covariance matrix Cxy has a dimension of p×q, highlighting the possible couplings between X and Y in spatial and temporal domains. Computationally, singular value decomposition (SVD), a matrix factorization method (Prohaska, 1976; Lanzante, 1984), is typically applied to decompose the cross-covariance matrix C xy so as to decouple the potential interactions between X and Y:

+ β × FC ˆy = +β β+ ×β UV× RH+ β+ ×β RH× T×+Uβ+×β U×+RHβ ×× VBLH+ β+ ×β BLH ×T×U t

Cxy = U T ΣV

(2)

(3)

V = {v1 (q), v2 (q), ... , vr (q)}

(4)

yk =

vkT

Y

1

2

8

3

4

9

5

6

10

+ β11 × T × V + β12 × V × BLH + β13 × V × FC + β14 × FC × U + β15 × U 2

(8)

where FC , T , RH , BLH , U , V and UV are explanatory variables of fire occurrence counts, air temperature, relative humidity, boundary layer height, Zonal wind, Meridional wind and UV radiation, respectively. β denotes the regression coefficient for each term. To better determine the regression coefficients, a constant value (also known as intercept and denoted as β0 in Eq. (8)) is oftentimes added in order to solve the above equation in a least squares manner. In spite of the linear regression form shown in Eq. (8), nonlinear features have been incorporated so as to enable the modeling scheme to better approximate the interactions between the response variable and explanatory variables. In this study, the associations between tropospheric ozone variations and seven potential explanatory variables were modeled in a manner of Eq. (8). To assess the overall response of tropospheric ozone to all these potent contributing factors, the tropospheric ozone time

By maximizing the covariance between two datasets, the leading mode of C xy (i.e., u1 and v1) can be obtained to represent the maximized X and Y patterns. Finally, the coupled modes between two data sets can be obtained by projecting the singular vectors onto each original data set domain to realize dimensionality reduction. For instance, time series xk and yk will be derived as follows:

xk = ukT X

0

7

where Σ is a diagonal matrix containing singular values in descending orders as σ1 ≥ σ2 ≥ σ3 ≥ … ≥ σr , and r is the number of singular values. U and V are two orthogonal matrices whose columns are singular vectors of X and Y respectively,

U = {u1 (p), u2 (p), ..., ur (p)}

(7)

Compared with many other methods like canonical correlation analysis (CCA) commonly used to decouple interactions between two geophysical fields, MCA has two distinct advantages. The first is associated with the fact that MCA maximizes covariance rather than correlation, whereas CCA modes may capture little variance (i.e., yield small covariance) because of high correlation. Secondly, the two singular vectors derived from MCA are orthogonal, and the projections xk and yk are in general correlated, whereas the vectors derived from methods like CCA are not generally orthogonal and the variates are uncorrelated (Wilks, 2006). In this study, MCA was used to decouple the possible interactions between tropospheric ozone over East Asia and each potential contributing factor (except for UV radiation), and the derived time series of each parameter will be then used as inputs for further statistical modeling framework. In practice, each data set was firstly centered before performing MCA so as to mitigate large biases (e.g., extreme large value of the first eigenvalue) associated with data magnitudes in calculating SCF.

3.1. Dimensionality reduction

Cxy =

σi2 . N ∑ j = 1 σ j2

(5) (6) 18

Atmospheric Environment 194 (2018) 14–30

M. Ma et al.

Fig. 2. The principal MCA modes between tropospheric ozone and fire occurrence counts over East Asia during 2004–2016. (a) Spatial patterns of tropospheric ozone associated with the first mode (59.5% total covariance explained) (b) same as in (a) but for fire occurrence counts, (c) temporal patterns of tropospheric ozone and fire occurrence counts associated with the first mode, (d) same as in (a) but for the second mode (32.7% total covariance explained), (e) same as in (b) but for the second mode, (f) same as in (c) but for the second mode. All data sets have been normalized by performing a z-score analysis (removing mean and then divided by the standard deviation) and is referred to as standardized data in each figure.

input features based on empirical experience) was created aiming to approximate the statistical relationship between tropospheric ozone and seven explanatory variables. The widely used Levenberg-Marquardt method was applied as the learning algorithm due to its fair performance in both learning speed and overall accuracy (Hagan and Menhaj, 1994; Svozil et al., 1997). In the training process, the preprocessed (i.e., normalized and de-correlated) seven input features and the relevant target time series were randomly divided into two portions, with 70% of the samples for training and the remaining 30% for validation. The training process continues unless the early stopping criterion, i.e., a coefficient of determination (R2) between the predicted validation data and the relevant actual observations greater than 0.9, is reached. To attain a robust modeling framework, a set of 100 models were simulated independently, and the median values of these 100 models were then used to represent the overall performance of all modeling results.

series coupled with these influential factors derived from MCA were averaged and then used as the total response in the modeling framework. To remove the potent modeling bias associated with the different dimension and amplitude of each explanatory variable, time series of all seven contributing factors were first normalized (z-score analysis) to make them dimensionless. Subsequently, the normalized time series were de-correlated (also known as whitening) to avoid the collinearity related bias in the further statistical modeling (see in Table S1 in the supplementary data). The two-sided Student-t test was performed for each factor (Table S2), and only those factors being significant at 95% confidence interval (i.e., p-value < 0.05) were retained to rebuild a robust linear regression model. 3.2.2. Artificial neural network ANN has been widely used to approximate complex relationships between variables. In this study, a single hidden layer feedforward neural network with 14 hidden neurons (two times of the number of 19

Atmospheric Environment 194 (2018) 14–30

M. Ma et al.

where the large amplitudes of tropospheric ozone agreed well with the fire occurrence counts peaks in every June (Fig. 2f). In terms of the time series associated with fire occurrence counts shown in Fig. 2f, the low amplitude resembled the observed fire occurrence counts minimum in South China, which exhibits a prominent antiphase pattern, namely, negative values shown in Fig. 2e. This can be ascribed to the fact that each data set has been centered prior to MCA, indicating fire occurrence counts in South China are lower than the regional average. In addition, the low amplitude of fire occurrence counts and tropospheric ozone was not coincident with each other, with a time delay of about three months between them. Hence, it is indicative of a possible linkage between increased biomass burning and elevated regional tropospheric ozone, since the positive amplitudes of both variables were coincident (mode 2). In order to quantify impacts of crop residue burning in the NCP on tropospheric ozone over East Asia and reducing contributions of biomass burning from other regions (e.g., South China), only fire occurrence counts extracted in the NCP (within the dotted blue rectangle in Fig. 2e) were used for further statistical modeling. Analogously, the leading MCA modes between tropospheric ozone and the other five potential explanatory variables (i.e., air temperature, relative humidity, boundary layer height, U, and V) over East Asia were also examined (Figs. S2–S4). Differing from fire occurrence counts, the leading MCA modes between tropospheric ozone and those five atmospheric factors were all well characterized by the first mode, explaining more than 90% of the total covariance. Hence, the associated time series were highly correlated (r ≈ 0.90). In the spatial domain, the results show that regional tropospheric ozone peaks over East Asia corresponded well to the area with positive temperature anomalies north of 35°N (Fig. S2b), higher humidity in the NCP (Fig. S2e), zonal wind (U) and meridional wind (V) anomalies (Figs. S3b and S3e) as well as the positive anomalies of boundary layer height in the northwest side of the NCP (Fig. S4b). Meanwhile, the temporal variations suggest that the anomalies with the maximum values in June for these factors (temperature, humidity, wind, and boundary layer height) all corresponded well to the high concentrations of tropospheric ozone over East Asia. Note that the zonal wind shown in Fig. S3c varied with an opposite phase against tropospheric ozone, and this is mainly due to the prescribed positive and negative signs associated with the direction of wind (negative signs for easterly wind in the later summer and autumn while the positive signs for westerly wind in most other months) in ERA-Interim reanalysis (i.e., does not imply a reduced wind speed).

3.3. Sensitivity analysis To assess the impacts of crop residue burning on tropospheric ozone variations, a sensitivity analysis scheme was performed based on the simulated statistical models, aiming to quantify tropospheric ozone response associated with each explanatory variable. As previously stated, fire occurrence counts time series were used here as the proxy for crop residue burning. Given the crop residue burning occurs mainly in June, another fire occurrence counts time series was then created by replacing fire occurrence counts data values in every June with the averaged value of May and July in the same year. In other words, one scenario without crop residue burning was simulated arbitrarily. The difference between fire occurrence counts in June and the averaged value of May and July can be found in Fig. S1. Subsequently, the tropospheric ozone response to crop residue burning (denoted as TOCresp in Eq. (11)) can be modeled as the tropospheric ozone differences between the original well simulated tropospheric ozone (denoted as TOCorig in Eq. (9) below) and the latter one simulated with masked fire occurrence counts (the fire occurrence counts in June has been replaced by the May/July average, while other factors remain unchanged, denoted as TOCmasked in Eq. (10)). The whole process can be formulated as follows:

TOCorig = ft (FC , T , RH , BLH , U , V , UV )

(9)

TOCmasked = ft (FC ∗, T , RH , BLH , U , V , UV )

(10)

TOCresp = TOCorig − TOCmasked

(11)

where FC , T , RH , BLH , U , V and UV are explanatory variables of fire occurrence counts, air temperature, relative humidity, boundary layer height, Zonal wind, Meridional wind and UV radiation, respectively. FC ∗ represents the masked fire occurrence counts time series. ft denotes the simulated models by MLR or ANN. TOCresp denotes the tropospheric ozone response to fire occurrence counts, specifically, the tropospheric ozone response to crop residue burning. Likewise, tropospheric ozone responses to other six explanatory variables were also evaluated through such a sensitivity analysis scheme. 4. Results and discussions 4.1. Covariance of tropospheric ozone and potential influential factors MCA was first used to detect and extract the leading modes of spatiotemporal interactions (known as principal modes hereafter) between tropospheric ozone and six potential explanatory variables (except UV radiation due to the lack of gridded product). The first two principal MCA modes between tropospheric ozone and fire occurrence counts over East Asia explained about 92% of the total covariance (Fig. 2). The first mode (left panel of Fig. 2) explained 59.5% of the covariance, while the extracted spatial pattern of tropospheric ozone and fire occurrence counts resembled an elevated ozone level in South China (Fig. 2a) and increased biomass burning in Southeast Asia (Fig. 2b), respectively. Meanwhile, the associated time series of these two patterns (mode 1) are highly correlated (r = 0.66 , p < 0.001), indicating a potential covarying pattern between tropospheric ozone in South China and crop residue burning over Southeast Asia. Such a deductive finding was also evidenced in previous studies (e.g., Kondo et al., 2004; Lin et al., 2009a; Lin et al., 2013; Chi et al., 2010; Yang et al., 2018). The second MCA mode explained about 32.7% of the total covariance. In contrast to the first mode, the tropospheric ozone pattern resembled a prominent tropospheric ozone elevation over East Asia, especially nearby the NCP (Fig. 2d), where the fire occurrence counts pattern depicted a significant fire occurrence count increase in the NCP (Fig. 2e). These patterns (mode 2) are in accordance with those observed characteristics shown in Fig. 1a and b. In the temporal domain, the associated time series of these two patterns (mode 2) were also highly correlated (r = 0.62 , p < 0.001),

4.2. Statistical linkage between tropospheric ozone and influential factors As stated in Section 3, both linear and nonlinear modeling methods were applied to quantitatively link tropospheric ozone variations to a set of influential factors. In terms of the response variable, tropospheric ozone time series associated with each MCA mode should be averaged to depict the overall ozone variations being modulated by these influential factors collectively. Although tropospheric ozone time series associated with each MCA mode depict a similar ozone variation pattern over East Asia, the regionally averaged (area within the dotted green rectangle shown in Fig. 1a) tropospheric ozone time series was used as the response variable rather than the standardized ozone time series because the latter one is dimensionless whereas our purpose is to quantify the relative amount of tropospheric ozone response to each potent influential factor. Further comparisons of standardized tropospheric ozone time series between the averaged time series derived from MCA modes and the regional averaged one over East Asia indicate that these two time series agreed well with each other with a correlation coefficient of 0.99 (Fig. S5). Differing from the response variable, standardized time series of each atmospheric factor derived from the leading MCA modes were directly used as predictors in the modeling framework except for the fire occurrence counts and UV radiation. The exact number of fire occurrence counts in the NCP was first extracted and then logarithmically transformed to reduce the large amplitudes. 20

Atmospheric Environment 194 (2018) 14–30

M. Ma et al.

Fig. 3. Comparisons between observed and (a) MLR- and (b) ANN-simulated tropospheric ozone time series. The observed tropospheric ozone time series are regional average over East Asia. The ANN simulated tropospheric ozone time series are median values of 100 ensemble trials outputs.

simulated results, the MLR-simulated results are prone to larger uncertainties. This effect could be ascribed to the prescribed linear relationship between tropospheric ozone and explanatory variables by MLR. Given all these seven explanatory variables have been de-correlated beforehand and the correlation coefficients between the observed tropospheric ozone and these variables are all lower than 0.5 (Table S1), we may deduce that the observed tropospheric ozone variations over East Asia can be well explained by these factors collectively since the R2 values of both modeling schemes are greater than 0.95 (i.e., more than 95% variance explained). Additionally, it is observed that simulations from ANN models approximate the observed tropospheric ozone better than those from the MLR model, particularly in predicting extreme values. This effect indicates that interactions between tropospheric ozone and the selected potent explanatory variables exhibit significant nonlinear characteristics that cannot be well characterized via a prescribed linear regression form. In general, all these results suggest that our modeling schemes approximate the observed tropospheric ozone variations in the NCP with excellent accuracy based on the selected influential factors.

