Skip to main content
NCBI home page
NASA Author Manuscripts logoLink to NASA Author Manuscripts
. Author manuscript; available in PMC: 2018 Apr 16.
Published in final edited form as: Aerosol Air Qual Res. 2017 Jan 1;17(1):253–261. doi: 10.4209/aaqr.2016.04.0145

Evaluation of PM2.5 surface concentration simulated by Version 1 of the NASA’s MERRA Aerosol Reanalysis over Israel and Taiwan

Simon Provençal 1,*, Virginie Buchard 2,3, Arlindo M da Silva 2, Richard Leduc 1, Nathalie Barrette 1, Emily Elhacham 4, Sheng-Hsiang Wang 5
PMCID: PMC5901751  NIHMSID: NIHMS954315  PMID: 29670645

Abstract

Version 1 of the NASA MERRA Aerosol Reanalysis (MERRAero) assimilates bias-corrected aerosol optical depth (AOD) data from MODIS-Terra and MODIS-Aqua, and simulates particulate matter (PM) concentration data to reproduce a consistent database of AOD and PM concentration around the world from 2002 to the end of 2015. The purpose of this paper is to evaluate MERRAero’s simulation of fine PM concentration against surface measurements in two regions of the world with relatively high levels of PM concentration but with profoundly different PM composition, those of Israel and Taiwan. Being surrounded by major deserts, Israel’s PM load is characterized by a significant contribution of mineral dust, and secondary contributions of sea salt particles, given its proximity to the Mediterranean Sea, and sulfate particles originating from Israel’s own urban activities and transported from Europe. Taiwan’s PM load is composed primarily of anthropogenic particles (sulfate, nitrate and carbonaceous particles) locally produced or transported from China, with an additional contribution of springtime transport of mineral dust originating from Chinese and Mongolian deserts. The evaluation in Israel produced favorable results with MERRAero slightly overestimating measurements by 6% on average and reproducing an excellent year-to-year and seasonal fluctuation. The evaluation in Taiwan was less favorable with MERRAero underestimating measurements by 42% on average. Two likely reasons explain this discrepancy: emissions of anthropogenic PM and their precursors are largely uncertain in China, and MERRAero doesn’t include nitrate particles in its simulation, a pollutant of predominately anthropogenic sources. MERRAero nevertheless simulates well the concentration of fine PM during the summer, when Taiwan is least affected by the advection of pollution from China.

Keywords: MERRAero, Evaluation, Fine particulate matter, Israel, Taiwan

INTRODUCTION

NASA’s Modern-Era Retrospective Analysis for Research and Application (MERRA, Rienecker et al., 2011) is a reanalysis tool integrating satellite observations from the Earth Observing System and model data from the 5th version of the Goddard Earth Observing System (GEOS-5) atmospheric model and data assimilation system (Rienecker et al., 2008) in order to produce a consistent database in both time and space of various environmental variables around the world since the beginning of the satellite era. Recently, bias-corrected aerosol optical depth (AOD) observations from the Moderate Resolution Imaging Spectroradiometers (MODIS, Remer et al., 2005) on board the Terra and Aqua satellites as well as the Goddard Chemistry, Aerosol, Radiation and Transport (GOCART) model (Chin et al., 2002) were included in MERRA to create a reanalysis of aerosols labelled “MERRAero”. GOCART simulates the sources, sinks, transport and concentration of sulfate (SO4), organic carbon (OC), black carbon (BC), dust (DS) and sea salt (SS) aerosols (Chin et al., 2002; Colarco et al., 2010). DS and SS emissions are a function of surface properties and wind speed at the surface, and their respective concentrations are classified in different diameter bins. Sources of other species are simulated from emission inventories, including their precursors. Sulfur dioxide (SO2, the precursor of SO4) anthropogenic emissions are input from the Emission Database for Global Atmospheric Research (EDGAR) version 4.1 inventory from 2005 and biomass burning emissions (primarily OC and BC) are input from the NASA Quick Fire Emission Dataset (QFED) version 2.1 (Buchard et al., 2015).

