Growing Urbanization and the Impact on Fine Particulate Matter (PM2.5) Dynamics

Article Growing Urbanization and the Impact on Fine Particulate Matter (PM2.5) Dynamics

Lijian Han, Weiqi Zhou * and Weifeng Li

State Key Laboratory of Urban and Regional Ecology, Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing 100085, China; [email protected] (L.H.); [email protected] (W.L.) * Correspondence: [email protected]; Tel.: +86-10-6291-5372

Received: 11 April 2018; Accepted: 18 May 2018; Published: 23 May 2018

Abstract: Changes in urban air quality and its relationship with growing urbanization provide an important insight into urban development strategies. Thus, we collected remotely sensed PM2.5 concentrations, as well as urban population datasets, and analyzed the scaling relationship between changes in urban population and concentrations of PM2.5. The majority of large cities in North America and Europe had PM2.5 concentrations which decreased significantly. Only 2.0% of large cities in the U.S. were found to have significant positive trends. PM2.5 concentration trends of less than 0.5 µg/m3·year were found in all large cities of Africa and Latin America. However, 3 PM2.5 concentration trends of more than 1.0 µg/m ·year were found in 56.7% of the large cities in Asia, where only 2.3% of the cities in China were found with significant negative trends, and no cities in India were found with significant negative trends. Large cities in Asia were found with contributions of 4.12 ± 4.27 µg/m3·year per million people, particularly large cities in China (5.40 ± 4.80 µg/m3·year per million people) and India (4.07 ± 3.07 µg/m3·year per million people). Significant negative or positive relationships were obtained between PM2.5 trends and population change rates in large cities of North America (R2 = 0.9195, p < 0.05) or Europe (R2 = 0.9161, p < 0.05). Moreover, a significant inverse “U-type” relationship (R2 = 0.8065, p < 0.05) was found between PM2.5 trends and population change rates in large cities of Asia. In addition, the positive or negative relationships between the trends in population and PM2.5 were obtained in typical low- and mid-income countries (e.g., China and India) or high-income countries (e.g., USA), respectively.

Keywords: air pollution; China; India; urbanization strategy; smart growth; PM2.5; scaling approach

1. Introduction More than 50% of the global population live in urban areas due to rapid urbanization derived from population migration occurring over the past several decades [1]. Compared to small cities, large cities can absorb higher populations because of its better employment opportunities, access to medical care, education levels, etc. Rapid population migrations and the intensive living and working conditions in high-density populated urban areas have impacted the ecology and the environment in these areas in many visible ways, including through environmental pollution and ecological change (e.g., air pollution and net primary productivity [2,3]). Thus, special concern is paid to the inﬂuence of world urbanization, especially the impact of increased population derived from intensive human activities on urban air quality, which is easily visible to both citizens and authorities [4]. As traditional air pollutants have been monitored and analyzed for quite some time, some traditional air pollutants (e.g., NO2) have been well examined in conjunction with changes in population in urban areas [5]. However, many other air pollutants, which may not be new but can now be monitored thanks to recent technological improvement, have not been well documented [4].

Sustainability 2018, 10, 1696; doi:10.3390/su10051696 www.mdpi.com/journal/sustainability Sustainability 2018, 10, 1696 2 of 9

Fine particulates with less than 2.5 µm in diameter (PM2.5), which have toxic components that can negatively impact public health and meteorological visibility, are important major air pollutants in urban areas [6,7]. Traditional operational air quality monitoring stations provide very accurate records at certain locations [7–10]. But, the use of these records are limited by the spatial availability of the stations, and cannot illustrate the full-covered spatial patterns of PM2.5 concentration in long time-series on a global scale. Thus, satellite remotely sensed records of PM2.5 concentration together with complex models validated/calibrated by ground observations is highly suggested to be employed in order to present a quantitative spatial analysis [11–13]. For global scale urbanization research, the size of an urban population is an important and easy-to-compare factor to illustrate urbanization. It could explain the general regulation of urban development models. Urban air pollution has been highly attributed to urbanization in terms of urban population size in previous studies [4,5]. Therefore, to investigate the scaling relationship between changes in PM2.5 and growing urbanization in large cities, by using the full-covered PM2.5 concentration and United Nation urban population records at a global scale.

2. Materials and Methodology

2.1. Populations and Large Cities In our research, large cities refer to cities with a population of more than 0.75 million, which is a general standard by the United Nations. The population records used, which are statistical records for each city in 2000 and 2010, were collected from World Urbanization Prospects (available at http://esa.un.org/unup/). The data provided the location of each large city, but without city boundary information. We then employed a moderate resolution imaging spectroradiometer (MODIS) global map of urban extent to overlay with the large city points [14,15]. A ﬁnal global large cities’ polygon layer, with a spatial resolution of 0.5 × 0.5 km, was selected where both urban extent and large cities were identiﬁed. Finally, 456 large cities were selected: 252 in Asia, 50 in Europe, 53 in Latin America, 55 in America, 41 in Africa, and 5 in other areas (See Supplementary Materials for the details of those cities).