Subsequently, a z-score normalization was performed to make it dimensionless. Likewise, the UV radiation time series recorded at Baoding and Tianjin stations were first averaged and then normalized for further modeling practices. MLR was first used to establish the relationship between the observed regional tropospheric ozone variations and seven influential factors along with their possible cross interactions. Statistics associated with the MLR modeling were summarized in Table S2. It is indicative that the contribution of many terms was not statistically significant (pvalue > 0.05). To reduce the complexity of the linear model and to reduce variability associated with those insignificant factors, a simplified linear model was thereby established by only including those significant terms. Comparisons of MLR simulated and observed tropospheric ozone time series are shown in Fig. 3a. The results show that both time series agree well both in the amplitude and seasonality, with an R2 of 0.95 and a root mean squared error (RMSE) of about 1.54 Dobson Units (DU). In contrast with the MLR model that used only significant factors, all seven influential factors were fed directly into the ANN structure to characterize all possible associations. It shows that the ANN-simulated tropospheric ozone time series (represented by the median value of simulated time series from 100 independent trials) are in fair accordance with the observed tropospheric ozone, with an R2 of 0.98 and RMSE of 1.12 DU (Fig. 3b). To assess the modeling uncertainties, the upper and lower bounds of two modeling results at 95% confidence interval were also estimated (Fig. S6). It indicates that both models proximate the observed tropospheric ozone variations with high confidence, with relative larger uncertainties observed primarily in predicting extremes (maximum and minimum). Compared with ANN-

4.3. Quantified tropospheric ozone responses to influential factors By following the sensitivity analysis scheme described in Section 3.3, the tropospheric ozone response to each influential factor was quantified by calculating the arithmetic deviations between two simulated tropospheric ozone time series, with the first one simulated by including all factors whereas another simulated by omitting one factor's 21

Atmospheric Environment 194 (2018) 14–30

M. Ma et al.

Fig. 4. Estimated tropospheric ozone response to fire occurrence counts in June from (a) MLR and (b) ANN modeling frameworks. The error bars depict the relevant uncertainties at 95% confidence interval. The uncertainties for the MLR simulated results were determined by the standard error for each regression coefficient under the 95% confidence interval while the uncertainties for the ANN-simulated results were characterized by the deviations of the simulated tropospheric ozone time series from 100 trials to their median values (median ± 1.96 × standard error).

6d), particularly for those two in 2011 and 2013. To assess the relative contribution of each influential factor to the observed total tropospheric ozone variations in June, the mean response (MR) was first estimated by calculating the arithmetic mean of all tropospheric ozone responses in June via

influence (i.e., replacing the data value in June with the average of May and July). Taking fire occurrence counts for instance, the tropospheric ozone response to fire occurrence counts was quantified as ozone residuals between two simulated ozone time series, with the former one predicted using all regular inputs whereas the latter with reduced fire occurrence counts (Fig. 4). The resulting positive ozone residuals from both MLR and ANN modeling schemes shown in the bottom panel of Fig. 4 resemble evident tropospheric ozone increases over East Asia, by a magnitude of 4–7 DU, accounting for about 8% of the observed total tropospheric ozone in June. The results indicate that the elevated fire occurrence counts in the NCP would result in considerable tropospheric ozone increase over East Asia. Apparently, the MLR-simulated tropospheric ozone response to fire occurrence counts is prescribed to the shape of differential fire occurrence counts time series in June. In other words, the estimated ozone response via MLR is more likely a scaled time series associated with each explanatory variable weighted by the corresponding regression coefficient if no interaction term is considered. However, this assumption does not apply for nonlinear modeling schemes. It is indicative that the ozone residuals derived from ANN differ from those of MLR, which can be further evidenced by different amplitudes and variability of the estimated tropospheric ozone residuals, particularly for those two in 2007 and 2011 (Fig. 4b). Similarly, tropospheric ozone responses to the remaining six factors were also evaluated (Figs. 5 and 6). By referring to the differential time series of each factor (Fig. S1), it is indicative that the estimated tropospheric ozone response associated with each factor is highly prescribed by the variability of differential time series. In regard to each influential factor, the resulting responses from MLR and ANN agree generally well in phases at most timescales except for the data amplitude. More specifically, tropospheric ozone responses derived from MLR have larger amplitudes than those from the ANN modeling scheme (except for fire occurrence counts). In other words, larger variations are observed associated with the MLR-derived tropospheric ozone responses. This effect could be largely attributed to the linear nature of MLR because it only enables to stretch (or compress) the original explanatory variable by one constant scaling factor and hence it is incapable of modulating the internal variations within time series. In terms of the ANN-based nonlinear modeling framework, such an effect was mitigated by a large extent, which can be further evidenced by the estimated ozone response to the boundary layer height (Figs. 5d and

MRj =

1 N

N

∑ Rij

(12)

i=1

where i denotes the sampling number of time and j represents the jth influential factor. Rij is the tropospheric ozone response to the jth influential factor in the ith year. Subsequently, the mean absolute response fraction (MARF) associated with each factor was calculated to assess the relative importance of each factor contributing to the overall variations by

MARFj =

MRj 7 ∑j = 1

MRj

∗100% (13)

Statistics of MR and MARF associated with each factor were summarized in Table 2. It is indicative that the estimated MR of tropospheric ozone associated with each factor agrees well in both data amplitudes and phases (positive or negative) from both modeling frameworks. Compared with other six influential factors, the fire occurrence counts played the first leading role in contributing to the observed ozone increases, with a MR of about +3.82 DU (+4.06 DU) derived from the ANN (MLR) modeling scheme (Table 2). Meanwhile, it explains a fraction of 39.14% (32.98%) total variances of the observed tropospheric ozone variations. The MARF of fire occurrence counts related tropospheric ozone fully resembles the predominant role of fire occurrence counts in contributing to the observed tropospheric ozone increases over East Asia. By contrast, zonal wind played the second leading role, which yielded an ozone increase by +2.52 DU (+2.69 DU from MLR), explaining a fraction of total variance by 25.82% (21.85%). Such an effect (i.e., a positive tropospheric ozone response to zonal wind) could be ascribed to the fact that the original zonal wind in June was weaker than that in May and July (see in Fig. S1c), hence the crop residue burning yielded ozone could accumulate there which in turn resulted in the elevation of ozone amount. In spite of the critical importance in ozone formation, UV radiation played the third leading role 22

Atmospheric Environment 194 (2018) 14–30

M. Ma et al.

Fig. 5. Quantified tropospheric ozone responses to the remaining six influential factors in June via MLR. (a) UV radiation, (b) Zonal wind, (c) Meridional wind, (d) boundary layer height, (e) temperature, and (f) humidity. The error bars depict the relevant uncertainties at 95% confidence interval.

the correlation between the response variable and predictors (Table 2). For instance, the smallest uncertainty is observed to be associated with UV radiation (with a standard error of 0.01) because it had the largest correlation with ozone time series (Table S1). Nonetheless, no such effect is observed in the ANN modeling framework. In spite of the slight difference between MLR- and ANN-based tropospheric ozone responses, the resulting statistics from both modeling schemes have well demonstrated the leading role of fire occurrence counts, zonal wind, UV radiation and meridional wind in modulating regional tropospheric ozone increases over East Asia.

in modulating the observed total tropospheric ozone variations over East Asia. Differing from crop residue burning and zonal wind, the UV radiation related tropospheric ozone response is negative (−1.58 DU from ANN and −2.79 DU from MLR) because the UV radiation in June is lower than the averaged UV radiation in May and July (Fig. S1b). Therefore, the ozone response to UV radiation in June will be negative once the original smaller value was replaced with a larger UV radiation in sensitivity analysis (smaller minus larger ozone value). Additionally, meridional wind (V) accounts for more than 10% total variance of the observed tropospheric ozone variations. In general, these four factors have collectively explained more than 90% of the observed total tropospheric ozone variations in June over East Asia. In contrast to the four identified primary contributing factors, the remaining three variables played a relatively weak role in modulating the observed tropospheric ozone increases. Although the ozone related photochemical process is highly related to ambient air temperature, the estimated tropospheric ozone response to temperature is almost negligible (within 1 DU at most time scales). The reason could be largely ascribed to the fact that the temperature in June differs slightly from that in May and July (Fig. S1f), thereby the temperature difference is not salient. Moreover, the air temperature is high enough in summer for the ozone photochemical process and thus small temperature difference would not result in considerable changes in total amount of tropospheric ozone. Similar effects could be also found for boundary layer height and humidity and such results are also consistent with previous findings (e.g., Haman et al., 2014; Porter et al., 2015; Xing et al., 2017). Comparisons of the estimated uncertainties (i.e., standard error) for MR from MLR and ANN modeling frameworks indicate that uncertainties in the MLR modeling framework are largely influenced by

4.4. Mechanisms linking tropospheric ozone increases over East Asia to crop residue burning in the NCP As evidenced by the sensitivity analysis results, fire occurrence counts, zonal wind, UV radiation, and meridional wind are four primary contributing factors associated with the observed tropospheric ozone increases in June over East Asia. Since the crop residue burning occurs primarily in the NCP whereas the observed tropospheric ozone increases present mainly over East Asia, we hypothesized that the observed ozone increases over East Asia could be related to the elevated concentrations of ozone precursors such as NOX and VOCs emitted from mass crop residue burning in the NCP. Such an assumption is reasonable as the summertime ozone photochemical process over East Asia is oftentimes determined by the relative concentrations of NOX and VOCs rather than other factors like UV radiation and temperature, thereby the elevated NOX concentration would result in significant tropospheric ozone increase therein. To corroborate the crop residue burning related NOX increase over 23

Atmospheric Environment 194 (2018) 14–30

M. Ma et al.

Fig. 6. Same as in Fig. 5 but for the ANN-derived estimates.

subtracting NOX emissions in April from that of June. As indicated in Fig. 7a, the spatial distribution of the averaged NOX emissions differences between June and April fully resembles the distribution of detected fire occurrence counts in the NCP shown in Fig. 1b, in particular the coupled mode in Fig. 2e. Likewise, the fire occurrence counts anomalies were derived by performing the same isolation scheme to fire occurrence counts in the NCP. It shows that both anomalies agreed well in the temporal domain, with a correlation coefficient of 0.88 between them (Fig. 7b). Such a finding is informative, revealing that crop residue burning in the NCP could result in large amount of NOX emissions in June, which is also in accordance with findings from previous studies (e.g., Guo et al., 2004; Yan et al., 2006; Zhang et al., 2008; Suthawaree et al., 2010; Huang et al., 2012; Li et al., 2016; Wu et al., 2016). By referring to the annual variability of tropospheric NO2 concentrations (the reason to use tropospheric NO2 concentrations rather than NOX emissions here is due to the former depicts the total NO2 amount whereas the latter mainly reflects the anthropogenic emissions without accounting for the baseline amount in nature), we may find that the concentration of NO2 oftentimes maintains a relative low value in June (Fig. S7). Such a regime renders the ozone production processes be NOX-limited (Archibald et al., 2011; Monks et al., 2015; Jin et al., 2017). Hence, once the crop residue burning emitted NOX enters into the atmosphere, it will result in net ozone production which in turn leads to the increase of ozone mixing ratios and thereby the elevation of tropospheric ozone concentrations therein. This induction is also in accordance with results from previous studies in the literature (e.g., Li et al., 2008; Yamaji et al., 2010; Ding et al., 2013; Kanaya et al., 2013; Tang et al., 2013; Baylon et al., 2015; Pan et al., 2015; Stavrakou et al., 2016; Lu et al., 2017; Kumari et al., 2018). In addition to the critical role of NOX, VOCs could also play an essential role in the observed