MERRAero simulates the concentration of the five aerosol species listed in the previous paragraph all over the world with a resolution of 0.5° latitude by 0.625° longitude and 72 vertical layers (from the surface to 80 km) from 2002 to the end of 2015 (Buchard et al., 2015). Considering that these aerosol species, also referred to airborne particulate matter (PM), affect public health and visibility differently depending on their size and chemistry (e.g., Laden et al., 2000; Schwartz and Neas, 2000; Groblicki et al., 1981), MERRAero’s differentiation of the aerosols’ chemical speciation is a significant improvement for studying a broad range of air quality issues around the world since very few monitoring networks make such a distinction of local PM observations, but especially in regions with unreliable or scarce monitoring.

Different components of MERRAero have been evaluated in different regions of the world. Its assimilation of AOD has been validated over Africa, South America, central and eastern Asia using many remote sensing instruments (Buchard et al., 2015); in the United States, the surface concentrations of PM2.5, their chemical speciation and SO2 has been thoroughly evaluated (Buchard et al., 2014; 2016); and in Europe, an evaluation of the surface concentrations of PM10, PM2.5 and some of their chemical speciation has been performed (Provençal et al., 2016). The concentrations of PM10, PM2.5 and SO4 were generally well simulated in both the U.S. and Europe but Buchard et al. (2016) and Provençal et al. (2016) noticed an underestimation of carbonaceous concentration in urban/suburban locations, particularly in winter, due to unresolved sources by MERRAero.

The U.S. and Europe have similar PM signatures in the sense that both regions are highly industrialized and therefore anthropogenic particles contribute significantly to their PM load. At the same time, implementation of air quality regulation has successfully reduced the emissions of various atmospheric pollutants across the U.S. and Europe over the last decades (e.g. Granier et al., 2011; Klimont et al., 2013; Hand et al., 2012; Xing et al., 2013; de Gouw et al., 2014; Vestreng et al., 2007) and, as a result, maintained relatively low levels of PM concentration. There are nevertheless important differences with respect to the chemical speciation of PM between the two regions such as a predominance of carbonaceous particles over the western U.S. due to summer wildfires and a predominance of dust particles over southern Europe due to its proximity to the Sahara desert.

The PM signature in the U.S. and Europe is not representative of many other regions in the world where PM sources and pollution control are profoundly different. In order for MERRAero to achieve optimal reliability for studying air quality issues around the world, the purpose of this article is to pursue MERRAero’s evaluation in regions with different and distinct aerosol signatures, those of Israel and Taiwan. The evaluation in Israel, a region with a heavy PM load due to its proximity to major deserts, will ascertain MERRAero’s ability to simulate the concentration of aerosol originating from natural sources. Taiwan being located in a region of the world which is routinely experiencing severe air pollution episodes, the evaluation there will provide insight on MERRAero’s applicability in highly polluted regions where its contribution would be most beneficial.

LOCATIONS AND METHODS

Israel and Taiwan

Israel is located in western Asia, surrounded by the Mediterranean Sea, the Sahara desert and the Middle Eastern deserts. Its PM concentration load is relatively high, composed largely of mineral DS (Kushelevsky et al., 1983; Malenky et al., 1983; Foner and Ganor, 1992) with an occasionally important contribution from SS particles when the wind is blowing inland (Foner and Ganor, 1992). PM concentration in urban areas such as the coastal city of Tel Aviv is even higher due to anthropogenic SO4 locally produced or transported from Europe (Foner and Ganor, 1992). Rural locations in Israel have also been impacted by the advection of SO4 particles originating from Europe (Luria et al., 1989).

Taiwan is an island located in eastern Asia, separated from mainland China by the Taiwan Strait. Its concentration level of PM is fairly high, especially in urban areas (Chen et al., 1999), caused by industrial and transportation activities within Taiwan but also due to wintertime synoptic features that transport polluted air from China (Lin et al., 2005). SO4, OC, BC, nitrate (NO3) and ammonium (NH4) particles together compose a large portion of PM concentration (Lin et al., 2008; Lin, 2002; Chen et al., 2003; Tsai and Kuo, 2005; Tsai and Cheng, 1999; 2004). Taiwan nevertheless enjoys cleaner air during the summer, coinciding with the typhoon season which sweeps the island with strong winds and heavy rain (Lin et al., 2008). In spring, Taiwan is also impacted by the advection of dust originating from Chinese and Mongolian deserts (Chen et al., 2004).