2.2. PM2.5 Concentration Trends in Large Cities

The PM2.5 concentrations were estimated with an optimal estimation algorithm based on top-of-atmosphere reflectance observed by MODIS products [12,13]. Practically, based on the simulation of a GEOS-Chem chemical transport model, the PM2.5 concentrations were estimated from a combination of MODIS and multi-angle imaging spectroradiometer (MISR) aerosol optical depth (AOD) with aerosol vertical profiles and scattering properties [11,12]. The global PM2.5 concentration dataset had a spatial resolution of 10 × 10 km as a three year moving average during 1999–2011, (available at http://fizz.phys.dal.ca/~atmos/martin/?page_id=140[ 12]). The dataset was downloaded and used in this study, and the PM2.5 trend was calculated as the significant (p < 0.05) slope of the linear regression model at each pixel’s time series. Positive/negative trends were then defined as trends larger than zero or smaller than zero, respectively. Finally, the PM2.5 concentration trends were resampled to meet the resolution (i.e., 0.5 × 0.5 km) with the global cities’ polygon and then overlaid with the layer of the global large cities’ layer to obtain the PM2.5 concentration trends at each large city.

2.3. Scaling Relationship Analysis

At a global scale, comparing the relationship between urbanization and PM2.5 dynamics in large cities is one of the most appropriate and easy ways to understand the types of impacts urbanization has on the eco-environment. It can provide knowledge for policy-makers to select appropriate ways to design their strategies on both urban development and eco-environmental conservation. Thus, the scaling analysis was selected and carried out for this research. The PM2.5 concentration trends Sustainability 2018, 10, 1696 3 of 9

Sustaiwerenability ﬁrst divided2018, 10, x and FOR averagedPEER REVIEW by the population change rates at every 10% interval to examine3 of the 9 variations of PM2.5 concentration change against urban population change. Then, correlation analysis analysis was carried out between the mean PM2.5 concentration trends and their urban population was carried out between the mean PM concentration trends and their urban population change change rates (which is the scaling relationship2.5 in our work) in large cities in different regions and rates (which is the scaling relationship in our work) in large cities in different regions and countries to countries to quantify the impact of growing urbanization on PM2.5 pollution change. quantify the impact of growing urbanization on PM2.5 pollution change.

2.4. Potential Contribution of Urbanization to PM2.5 Concentration 2.4. Potential Contribution of Urbanization to PM2.5 Concentration Increasing PM2.5 concentration is expected to come from growing human activities that emit Increasing PM2.5 concentration is expected to come from growing human activities that emit pollutantpollutants.s. However, However, an an increase increase in in human human activities activities that that polluted polluted did did not not always always come come from from an an increaseincrease inin urbanurban population population for for each each city. city In. low-technical, In low-technical labor-based, labor-based cities, cities,human human activities activities directly directly resulted from an increase in urban population, and urban population increase led to higher resulted from an increase in urban population, and urban population increase led to higher PM2.5 PMconcentrations.2.5 concentration However,s. However, in the casein the of high-technical,case of high-technical non-labor, non based-labor cities, based urban cities population, urban population increase did not necessarily lead to increases in PM2.5 concentrations. At a global scale, to increase did not necessarily lead to increases in PM2.5 concentrations. At a global scale, to simplify simplify the potential contribution of growing urbanization, in terms of urban population change, to the potential contribution of growing urbanization, in terms of urban population change, to PM2.5 PM2.5 concentration, we assumed that the increase of urban PM2.5 concentration came from an concentration, we assumed that the increase of urban PM2.5 concentration came from an increase increase in urban population, and therefore, took the PM2.5 concentration trend per increased capita in urban population, and therefore, took the PM2.5 concentration trend per increased capita as the as the potential contribution of urbanization to PM2.5 concentration. potential contribution of urbanization to PM2.5 concentration.

3.3. Result Resultss and and Discussion Discussion

The significant trends (i.e., p > 0.05) of PM2.5 concentration were found to vary among regions The significant trends (i.e., p > 0.05) of PM2.5 concentration were found to vary among regions ((FigureFigure 11).). FourFour areasareas couldcould bebe highlightedhighlighted duedue toto theirtheir significantsignificant negative/positivenegative/positive trends. Central andand Eastern Eastern U.S U.S.. were were observed observed with with significant significant negative negative trends, trends, indicating indicating the the improvement improvement of of air air quality.quality. The The n northeast,ortheast, e east,ast, south, south, and and southeast southeast of of China, China, India, India, and and the the Arabian Arabian Peninsula Peninsula were were foundfound to to have have significant significant positive positive trends, trends, indicating t thehe degeneration of air quality.

FigFigureure 1. GlobalGlobal PM PM2.52.5 trendtrendss from from 1999 1999–2011–2011 in in large large cities. cities.

SimilarSimilar result resultss were were reported reported in in some some of of the the previous previous research research [12,13,1 [12,13,166,1,177]],, and and the the changes changes in in EastEast China were explainedexplained toto havehave been been primarily primarily caused caused by by the the increase increase in in urban urban population population and and the thedecrease decrease in wind in wind speeds speed [18s,19 [18,19]]. However. However in the in U.S., the India, U.S., India and the, and Arabian the Arabian Peninsula, Peninsula, the reasons the reasonsfor the changesfor the changes are still are unknown still unknown and need and further need analysis.further analysis. The significant trends of PM2.5 concentration in large cities were also varied (Figure 2). In The significant trends of PM2.5 concentration in large cities were also varied (Figure2). continentsIn continents dominateddominated by byhigh high-income-income countries countries (i.e., (i.e., North North America America and and Europe Europe),), most most cities cities observed significant decreases in PM2.5 concentrations, and only 1.8% and 28% of large cities in these observed significant decreases in PM2.5 concentrations, and only 1.8% and 28% of large cities two areas, respectively, observed significant increases of PM2.5 concentrations at less than 0.5 in these two areas, respectively, observed significant increases of PM2.5 concentrations at less μg/m3·year. All3 large cities in Africa and Latin America were found to have PM2.5 concentration than 0.5 µg/m ·year. All large cities in Africa and Latin America were found to have PM2.5 3 trendsconcentration of less than trends 0.5 of μg/m less than·year,0.5 meanwhileµg/m3·year 53.,0 meanwhile% and 66.0% 53.0% of their and 66.0% large cities, of their respectively, large cities, were found with increasing trends. Only 4.0% of the large cities in Asia had PM2.5 concentrations with significant negative trends, but 8.0%, 16.4%, and 32.3% of the large cities were found to have Sustainability 2018, 10, 1696 4 of 9

