1 Introduction

Recent reports from China document severe levels of air pollution where poor air quality dramatically lowered visibility, and airports and schools were closed (Armstrong 2013; Wong 2014). The Beijing Olympics of 2008 brought further attention to air quality in China, as efforts to lower pollution in the city were largely successful, although dependent on the continued implementation of abatement policies (Chen et al. 2013). While the health burden from ambient air pollution has declined in recent decades in many parts of the world, in China, ambient particulate matter’s contribution to mortality increased from 1990 (11.58 %) to 2010 (14.86 %) (Global Burden of Disease 2014). The Global Burden of Disease project estimates that over 1.2 million persons died from ambient particulate matter in China in 2010. Ambient particulate matter is the fourth leading cause of death and disability-adjusted life years in China. Studies of the human health effects of air pollution in China have identified associations with cardiopulmonary effects (Wu et al. 2014 (in press)), blood pressure (Zhao et al. 2014), hospital admissions (Zhang et al. 2013), and years of life lost (Guo et al. 2013). Other concerns for air pollution in China include emissions of greenhouse gases (Kurokawa et al. 2013), agricultural losses (Wei et al. 2014), economic consequences (Guo et al. 2010), soil acidification (Zhao et al. 2013), and long-range transport to other regions such as the USA (Lin et al. 2014).

China is not alone with respect to air quality problems. At least one fourth of the world’s population is exposed to unhealthy concentrations of ambient air pollutants (WHO 2006). Air pollution has been linked with a range of adverse human health effects, including aggravated respiratory and cardiovascular disease, reduced lung function, changes in lung tissue, acute respiratory infections, adverse pregnancy outcomes, and premature mortality (Bell et al. 2006, 2007a; US National Library of Medicine 2012). Most studies of air pollution and health were conducted for North America and Europe, with comparatively few studies examining health impacts in Latin America, and particularly Africa and parts of Asia (O’Neill et al. 2008; Wong et al. 2008). Asia is estimated to account for 65 % of all deaths from outdoor air pollution (Cohen et al. 2005). The relatively smaller contribution of scientific evidence from China is pronounced, given the high pollution levels, large population exposed, and increasing contribution of air pollution to overall mortality even using estimates based on studies from other parts of the world (Global Burden of Disease 2010).

When studies of air pollution and health do not exist for a particular area of interest, results from other regions are often extrapolated. A recent example of this method is the Global Burden of Disease project, which estimated mortality and loss of health by age and sex for regions of the world, using exposure-response estimates from one region to estimate health consequences in another, as local studies have not been conducted for all health outcomes and exposures (Lopez et al. 2006). A study of the economic costs of health impacts from particulate air pollution in Lanzhou for 2002 to 2009 used epidemiological associations of pollution and health from other areas of the world, finding economic losses over a billion renminbi (RMB) in most years and the highest loss of 1.66 billion RMB in 2009 (Hou et al. 2011).

This approach of applying estimates of health impacts or cost from another area, while useful, is limited because of differences in populations (e.g., demographics, health care systems, susceptible subpopulations, and activity patterns) and because of variation in air pollution characteristics (O’Neill et al. 2003). For example, earlier work has shown that the chemical structure of particulate matter varies substantially by region (Bell et al. 2007b; Son et al. 2012), and that the health impacts from particles also exhibit spatial patterns (Dominici et al. 2003, 2006).

Air pollution monitoring networks also differ across areas in terms of the pollutants measured, monitor density, and strategy for selecting monitoring sites, which can dramatically affect estimates of exposure (Bravo and Bell 2011). In many regions, air pollution monitors are placed in urban areas where air pollution levels are anticipated to be high, in order to track compliance with regulatory standards. The time when air pollution monitoring began, frequency of measurement, and population characteristics near monitoring stations can impact the design and results of studies on air pollution. Further, local studies can be more meaningful to decision makers who develop and implement efforts to protect public health from air pollution.

Although a growing number of air pollution studies have been conducted in Asia, this region remains understudied given its population and air pollution burden, especially in comparison with the depth of research in North America and Europe. A first step toward understanding the consequences of air pollution is to examine the air quality itself in terms of the overall pollution levels, spatial distribution, and pollution patterns. Recent trends in China have resulted in unprecedented access to air quality data, such as the real-time reports from air quality monitors in many areas. Still, air quality in major Chinese cities is less understood than in many major Western urban areas. Analysis of air pollution index (API) values for particulate matter with an aerodynamic diameter less than or equal to 10 μm (PM10), and monitoring data in Beijing suggest potential discrepancies between measurements and reported API values and indicate the need to investigate original monitoring data (Andrews 2008).