Evaluation method

MERRAero simulates the concentration of five PM2.5 (PM with diameter ≤ 2.5 µm) species every hour: SO4, OC, BC, DS2.5 and SS2.5. From these, it is possible to apply a mass reconstruction method to estimate the total concentration of PM2.5. Chow et al. (2015) reviewed 11 commonly used equations to reconstruct PM mass from speciation measurements which are usually determined by the measurements available. The equations usually took the following form:

PM=Inorganic ions+Organic matter+BC+DS+SS (1)

Inorganic ions include SO4, NO3 and NH4 ions. When NH4 measurements were lacking, SO4 and NO3 were assumed to be fully neutralized by NH4 in the form of ammonium sulfate ((NH4)2SO4) and ammonium nitrate (NH4NO3) by multiplying their respective concentrations by 1.375 and 1.29; (NH4)2SO4 being composed of 73% of SO4 by mass and NH4NO3 being similarly composed of 78% of NO3. The concentration of inorganic ions was ultimately estimated by: 1.375 × [SO4] + 1.29 × [NO3] (brackets denote concentration). [NH4NO3] was occasionally omitted altogether when NO3 measurements were lacking or unreliable (e.g., Malm et al., 1994).

The concentration of particulate organic matter (POM) was estimated through OC measurements multiplied by a coefficient which took into account other organic compounds found in POM but not measured. Commonly and historically, a coefficient of 1.4 was used (Chow et al., 2015; Turpin and Lim, 2001), but Turpin and Lim (2001) argued that such a value is often too low. They recommended a value of 1.6 ± 0.2 for urban carbonaceous particles, 2.1 ± 0.2 for aged (non-urban) particles and a value as high as 2.6 for biomass burning particles.

Taking into consideration the PM species simulated by MERRAero and given that this evaluation is performed in a combination of urban and non-urban locations, the following reconstruction is used:

[PM2.5]=1.375×[SO4]+1.8×[OC]+[BC]+[DS2.5]+[SS2.5] (2)

Eq. 2 lacks the concentration of NO3 particles whose sources are predominantly anthropogenic in nature (Delmas et al., 1997).

MERRAero’s simulation at the surface is compared to hourly observations of [PM2.5] measured at 11 locations in Israel between 2003 and 2014, and 13 locations in and around Taiwan between 2005 and 2014 (Fig. 1). A spatial consistency algorithm is applied to assure reliability of the observed and simulated data which goes as follows: since trace concentrations are usually lognormally distributed, the bias between log-simulated concentration and log-observed concentration (Blog = log(Cs) − log(Co); Cs: simulated concentration, Co: observed concentration) is calculated at all locations within each study areas on a given hour; the average and standard deviation of Blog are calculated and used to define a reliability interval which justifies ~95% of the normal distribution: Blog¯±2σBlog; all data pairs that fall outside this interval are excluded.

Fig. 1.

Fig. 1

Open in a new tab

Location of monitoring stations.

Performance statistics are calculated to quantify MERRAero’s accuracy: the mean fraction F¯=Cs¯/Co¯, the mean bias B¯=Cs¯−Co¯, the standard deviation of the bias (SD-B) and the correlation coefficient (R). Given that trace concentrations are lognormally distributed, it is also relevant to compute log-transformed statistics: Blog¯, SD-Blog and Rlog. Willmott (1982) criticized the use of R to evaluate model performance since it doesn’t directly compare simulated with observed data. Therefore, Chang and Hanna (2004) recommended as a rigorous index to evaluate air quality models the proportion of simulated data which falls within a factor of 2 of observed data (FAC2, i.e. proportion of the data which satisfies 0.5 ≤ Co / Cs ≤ 2.0) since this index is not disproportionately sensitive to extreme values and is unaffected by simplification of errors. Chang and Hanna (2004) considered a model’s performance to be reasonably good if FAC2 ≥ 0.5.