Table 2 Estimated total tropospheric ozone response in June to each influential factor from MLR and ANN modeling frameworks. Uncertainties for mean response (MR) were characterized by one standard error and showed in brackets. Explanatory Variables

Fire occurrence counts Zonal wind (U) UV radiation Meridional wind (V) Boundary layer height Relative humidity Temperature

MLR

ANN

MR (DU)

MARF

MR (DU)

MARF

+4.06 +2.69 −2.79 +1.62 −0.47 +0.48 −0.20

32.98% 21.85% 22.66% 13.16% 3.82% 3.90% 1.63%

+3.82 +2.52 −1.58 +1.00 −0.53 +0.15 −0.16

39.14% 25.82% 16.19% 10.25% 5.43% 1.54% 1.63%

(0.63) (0.09) (0.01) (0.41) (0.16) (0.09) (0.02)

(0.17) (0.19) (0.23) (0.18) (0.19) (0.18) (0.18)

the NCP, the estimated monthly NOX emission inventory over the NCP from 2005 (the ozone record starts from October 2004 and thus no data is available in June 2004) to 2014 were extracted and compared with the variability of fire occurrence counts. Given the estimated NOX concentrations are total amounts of all types of emissions in each month, this product cannot be directly applied to depict crop residue burning related NOX emissions. By taking the seasonality of NOX emissions and crop residue burning into account, NOX emissions in April were subtracted from that of June to isolate crop residue burning related NOX emissions over the NCP. The reasons to choose NOX emissions in April as a baseline can be mainly ascribed to: (1) NOX emissions in April are less likely to be influenced by prominent heatingrelated NOX emissions because of wintertime coal burning, and (2) crop residue burning is seldomly detected in April. Therefore, crop residue burning related NOX emissions over the NCP could be quantified by 24

Atmospheric Environment 194 (2018) 14–30

M. Ma et al.

Fig. 7. Estimated crop residue burning related NOX emissions over the NCP during 2005–2014. (a) Spatial distribution of the averaged NOX emissions differences between June and April during 2005–2014, and (b) temporal variability of averaged crop residue burning related NOX emissions and fire occurrence counts over the NCP (the blue dotted region). In terms of fire occurrence counts time series, data values in April were also subtracted from those in June and then averaged over the NCP. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

and informative. By taking advantage of this statistical modeling scheme, the impacts of summertime crop residue burning in the NCP on regional tropospheric ozone increase over East Asia in June were quantitatively assessed. Relative to other potent influential factors such as temperature, humidity, and boundary layer height that were associated with ozone variations, crop residue burning in the NCP played the predominant role in contributing to the observed regional tropospheric ozone increase therein, by yielding a tropospheric ozone elevation of about 4.0 DU on average, accounting for nearly 8% of the observed tropospheric ozone column in June. This result clearly demonstrates the linkage between crop residue burning in the NCP and the summertime tropospheric ozone increase over East Asia. Additionally, UV radiation, zonal and meridional winds were proved to be another three leading factors largely contributing to the observed regional tropospheric ozone increases. These four essential contributing factors collectively explained more than 90% of the observed tropospheric ozone variations. Further investigations revealed a mechanism that the elevated NOX concentrations resulting from summertime crop residue burning in the NCP yielded significant net ozone production in a NOX-limited Ozone-VOCNOX regime. Subsequently, the increased ozone productions were transported to East Asia via atmospheric circulation and then accumulated therein, which in turn resulted in regional tropospheric ozone peaks over there. In this study, MCA was used to decouple complex interactions between tropospheric ozone and a set of influential factors in the spatiotemporal domain for dimensionality reduction. Unlike other methods like CCA, MCA maximizes covariance rather than correlation between two variables, which renders MCA great superiority in environmental modeling. Nevertheless, expert knowledge is oftentimes required to judge which coupled mode resembles the desired pattern when

tropospheric ozone increase. However, the tropospheric ozone response to VOC was not evaluated in this study due to the lack of proper VOC data sets. To examine the spatial linkage between crop residue burning related NOX emissions in the NCP and tropospheric ozone increases over East Asia in June, the monthly mean wind fields at 850 hPa in June during 2004–2016 over the study region were estimated (Fig. 8). A prevailing south westerly wind was observed in the low troposphere over the NCP and the adjacent coastal regions, which transported ozone and its precursors from the NCP to its downwind regions (i.e., East Asia). Such a finding may explain why crop residue burning and the regional ozone increases are spatially inconsistent. Specifically, the crop residue burning yielded ozone and its precursor emissions in the NCP were quickly transported to East Asia through atmospheric circulation and then accumulated therein (due to relative weak zonal wind therein), which resulted in the observed regional tropospheric ozone increases in turn. 5. Conclusions Using a synthesis of remotely sensed tropospheric ozone measurements from OMI, fire occurrence counts from MODIS, meteorological fields from ERA-Interim reanalysis, and estimated NOX emission from inverse analysis studies, the associations between tropospheric ozone increase over East Asia and crop residue burning in the NCP in June were fairly evaluated based on a delicate statistical modeling framework. Differing from many studies applying complex chemical transport models in conjunction with meteorological models such as Weather Research and Forecasting model for atmospheric environment simulation and attribution, such an evidence-based modeling framework is not only computational efficient but also physically meaningful 25

Atmospheric Environment 194 (2018) 14–30

M. Ma et al.

Fig. 8. Monthly mean wind fields at 850 hpa in June over East Asia during 2004–2016. The region within the dotted red rectangle depicts the NCP region where significant crop residue burning presents. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Appendix A. Supplementary data

performing MCA analysis. To link the observed response variable to influential factors, both linear (MLR) and nonlinear (ANN) modeling methods were employed. In terms of MLR modeling scheme, quadratic interactions among explanatory variables were also considered in order to better approximate the observed response variable. Overall, our results provide an evidence-based mechanism linking the observed tropospheric ozone increase to crop residue burning over the NCP in June, which can be further justified by using simulations from physical meaningful atmospheric photochemical models.

Supplementary data to this article can be found online at https:// doi.org/10.1016/j.atmosenv.2018.09.018. References Ainsworth, E.A., Yendrek, C.R., Sitch, S., Collins, W.J., Emberson, L.D., 2012. The effects of tropospheric ozone on net primary productivity and implications for climate change. Annu. Rev. Plant Biol. 63, 637–661. https://doi.org/10.1146/annurevarplant-042110-103829. Akagi, S.K., Craven, J.S., Taylor, J.W., McMeeking, G.R., Yokelson, R.J., Burling, I.R., Urbanski, S.P., Wold, C.E., Seinfeld, J.H., Coe, H., Alvarado, M.J., Weise, D.R., 2012. Evolution of trace gases and particles emitted by a chaparral fire in California. Atmos. Chem. Phys. 12, 1397–1421. https://doi.org/10.5194/acp -12-1397-2012. Allaart, M., van Weele, M., Fortuin, P., Kelder, H., 2004. An empirical model to predict the UV-index based on solar zenith angles and total ozone. Meteorol. Appl. 11, 59–65. https://doi.org/10.1017/S1350482703001130. Alvarado, M.J., Logan, J.A., Mao, J., Apel, E., Riemer, D., Blake, D., Cohen, R.C., Min, K.E., Perring, A.E., Browne, E.C., Wooldridge, P.J., Diskin, G.S., Sachse, G.W., Fuelberg, H., Sessions, W.R., Harrigan, D.L., Huey, G., Liao, J., Case-Hanks, A., Jimenez, J.L., Cubison, M.J., Vay, S.A., Weinheimer, A.J., Knapp, D.J., Montzka, D.D., Flocke, F.M., Pollack, I.B., Wennberg, P.O., Kurten, A., Crounse, J., St Clair, J.M., Wisthaler, A., Mikoviny, T., Yantosca, R.M., Carouge, C.C., Le Sager, P., 2010. Nitrogen oxides and PAN in plumes from boreal fires during ARCTAS-B and their impact on ozone: an integrated analysis of aircraft and satellite observations. Atmos. Chem. Phys. 10, 9739–9760. https://doi.org/10.5194/acp-10-9739-2010. Alvarado, M.J., Lonsdale, C.R., Yokelson, R.J., Akagi, S.K., Coe, H., Craven, J.S., Fischer, E.V., McMeeking, G.R., Seinfeld, J.H., Soni, T., Taylor, J.W., Weise, D.R., Wold, C.E., 2015. Investigating the links between ozone and organic aerosol chemistry in a biomass burning plume from a prescribed fire in California chaparral. Atmos. Chem. Phys. 15, 6667–6688. https://doi.org/10.5194/acp-15-6667-2015. Amiridis, V., Zerefos, C., Kazadzis, S., Gerasopoulos, E., Eleftheratos, K., Vrekoussis, M., Stohl, A., Mamouri, R.E., Kokkalis, P., Papayannis, A., Eleftheriadis, K., Diapouli, E., Keramitsoglou, I., Kontoes, C., Kotroni, V., Lagouvardos, K., Marinou, E., Giannakaki, E., Kostopoulou, E., Giannakopoulos, C., Richter, A., Burrows, J.P., Mihalopoulos, N., 2012. Impact of the 2009 Attica wild fires on the air quality in urban Athens. Atmos. Environ. 46, 536–544. https://doi.org/10.1016/j.atmosenv.2011.07.056. Andersson, E., Bauer, P., Beljaars, A., Chevallier, F., Hólm, E., Janisková, M., Kallberg, P., Kelly, G., Lopez, P., McNally, A., Moreau, E., Simmons, A.J., Thépaut, J.N., Tompkins, A.M., 2005. Assimilation and modeling of the atmospheric hydrological cycle in the ECMWF forecasting system. Bull. Am. Meteorol. Soc. 86, 387–402. https://doi.org/10.1175/BAMS-86-3-387. Archibald, A.T., Levine, J.G., Abraham, N.L., Cooke, M.C., Edwards, P.M., Heard, D.E., Jenkin, M.E., Karunaharan, A., Pike, R.C., Monks, P.S., Shallcross, D.E., Telford, P.J., Whalley, L.K., Pyle, J.A., 2011. Impacts of HOx regeneration and recycling in the oxidation of isoprene: consequences for the composition of past, present and future

Conflicts of interest The authors declare no conflict of interest.

Acknowledgements The authors would like to acknowledge three anonymous reviewers for their constructive comments and valuable suggestions in helping improve the manuscript. This work was supported by the National Key Research and Development Program of China (2016YFC1302602), Shanghai Municipal Commission of Health and Family Planning (15GWZK0201) and the Fundamental Research Funds for the Central Universities (East China Normal University). The authors acknowledge NASA Goddard Space Flight Center for providing tropospheric ozone (https://acd-ext.gsfc.nasa.gov/Data_services/cloud_slice/new_data. html) and NO2 data (https://mirador.gsfc.nasa.gov/cgi-bin/mirador/ collectionlist.pl), the Land Processes Distributed Active Archive Center (LP-DAAC) for providing fire occurrence counts product (https:// ladsweb.modaps.eosdis.nasa.gov/search/order/1/MOD14A1–0), ECMWF for providing ERA-Interim reanalysis (http://apps.ecmwf.int/ datasets/data/interim-full-daily/levtype=pl/), the TEMIS group for providing UV dose data (http://www.temis.nl/uvradiation/UVarchive/ stations_uv_msr2.html), and the Peking University group for providing NOX emission inventory (http://inventory.pku.edu.cn/download/ download.html).