The purpose of this paper is to investigate air pollution in Lanzhou, a major industrial city in China, using the available data for several common ambient air pollutants. We examine the spatial heterogeneity of pollutant concentrations, as previous work demonstrated that some pollutants are more heterogeneous than others and such heterogeneity can vary by season and location (Jo et al. 2000; Zhang et al. 2006). These analyses have implications for the choice of method to estimate exposure to outdoor air pollution for future studies on health. We investigate whether the spatial patterns of air pollution, for various pollutants, support assessment of exposure based on a city-wide average or if strong spatial variation exists such that more localized exposure assessment would be needed. Further, this information could be used to aid the design of air monitoring networks. We also explore whether air pollution patterns differ by season or day of the week, which has been shown in other parts of the world due to differences in weather and sources (Marr and Harley 2002). Such findings would inform the degree of potential confounding for these variables for epidemiological analysis and can provide insights into differences in sources and patterns of emissions. Pollutant levels were compared to air quality standards and guidelines to help demonstrate the overall air quality in this area. Finally, we summarize the existing, limited scientific literature on outdoor air pollution in Lanzhou.

2 Materials and Methods

2.1 Data and Study Area

The city of Lanzhou is in the Gansu Province in northern China with a population of approximately 3.5 million. It is the Gansu Province’s largest city and capital. A number of industries are based in the city including an oil refinery, petrochemical industries, machinery and metallurgical industries, textile mills, food processing centers, cement manufacturing, coal mining, rubber processing, electrical power generation, medical industries, lead and zinc mining and smelting (Zhang et al. 2012), and fertilizer plants (Chen and Sun 2010; Costabile et al. 2010; Zhang et al. 2008). In addition to local pollution, dust transported from other regions affects air quality in this area. Because the city is situated in the Lanzhou Valley surrounded by mountains, pollution levels are often elevated by a stable atmosphere (Fig. 1). The annual rainfall is about 315 mm, and almost all of the rainfall occurs from May to October. Average temperature is 11.7 °C based on a weather monitor in Lanzhou (April 2009 to December 2012). Lanzhou has some of the highest air pollution levels in China and in the world (Wang et al. 2009).

Fig. 1
figure 1

Elevation of Lanzhou City center and placement of monitors

Full size image

Air pollution data included daily observations of PM10, sulfur dioxide (SO2), and nitrogen dioxide (NO2) for four stations in Lanzhou. Particulate matter with aerodynamic diameter less than or equal to 2.5 μm (PM2.5) was not measured in Lanzhou before 2013. These data were obtained from Lanzhou air monitoring stations and were available as 24-h averages. The four monitoring locations are shown in Fig. 1 and in the online supporting material in Fig. S1: (1) Huanghebei in the Chengguan District, a residential area near the Lanzhou Institute of Biological Products, which performs research and manufacturing of biological products; (2) Tieluju in the Chengguan District, residential and commercial areas near a machinery industry; (3) Xizhan in the Qilihe District, residential and commercial areas near the Lanzhou Petro-Chemical Machinery Plant, which produces large machines for oil production; and (4) Xigu in the Xigu District, an industrial area with an oil refinery company, petrochemical plant, textile mill, and other small industries. The monitors were located in the southern part of the city in the metropolitan area with high population density (Fig. 2).

Fig. 2
figure 2

Population of Lanzhou City center and placement of monitors

Full size image

Two monitoring locations, Xigu and Huanghebei, had 45 months of daily observations of PM10, SO2, and NO2 (April 1, 2009–December 31, 2012). The other stations, Tieluju and Xizhan, had daily monitoring data for 24 months (January 1, 2011–December 31, 2012). Table 1 contains the start and end dates and percent of days with observations recorded for each station. In total, 4,114 daily observations were included in the study for SO2, 4,138 observations for PM10, and 4,137 observations for NO2.

Table 1 Summary of 24-h pollutant concentrations and data in the study period
Full size table

2.2 Analysis of Air Pollution

Since relatively little is known about ambient air pollution concentrations in China compared to some other parts of the world, identifying average pollutant levels and possible trends in pollution levels over time is of interest. Temporal trends in pollutant concentrations were evaluated by calculating monthly average concentrations for each pollutant and monitoring station. Pollutant levels were compared to multiple health-based guidelines and regulations: the US National Ambient Air Quality Standards (NAAQS), Chinese ambient air quality standards, and the World Health Organization (WHO) air quality guidelines for PM10, SO2, and NO2 (Chan and Yao 2008; US Environmental Protection Agency 2012; WHO 2006).