RESULTS AND DISCUSSION

Israel

The spatial consistency algorithm excluded 5% of the data in Israel. At 22.5 µg m−3, the PM2.5 load in Israel is high (Table 1) compared to Europe (Provençal et al., 2016), rural and suburban U.S. (Buchard et al., 2016). Overall, MERRAero simulates [PM2.5] very well in Israel by slightly overestimating its average concentration by 6% or 1.4 µg m−3 (Table 1). However, the high SD-B value and modest R suggest significant scatter within the data and a low bias resulting from simplification of errors. On the other hand, it is worth mentioning that SD’s are disproportionately impacted by extreme data pairs. For instance, if such data which fall outside a factor of 5 between observed and simulated concentrations, which represent 2.6% of the sample, B and SD-B are reduced to 1.0 µg m−3 and 19.6 µg m−3, respectively. Furthermore, the density scatter plot of Fig. 2a reveals that although there is some scatter, the bulk of the data is generally well simulated. This is further supported by a high FAC2 value of 76%. The log-transformed data (Table 1; Fig. 2b) support a similar analysis. Additionally, Fig. 3 compares the annual and monthly fluctuations between simulated and observed data, and illustrates an excellent identity between both datasets.

Table 1.

Performance statistics for the ensemble of locations in Israel and Taiwan. AOC stands for “average observed concentration”.

Israel Taiwan
n 1,016,778 1,024,992
AOC (µg m−3) 22.5 29.8
F 1.06 0.70
B¯(µg  m−3)
 1.4 −8.8
SD-B (µg m−3) 23.8 22.7
Blog¯
 0.07 −0.27
SD-Blog 0.66 0.83
R 0.56 0.27
Rlog 0.50 0.30
FAC2 0.76 0.59
Open in a new tab

Fig. 2.

Fig. 2

Open in a new tab

Density scatter plot for (a) observed and simulated [PM2.5], and (b) log transformed observed and simulated [PM2.5] for the ensemble of locations in Israel.

Fig. 3.

Fig. 3

Open in a new tab

(a) Yearly and (b) monthly average of [PM2.5] observation, simulation and bias, and (c–d) similarly for the SD and FAC2, for the ensemble of locations in Israel.

MERRAero’s ability to accurately estimate [PM2.5] in Israel relies predominantly on its ability to simulate [DS2.5] since [PM2.5] is largely composed of this species (Table 2) and, to a lesser extent, its ability to simulate [SO4] and [SS2.5]. The evaluation of [DS2.5] in the U.S. revealed important seasonal biases without much impact on the evaluation of [PM2.5] given its small contribution to [PM2.5] there (Buchard et al., 2016). While the U.S. is mostly impacted by long range transport of DS, this evaluation in Israel would suggest that MERRAero performs well in simulating [DS2.5] originating from local sources. [SO4] has been shown to be well simulated in the U.S. and in Europe (Buchard et al., 2016; Provençal et al., 2016), we therefore have no reason to believe otherwise in this region. MERRAero largely overestimated [SS2.5] in both the U.S. and Europe due in part to measurement biases which could very well be the case in Israel. An overestimation of [SS2.5] could compensate the lack of nitrate particles in the simulation. In any case, the lack of nitrate particles is likely a minor shortcoming given that they are less abundant than SO4 and probably contribute little to [PM2.5].

Table 2.

Average concentration simulated by MERRAero for the ensemble of locations in Fig. 1 over the study period.

Species Israel Taiwan
Average concentration
(µg m−3)		 Proportion of PM2.5
concentration (%)		 Average concentration
(µg m−3)		 Proportion of PM2.5
concentration (%)
PM2.5 24.0 – 20.9 –
(NH4)2SO4 4.5 18.7 9.3 44.3
POM 1.3 5.4 3.0 14.1
BC 0.4 1.7 0.7 3.4
DS2.5 13.9 57.9 2.0 9.3
SS2.5 3.9 16.3 6.1 28.9
Open in a new tab

Taiwan

The spatial consistency algorithm excluded 3% of the data in Taiwan. The PM2.5 load in Taiwan is higher than in Israel (29.8 µg m−3; Table 1). Despite a FAC2 of 59%, MERRAero’s performance in Taiwan is much less encouraging. On average, MERRAero underestimates total [PM2.5] by 8.8 µg m−3, a factor of 1.42. The SD-B is also high and R is positive but low. Fig. 4a–b reveals that the bulk of the simulated data, 66% to be precise, is indeed underestimated.