26

Atmospheric Environment 194 (2018) 14–30

M. Ma et al.

https://doi.org/10.1016/j.atmosenv.2008.08.013. Derrien, M., Le Gléau, H., 2005. MSG/SEVIRI cloud mask and type from SAFNWC. Int. J. Rem. Sens. 26, 4707–4732. https://doi.org/10.1080/01431 160500166128. Dee, D.P., Uppala, S.M., Simmons, A.J., Berrisford, P., Poli, P., Kobayashi, S., Andrae, U., Balmaseda, M.A., Balsamo, G., Bauer, P., Bechtold, P., Beljaars, A.C.M., van de Berg, L., Bidlot, J., Bormann, N., Delsol, C., Dragani, R., Fuentes, M., Geer, A.J., Haimberger, L., Healy, S.B., Hersbach, H., Hólm, E.V., Isaksen, L., Kållberg, P., Köhler, M., Matricardi, M., Mcnally, A.P., Monge-Sanz, B.M., Morcrette, J.J., Park, B.K., Peubey, C., de Rosnay, P., Tavolato, C., Thépaut, J.N., Vitart, F., 2011. The ERAInterim reanalysis: configuration and performance of the data assimilation system. Q. J. R. Meteorol. Soc. 137, 553–597. https://doi.org/10.1002/qj.828. Derwent, R.G., Utembe, S.R., Jenkin, M.E., Shallcross, D.E., 2015. Tropospheric ozone production regions and the intercontinental origins of surface ozone over Europe. Atmos. Environ. 112, 216–224. https://doi.org/10.1016/j.atmosenv.2015.04.049. Ding, A.J., Wang, T., Thouret, V., Cammas, J.-P., Nédélec, P., 2008. Tropospheric ozone climatology over Beijing: analysis of aircraft data from the MOZAIC program. Atmos. Chem. Phys. 8, 1–13. https://doi.org/10.5194/acp-8-1-2008. Ding, A.J., Fu, C.B., Yang, X.Q., Sun, J.N., Zheng, L.F., Xie, Y.N., Herrmann, E., Nie, W., Petaja, T., Kerminen, V.M., Kulmala, M., 2013. Ozone and fine particle in the western Yangtze River Delta: an overview of 1 yr data at the SORPES station. Atmos. Chem. Phys. 13, 5813–5830. https://doi.org/10.5194/acp-13-5813-2013. Ding, Q., Steig, E.J., Battisti, D.S., Küttel, M., 2011. Winter warming in West Antarctica caused by central tropical Pacific warming. Nat. Geosci. 4, 398–403. https://doi.org/ 10.1038/ngeo1129. Duncan, B.N., Bey, I., Chin, M., Mickley, L.J., Fairlie, T.D., Martin, R.V., Matsueda, H., 2003. Indonesian wildfires of 1997: impact on tropospheric chemistry. J. Geophys. Res. 108, 4458. https://doi.org/10.1029/2002JD 003195. Dufour, G., Eremenko, M., Orphal, J., Flaud, J.-M., 2010. IASI observations of seasonal and day-to-day variations of tropospheric ozone over three highly populated areas of China: Beijing, Shanghai, and Hong Kong. Atmos. Chem. Phys. 10, 3787–3801. https://doi.org/10.5194/acp-10-3787-2010. Ebi, K.L., McGregor, G., 2008. Climate change, tropospheric ozone and particulate matter, and health impacts. Environ. Health Perspect. 116, 1449–1455. https://doi.org/10. 1289/ehp.11463. FAOSTAT, 2014. Statistics Division of the Food Agricultural Organisation (FAO). Live Animal Numbers, Crop Production, Total Nitrogen Fertiliser Consumption Statistics till 2012. pp. 15. http://www.fao.org/faostat/en/#data/QD info retrieved 10.2016, 2016a. Fishman, J., Watson, C.E., Larsen, J.C., Logan, J.A., 1990. Distribution of tropospheric ozone determined from satellite data. J. Geophys. Res. 95, 3599. https://doi.org/10. 1029/JD095iD04p03599. Fishman, J., Wozniak, A.E., Creilson, J.K., 2003. Global distribution of tropospheric ozone from satellite measurements using the empirically corrected tropospheric ozone residual technique: identification of the regional aspects of air pollution. Atmos. Chem. Phys. Atmospheric Chemistry and Physics 3, 893–907. https://doi.org/10.5194/ acpd-3-1453-2003. Giglio, L., 2015. MODIS Collection 6 Active Fire Product User's Guide Revision A. Department of Geographical Sciences, University of Maryland Unpublished manuscript. https://lpdaac.usgs.gov/sites/default/files/public/product_documentation/ mod14_user_guide.pdfs. Giglio, L., Schroeder, W., Justice, C.O., 2016. The collection 6 MODIS active fire detection algorithm and fire products. Rem. Sens. Environ.Environ, 178, 31–41. https://doi. org/10.1016/j.rse.2016.02.054. Guo, H., Wang, T., Simpson, I.J., Blake, D.R., Yu, X.M., Kwok, Y.H., Li, Y.S., 2004. Source contributions to ambient VOCs and CO at a rural site in eastern China. Atmos. Environ. 38, 4551–4560. https://doi.org/10.1016/j.atmosenv.2004.05.004. Guo, J., Miao, Y., Zhang, Y., Liu, H., Li, Z., Zhang, W., He, J., Lou, M., Yan, Y., Bian, L., Zhai, P., 2016. The climatology of planetary boundary layer height in China derived from radiosonde and reanalysis data. Atmos. Chem. Phys. 16, 13309–13319. https:// doi.org/10.5194/acp-16-13309-2016. Guenther, A.B., Jiang, X., Heald, C.L., Sakulyanontvittaya, T., Duhl, T., Emmons, L.K., Wang, X., 2012. The model of emissions of gases and aerosols from nature version 2.1 (MEGAN2.1): an extended and updated framework for modeling biogenic emissions. Geosci. Model Dev. (GMD) 5, 1471–1492. https://doi.org/10.5194/gmd-5-14712012. Haman, C.L., Couzo, E., Flynn, J.H., Vizuete, W., Heffron, B., Lefer, B.L., 2014. Relationship between boundary layer heights and growth rates with ground-level ozone in Houston, Texas. J. Geophys. Res.: Atmosphere 119, 6230–6245. https://doi. org/10.1002/2013JD020473. Hagan, M.T., Menhaj, M.B., 1994. Training Feedfoward networks with the Marquardt algorithn. Neural network. IEEE Trans. 5, 989–993. https://doi.org/10.1109/72. 329697. He, Y.J., Uno, I., Wang, Z.F., Pochanart, P., Li, J., Akimoto, H., 2008. Significant impact of the East Asia monsoon on ozone seasonal behavior in the boundary layer of Eastern China and the west Pacific region. Atmos. Chem. Phys. 8, 7543–7555. https://doi. org/10.5194/acpd-8-14927-2008. Huang, X., Li, M., Li, J., Song, Y., 2012. A high-resolution emission inventory of crop burning in fields in China based on MODIS Thermal Anomalies/Fire products. Atmos. Environ. 50, 9–15. https://doi.org/10.1016/j.atmosenv.2012.01.017. Huang, G., Liu, X., Chance, K., Yang, K., Bhartia, P.K., Cai, Z., Allaart, M., Ancellet, G., Calpini, B., Coetzee, G.J.R., Cuevas-Agulló, E., Cupeiro, M., De Backer, H., Dubey, M.K., Fuelberg, H.E., Fujiwara, M., Godin-Beekmann, S., Hall, T.J., Johnson, B., Joseph, E., Kivi, R., Kois, B., Komala, N., König-Langlo, G., Laneve, G., Leblanc, T., Marchand, M., Minschwaner, K.R., Morris, G., Newchurch, M.J., Ogino, S.Y., Ohkawara, N., Piters, A.J.M., Posny, F., Querel, R., Scheele, R., Schmidlin, F.J., Schnell, R.C., Schrems, O., Selkirk, H., Shiotani, M., Skrivánková, P., Stübi, R., Taha,

atmospheres. Geophys. Res. Lett. 38, 1–6. https://doi.org/10.1029/2010GL046520. Bai, K., Chang, N.B., Gao, W., 2016. Quantification of relative contribution of Antarctic ozone depletion to increased austral extratropical precipitation during 1979–2013. J. Geophys. Res.: Atmosphere 121, 1459–1474. https://doi.org/10.1002/ 2015JD024247. Bai, K., Chang, N.B., Shi, R., Yu, H., Gao, W., 2017. An intercomparison of multidecadal observational and reanalysis data sets for global total ozone trends and variability analysis. J. Geophys. Res.: Atmosphere 122, 7119–7139. https://doi.org/10.1002/ 2016JD025835. Bärtsch-Ritter, N., Keller, J., Dommen, J., Prévôt, a. S.H., 2004. Effects of various meteorological conditions and spatial emission resolutions on the ozone concentration & ROG/NOx limitation in the Milan area (I). Atmos. Chem. Phys. Discuss. 3, 733–768. https://doi.org/10.5194/acp-4-423-2004. Bethan, S., Vaughan, G., Reid, S.J., 1996. A comparison of ozone and thermal tropopause heights and the impact of tropopause definition on quantifying the ozone content of the troposphere. Q. J. R. Meteorol. Soc. 122, 929–944. https://doi.org/10.1256/ smsqj.53206. Baylon, P., Jaffe, D.A., Wigder, N.L., Gao, H., Hee, J., 2015. Ozone enhancement in western US wildfire plumes at the Mt. Bachelor Observatory: the role of NOx. Atmos. Environ. 109, 297–304. https://doi.org/10.1016/j.atmosenv.2014.09.013. Bormann, N., Thépaut, J.-N., 2004. Impact of MODIS polar winds in ECMWF's 4DVAR data assimilation system. Mon. Weather Rev. 132, 929–940. https://doi.org/10. 1175/1520-0493(2004)132<0929:IOMPWI>2.0.CO;2. Bretherton, C.S., Smith, C., Wallace, J.M., 1992. An intercomparison of methods for finding coupled patterns in climate data. J. Clim. 5, 541–560. https://doi.org/10. 1175/1520-0442(1992)005<0541:AIOMFF>2.0.CO;2. Chan, C.Y., Chan, L.Y., Harris, J.M., Oltmans, S.J., Blake, D.R., Qin, Y., Zheng, Y.G., Zheng, X.D., 2003. Characteristics of biomass burning emission sources, transport, and chemical speciation in enhanced springtime tropospheric ozone profile over Hong Kong. J. Geophys. Res. 108, 4015. https://doi.org/10.1029/2001jd001555. Chan, C.Y., Wong, K.H., Li, Y.S., Chan, L.Y., Zheng, X.D., 2006. The effects of Southeast Asia fire activities on tropospheric ozone, trace gases and aerosols at a remote site over the Tibetan Plateau of Southwest China. Tellus Ser. B Chem. Phys. Meteorol. 58, 310–318. https://doi.org/10.1111/j.1600-0889.2006.00187.x. Chandra, S., Ziemke, J.R., Bhartia, P.K., Martin, R.V., 2002. Tropical tropospheric ozone: implications for dynamics and biomass burning. J. Geophy. Res. Atmos. 107, 1–17. https://doi.org/10.1029/2001JD000447. Chandra, S., Ziemke, J.R., Martin, R.V., 2003. Tropospheric ozone at tropical and middle latitudes derived from TOMS/MLS residual: comparison with a global model. J. Geophys. Res.: Atmosphere 108https://doi.org/10.1029/2002JD002912. n/a–n/a. Chandra, S., Ziemke, J.R., Duncan, B.N., Diehl, T.L., Livesey, N.J., Froidevaux, L., 2009. Effects of the 2006 El Niño on tropospheric ozone and carbon monoxide: implications for dynamics and biomass burning. Atmos. Chem. Phys. 9, 4239–4249. https://doi. org/10.5194/acp-9-4239-2009. Chen, X., Škerlak, B., Rotach, M.W., Añel, J.A., Su, Z., Ma, Y., Li, M., 2016. Reasons for the extremely high-ranging planetary boundary layer over the western Tibetan plateau in winter. J. Atmos. Sci. 73, 2021–2038. https://doi.org/10.1175/JAS-D-150148.1. Chen, Y., Velicogna, I., Famiglietti, J.S., Randerson, J.T., 2013. Satellite observations of terrestrial water storage provide early warning information about drought and fire season severity in the Amazon. J. Geophys. Res.: Biogeosciences 118, 495–504. https://doi.org/10.1002/jgrg.20046. Chi, K.H., Lin, C.Y., Yang, C.F.O., Wang, J.L., Lin, N.H., Sheu, G.R., Lee, C. Te, 2010. PCDD/F measurement at a high-altitude station in central Taiwan: evaluation of longrange transport of PCDD/FS during the southeast Asia biomass burning event. Environ. Sci. Technol. 44, 2954–2960. https://doi.org/10.1021/es1000984. Chuvieco, E., Giglio, L., Justice, C., 2008. Global characterization of fire activity: toward defining fire regimes from Earth observation data. Global Change Biol. 14, 1488–1502. https://doi.org/10.1111/j.1365-2486.2008.01585.x. Cristofanelli, P., Bracci, A., Sprenger, M., Marinoni, A., Bonafé, U., Calzolari, F., Duchi, R., Laj, P., Pichon, J.M., Roccato, F., Venzac, H., Vuillermoz, E., Bonasoni, P., 2010. Tropospheric ozone variations at the Nepal climate Observatory-Pyramid (Himalayas, 5079 m a.s.l.) and influence of deep stratospheric intrusion events. Atmos. Chem. Phys. 10, 6537–6549. https://doi.org/10.5194/acp-10-6537-2010. Cristofanelli, P., Fierli, F., Marinoni, A., Calzolari, F., Duchi, R., Burkhart, J., Stohl, A., Maione, M., Arduini, J., Bonasoni, P., 2013. Influence of biomass burning and anthropogenic emissions on ozone, carbon monoxide and black carbon at the Mt. Cimone GAW-WMO global station (Italy, 2165ma.s.l.). Atmos. Chem. Phys. 13, 15–30. https://doi.org/10.5194/acp-13-15-2013. Crippa, M., Janssens-Maenhout, G., Dentener, F., Guizzardi, D., Sindelarova, K., Muntean, M., Van Dingenen, R., Granier, C., 2016. Forty years of improvements in European air quality: regional policy-industry interactions with global impacts. Atmos. Chem. Phys. 16, 3825–3841. https://doi.org/10.5194/acp- 16-3825-2016. Crippa, M., Guizzardi, D., Muntean, M., Schaaf, E., Dentener, F., Aardenne, J.A. va, Monni, S., Doering, U., Olivier, J.G.J., Pagliari, V., Janssens-Maenhout, G., 2018. Gridded emissions of air pollutants for the period 1970-2012 within EDGAR v4.3.2. Earth Syst. Sci. Data Discuss. 1–40. https://doi.org/10.5194/essd -2018-31. David, L.M., Girach, I.A., Nair, P.R., 2011. Distribution of ozone and its precursors over Bay of Bengal during winter 2009: role of meteorology. Ann. Geophys. 29, 1613–1627. https://doi.org/10.5194/angeo-29-1613-2011. Dawson, J.P., Adams, P.J., Pandis, S.N., 2007. Sensitivity of ozone to summertime climate in the eastern USA: a modeling case study. Atmos. Environ. 41, 1494–1511. https:// doi.org/10.1016/j.atmosenv.2006.10.033. Deng, X., Tie, X., Zhou, X., Wu, D., Zhong, L., Tan, H., Li, F., Huang, X., Bi, X., Deng, T., 2008. Effects of Southeast Asia biomass burning on aerosols and ozone concentrations over the Pearl River Delta (PRD) region. Atmos. Environ. 42, 8493–8501.