Sustainability 2018, 10, x FOR PEER REVIEW 4 of 9 respectively, were found with increasing trends. Only 4.0% of the large cities in Asia had PM2.5 concentrations with signiﬁcant negative trends, but 8.0%, 16.4%, and 32.3% of the large cities were positive trends of more than 3.0 μg/m3·year, from 2.0–3.0 μg/m3·year, and from 1.0–2.0 μg/m3·year, found to have positive trends of more than 3.0 µg/m3·year, from 2.0–3.0 µg/m3·year, and from respectively. 1.0–2.0 µg/m3·year, respectively.

100%

80%

60%

trend 2.5

40% of PM of 20%

0% Portion of cities cities in ofdifferent Portion groups North Latin Africa Asia Europe U.S.A China India America America > 3.0 8.0% 15.6% 2.5 - 3.0 8.0% 15.6% 2.0 - 2.5 8.4% 14.1% 1.7% 1.5 - 2.0 12.0% 14.1% 12.1% 1.0 - 1.5 20.3% 15.6% 39.7% 0.5 - 1.0 19.5% 2.0% 12.5% 36.2% 0 - 0.5 53.7% 1.8% 19.9% 26.0% 66.0% 2.0% 10.2% 10.3% -0.5 - 0 26.8% 83.6% 4.0% 66.0% 32.1% 82.0% 2.3% -1 - -0.5 12.2% 14.5% 6.0% 1.9% 16.0% -1.5 - -1 4.9% <-1.5 2.4%

FigureFigure 2.2. PortionPortion of of large large cities cities in differentin different PM 2.5PMtrend2.5 trend groups groups at continental at continent andal typical and typical country countr scales.y scales.

The significant trends of PM2.5 concentration in large cities were also different between large cities in typicalThe significant high-income trends countries of PM (e.g.,2.5 concentration U.S.) and typical in large low- cities and were mid-income also different countries between (e.g., China large andcities India) in typical (Figure high2).-income Only 2.0% countr of theies cities(e.g., inU.S. the) and U.S. typical were found low- withand mid significant-income positive countries trends, (e.g., whichChina and were India less) than (Figure 0.5 µ2)g/m. Only3·year. 2.0% However,of the cities only in the 2.3% U.S. of were the cities found in with China signific wereant found positive with 3 significanttrends, which negative were trends,less than and 0.5 no μg/m cities·year. in India However, were found only with2.3% significantof the cities negative in China trends. were found with Rapidsignificant urbanization negative intrends, large and cities no on cities continents in India such were as found Europe with and significant North America negative has trends. largely finished,Rapid and urbanization therefore, there in large has beencitiesless on conti of annents impact such on theas Europe environment. and North Anti-pollution America has measures largely finished, and therefore, there has been less of an impact on the environment. Anti-pollution in the U.S. and Canada have contributed to the decrease of PM2.5. In addition, the economic crisis measures in the U.S. and Canada have contributed to the decrease of PM2.5. In addition, the economic (2007–2009) may have also affected the trend of PM2.5 [20,21]. Similar conditions can be found in Latin America,crisis (200 where7–2009 80%) may of have the populationalso affected lives the intrend cities. of PM However2.5 [20,21] large. Similar cities condition in East Asia,s can particular be found Chinain Latin and Americ India,a,are where still 80%undergoing of the population rapid urbanization lives in cities. which However brings strong large negative cities in impacts East Asia, to particular China and India, are still undergoing rapid urbanization which brings strong negative the environment. Thus, the change of PM2.5 concentration should be a useful indicator to explain a impacts to the environment. Thus, the change of PM2.5 concentration should be a useful indicator to development strategy that balances urbanization and environmental protection at the country level. explain a development strategy that balances urbanization and environmental protection at the Such information would highlight the development strategy which should be given more attention country level. Such information would highlight the development strategy which should be given for environmental protection in countries experiencing rapid urbanization. Moreover, we believe that more attention for environmental protection in countries experiencing rapid urbanization. cooperation between the government and the public to mitigate the PM2.5 hazard would ensure a Moreover, we believe that cooperation between the government and the public to mitigate the PM2.5 sustainable future, while accommodating low- and mid-income countries’ growing urbanization. hazard would ensure a sustainable future, while accommodating low- and mid-income countries’ The contribution of population to PM2.5 changes were varied among continents, and between growing urbanization. high-income and low- and mid-income countries (Figure3). Large cities in Asia had an average and The contribution of population to PM2.5 changes were varied among continents, and between standard deviation of 4.12 ± 4.27 µg/m3·year per million people increase, while all large cities on the high-income and low- and mid-income countries (Figure 3). Large cities in Asia had an average and standard deviation of 4.12 ± 4.27 μg/m3·year per million people increase, while all large cities on the rest of the continents were found with contributions of less than or near the average and standard deviation of 4.12 ± 4.27 μg/m3·year per million people. Although large cities in the U.S. had a Sustainability 2018, 10, 1696 5 of 9 Sustainability 2018, 10, x FOR PEER REVIEW 5 of 9