Fig. 4.

Fig. 4

Open in a new tab

Density scatter plot for (a) observed and simulated [PM2.5], and (b) log transformed observed and simulated [PM2.5] for the ensemble of locations in Taiwan.

MERRAero’s simulation of [PM2.5] in Taiwan is mostly anthropogenic in nature (Table 2) with a significant proportion attributed to SO4. This information coupled with Fig. 5 which compares annually and monthly averaged simulated and observed concentration reveals a few clues as to why MERRAero’s performance is less favorable in Taiwan. The evaluation performs well during the summer (typhoon) season but deteriorates during the rest of the year when Taiwan is most impacted by the advection of pollution from China. The use of a constant inventory of SO2 emissions from 2005 is problematic in the long term since it is increasingly becoming antiquated with every passing year. Indeed, in 2005, China successfully implemented comprehensive policies to reduce SO2 emissions. As a result, SO2 emissions and, by extension, SO4 concentrations have been decreasing since 2006 (Lu et al., 2010; 2011; Wang and Hao, 2012; Zhang et al., 2012; Klimont et al., 2013; B. Zhao et al., 2013; Y. Zhao et al., 2013). This is reflected in Fig. 5 with a near constant decrease of [PM2.5] observations as opposed to the near constant year to year concentrations simulated by MERRAero. SO2 emission estimates from China are also crippled with uncertainties (Smith et al., 2011). The lack of nitrate particles in the simulation is much more troublesome in Taiwan given that [PM2.5] is mostly composed of anthropogenic particles. This would certainly explain a significant portion of the underestimation. Another possible explanation for the wintertime discrepancy, one that’s also been highlighted by Provençal et al. (2016) for the evaluation in Europe, is local sources of pollution unresolved by MERRAero. While MERRAero’s simulation takes into account urban sources of pollution, its resolution is too coarse to capture the urban core of cities. Some monitoring stations in Taiwan (Fig. 1) are located in or around large cities, but none of them are located in their downtown core. We therefore don’t expect them to be overly influenced by local sources of pollution. Nevertheless, some influence of unresolved sources should be anticipated. The springtime maximum observed in Fig. 5 is the likely contribution of long range transport of DS, well captured by MERRAero.

Fig. 5.

Fig. 5

Open in a new tab

(a) Yearly and (b) monthly average of [PM2.5] observation, simulation and bias, and (c–d) similarly for the SD and FAC2, for the ensemble of locations in Taiwan.

CONCLUSION

We evaluated version 1 of the MERRA Aerosol Reanalysis’ ability to simulate the concentration of PM2.5 in two regions of the world with relatively high levels of PM concentration but with profoundly different PM composition. Israel is characterized by a high concentration of PM2.5 due to its proximity to major deserts and to the highly saline Mediterranean Sea. Its PM2.5 load is composed mostly of natural particles (mineral dust and sea salt) with some contribution of anthropogenic particles (sulfate) originating from Israel’s urban activities and advection from Europe. Taiwan’s high PM2.5 concentration is mostly anthropogenic in nature due to Taiwan’s own industrial activities and to advection of polluted air from China with some contribution of dust particles originating from east Asian deserts.

The evaluation reproduced favorable results in Israel where MERRAero slightly overestimated actual PM2.5 concentration by 6% on average. Although there is scatter within the distribution, most of the simulation is reasonably accurate with over 75% of the simulated data falling within a factor of 2 of measurements. Given that most of PM2.5 in Israel is mineral dust, this evaluation supports the assumption that MERRAero performs well in simulating the concentration of fine dust originating from local and regional sources throughout the year.