27

Atmospheric Environment 194 (2018) 14–30

M. Ma et al.

the Indo-Gangetic Plain. Environ. Sci. Pollut. Control Ser. 1–15. https://doi.org/10. 1007/s11356-018-2034-y. Lanzante, J.R., 1984. A rotated Eigenanalysis of the correlation between 700 mb heights and sea surface temperatures in the Pacific and Atlantic. Mon. Weather Rev. 112, 2270–2280. https://doi.org/10.1175/1520-0493(1984)112<2270:AREOTC>2.0. CO;2. Leventidou, E., Eichmann, K.U., Weber, M., Burrows, J.P., 2016. Tropical tropospheric ozone columns from nadir retrievals of GOME-1/ERS-2, SCIAMACHY/Envisat, and GOME-2/MetOp-A (1996-2012). Atmos. Measur. Tech. 9, 3407–3427. https://doi. org/10.5194/amt-9-3407-2016. Li, J., Wang, Z., Akimoto, H., Gao, C., Pochanart, P., Wang, X., 2007. Modeling study of ozone seasonal cycle in lower troposphere over east Asia. J. Geophy. Res. Atmos. 112, 1–15. https://doi.org/10.1029/2006JD008209. Li, J., Wang, Z., Akimoto, H., Yamaji, K., Takigawa, M., Pochanart, P., Liu, Y., Tanimoto, H., Kanaya, Y., 2008. Near-ground ozone source attributions and outflow in Central Eastern China during MTX2006. Atmos. Chem. Phys. 8, 7335–7351. https://doi.org/ 10.5194/acp-8-7335-2008. Li, J., Li, Y., Bo, Y., Xie, S., 2016. High-resolution historical emission inventories of crop residue burning in fields in China for the period 1990-2013. Atmos. Environ. 138, 152–161. https://doi.org/10.1016/j.atmosenv.2016.05.002. Lim, S.S., Vos, T., Flaxman, A.D., Danaei, G., Shibuya, K., Adair-Rohani, H., Amann, M., Anderson, H.R., Andrews, K.G., Aryee, M., Atkinson, C., Bacchus, L.J., Bahalim, A.N., Balakrishnan, K., Balmes, J., Barker-Collo, S., Baxter, A., Bell, M.L., Blore, J.D., Blyth, F., Bonner, C., Borges, G., Bourne, R., Boussinesq, M., Brauer, M., Brooks, P., Bruce, N.G., Brunekreef, B., Bryan-Hancock, C., Bucello, C., Buchbinder, R., Bull, F., Burnett, R.T., Byers, T.E., Calabria, B., Carapetis, J., Carnahan, E., Chafe, Z., Charlson, F., Chen, H., Chen, J.S., Cheng, A.T.A., Child, J.C., Cohen, A., Colson, K.E., Cowie, B.C., Darby, S., Darling, S., Davis, A., Degenhardt, L., Dentener, F., Des Jarlais, D.C., Devries, K., Dherani, M., Ding, E.L., Dorsey, E.R., Driscoll, T., Edmond, K., Ali, S.E., Engell, R.E., Erwin, P.J., Fahimi, S., Falder, G., Farzadfar, F., Ferrari, A., Finucane, M.M., Flaxman, S., Fowkes, F.G.R., Freedman, G., Freeman, M.K., Gakidou, E., Ghosh, S., Giovannucci, E., Gmel, G., Graham, K., Grainger, R., Grant, B., Gunnell, D., Gutierrez, H.R., Hall, W., Hoek, H.W., Hogan, A., Hosgood, H.D., Hoy, D., Hu, H., Hubbell, B.J., Hutchings, S.J., Ibeanusi, S.E., Jacklyn, G.L., Jasrasaria, R., Jonas, J.B., Kan, H., Kanis, J.A., Kassebaum, N., Kawakami, N., Khang, Y.H., Khatibzadeh, S., Khoo, J.P., Kok, C., Laden, F., Lalloo, R., Lan, Q., Lathlean, T., Leasher, J.L., Leigh, J., Li, Y., Lin, J.K., Lipshultz, S.E., London, S., Lozano, R., Lu, Y., Mak, J., Malekzadeh, R., Mallinger, L., Marcenes, W., March, L., Marks, R., Martin, R., McGale, P., McGrath, J., Mehta, S., Mensah, G.A., Merriman, T.R., Micha, R., Michaud, C., Mishra, V., Hanafiah, K.M., Mokdad, A.A., Morawska, L., Mozaffarian, D., Murphy, T., Naghavi, M., Neal, B., Nelson, P.K., Nolla, J.M., Norman, R., Olives, C., Omer, S.B., Orchard, J., Osborne, R., Ostro, B., Page, A., Pandey, K.D., Parry, C.D.H., Passmore, E., Patra, J., Pearce, N., Pelizzari, P.M., Petzold, M., Phillips, M.R., Pope, D., Pope, C.A., Powles, J., Rao, M., Razavi, H., Rehfuess, E.A., Rehm, J.T., Ritz, B., Rivara, F.P., Roberts, T., Robinson, C., Rodriguez-Portales, J.A., Romieu, I., Room, R., Rosenfeld, L.C., Roy, A., Rushton, L., Salomon, J.A., Sampson, U., Sanchez-Riera, L., Sanman, E., Sapkota, A., Seedat, S., Shi, P., Shield, K., Shivakoti, R., Singh, G.M., Sleet, D.A., Smith, E., Smith, K.R., Stapelberg, N.J.C., Steenland, K., Stöckl, H., Stovner, L.J., Straif, K., Straney, L., Thurston, G.D., Tran, J.H., Van Dingenen, R., Van Donkelaar, A., Veerman, J.L., Vijayakumar, L., Weintraub, R., Weissman, M.M., White, R.A., Whiteford, H., Wiersma, S.T., Wilkinson, J.D., Williams, H.C., Williams, W., Wilson, N., Woolf, A.D., Yip, P., Zielinski, J.M., Lopez, A.D., Murray, C.J.L., Ezzati, M., 2013. A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990-2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 380, 2224–2260. https://doi.org/10. 1016/S0140-6736(1261766-8). Lin, M., Horowitz, L.W., Oltmans, S.J., Fiore, A.M., Fan, S., 2014. Tropospheric ozone trends at Mauna Loa Observatory tied to decadal climate variability. Nat. Geosci. 7, 136–143. https://doi.org/10.1038/ngeo2066. Lin, W., Xu, X., Zhang, X., Tang, J., 2008. Contributions of pollutants from north China plain to surface ozone at the Shangdianzi GAW station, Atmos. Atmos. Chem. Phys. 8, 5889–5898. https://doi.org/10.5194/acpd-8-9139-2008. Lin, C.Y., Hsu, H.M., Lee, Y.H., Kuo, C.H., Sheng, Y.F., Chu, D.A., 2009a. A new transport mechanism of biomass burning from Indochina as identified by modeling studies. Atmos. Chem. Phys. 9, 7901–7911. https://doi.org/10.5194/acp-9-7901-2009. Lin, W., Xu, X., Ge, B., Zhang, X., 2009b. Characteristics of gaseous pollutants at Gucheng, a rural site southwest of Beijing. J. Geophy. Res. Atmos. 114. https://doi.org/10. 1029/2008JD010339. Lin, Y.C., Lin, C.Y., Lin, P.H., Engling, G., Lin, Y.C., Lan, Y.Y., June Chang, C.W., Kuo, T.H., Hsu, W.T., Ting, C.C., 2013. Influence of Southeast Asian biomass burning on ozone and carbon monoxide over subtropical Taiwan. Atmos. Environ. 64, 358–365. https://doi.org/10.1016/j.atmosenv.2012.09.050. Liu, H., Chang, W.L., Oltmans, S.J., Chan, L.Y., Harris, J.M., 1999. On springtime high ozone events in the lower troposphere from Southeast Asian biomass burning. Atmos. Environ. 33, 2403–2410. https://doi.org/10.1016/S1352-2310 (98)00357-4. Liu, X., Bhartia, P.K., Chance, K., Spurr, R.J.D., Kurosu, T.P., 2010a. Ozone profile retrievals from the ozone monitoring instrument. Atmos. Chem. Phys. 10, 2521–2537. https://doi.org/10.5194/acp-10-2521-2010. Liu, J., Logan, J.A., Jones, D.B.A., Livesey, N.J., Megretskaia, I., Carouge, C., Nedelec, P., 2010b. Analysis of CO in the tropical troposphere using Aura satellite data and the GEOS–Chem model: insights into transport characteristics of the GEOS meteorological products. Atmos. Chem. Phys. 10, 12207–12232. https://doi.org/10.5194/ acp-10-12207-2010. Logan, J.A., Megretskaia, I., Nassar, R., Murray, L.T., Zhang, L., Bowman, K.W., Worden, H.M., Luo, M., 2008. Effects of the 2006 el Niño on tropospheric composition as revealed by data from the tropospheric emission spectrometer (TES). Geophys. Res.