Sustairestnegative ofnability the contribution 201 continents8, 10, x FOR were toPEER PM found REVIEW2.5 concentration with contributionss (−1.32 of± less1.37 than μg/m or3·year near per the millionaverage p andeople), standard 5 large of 9 deviationcities in China of 4.12 (5.40± 4.27 ± µ4.80g/m μg/m3·year3·year per million per million people. people) Although and large India cities (4.07 in the± 3.07 U.S. μg/m had a3·year negative per negative contribution to PM2.5 concentrations (−1.32 ± 1.373 μg/m3·year per million people), large contributionmillion people) to had PM2.5 a positiveconcentrations contribution (−1.32 to± PM1.372.5 concentrationµg/m ·year pers. million people), large cities in 3 3 citiesChina in (5.40 China± 4.80 (5.40µ g/m± 4.803·year μg/m per·year million per people) million andpeople) India and (4.07 India± 3.07 (4.07µ g/m± 3.073·year μg/m per·year million per million people) had a positive contribution to PM2.5 concentrations.

people) had a positive12.00 contribution to PM2.5 concentrations. )

12.008.00 )

8.004.00

percapita million person million

4.000.00

per

trend

percapita million person million

2.5 -0.004.00

•year

per PM

trend -8.00 g/m

2.5 -4.00

(

•year

3 PM

-12.00-8.00 g/m

μ Africa North Asia Europe Latin U.S.A China India ( -12.00 America America

Africa North Asia Europe Latin U.S.A China India Figure 3. PM2.5 trends perAmerica capita in large cities inAmerica different countries and continents.

Figure 3. Similar Fig to ure our 3. PM previous2.5 trendtrends sassertion per per capita capita inthat in large large population cities cities in in different different increase countries countries did andnot continents continents. necessarily. result in environmental degeneration, the type of development could be the key factor. The contribution of populationSimilar increase to our to previous PM2.5 concentration assertion thatchanges populationpopulation would suggest increase a sustainable did not necessarily development result model in environmentalbetween urbanization degeneration, and environmentalthe type of development protection could , particular be thely key for factor. large The cities contribution in rapid ofly populationurbanizing increasecountries, to ( PMe.g.2.52.5, China concentrationconcentration and India changes changes). The types wou would ldof suggest development a sustainable models developmentin Europe, U.S, model and betweenLatin America urbanization urbanization have alr and eady and environmental provided environmental a better protection, protectionstrategy particularly for, developmentparticular for largely forwithout cities large in bring rapidly citiesing inurbanizingsignificant rapidly urbanizcountries,negativeing impact (e.g., countries, Chinas to the and(e.g. environment. India)., China The and types IndiaIn addition, of). developmentThe types the valuesof development models and in patterns Europe, models of U.S, theinand Europe, contribution Latin U.S,America ands in Latinhavedifferent alreadyAmerica regions provided have and alr countries aeady better provided strategy were asimilar for better development tostrategy the previous for without development analysis bringing, whichwithout signiﬁcant used bring negativeonlying significantone impacts year’s negativetoworth the environment.of dataimpact froms to the Inthe addition,large environment. cities the [4] values. ItIn proved addition, and patternsagain the that values of a the method contributionsand patterns of spatial inof sequence differentthe contribution regionsinstead andsof in a differentcountriestime succession regions were similar sequenceand countries to the can previous workwere similar when analysis, doingto whichthe researchprevious used only oanalysisn onethe year’simpact, whichworth ofused population ofonly data one from onyear the air’s worthlargepollution. cities of data [4]. from It proved the large again cities that [4] a method. It proved of spatial again sequencethat a method instead of ofspatial a time sequence succession instead sequence of a timecan work Diversesuccession when relationships doing sequence research betweencan onwork the P impactM when2.5 trends ofdoing population and research population on o airn pollution.the change impact rate ofs in population large cities on were air pollution.foundDiverse in different relationships areas between (Figure PM4).2.5 Significanttrends and populationnegative relationships change rates in were large mainly cities were found found in incontinents differentDiverse withareas relationships (Figurehigh-income4). between Signiﬁcants, while PM negative inverse2.5 trends, relationships“ andU-type population” relationship were change mainlys were ratefounds mainlyin in large continents obtained cities with were in foundhigh-incomes,mid-income in different continents. while inverse,areas (Figure “U-type” 4). relationshipsSignificant negative were mainly relationships obtained in were mid-income mainly continents. found in continents with high-incomes, while inverse, “U-type” relationships were mainly obtained in mid-income0 continents. 0.2

0.1

-0.20 0.2 •year)

•year) 0.0

3 3

0.1

g/m g/m

μ -0.1 μ

-0.20.4 •year)

•year) 0.0 3

3 -0.2

trend(

g/m

2.5 μ 2.5 -0.1 μ -0.3

PM -0.4

-0.6 PM y = -0.0739x - 0.0373 -0.2

trend( y = 0.0503x - 0.2258 R² = 0.9195, P < 0.05 trend( -0.4 2.5 R² = 0.9161, P < 0.05