The evaluation is not as favorable in Taiwan where MERRAero significantly underestimated measured PM2.5 concentration by 42% on average. Given that PM2.5 in Taiwan is mostly composed of anthropogenic particles, many of which originate from China, two likely reasons explain this outcome: the uncertainty with respect to Chinese emissions and the lack of nitrate particles in the simulation. The simulation was indeed better during the summer when Taiwan is least impacted by advection of polluted air from China.

Acknowledgments

The authors are grateful to Israel’s Ministry of Environmental Protection and Taiwan’s Environmental Protection Administration for providing the PM2.5 concentration data in their respective countries.

References

  1. Buchard V, da Silva AM, Colarco P, Krotkov N, Dickerson RR, Stehr JW, Mount G, Spinei E, Arkinson HL, He H. Evaluation of GEOS-5 sulfur dioxide simulations during the Frostburg, MD 2010 field campaign. Atmos. Chem. Phys. 2014;14:1929–1941. [Google Scholar]
  2. Buchard V, da Silva AM, Colarco PR, Darmenov A, Randles CA, Govindaraju R, Torres O, Campbell J, Spurr R. Using the OMI aerosol index and absorption aerosol optical depth to evaluate the NASA MERRA Aerosol Reanalysis. Atmos. Chem. Phys. 2015;15:5743–5760. [Google Scholar]
  3. Buchard V, da Silva AM, Randles CA, Colarco P, Ferrare R, Hair J, Hostetler C, Tackett J, Winker D. Evaluation of the surface PM2.5 in Version 1 of the NASA MERRA Aerosol Reanalysis over the United States. Atmos. Environ. 2016;125:100–111. [Google Scholar]
  4. Chang JC, Hanna SR. Air quality model performance evaluation. Meteorol. Atmos. Phys. 2004;87:167–196. [Google Scholar]
  5. Chen KS, Lin CF, Chou YM. Determination of source contribution to ambient PM2.5 in Kaohsiung, Taiwan, using a receptor model. J. Air Waste Manage. Assoc. 2003;51:489–498. doi: 10.1080/10473289.2001.10464287. [DOI] [PubMed] [Google Scholar]
  6. Chen ML, Mao TF, Lin IK. The PM2.5 and PM10 particles in urban areas of Taiwan. Sci. Total Environ. 1999;226:227–235. doi: 10.1016/s0048-9697(98)00407-0. [DOI] [PubMed] [Google Scholar]
  7. Chen YS, Sheen PC, Chen ER, Liu YK, Wu TN, Yang CY. Effects of Asian dust storm events on daily mortality in Taipei, Taiwan. Environ. Res. 2004;95:151–155. doi: 10.1016/j.envres.2003.08.008. [DOI] [PubMed] [Google Scholar]
  8. Chin M, Ginoux P, Kinne S, Torres O, Holben BN, Duncan BN, Martin RV, Logan JA, Higurashi A, Nakajima T. Tropospheric aerosol optical thickness from the GOCART model and comparisons with satellite and sun photometer measurements. J. Atmos. Sci. 2002;59:461–483. [Google Scholar]
  9. Chow JC, Lowenthal DH, Antony Chen LW, Wang X, Watson JG. Mass reconstruction methods for PM2.5: a review. Air Qual. Atmos. Health. 2015;8:243–263. doi: 10.1007/s11869-015-0338-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. de Gouw JA, Parrish DD, Frost GJ, Trainer M. Reduced emissions of CO2 NOX and SO2 from U.S. power plants owing to switch from coal to natural gas with combined cycle technology. Earth’s Future. 2014;2:75–82. [Google Scholar]
  11. Delmas R, Serça D, Jambert C. Global inventory of NOX sources. Nutr. Cycl. Agroecosyst. 1997;48:51–60. [Google Scholar]
  12. Foner HA, Ganor E. The chemical and mineralogical composition of some urban atmospheric aerosols in Israel. Atmos. Environ. 1992;26B:125–133. [Google Scholar]