The overall purpose of these standards and guidelines are similar, to protect public health from harmful air pollutants; however, their design and implementation differ. The US NAAQS are regulatory standards set at a level intended to protect human health including the protection of sensitive populations, such as children, asthmatics, and older persons. The Chinese regulatory standards are based on different levels of stringency by type of area (e.g., nature reserves, residential areas, and industrial areas). In contrast, the WHO guidelines are not enforceable standards but were published by WHO in order to provide recommendations to help lower the public health burden of air pollution by providing policy makers with targets for air quality. For this analysis, we compare air quality guidelines and standards for particulate matter to levels of PM10, although guidelines and standards exist for other forms of particulate matter as well. The US regulates PM2.5, as well as PM10. The Chinese ambient air quality standards control particles through PM10, PM2.5, and total suspended particles (TSP). The WHO provides guidelines for both PM2.5 and PM10.

In addition to potential time trends and comparison to air quality standards, spatial heterogeneity of pollutants was considered. To evaluate possible spatial heterogeneity, for each monitor pair and pollutant, Pearson correlation coefficients and coefficient of divergence (COD) values were calculated and compared to the distance between monitors. Latitude and longitude coordinates of each monitoring station were used to determine distances between individual monitor pairs. The COD is defined as follows:

$$ CO{D}_{jk}={\left(\frac{1}{p}{\displaystyle {\sum}_{i=1}^p{\displaystyle {\left[\frac{x_{ij}-{x}_{ik}}{x_{ij}+{x}_{ik}}\right]}^2}}\right)}^{1/2} $$
(1)

where x ij and x ik are the concentration of a specific pollutant (24-h average) for day i at sites j and k, and p is the number of observations (i.e., days with data for that pollutant for both monitors) (Pinto et al. 2004; Wongphatarakul et al. 1998). A low COD value indicates small differences between concentrations at sampling sites, while a value close to 1 signifies higher disparity between pollutant concentrations. COD values were calculated for each pollutant and monitor pair (i.e., six values for each pollutant based on the four monitors).

Correlation coefficients and CODs provide different types of information regarding the spatial distribution of pollutants. A correlation coefficient shows whether pollutant levels at different locations co-vary over time but does not provide information on whether the absolute levels are similar. For example, a monitor that consistently has values exactly twice that of another monitor would have different absolute levels but a correlation of 1.0. Conversely, monitors could have similar absolute levels but a low correlation. While correlation coefficients are commonly used to assess uniformity of pollutant concentrations, studies in the US indicate that they poorly predict concentration uniformity because they track temporal similarity of paired sites but do not exhibit a strong relationship with spatial homogeneity of pollutant concentrations (Pinto et al. 2004; Wilson et al. 2006). This study considers both how pollutant levels co-vary (correlations) and the relationship among absolute levels (COD). Because we compare correlations and COD values across time within monitor pairs, both spatial and temporal variations are incorporated. Both types of statistics have been used to estimate relationships among air pollutant levels in previous studies (Bell et al. 2011; Bravo and Bell 2011; Krudysz et al. 2009). All analyses were conducted in R statistical software version 2.11.1.

2.3 Review of Studies of Air Pollution and Health in Lanzhou

To better understand the state of scientific literature on human health and air pollution in Lanzhou, we performed a systematic review. We identified studies of outdoor air pollution and human health in Lanzhou by searching two scientific literature databases: (1) PubMed of the US National Library of Medicine, National Institutes of Health (US National Library of Medicine 2012), and (2) Scopus, an Elsevier database of research literature (SciVerse 2012). The final search was performed February 17, 2014. Search terms were for the title or abstract in the PubMed, and the title, abstract, or key words in Scopus. Searches were performed for the following terms: (1) Lanzhou; AND (2) “air pollution” or “air pollutant” or “air pollutants” or “particulate matter” or PM10 or SO2 or NO2; AND (3) health or respiratory or cardiovascular or mortality or hospital*, where * represents a wildcard symbol (e.g., hospital* may be hospitals or hospitalizations).

Non-English articles were included. Due to the scarcity of studies on air pollution and health in Lanzhou, we included conference proceedings.