2.5 -0.3 PM -0.60.8 PM -0.5

y = -0.0739x - 0.0373

20 30 50 40 10

<0 y = 0.0503x - 0.2258

- - - -

- -

>50 -0.4 >20

R² = 0.91950 , P < 0.05

0 R² = 0.9161, P < 0.05

10 20 40 30 10

North-0.8 America Population change rate (% per decade) Europe-0.5 Population change rate (% per decade)

20 30 50 40

- - - -

>50

>20

10 20 40 30 10 North America Population change rate (% per decade) Figure 4. ContEurope. Population change rate (% per decade) Sustainability 2018, 10, 1696 6 of 9 Sustainability 2018, 10, x FOR PEER REVIEW 6 of 9 Sustainability 2018, 10, x FOR PEER REVIEW 6 of 9

5.0 1.0 5.0 y = -0.0342x2 + 0.5352x - 0.3769 1.0 y = -0.0232x - 0.0325 R²y = = - 0.0563,0.0232x P - <0.0325 0.05 4.0 y = -0.0342xR² = 0.80652 + 0.5352x, P < 0.05 - 0.3769 0.5 R² = 0.0563, P < 0.05 4.0 R² = 0.8065, P < 0.05

•year) 0.5

•year)

3 3

3.0 •year)

•year)

3 3 g/m 0.0

g/m 3.0

μ μ

g/m 0.0

g/m μ

μ 2.0 2.0

trend( -0.5

trend(

2.5 trend(

2.5 -0.5

trend( 1.0

2.5 PM 2.5 1.0

PM -1.0 PM PM 0.0 -1.0 0.0

-1.0 -1.5

20 30 40 50 60 70 80 90

30 40 50 60 70 80 90

20 -1.5

------

-1.0 -

100

------

100

>100

20 30 40 50 60 70 80 90

>100

20 30 40 50 60 70 80 90

10 20 30 40 50 60 70 80

------

10 20 30 40 50 60 70 80

100

------

100

>100

10 20 30 40 50 60 70 80

Asia Africa Population change rate (% per decade) 90 Population change rate (% per decade) 90 Africa Asia Population change rate (% per decade) Population change rate (% per decade) 0.4 y = -0.0068x + 0.0636 0.4 R²y = = - 0.0068x0.0524, P+ <0.0636 0.05 0.3 R² = 0.0524, P < 0.05 0.3

•year) 0.2 3

•year) 0.2

3 g/m

μ 0.1 g/m

μ 0.1 0.0

trend( 0.0 2.5

trend( -0.1 2.5 PM -0.1 PM -0.2 -0.2

-0.3

30 20 10

-0.3 <0

- -

>30

30 20

20 10

- -

>30

0 20 Latin America Population change rate10 (% per decade) Latin America Population change rate (% per decade) Figure 4. Relationship between PM2.5 trends (Y-axis; μg/m33·year) and population change rates Figure 4. Relationship between PM2.5 trends (Y-axis; µg/m3 ·year) and population change rates Figure 4. Relationship between PM2.5 trends (Y-axis; μg/m ·year) and population change rates (X--axis;axis; % % per decade) in large cities on different continents. The The bars bars in in each each figures figures indicate the the (X-axis; % per decade) in large cities on different continents. The bars in each figures indicate the standardstandard deviations of each group. Significant Significant negative relationshiprelationshipss (R 22 == 0 0.9195,.9195, pp << 0.05) were standard deviations of each group. Significant negative relationships (R2 = 0.9195, p < 0.05) were obtained between PM2.5 trends and population change rates in large cities of North America. obtained between PM2.5 trends and population change rates in large cities of North America. However, obtained between PM2.5 trends and population change rates in large cities of North America. 2 2 2.5 However,significant significant positive relationships positive relationship (R = 0.9161,s (R = p0.9161,< 0.05) p < were 0.05) obtainedwere obtained between between PM2.5 PMtrends trends and However, significant positive relationships (R2 = 0.9161, p < 0.05) were obtained between PM2.5 trends andpopulation population change change rates in rate larges in cities large of cities Europe. of Moreover,Europe. Moreover, a significant a significantinverse “U-type” inverse relationship “U-type” and population change rates in large cities of Europe. Moreover, a significant inverse “U-type” 2 2 relationship(R = 0.8065, (Rp < = 0.05) 0.8065, was pfound < 0.05) between was found PM 2.5betweentrends andPM2.5 population trends and change population rates inchange large rate citiess in of relationship (R2 = 0.8065, p < 0.05) was found between PM2.5 trends and population change rates in largeAsia. However,cities of Asia. no significant However, relationships no significant were relationships found between were PM found2.5 trends between and population PM2.5 trends change and large cities of Asia. However, no significant relationships were found between PM2.5 trends and populationrates in large change cities ofrate Africas in large and cities Latin of America. Africa and Latin America. population change rates in large cities of Africa and Latin America. Similar to the scaling relationship among continents, the relationships were also different Similar to to the the scaling scaling relationship relationship among among continents, continents, the relationships the relationship weres also were different also different between between typical low- and mid-income countries (e.g., China and India) and high-income countries typicalbetween low- typical and low mid-income- and mid countries-income countries (e.g., China (e.g., and China India) and and India) high-income and high countries-income (e.g.,countries U.S) (e.g., U.S) (Figure 5). Significant negative relationships were mainly found in countries with (Figure(e.g., U.S)5). Significant (Figure 5) negative. Significant relationships negative were relationships mainly found were in countries mainly foundwith high-incomes, in countries while with high-incomes, while “U-type” relationships were mainly obtained in mid-income countries. “U-type”high-income relationshipss, while “U were-type mainly” relationship obtaineds w inere mid-income mainly obtained countries. in mid-income countries.