  13. Granier C, Bessagnet B, Bond T, d’Angiola A, van der Gon HD, Frost GJ, Heil A, Kaiser JW, Kinne S, Klimont K, Kloster S, Lamarque JF, Liousse C, Masui T, Meleux F, Mieville A, Ohara T, Raut JC, Riahi K, Schultz MG, Smith SJ, Thompson A, van Aardenne J, van der Werf GR, van Vuuren DP. Evolution of anthropogenic and biomass burning emissions of air pollutants at global and regional scales during the 1980–2010 period. Clim. Chang. 2011;109:163–190. [Google Scholar]
  14. Groblicki PJ, Wolff GT, Countess RJ. Visibility-reducing species in the Denver “brown cloud”—I. Relationships between extinction and chemical composition. Atmos. Environ. 1981;15:2473–2484. [Google Scholar]
  15. Hand JL, Schichtel BA, Malm WC, Pitchford ML. Particulate sulfate ion concentration and SO2 emission trends in the United States from the early 1990s through 2010. Atmos. Chem. Phys. 2012;12:10353–10365. [Google Scholar]
  16. Klimont Z, Smith SJ, Cofala J. The last decade of global anthropogenic sulfur dioxide: 2000–2011 emissions. Environ. Res. Lett. 2013;8:014003. [Google Scholar]
  17. Kushelevsky A, Shani G, Haccoun A. Effect of meteorological conditions on total suspended particulate (TSP) levels and elemental concentration of aerosols in a semi-arid zone (Beer-Sheva, Israel) Tellus. 1983;35B:55–64. [Google Scholar]
  18. Laden F, Neas LM, Dockery DW, Schwartz J. Association of fine particulate matter from different sources with daily mortality in six U.S. cities. Environ. Health Persp. 2000;108:941–947. doi: 10.1289/ehp.00108941. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Lin CH, Wu YL, Lai CH, Watson JG, Chow JC. Air quality measurements from the southern particulate matter supersite in Taiwan. Aerosol Air Qual. Res. 2008;8:233–264. [Google Scholar]
  20. Lin CY, Liu SC, Chou CCK, Huang SJ, Liu CM, Kuo CH, Young CY. Long-range transport of aerosols and their impact on the air quality of Taiwan. Atmos. Environ. 2005;39:6066–6076. [Google Scholar]
  21. Lin JJ. Characterization of the major chemical species in PM2.5 in the Kaohsiung City, Taiwan. Atmos. Environ. 2002;36:1911–1920. [Google Scholar]
  22. Lu Z, Streets DG, Zhang Q, Wang S, Carmichael GR, Cheng YF, Wei C, Chin M, Diehl T, Tau Q. Sulfur dioxide emissions in China and sulfur trends in East Asia since 2000. Atmos. Chem. Phys. 2010;10:6311–6331. [Google Scholar]
  23. Lu Z, Zhang Q, Streets DG. Sulfur dioxide and primary carbonaceous aerosol emissions in China and India, 1996–2010. Atmos. Chem. Phys. 2011;11:9839–9864. [Google Scholar]
  24. Luria M, Lifschitz B, Peleg M. Particulate sulfate levels at a rural site in Israel. J. Atmos. Chem. 1989;8:241–250. [Google Scholar]
  25. Malenky B, van Grieken R, Van’t Dack L, Luria M. Atmospheric trace element concentration in Jerusalem, Israel. Atmos. Environ. 1983;17:819–822. [Google Scholar]
  26. Malm WC, Sisler JF, Huffman D, Eldred RA, Cahill TA. Spatial and seasonal trends in particle concentration and optical extinction in the United States. J. Geophys. Res. 1994;99:1347–1370. [Google Scholar]
  27. Provençal S, Buchard V, da Silva AM, Leduc R, Barrette N. Evaluation of PM surface concentrations simulated by Version 1 of the NASA MERRA Aerosol Reanalysis over Europe. Atmos. Pollut. Res. 2016 doi: 10.1016/j.apr.2016.10.009. submitted. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Remer LA, Kaufman YJ, Tanré D, Mattoo S, Chu DA, Martins JV, Li RR, Ichoku C, Levy RC, Kleidman RH, Eck TF, Vermote E, Holben BN. The MODIS aerosol algorithm, products, and validation. J. Atmos. Sci. 2005;62:947–973. [Google Scholar]