G., Tarasick, D.W., Thompson, A.M., Thouret, V., Tully, M.B., Van Malderen, R., Vömel, H., Von Der Gathen, P., Witte, J.C., Yela, M., 2017a. Validation of 10-year SAO OMI Ozone Profile (PROFOZ) product using ozonesonde observations. Atmos. Measur. Tech. 10, 2455–2475. https://doi.org/10.5194/amt-10-2455-2017. Huang, T., Zhu, X., Zhong, Q., Yun, X., Meng, W., Li, B., Ma, J., Zeng, E.Y., Tao, S., 2017b. Spatial and temporal trends in global emissions of nitrogen oxides from 1960 to 2014. Environ. Sci. Technol. 51, 7992–8000. https://doi.org/10.1021/acs.est.7b02235. Huang, G., Liu, X., Chance, K., Yang, K., Cai, Z., 2018. Validation of 10-year SAO OMI ozone profile (PROFOZ) product using Aura MLS measurements. Atmos. Measur. Tech. 11, 17–32. https://doi.org/10.5194/amt-11-17-2018. Hudson, R.D., Thompson, A.M., 1998. Tropical tropospheric ozone from total ozone mapping spectrometer by a modified residual method. J. Geophys. Res.-Atmos. 103, 22129–22145. https://doi.org/10.1029/98JD00729. Huo, H., Yao, Z., Zhang, Y., Shen, X., Zhang, Q., Ding, Y., He, K., 2012. On-board measurements of emissions from light-duty gasoline vehicles in three mega-cities of China. Atmos. Environ. 49, 371–377. https://doi.org/10.1016/j.atmosenv.2011.11. 005. IEA, 2014. Energy Balances of OECD Aned Non-OECD Countries, International Energy 5 Y Agency, Paris, beyond 2020. Online Databas. http://data.iea.org. Jacob, D.J., Logan, J.A., Gardner, G.M., Yevich, R.M., Spivakovsky, C.M., Wofsy, S.C., Sillman, S., Prather, M.J., 1993. Factors regulating ozone over the United States and its export to the global atmosphere. J. Geophys. Res. 98, 14817. https://doi.org/10. 1029/98JD01224. Jaffe, D., Bertschi, I., Jaeglé, L., Novelli, P., Reid, J.S., Tanimoto, H., Vingarzan, R., Westphal, D.L., 2004. Long-range transport of Siberian biomass burning emissions and impact on surface ozone in western North America. Geophys. Res. Lett. 31, 6–9. https://doi.org/10.1029/2004GL020093. Jaffe, D.A., Wigder, N.L., 2012. Ozone production from wildfires: a critical review. Atmos. Environ. 51, 1–10. https://doi.org/10.1016/j.atmosenv.2011.11.063. Janssens-Maenhout, G., Crippa, M., Guizzardi, D., Muntean, M., Schaaf, E., Dentener, F., Bergamaschi, P., Pagliari, V., Olivier, J.G.J., Peters, J.A.H.W., van Aardenne, J.A., Monni, S., Doering, U., Petrescu, A.M.R., 2017. EDGAR v4.3.2 global atlas of the three major greenhouse gas emissions for the period 1970-2012. Earth Syst. Sci. Data Discuss. 1–55. https://doi.org/10.5194/essd-2017-79. Jerrett, M., Burnett, R.T., Pope, C.A., Ito, K., Thurston, G., Krewski, D., Shi, Y., Calle, E., Thun, M., 2009. Long-term ozone exposure and mortality. N. Engl. J. Med. 360, 1085–1095. https://doi.org/10.1056/NEJMoa0803894. Jin, X., Fiore, A.M., Murray, L.T., Valin, L.C., Lamsal, L.N., Duncan, B., Folkert Boersma, K., De Smedt, I., Abad, G.G., Chance, K., Tonnesen, G.S., 2017. Evaluating a spacebased indicator of surface ozone-NOx-VOC sensitivity over Midlatitude source regions and application to decadal trends. J. Geophys. Res.: Atmosphere 122, 10439–10461. https://doi.org/10.1002/2017JD026720. Jonquières, I., Marenco, A., Maalej, A., Rohrer, F., 1998. Study of ozone formation and transatlantic transport from biomass burning emissions over West Africa during the airborne Tropospheric Ozone Campaigns TROPOZ I and TROPOZ II. J. Geophys. Res.: Atmosphere 103, 19059–19073. https://doi.org/10.1029/98JD00819. Jourdain, L., Worden, H.M., Worden, J.R., Bowman, K., Li, Q., Eldering, A., Kulawik, S.S., Osterman, G., Boersma, K.F., Fisher, B., Rinsland, C.P., Beer, R., Gunson, M., 2007. Tropospheric vertical distribution of tropical Atlantic ozone observed by TES during the northern African biomass burning season. Geophys. Res. Lett. 34, 1–6. https:// doi.org/10.1029/2006GL028284. Kanaya, Y., Akimoto, H., Wang, Z.F., Pochanart, P., Kawamura, K., Liu, Y., Li, J., Komazaki, Y., Irie, H., Pan, X.L., Taketani, F., Yamaji, K., Tanimoto, H., Inomata, S., Kato, S., Suthawaree, J., Okuzawa, K., Wang, G., Aggarwal, S.G., Fu, P.Q., Wang, T., Gao, J., Wang, Y., Zhuang, G., 2013. Overview of the Mount Tai experiment (MTX2006) in central east China in June 2006: studies of significant regional air pollution. Atmos. Chem. Phys. 13, 8265–8283. https://doi.org/10.5194/acp-138265-2013. Kaufman, Y.J., Setzer, A., Ward, D., Tanre, D., Holben, B.N., Menzel, P., Pereira, M.C., Rasmussen, R., 1992. Biomass burning airborne and Spaceborne experiment in the Amazonas (Base-a). J. Geophys. Res.: Atmosphere 97, 14581–14599. https://doi.org/ 10.1029/92JD00275. Kalmus, P., Wong, S., Teixeira, J., 2015. The Pacific subtropical cloud transition: a MAGIC assessment of AIRS and ecmwf thermodynamic structure. Geosci. Rem. Sens. Lett. IEEE 12, 1586–1590. https://doi.org/10.1109/LGRS.2015.2413771. Kim, J.H., Newchurch, M.J., Han, K., 2001. Distribution of tropical tropospheric ozone determined by the scan-angle method applied to TOMS measurements. J. Atmos. Sci. 58, 2699–2708. https://doi.org/10.1175/1520-0469(2001)058<2699: DOTTOD >2.0.CO;2. Konovalov, I.B., Beekmann, M., Kuznetsova, I.N., Yurova, A., Zvyagintsev, A.M., 2011. Atmospheric impacts of the 2010 Russian wildfires: integrating modelling and measurements of an extreme air pollution episode in the Moscow region. Atmos. Chem. Phys. 11, 10031–10056. https://doi.org/10.5194/acp-11- 10031-2011. Kondo, Y., Morino, Y., Takegawa, N., Koike, M., Kita, K., Miyazaki, Y., Sachse, G.W., Vay, S.A., Avery, M.A., Flocke, F., Weinheimer, A.J., Eisele, F.L., Zondlo, M.A., Weber, R.J., Singh, H.B., Chen, G., Crawford, J., Blake, D.R., Fuelberg, H.E., Clarke, A.D., Talbot, R.W., Sandholm, S.T., Browell, E.V., Streets, D.G., Liley, B., 2004. Impacts of biomass burning in Southeast Asia on ozone and reactive nitrogen over the western Pacific in spring. J. Geophys. Res. Atmos. 109, 1–22. https://doi.org/10.1029/ 2003JD004203. Kumar, R., Naja, M., Satheesh, S.K., Ojha, N., Joshi, H., Sarangi, T., Pant, P., Dumka, U.C., Hegde, P., Venkataramani, S., 2011. Influences of the springtime northern Indian biomass burning over the central Himalayas. J. Geophy. Res. Atmos. 116, 1–14. https://doi.org/10.1029/2010JD015509. Kumari, S., Verma, N., Lakhani, A., Tiwari, S., Kandikonda, M.K., 2018. Tropospheric ozone enhancement during post-harvest crop-residue fires at two downwind sites of

28

Atmospheric Environment 194 (2018) 14–30

M. Ma et al.

Malaysia. Atmos. Environ. 71, 36–43. https://doi.org/10.1016/j.atmosenv.2013.01. 019. Real, E., Law, K.S., Weinzierl, B., Fiebig, M., Petzold, A., Wild, O., Methven, J., Arnold, S.A., Stohl, A., Huntrieser, H., Roiger, A.E., Schlager, H., Stewart, D., Avery, M.A., Sachse, G.W., Browell, E.V., Ferrare, R.A., Blake, D., 2007. Processes influential ozone levels in Alaskan forest fire plumes during long-range transport over the North Atlantic. J. Geophy. Res. Atmos. 112, 1–19. https://doi.org/10.1029/ 2006JD007576. Real, E., Orlandi, E., Law, K.S., Fierli, F., Josset, D., Cairo, F., Schlager, H., Borrmann, S., Kunkel, D., Volk, C.M., McQuaid, J.B., Stewart, D.J., Lee, J., Lewis, A.C., Hopkins, J.R., Ravegnani, F., Ulanovski, A., Liousse, C., 2010. Cross-hemispheric transport of central African biomass burning pollutants: implications for downwind ozone production. Atmos. Chem. Phys. 10, 3027–3046. https://doi.org/10.5194/acp-10-30272010. Rotach, M.W., Zardi, D., 2007. On the boundary‐layer structure over highly complex terrain: Key findings from MAP. Q. J. R. Meteorol. Soc. 133, 937–948. https://doi. org/10.1002/qj. Safieddine, S., Boynard, A., Hao, N., Huang, F., Wang, L., Ji, D., Barret, B., Ghude, S.D., Coheur, P.F., Hurtmans, D., Clerbaux, C., 2015. Tropospheric ozone variability during the East Asian summer monsoon as observed by satellite (IASI), aircraft (MOZAIC) and ground stations. Atmos. Chem. Phys. 16, 10489–10500. https://doi. org/10.5194/acp-16-10489-2016. Sanhueza, E., Crutzen, P.J., Fernández, E., 1999. Production of boundary layer ozone from tropical American Savannah biomass burning emissions. Atmos. Environ. 33, 4969–4975. https://doi.org/10.1016/S1352-2310(99)00301-5. Sarangi, T., Naja, M., Ojha, N., Kumar, R., Lal, S., Venkataramani, S., Kumar, A., Sagar, R., Chandola, H.C., 2014. First simultaneous measurements of ozone, CO, and NOy at a high‐altitude regional representative site in the central Himalayas. J. Geophys. Res.: Atmosphere 119, 1592–1611. https://doi.org/10.1002/2014JD022994. Sauvage, B., Thouret, V., Cammas, J., Gheusi, F., Athier, G., Nédélec, P., 2005. Tropospheric ozone over Equatorial Africa : regional aspects from the MOZAIC data. Atmos. Chem. Phys. 5, 311–335. https://hal.archives-ouvertes.fr/hal-00327887. Schultz, M.G., Jacob, D.J., Wang, Y., Logan, J.A., Atlas, E.L., Blake, D.R., Blake, N.J., Bradshaw, J.D., Browell, E.V., Fenn, M.A., Flocke, F., Gregory, G.L., Heikes, B.G., Sachse, G.W., Sandholm, S.T., Shetter, R.E., Singh, H.B., Talbot, R.W., 1999. On the origin of tropospheric ozone and NO x over the tropical South Pacific. J. Geophys. Res.: Atmosphere 104, 5829–5843. https://doi.org/10.1029/98JD02309. Schroeder, W., Prins, E., Giglio, L., Csiszar, I., Schmidt, C., Morisette, J., Morton, D., 2008. Validation of GOES and MODIS active fire detection products using ASTER and ETM + data. Rem. Sens. Environ. 112, 2711–2726. https://doi.org/10.1016/j.rse.2008. 01.005. Shindell, D., Kuylenstierna, J.C.I., Vignati, E., van Dingenen, R., Amann, M., Klimont, Z., Anenberg, S.C., Muller, N., Janssens-Maenhout, G., Raes, F., Schwartz, J., Faluvegi, G., Pozzoli, L., Kupiainen, K., Hoglund-Isaksson, L., Emberson, L., Streets, D., Ramanathan, V., Hicks, K., Oanh, N.T.K., Milly, G., Williams, M., Demkine, V., Fowler, D., 2012. Simultaneously mitigating near-term climate change and improving human health and Food security. Science 335, 183–189. https://doi.org/10. 1126/science.1210026. Sillman, S., Samson, P.J., 1995. Impact of temperature on oxidant photochemistry in urban, polluted rural and remote environments. J. Geophys. Res.-Atmos. 100, 11497–11508. https://doi.org/10.1029/94JD02146. Sillman, S., 1999. The relation between ozone, NO(x) and hydrocarbons in urban and polluted rural environments. Atmos. Environ. 33, 1821–1845. https://doi.org/10. 1016/S1352-2310(98)00345-8. Sillman, S., He, D., 2002. Some theoretical results concerning O3-NOx-VOC chemistry and NOx‐VOC indicators. J. Geophys. Res. 107, 4659. https://doi.org/10.1029/2001 JD001123. Sinha, V., Kumar, V., Sarkar, C., 2014. Chemical composition of pre-monsoon air in the Indo-Gangetic Plain measured using a new air quality facility and PTR-MS: high surface ozone and strong influence of biomass burning. Atmos. Chem. Phys. 14, 5921–5941. https://doi.org/10.5194/acp-14-5921-2014. Skeie, R.B., Berntsen, T.K., Myhre, G., Tanaka, K., Kvaleväg, M.M., Hoyle, C.R., 2011. Anthropogenic radiative forcing time series from pre-industrial times until 2010. Atmos. Chem. Phys. 11, 11827–11857. https://doi.org/10.5194/acp-11-11827-2011. Sonkaew, T., Macatangay, R., 2015. Determining relationships and mechanisms between tropospheric ozone column concentrations and tropical biomass burning in Thailand and its surrounding regions. Environ. Res. Lett. 10, 065009. https://doi.org/10. 1088/1748-9326/10/6/065009. Sousa, S.I.V., Martins, F.G., Alvim-Ferraz, M.C.M., Pereira, M.C., 2007. Multiple linear regression and artificial neural networks based on principal components to predict ozone concentrations. Environ. Model. Software 22, 97–103. https://doi.org/10. 1016/j.envsoft.2005.12.002. Stajner, I., Wargan, K., Pawson, S., Hayashi, H., Chang, L.P., Hudman, R.C., Froidevaux, L., Livesey, N., Levelt, P.F., Thompson, A.M., Tarasick, D.W., Stübi, R., Andersen, S.B., Yela, M., König-Langlo, G., Schmidlin, F.J., Witte, J.C., 2008. Assimilated ozone from EOS-Aura: evaluation of the tropopause region and tropospheric columns. J. Geophy. Res. Atmos. 113, 1–17. https://doi.org/10.1029/2007JD008863. Stavrakou, T., Müller, J.F., Bauwens, M., De Smedt, I., Lerot, C., Van Roozendael, M., Coheur, P.F., Clerbaux, C., Boersma, K.F., Van Der, A.R., Song, Y., 2016. Substantial underestimation of post-harvest burning emissions in the north China plain revealed by multi-species space observations. Sci. Rep. 6, 1–11. https://doi.org/10.1038/ srep32307. Stevenson, D.S., Young, P.J., Naik, V., Lamarque, J.F., Shindell, D.T., Voulgarakis, A., Skeie, R.B., Dalsoren, S.B., Myhre, G., Berntsen, T.K., Folberth, G.A., Rumbold, S.T., Collins, W.J., MacKenzie, I.A., Doherty, R.M., Zeng, G., Van Noije, T.P.C., Strunk, A., Bergmann, D., Cameron-Smith, P., Plummer, D.A., Strode, S.A., Horowitz, L., Lee,