0.0 3.5 0.0 3.5 3.0 -0.1 3.0 -0.1 2.5

-0.2 •year) •year)

3 2.5

3 2.0

-0.2 •year)

•year)

3 3

g/m 2.0 g/m

-0.3 μ 1.5

g/m g/m

-0.3 μ 1.5 μ

-0.4 1.0 trend(

trend( 1.0

-0.4 2.5

trend( 0.5

2.5 trend(

-0.5 2.5 0.5

PM 2.5

PM -0.5 0.0 PM

PM 0.0 -0.6 y = -0.0565x - 0.1269 -0.5 y = -0.0345x2 + 0.3344x + 1.225 -0.6 R²y = = - 0.84980.0565x, P - <0.1269 0.05 -0.5 2 -1.0 y = -0.0345xR² = 0.6881+ 0.3344x, P < 0.05 + 1.225

-0.7 R² = 0.8498, P < 0.05 R² = 0.6881, P < 0.05

20 30 50 60 70 80 90

-1.0 40

------

20 30 40

10 100

-0.7 <0

- - -

>40

>100

20 30 50 60 70 80 90 40

10 20 40 50 60 70 80 30

------

20 30 40

100

10 20 30

- - -

>40

>100

10 20 40 50 60 70 80

U.S.A China 30

10 20 30 Population change rate (% per decade) Population change rate (% per decade) 90 U.S.A Population change rate (% per decade) China Population change rate (% per decade)

Figure 5. Cont. Sustainability 2018, 10, 1696 7 of 9 Sustainability 2018, 10, x FOR PEER REVIEW 7 of 9

2.0 1.8 1.6

•year) 1.4 3

g/m 1.2 μ 1.0

trend( 0.8

2.5 0.6 PM 0.4 y = -0.081x + 1.2913 0.2 R² = 0.7129, P < 0.05

0.0

20 30 40

- - -

>40

10 20 30 India Population change rate (% per decade)

Figure 5. Relationship between PM2.5 trends (Y-axis; μg/m3 3·year) and population change rates Figure 5. Relationship between PM2.5 trends (Y-axis; µg/m ·year) and population change rates (X-axis; (%X- peraxis; decade) % per decade) in large in cities large of cities typical of high-incometypical high-income and low- and and low mid-income- and mid-income countries. countries. The bars The in barseach in ﬁgures each figures indicate indicate the standard the standard deviations deviations of each of group. each group.

Significant negative relationships were found between PM2.5 trends and population change Significant negative relationships were found between PM trends and population change rates rates in large cities in both India (R2 = 0.7129, p < 0.05) and U.S.2.5 (R2 = 0.8498, p < 0.05); however, in large cities in both India (R2 = 0.7129, p < 0.05) and U.S. (R2 = 0.8498, p < 0.05); however, population population increases made large cities’ PM2.5 increase in India, but decrease in the U.S. China had a increases made large cities’ PM2.5 increase in India, but decrease in the U.S. China had a unique inverse unique inverse “U-type” relationship between PM2.5 trends and population change rates in large “U-type” relationship between PM2.5 trends and population change rates in large cities; cities with cities; cities with increases in PM2.5 trends were found to have population increase rates of less than increases in PM2.5 trends were found to have population increase rates of less than 60%, but cities with 60%, but cities with decrease of PM2.5 trends were found with population increase rates more than decrease of PM trends were found with population increase rates more than 60%. 60%. 2.5 These scaling relationships explain the relationship between environmental pollution and These scaling relationships explain the relationship between environmental pollution and economic development. This follows the Environmental Kuznets Curve (EKC) hypothesis, which states economic development. This follows the Environmental Kuznets Curve (EKC) hypothesis, which that environmental pollution first increases and then decreases [22]. The results found in this work also states that environmental pollution first increases and then decreases [22]. The results found in this follow the EKC hypothesis. For instance, in high-income continents/countries, population increased work also follow the EKC hypothesis. For instance, in high-income continents/countries, population but PM2.5 concentration decreased, indicating a high level of development with consideration of both increased but PM2.5 concentration decreased, indicating a high level of development with urban development and environmental improvement. In mid- and low-income continents/countries, consideration of both urban development and environmental improvement. In mid- and the EKC hypothesis can also be found. For the case of China, when population change rates increased, low-income continents/countries, the EKC hypothesis can also be found. For the case of China, when the PM2.5 trend increased first and then decreased, indicating those cities were still before the peak population change rates increased, the PM2.5 trend increased first and then decreased, indicating points of their EKC. The large cities with larger population increase rates paid more attention to air those cities were still before the peak points of their EKC. The large cities with larger population pollution, making the PM2.5 increases smaller than the cities with lower population increase rates. increase rates paid more attention to air pollution, making the PM2.5 increases smaller than the cities This is highly attributed to the intensive environmental management in cities with rapid urbanization; with lower population increase rates. This is highly attributed to the intensive environmental some examples include Shenzhen and Guangzhou in South China, where their air pollution days and management in cities with rapid urbanization; some examples include Shenzhen and Guangzhou in pollutants’ concentration decreased during the period. South China, where their air pollution days and pollutants’ concentration decreased during the Finally, the analysis in this work has some limitations which need to be considered in future period. research. The analysis in this work used a global scale dataset, which may have introduced errors into Finally, the analysis in this work has some limitations which need to be considered in future the results. First, the MODIS global map of urban extent introduces low errors for regions with low research. The analysis in this work used a global scale dataset, which may have introduced errors speeds of urban expansion (e.g., U.S. and Europe), but introduces considerable error in regions with into the results. First, the MODIS global map of urban extent introduces low errors for regions with high speeds of urban expansion (e.g., China). We therefore suggest to adopt new urban extent results, low speeds of urban expansion (e.g., U.S. and Europe), but introduces considerable error in regions which could represent changes from 2000 to 2010 in the future, when such datasets are available. with high speeds of urban expansion (e.g., China). We therefore suggest to adopt new urban extent Furthermore, we are currently using the PM concentration at the near-surface total amount, which results, which could represent changes from2.5 2000 to 2010 in the future, when such datasets are could be impacted by both nature (e.g., temperature, wind, precipitation, and humidity [23–25]) available. Furthermore, we are currently using the PM2.5 concentration at the near-surface total and anthropogenic emissions, which could be the direct emission of PM and secondary PM by amount, which could be impacted by both nature (e.g., temperature,2.5 wind, precipitation, 2.5 and emission of SO2, NO2, etc. (e.g., Reference [26]). Further detailed analysis in certain areas should also humidity [23–25]) and anthropogenic emissions, which could be the direct emission of PM2.5 and take into concern the source and meteorological factors beyond population amounts. secondary PM2.5 by emission of SO2, NO2, etc. (e.g., Reference [26]). Further detailed analysis in certain4. Conclusions areas should also take into concern the source and meteorological factors beyond population amounts. Urban air quality changes and their relationships with urbanization provide an important insight 4.for Conclusions understanding the urban development strategies of large cities in different countries. Moreover, such information would suggest that low- and mid-income countries with rapid urbanization should Urban air quality changes and their relationships with urbanization provide an important insight for understanding the urban development strategies of large cities in different countries. Sustainability 2018, 10, 1696 8 of 9 adopt better strategies to counteract increasing urban air pollution. Thus, we analyzed the scaling relationship between changes in urban population and PM2.5 concentrations. Our conclusions are:

(1) Trends in PM2.5 varied among global regions. Central and Eastern U.S. were observed to have signiﬁcant negative trends. The northeast, east, south, and southeast of China, India, and the Arabian Peninsula were found to have signiﬁcant positive trends.

(2) The contribution of population to PM2.5 changes were greatly varied among continents, and between high-income and low- and mid-income countries. Except for large cities in Asia, which had a 4.12 ± 4.27 µg/m3·year per million population increase, all large cities on the rest of the continents were found with contributions less than or near to 4.12 ± 4.27 µg/m3·year per million population. While, large cities in the U.S. had a negative contribution to PM2.5 concentrations, large cities in China and India had positive contributions to PM2.5 concentrations. (3) Different global areas identiﬁed diverse relationships between PM2.5 trends and population change rates in large cities. Other than the relationships in large cities in North America and Europe, a signiﬁcant inverse “U-type” relationship (p < 0.05) was found between PM2.5 trends and population change rates in large cities in Asia. Particularly was China’s unique inverse “U-type” relationship between PM2.5 trends and population change rates in large cities. Cities with increasing PM2.5 trends were observed to have population increase rates of less than 60%, but cities with decreasing PM2.5 trends were observed to have population increase rates of more than 60%.

Supplementary Materials: The following are available online at http://www.mdpi.com/2071-1050/10/5/1696/ s1, Figure S1: title, Table S1: title, Video S1: title. Author Contributions: L.H. review the literature, designed the research, performed the research, analyzed the data, and wrote the paper; W.Z. review the literaure, analyzed the data, wrote and revised the paper; W.L. wrote and revised the paper. Acknowledgments: This research was supported by the National Natural Science Foundation of China (41590841, and 41771201). In addition, the research received financial support from the National Key Research and Development Plan of China (Grant No. 2016YFC0503004); and the Key Research Project of Frontier Science of the Chinese Academy of Sciences (Grant No. QYZDB-SSW-DQC034-2). Conflicts of Interest: The authors declare no conflict of interest. The founding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the decision to publish the results.

References

1. Population Division, Department of Economic and Social Affairs, United Nations. World Urbanization Prospects: The 2011 Revision; Population Division; Department of Economic and Social Affairs, United Nations: New York, NY, USA, 2012. 2. Peng, J.; Shen, H.; Wu, W.; Liu, Y.; Wang, Y. Net primary productivity (NPP) dynamics and associated urbanization driving forces in metropolitan areas: A case study in Beijing City, China. Landsc. Ecol. 2016, 31, 1077–1092. [CrossRef] 3. OECD. OECD Environmental Outlook to 2050: The Consequences of Inaction; OECD Publishing: Paris, France, 2012. 4. Han, L.; Zhou, W.; Li, W. An optimum city size? The scaling relationship for urban population and ﬁne

particulate (PM2.5) concentration. Environ. Pollut. 2016, 208, 96–101. [CrossRef][PubMed] 5. Lamsal, L.; Martin, R.; Parrish, D.; Krotkov, N. Scaling Relationship for NO2 pollution and urban population size: A satellite perspective. Environ. Sci. Technol. 2013, 47, 7855–7861. [CrossRef][PubMed] 6. Lim, S.S.; Vos, T.; Flaxman, A.D.; Danaei, G.; Shibuya, K.; Adair-Rohani, H.; AlMazroa, M.A.; Amann, M.; Anderson, H.R.; Andrews, K.G.; et al. 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 2012, 380, 2224–2260. [CrossRef] 7. Han, L.; Zhou, W.; Li, W.; Li, L. Impact of urbanization level on urban air quality: A case of ﬁne particles