  29. Rienecker MM, Suarez MJ, Todling R, Bacmeister J, Takacs L, Liu HC, Sienkiewick M, Koster RD, Gelaro R, Stajner I, Nielsen JE. The GEOS-5 data assimilation system—documentation of versions 5.0.1, 5.1.0, and 5.2.0. Technical Report Series on Global Modeling and Data Assimilation. 2008;27 NASA/TM-2008-104606. [Google Scholar]
  30. Rienecker MM, Suarez MJ, Gelaro R, Todling R, Bacmeister J, Liu E, Bosilovich MG, Schubert SD, Takacs L, Kim GK, Bloom S, Chen J, Collins D, Conaty A, da Silva A, Gu W, Joiner J, Koster RD, Lucchesi R, Molod A, Owens T, Pawson S, Pegion P, Redder CR, Reichle R, Robertson FR, Ruddick AG, Sienkiewick M, Woollen J. MERRA: NASA’s Modern-Era Retrospective Analysis for Research and Application. J. Clim. 2011;24:3624–3648. doi: 10.1175/JCLI-D-16-0758.1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Schwartz J, Neas LM. Fine particles are more strongly associated than coarse particles with acute respiratory health effects in schoolchildren. Epidemiology. 2000;11:6–10. doi: 10.1097/00001648-200001000-00004. [DOI] [PubMed] [Google Scholar]
  32. Smith SJ, van Aardenne J, Klimont K, Andres RJ, Volke A, Delgado Arias S. Anthropogenic sulfur dioxide emissions: 18502IJ005. Atmos. Chem. Phys. 2011;11:1101–1116. [Google Scholar]
  33. Tsai YI, Cheng MT. Visibility and aerosol chemical composition near the coastal area in central Taiwan. Sci. Total Environ. 1999;231:37–51. [Google Scholar]
  34. Tsai YI, Cheng MT. Characterization of chemical species in atmospheric aerosols in a metropolitan basin. Chemosphere. 2004;54:1171–1181. doi: 10.1016/j.chemosphere.2003.09.021. [DOI] [PubMed] [Google Scholar]
  35. Tsai YI, Kuo SC. PM2.5 aerosol water content and chemical composition in a metropolitan and a coastal area in southern Taiwan. Atmos. Environ. 2005;39:4827–4839. [Google Scholar]
  36. Turpin BJ, Lim HJ. Species contribution to PM2.5 mass concentration: revisiting common assumptions for estimating organic mass. Aerosol Sci. Technol. 2001;35:602–610. [Google Scholar]
  37. Vestreng V, Myhre G, Fagerli H, Reis S, Tarrasón L. Twenty-five years of continuous sulfur dioxide emission reduction in Europe. Atmos. Chem. Phys. 2007;7:3663–3681. [Google Scholar]
  38. Wang S, Hao J. Air quality management in China: issues, challenges, and options. J. Environ. Sci. 2012;21:2–13. doi: 10.1016/s1001-0742(11)60724-9. [DOI] [PubMed] [Google Scholar]
  39. Willmott CJ. Some comments on the evaluation of model performance. Bull. Am. Meteorol. Soc. 1982;63:1309–1313. [Google Scholar]
  40. Xing J, Pleim J, Mathur R, Pouliot G, Hogrefe C, Gan CM, Wei C. Historical gaseous and primary aerosol emissions in the United States from 1990–2010. Atmos. Chem. Phys. 2013;13:7531–7549. [Google Scholar]
  41. Zhang Q, He K, Huo H. Cleaning China’s air. Nature. 2012;484:161–162. doi: 10.1038/484161a. [DOI] [PubMed] [Google Scholar]
  42. Zhao B, Wang S, Dong X, Wang J, Duan L, Fu X, Hao J, Fu J. Environmental effects of the recent emission changes in China: implications for particulate matter pollution and soil acidification. Environ. Res. Lett. 2013;8:024031. [Google Scholar]
  43. Zhao Y, Zhang J, Nielsen CP. The effects of recent control policies on trends in emissions of anthropogenic atmospheric pollutants and CO2 in China. Atmos. Chem. Phys. 2013;13:487–508. [Google Scholar]

RESOURCES