Lett. 35, 1–5. https://doi.org/10.1029/2007GL031698. Lou, S., Liao, H., Yang, Y., Mu, Q., 2015. Simulation of the interannual variations of tropospheric ozone over China: roles of variations in meteorological parameters and anthropogenic emissions. Atmos. Environ. 122, 839–851. https://doi.org/10.1016/j. atmosenv.2015.08.081. Lu, X., Chen, N., Wang, Y., Cao, W., Zhu, B., Yao, T., Fung, J.C.H., Lau, A.K.H., 2017. Radical budget and ozone chemistry during autumn in the atmosphere of an urban site in central China. J. Geophys. Res. 122, 3672–3685. https://doi.org/10.1002/ 2016JD025676. Martin, R.V., Jacob, D.J., Logan, J.A., Bey, I., Yantosca, R.M., Staudt, A.C., Li, Q., Fiore, A.M., Duncan, B.N., Liu, H., Ginoux, P., Thouret, V., 2002. Interpretation of TOMS observations of tropical tropospheric ozone with a global model and in situ observations. J. Geophy. Res. Atmos. 107. https://doi.org/10.1029/2001JD001480. Mauzerall, D.L., Logan, J.A., Jacob, D.J., Anderson, B.E., Blake, D.R., Bradshaw, J.D., Heikes, B., Sachse, G.W., Singh, H., Talbot, B., 1998. Photochemistry in biomass burning plumes and implications for tropospheric ozone over the tropical South Atlantic. J. Geophys. Res. 103, 8401. https://doi.org/10.1029/97JD02612. McMillan, N., Bortnick, S.M., Irwin, M.E., Berliner, L.M., 2005. A hierarchical Bayesian model to estimate and forecast ozone through space and time. Atmos. Environ. 39, 1373–1382. https://doi.org/10.1016/j.atmosenv.2004.10.047. McPeters, R., Kroon, M., Labow, G., Brinksma, E., Balis, D., Petropavlovskikh, I., Veefkind, J.P., Bhartia, P.K., Levelt, P.F., 2008. Validation of the aura ozone monitoring instrument total column ozone product. J. Geophys. Res. 113, 1–9. https:// doi.org/10.1029/2007JD008802. Meng, Z.Y., Xu, X.B., Yan, P., Ding, G.A., Tang, J., Lin, W.L., Xu, X.D., Wang, S.F., 2009. Characteristics of trace gaseous pollutants at a regional background station in Northern China. Atmos. Chem. Phys. 9, 927–936. https://doi.org/10.5194/acp-9927-2009. Menut, L., 2003. Adjoint modeling for atmospheric pollution process sensitivity at regional scale. J. Geophys. Res. 108, 8562. https://doi.org/10.1029/2002JD002549. Mielonen, T., De Haan, J.F., Van Peet, J.C.A., Eremenko, M., Veefkind, J.P., 2015. Towards the retrieval of tropospheric ozone with the Ozone Monitoring Instrument (OMI). Atmos. Measur. Tech. 8, 671–687. https://doi.org/10.5194/amt-8-671-2015. Miyazaki, K., Eskes, H.J., Sudo, K., Takigawa, M., Van Weele, M., Boersma, K.F., 2012. Simultaneous assimilation of satellite NO 2, O 3, CO, and HNO 3 data for the analysis of tropospheric chemical composition and emissions. Atmos. Chem. Phys. 12, 9545–9579. https://doi.org/10.5194/acp-12-9545-2012. Monks, P.S., Archibald, A.T., Colette, A., Cooper, O., Coyle, M., Derwent, R., Fowler, D., Granier, C., Law, K.S., Mills, G.E., Stevenson, D.S., Tarasova, O., Thouret, V., Von Schneidemesser, E., Sommariva, R., Wild, O., Williams, M.L., 2015. Tropospheric ozone and its precursors from the urban to the global scale from air quality to shortlived climate forcer. Atmos. Chem. Phys. 15, 8889–8973. https://doi.org/10.5194/ acp-15-8889-2015. Myhre, G., Shindell, D., Bréon, F.-M., Collins, W., Fuglestvedt, J., Huang, J., Koch, D., Lamarque, J.-F., Lee, D., Mendoza, B., Nakajima, T., Robock, A., Stephens, G., Takemura, T., Zhang, H., 2013. Anthropogenic and Natural Radiative Forcing. Climate Change 2013: the Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. pp. 659–740. https://doi.org/10.1017/CBO9781107415324.018. Neftel, A., Spirig, C., Prévôt, A.S.H., Furger, M., Stutz, J., Vogel, B., Hjorth, J., 2002. Sensitivity of photooxidant production in the Milan Basin: an overview of results from a EUROTRAC-2 Limitation of Oxidant Production field experiment. J. Geophy. Res. Atmos. 107, 1–10. https://doi.org/10.1029/2001JD001263. Newchurch, M.J., Liu, X., Kim, J.H., 2001. Lower-Tropospheric Ozone (LTO) derived from TOMS near mountainous regions. J. Geophys. Res.-Atmos. 106, 20403–20412. https://doi.org/10.1029/2000JD000162. Palm, S.P., Benedetti, A., Spinhirne, J., 2005. Validation of ECMWF global forecast model parameters using GLAS atmospheric channel measurements. Geophys. Res. Lett. 32, 1–5. https://doi.org/10.1029/2005GL023535. Pan, X., Kanaya, Y., Tanimoto, H., Inomata, S., Wang, Z., Kudo, S., Uno, I., 2015. Examining the major contributors of ozone pollution in a rural area of the Yangtze River Delta region during harvest season. Atmos. Chem. Phys. 15, 6101–6111. https://doi.org/10.5194/acp-15-6101-2015. Parrington, M., Palmer, P.I., Henze, D.K., Tarasick, D.W., Hyer, E.J., Owen, R.C., Helmig, D., Clerbaux, C., Bowman, K.W., Deeter, M.N., Barratt, E.M., Coheur, P.F., Hurtmans, D., Jiang, Z., George, M., Worden, J.R., 2012. The influence of boreal biomass burning emissions on the distribution of tropospheric ozone over North America and the North Atlantic during 2010. Atmos. Chem. Phys. 12, 2077–2098. https://doi.org/ 10.5194/acp-12-2077-2012. Parrington, M., Palmer, P.I., Lewis, A.C., Lee, J.D., Rickard, A.R., Di Carlo, P., Taylor, J.W., Hopkins, J.R., Punjabi, S., Oram, D.E., Forster, G., Aruffo, E., Moller, S.J., Bauguitte, J.B.S., Allan, J.D., Coe, H., Leigh, R.J., 2013. Ozone photochemistry in boreal biomass burning plumes. Atmos. Chem. Phys. 13, 7321–7341. https://doi.org/ 10.5194/acp-13-7321-2013. Peterson, D., Hyer, E., Wang, J., 2014. Quantifying the potential for high-altitude smoke injection in the North American boreal forest using the standard MODIS fire products and subpixel-based methods. Geophys. Res. Atmos. 119, 3401–3419. https://doi.org/ 10.1002/2013JD021067. Porter, W.C., Heald, C.L., Cooley, D., Russell, B., 2015. Investigating the observed sensitivities of air-quality extremes to meteorological drivers via quantile regression. Atmos. Chem. Phys. 15, 10349–10366. https://doi.org/10.5194/acp-15-10349-2015. Prohaska, J.T., 1976. A technique for analyzing the linear relationships between two meteorological fields. Mon. Weather Rev. 104, 1345–1353. https://doi.org/10.1175/ 1520-0493(1976)104<1345:ATFATL>2.0.CO;2. Rajab, J.M., MatJafri, M.Z., Lim, H.S., 2013. Combining multiple regression and principal component analysis for accurate predictions for column ozone in Peninsular