(PM2.5) in Chinese cities. Environ. Pollut. 2014, 194, 163–170. [CrossRef][PubMed] Sustainability 2018, 10, 1696 9 of 9

8. Gupta, S.; Pebesma, E.; Mateu, J.; Degbelo, A. Air quality monitoring network design optimization for robust land use regression models. Sustainability 2018, 10, 1–27. [CrossRef] 9. Buke, T.; Kone, A. Assessing air quality in Tukey: A proposed, air quality index. Sustainability 2016, 8, 73. [CrossRef] 10. Wang, Z.; Ke, L.; Cui, X.; Yin, Q.; Liao, L.; Gao, L.; Wang, Z. Monitoring environmental quality by sniffing social media. Sustainability 2017, 9, 85. [CrossRef] 11. Van Donkelaar, A.; Martin, R.V.; Brauer, M.; Kahn, R.; Levy, R.; Verduzco, C.; Villeneuve, P.J. Global estimates of ambient fine particulate matter concentrations from satellite-based aerosol optical depth: Development and application. Environ. Health Perspect. 2010, 118, 847–855. [CrossRef][PubMed] 12. Van Donkelaar, A.; Martin, R.; Brauer, M.; Boys, B. Use of satellite observation for long-term exposure assessment of global concentration of fine particulate matter. Environ. Health Perspect. 2015, 123, 135–143. [CrossRef][PubMed] 13. Boys, B.L.; Martin, R.V.; van Donkelaar, A.; MacDonell, R.; Hsu, N.C.; Cooper, M.J.; Yantosca, R.M.; Lu, Z.; Streets, D.G.; Zhang, Q.; et al. Fifteen-year global time series of satellite-derived fine particulate matter. Environ. Sci. Technol. 2014, 48, 11109–11118. [CrossRef][PubMed] 14. Schneider, A.; Friedl, M.; Potere, D. A new map of global urban extent from MODIS satellite data. Environ. Res. Lett. 2009, 4, 044003. [CrossRef] 15. Schneider, A.; Friedl, M.; Potere, D. Monitoring urban areas globally using MODIS 500m data: New methods and datasets based on ‘urban ecoregions’. Remote Sens. Environ. 2010, 114, 1733–1746. [CrossRef]

16. Han, L.; Zhou, W.; Li, W. City as a major source area of PM2.5 in China. Environ. Pollut. 2015, 206, 183–187. [CrossRef][PubMed]

17. Peng, J.; Chen, S.; Lv, H.; Liu, Y.; Wu, J. Spatiotemporal pattern of remotely sensed PM2.5 concentration in China from 1999 to 2011. Remote Sens. Environ. 2016, 174, 109–121. [CrossRef]

18. Han, L.; Zhou, W.; Li, W. Increasing impact of urban PM2.5 in China. Sci. Rep. 2015, 5, 12467. [CrossRef] [PubMed] 19. Lin, C.; Yang, K.; Huang, J.; Tang, W.; Qin, J.; Niu, X.; Chen, Y.; Chen, D.; Lu, N.; Fu, R. Impact of wind stilling on solar radiation variability in China. Sci. Rep. 2015, 5, 15135. [CrossRef][PubMed] 20. Robichaud, A.; Ménard, R. Multi-year objective analyses of warm season ground-level ozone and PM2.5 over North America using real-time observations and Canadian operational air quality models. Atmos. Chem. Phys. 2014, 14, 1769–1800. [CrossRef] 21. Castellanos, P.; Folkert Boersma, F. Reductions in nitrogen oxides over Europe driven by evironmental policy and economic recession. Sci. Rep. 2012, 2, 265. [CrossRef][PubMed] 22. Stern, D.; Common, M.; Barbier, E. Economic growth and environmental degradation: The envrionmental Kuznets curve and sustainable development. World Dev. 1996, 24, 1151–1160. [CrossRef] 23. Jacob, D.; Winner, D. Effect of climate change on air quality. Atmos. Environ. 2009, 43, 51–63. [CrossRef]

24. Tai, A.; Mickley, L.; Jacob, D. Correlations between ﬁne particulate matter (PM2.5) and meteorological variables in the United States: Implications for the sensitivity of PM2.5 to climate change. Atmos. Environ. 2010, 44, 3976–3984. [CrossRef] 25. Jeong, J.; Park, R. Effect of the meteorological variability on regional air quality in East Asia. Atmos. Environ. 2013, 69, 46–55. [CrossRef] 26. Huang, R.J.; Zhang, Y.; Bozzetti, C.; Ho, K.F.; Cao, J.J.; Han, Y.; Daellenbach, K.R.; Slowik, J.G.; Platt, S.M.; Canonaco, F.; et al. High secondary aerosol contribution to particulate pollution during haze events in China. Nature 2014, 514, 218–222. [CrossRef][PubMed]

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).