29

Atmospheric Environment 194 (2018) 14–30

M. Ma et al.

R.A., Girard, M.A., Jiang, J.H., Jiang, Y., Knosp, B.W., Labelle, R.C., Lam, J.C., Lee, K.A., Miller, D., Oswald, J.E., Patel, N.C., Pukala, D.M., Quintero, O., Scaff, D.M., Van Snyder, W., Tope, M.C., Wagner, P.A., Walch, M.J., 2006. The Earth observing system microwave limb sounder (EOS MLS) on the aura satellite. IEEE Trans. Geosci. Rem. Sens. 44, 1075–1092. https://doi.org/10.1109/TGRS.2006.873771. Wilks, D.S., 2006. Canonical correlation analysis (CCA). In: Statistical Methods in the Atmospheric Sciences, second ed. Academic Press, London, UK. Wittig, V.E., Ainsworth, E.A., Naidu, S.L., Karnosky, D.F., Long, S.P., 2009. Quantifying the impact of current and future tropospheric ozone on tree biomass, growth, physiology and biochemistry: a quantitative meta-analysis. Global Change Biol. 15, 396–424. https://doi.org/10.1111/j.1365-2486.2008.01774.x. Wu, R., Bo, Y., Li, J., Li, L., Li, Y., Xie, S., 2016. Method to establish the emission inventory of anthropogenic volatile organic compounds in China and its application in the period 2008-2012. Atmos. Environ. 127, 244–254. https://doi.org/10.1016/j. atmosenv.2015.12.015. Xing, C., Liu, C., Wang, S., Chan, K.L., Gao, Y., Huang, X., Su, W., 2017. Observations of the vertical distributions of summertime atmospheric pollutants and the corresponding ozone production in Shanghai, China. Atmos. Chem. Phys. 17, 14275–14289. https://doi.org/10.5194/acp-17-14275-2017. Xue, Y., Xu, H., Guang, J., Mei, L., Guo, J., Li, C., Mikusauskas, R., He, X., 2014. Observation of an agricultural biomass burning in central and east China using merged aerosol optical depth data from multiple satellite missions. Int. J. Rem. Sens. 35, 5971–5983. https://doi.org/10.1080/2150704X.2014.943321. Yadav, I.C., Linthoingambi Devi, N., Li, J., Syed, J.H., Zhang, G., Watanabe, H., 2017. Biomass burning in Indo-China peninsula and its impacts on regional air quality and global climate change-a review. Environ. Pollut. 227, 414–427. https://doi.org/10. 1016/j.envpol.2017.04.085. Yamaji, K., Li, J., Uno, I., Kanaya, Y., Irie, H., Takigawa, M., Komazaki, Y., Pochanart, P., Liu, Y., Tanimoto, H., Ohara, T., Yan, X., Wang, Z., Akimoto, H., 2010. Impact of open crop residual burning on air quality over central eastern China during the mount tai experiment 2006 (MTX2006). Atmos. Chem. Phys. 10, 7353–7368. https://doi.org/ 10.5194/acp-10-7353-2010. Yan, X., Ohara, T., Akimoto, H., 2006. Bottom-up estimate of biomass burning in mainland China. Atmos. Environ. 40, 5262–5273. https://doi.org/10.1016/j.atmosenv. 2006.04.040. Yang, X., Xu, J., Bi, F., Zhang, Z., Chen, Y., He, Y., Han, F., Zhi, G., Liu, S., Meng, F., 2018. Aircraft measurement over the Gulf of Tonkin capturing aloft transport of biomass burning. Atmos. Environ. 182, 41–50. https://doi.org/10.1016/j. atmosenv.2018.03. 020. Yokelson, R.J., Crounse, J.D., Decarlo, P.F., Karl, T., Urbanski, S., Atlas, E., Campos, T., Shinozuka, Y., Kapustin, V., Clarke, A.D., Weinheimer, A., Knapp, D.J., Montzka, D.D., Holloway, J., Weibring, P., Flocke, F., Zheng, W., Toohey, D., Wennberg, P.O., Wiedinmyer, C., Mauldin, L., Fried, A., Richter, D., Walega, J., Jimenez, J.L., Adachi, K., Buseck, P.R., Hall, S.R., Shetter, R., 2009. Emissions from biomass burning in the Yucatan. Atmos. Chem. Phys. 9, 5785–5812. https://doi.org/10.5194/acp-9-57852009. Yue, X., Unger, N., 2014. Ozone vegetation damage effects on gross primary productivity in the United States. Atmos. Chem. Phys. 14, 9137–9153. https://doi.org/10.5194/ acp-14-9137-2014. Zempila, M.M., van Geffen, J.H., Taylor, M., Fountoulakis, I., Koukouli, M.E., van Weele, M., van der A, R.J., Bais, A., Meleti, C., Balis, D., 2017. TEMIS UV product validation using NILU-UV ground-based measurements in Thessaloniki, Greece. Atmos. Chem. Phys. 17, 7157–7174. https://doi.org/10.5194/acp -17-7157-2017. Zhang, H., Ye, X., Cheng, T., Chen, J., Yang, X., Wang, L., Zhang, R., 2008. A laboratory study of agricultural crop residue combustion in China: emission factors and emission inventory. Atmos. Environ. 42, 8432–8441. https://doi.org/10.1016/j.atmosenv. 2008.08.015. Zhang, L., Li, Q.B., Jin, J., Liu, H., Livesey, N., Jiang, J.H., Mao, Y., Chen, D., Luo, M., Chen, Y., 2011. Impacts of 2006 Indonesian fires and dynamics on tropical upper tropospheric carbon monoxide and ozone. Atmos. Chem. Phys. 11, 10929–10946. https://doi.org/10.5194/acp-11-10929-2011. Ziemke, J.R., Chandra, S., Duncan, B.N., Froidevaux, L., Bhartia, P.K., Levelt, P.F., Waters, J.W., 2006. Tropospheric ozone determined from aura OMI and MLS: evaluation of measurements and comparison with the global modeling Initiative's chemical transport model. J. Geophy. Res. Atmos. 111, 1–18. https://doi.org/10.1029/ 2006JD007089. Ziemke, J.R., Chandra, S., Duncan, B.N., Schoeberl, M.R., Torres, O., Damon, M.R., Bhartia, P.K., 2009. Recent biomass burning in the tropics and related changes in tropospheric ozone. Geophys. Res. Lett. 36, 1–5. https://doi.org/10.1029/ 2009GL039303. Ziemke, J.R., Chandra, S., Labow, G.J., Bhartia, P.K., Froidevaux, L., Witte, J.C., 2011. A global climatology of tropospheric and stratospheric ozone derived from Aura OMI and MLS measurements. Atmos. Chem. Phys. 11, 9237–9251. https://doi.org/10. 5194/acp-11-9237-2011. Ziemke, J.R., Olsen, M.A., Witte, J.C., Douglass, A.R., Strahan, S.E., Wargan, K., Liu, X., Schoeberl, M.R., Yang, K., Kaplan, T.B., Pawson, S., Duncan, B.N., Newman, P.A., Bhartia, P.K., Heney, M.K., 2014. Assessment and applications of NASA Ozone data products derived from Aura OMI/MLS satellite measurements in context of the GMI chemical transport model. J. Geophys. Res.: Atmosphere 119, 5671–5699. https:// doi.org/10.1002/2013JD020914.

Y.H., Szopa, S., Sudo, K., Nagashima, T., Josse, B., Cionni, I., Righi, M., Eyring, V., Conley, A., Bowman, K.W., Wild, O., Archibald, A., 2013. Tropospheric ozone changes, radiative forcing and attribution to emissions in the atmospheric chemistry and climate model intercomparison Project (ACCMIP). Atmos. Chem. Phys. 13, 3063–3085. https://doi.org/10.5194/acp-13-3063-2013. Stohl, A., Berg, T., Burkhart, J.F., Fjæraa, A.M., Forster, C., Herber, A., Hov, Lunder, C., McMillan, W.W., Oltmans, S., Shiobara, M., Simpson, D., Solberg, S., Stebel, K., Ström, J., Tørseth, K., Treffeisen, R., Virkkunen, K., Yttri, K.E., 2007. Arctic smoke record high air pollution levels in the European Arctic due to agricultural fires in Eastern Europe in spring 2006. Atmos. Chem. Phys. 7, 511–534. https://doi.org/10. 5194/acp-7-511-2007. Suthawaree, J., Kato, S., Okuzawa, K., Kanaya, Y., Pochanart, P., Akimoto, H., Wang, Z., Kajii, Y., 2010. Measurements of volatile organic compounds in the middle of Central East China during Mount Tai Experiment 2006 (MTX2006): observation of regional background and impact of biomass burning. Atmos. Chem. Phys. 10, 1269–1285. https://doi.org/10.5194/acp-10- 1269-2010. Svozil, D., Kvasnička, V., Pospíchal, J., 1997. Introduction to multi-layer feed-forward neural networks. Chemometr. Intell. Lab. Syst. 39, 43–62. https://doi.org/10.1016/ S0169-7439(97)00061-0. Tang, Q., Prather, M.J., 2012. Tropospheric column ozone: matching individual profiles from Aura OMI and TES with a chemistry-transport model. Atmos. Chem. Phys. 12, 10441–10452. https://doi.org/10.5194/acp-12- 10441-2012. Tang, H., Liu, G., Zhu, J., Han, Y., Kobayashi, K., 2013. Seasonal variations in surface ozone as influenced by Asian summer monsoon and biomass burning in agricultural fields of the northern Yangtze River Delta. Atmos. Res. 122, 67–76. https://doi.org/ 10.1016/j. atmosres.2012.10.030. Thompson, A.M., 2001. Tropical tropospheric ozone and biomass burning. Science 291, 2128–2132. https://doi.org/10.1126/science.291.5511.2128. Thompson, A.M., 2003. Southern Hemisphere Additional Ozonesondes (SHADOZ) 1998–2000 tropical ozone climatology 2. Tropospheric variability and the zonal wave-one. J. Geophys. Res. 108, 8241. https://doi.org/10.1029/2002JD002241. Thompson, M.L., Reynolds, J., Cox, L.H., Guttorp, P., Sampson, P.D., 2001. A review of statistical methods for the meteorological adjustment of tropospheric ozone. Atmos. Environ. 35, 617–630. https://doi.org/10.1016/S1352-2310(00)00261-2. Tompkins, A.M., Gierens, K., Rädel, G., 2007. Ice supersaturation in the ECMWF integrated forecast system. Q. J. R. Meteorol. Soc. 133, 53–63. https://doi.org/10. 1002/qj.14. Toh, Y.Y., Lim, S.F., von Glasow, R., 2013. The influence of meteorological factors and biomass burning on surface ozone concentrations at Tanah Rata, Malaysia. Atmos. Environ. 70, 435–446. https://doi.org/10.1016/j.atmosenv.2013.01.018. Trentmann, J., Yokelson, R.J., Hobbs, P.V., Winterrath, T., Christian, T.J., Andreae, M.O., Mason, S.A., 2005. An analysis of the chemical processes in the smoke plume from a savanna fire. J. Geophys. Res. Atmos. 110, 1–20. https://doi.org/10.1029/ 2004JD005628. Turquety, S., Logan, J.A., Jacob, D.J., Hudman, R.C., Leung, F.Y., Heald, C.L., Yantosca, R.M., Wu, S., Emmons, L.K., Edwards, D.P., Sachse, G.W., 2007. Inventory of boreal fire emissions for North America in 2004: importance of peat burning and pyroconvective injection. J. Geophy. Res. Atmos. 112, 1–13. https://doi.org/10.1029/ 2006JD007281. Vadrevu, K.P., Ellicott, E., Giglio, L., Badarinath, K.V.S., Vermote, E., Justice, C., Lau, W.K.M., 2012. Vegetation fires in the himalayan region - aerosol load, black carbon emissions and smoke plume heights. Atmos. Environ. 47, 241–251. https://doi.org/ 10.1016/j. atmosenv.2011.11.009. Val Martín, M., Honrath, R.E., Owen, R.C., Pfister, G., Fialho, P., Barata, F., 2006. Significant enhancements of nitrogen oxides, black carbon, and ozone in the North Atlantic lower free troposphere resulting from North American boreal wildfires. J. Geophy. Res. Atmos. 111, 1–17. https://doi.org/10.1029/2006JD007530. Verma, S., Worden, J., Pierre, B., Jones, D.B.A., Al-Saadi, J., Boersma, F., Bowman, K., Eldering, A., Fisher, B., Jourdain, L., Kulawik, S., Worden, H., 2009. Ozone production in boreal fire smoke plumes using observations from the Tropospheric Emission Spectrometer and the Ozone Monitoring Instrument. J. Geophy. Res. Atmos. 114. https://doi.org/10.1029/2008JD010108. Von Engeln, A., Teixeira, J., 2013. A planetary boundary layer height climatology derived from ECMWF reanalysis data. J. Clim. 26, 6575–6590. https://doi.org/10.1175/JCLID-12-00385.1. Wallace, J.M., Smith, C., Bretherton, C.S., 1992. Singular value decomposition of wintertime sea surface temperature and 500-mb height anomalies. J. Clim. 5, 561–576. https://doi.org/10.1175/1520-0442(1992)005<0561:SVDOWS>2.0.CO;2. Wang, Y., Hao, J.I., Mcelroy, M., Munger, W., Ma, H., Nielsen, C., Zhang, Y., 2010. Year round measurements of O3 and CO at a rural site near Beijing: variations in their correlations. Tellus B 62, 228–241. https://doi.org/10.1111/j.1600-0889.2010. 00464.xs. Watson, C.E., Fishman, J., Reichle, H.G., 1990. The significance of biomass burning as a source of carbon-monoxide and ozone in the southern-hemisphere tropics - a satellite analysis. J. Geophys. Res.-Atmos. 95, 16443–16450. https://doi.org/10.1029/ JD095iD10p16443. Waters, J.W., Froidevaux, L., Harwood, R.S., Jarnot, R.F., Pickett, H.M., Read, W.G., Siegel, P.H., Cofield, R.E., Filipiak, M.J., Flower, D.A., Holden, J.R., Lau, G.K., Livesey, N.J., Manney, G.L., Pumphrey, H.C., Santee, M.L., Wu, D.L., Cuddy, D.T., Lay, R.R., Loo, M.S., Perun, V.S., Schwartz, M.J., Stek, P.C., Thurstans, R.P., Boyles, M.A., Chandra, K.M., Chavez, M.C., Chen, G.S., Chudasama, B.V., Dodge, R., Fuller,

30