3 Results

3.1 Characteristics of Air Pollution in Lanzhou

Table 1 summarizes concentrations for each pollutant and monitor location in terms of minimum, maximum, and average 24-h values. Boxplots of each pollutant and station are provided in Fig. 3. Median levels of PM10 recorded at each station were similar (Fig. 3a), and two stations (Xigu and Huanghebei) had one day with PM10 levels greater than 1,500 μg/m3. One of the stations (Huanghebei) had outliers of high concentrations on 80 days (6 % of days in the study period). Here, outliers were defined as any observation greater than 1.5 times the interquartile range added to the 75th percentile value or any observation less than 1.5 times the interquartile range subtracted from the 25th percentile. For NO2, median levels recorded at each station were similar, and one of the four stations (Huanghebei) had outliers of high concentrations on nearly 5 % of days in the study period (Fig. 3b). In the case of SO2, median values were somewhat similar across stations, although Xigu had slightly higher median SO2 concentration than the other three stations (Fig. 3c). Two of the four stations had outliers of high concentrations for SO2 on at least 7 % of days in the study period.

Fig. 3
figure 3

Pollutant levels over the study period, by station. a PM10; b NO2; c SO2. Note: The upper and lower horizontal lines of the box represent the 75th and 25th percentiles, respectively. The horizontal line within a box represents the median. The open circles (outliers) represent observations less than the 25th percentile − 1.5 × interquartile range (IQR), or larger than the 75th percentile + 1.5 × IQR. The vertical line above each box represents the 75th percentile to the lowest observation that exceeds the 75th percentile, but is not an outlier. The vertical line below each box represents the 25th percentile to the highest observation that is lower than the 25th percentile, but is not an outlier

Full size image

Monthly average pollutant levels were calculated to evaluate temporal trends within the study period. Figures S2, S3, and S4 in the supporting material provide results for each station for PM10, NO2, and SO2, respectively. Figure 4 shows temporal trends averaged across all monitors for each pollutant. Average levels for all pollutants fluctuated from month to month, and temporal trends were not easily identifiable. Findings suggest some seasonality in pollution levels, particularly for SO2, which tends to have higher concentrations during colder months (e.g., October–March) relative to warmer months. However, analysis is limited by the availability of air pollutant data, for which we have three years of data for the months January to March and four years of data for the months April to December. While our findings indicate seasonal trends, data are available for only three to four sets of data per season; thus, a longer time frame of data is needed to evaluate seasonality fully. In general, the highest PM10 levels occurred in November–December, with higher levels also occurring in October–May, and the lowest levels in June–August. Average concentrations of SO2 are also highest between November and February. Monthly or seasonal differences in NO2 are less pronounced, though concentrations of NO2 during warmer months (e.g., July–September) tend to be lower relative to cooler months (Fig. 4).

Fig. 4
figure 4

Average pollutant concentrations across study area and period

Full size image

Pollutant levels can vary by day of the week in some locations because of differences in emissions from industry and traffic. For this city, analysis by day of the week did not demonstrate strong differences among concentrations across days (Fig. S5). Concentrations were similar across different days, although highest on Fridays compared to other days for all pollutants, lowest on Mondays for SO2 or NO2, and lowest on Tuesdays for PM10.

The spatial heterogeneity of pollutant levels was examined by calculating correlation coefficients and CODs for daily (24-h) average concentrations observed at monitor pairs for each pollutant. Correlation coefficients between monitoring locations for PM10, SO2, and NO2 plotted against the distance between monitor locations are shown in Fig. 5, and CODs for these three pollutants are shown in Fig. S6 in the supporting material. In general, correlation coefficients declined with increasing distance between monitors, while CODs increased with increasing distance between monitors. However, these trends were not statistically significant (p > 0.05). Using a linear model for the data in Fig. 5, the average correlation for a distance of 10 km is 0.85, 0.71, and 0.57 for PM10, NO2, and SO2, respectively, indicating less spatial heterogeneity for PM10 than the other pollutants. The relatively high correlations among the pollutants within the city are evidence against the hypothesis of spatial heterogeneity, especially for PM10 and NO2.

Fig. 5
figure 5

Correlation between daily average observed pollution levels versus distance between monitor locations

Full size image

3.2 Comparison to Air Quality Regulations and Guidelines

Table 2 contains the values of PM10, NO2, and SO2 averaged across all four stations over the duration of the study period, as well as the US NAAQS, Chinese ambient air quality standards, and WHO guidelines. Figure 6 depicts the average levels for each station in comparison to US and Chinese standards and WHO guidelines. The Chinese regulations have three grades of ambient air quality standards. Grade I (first level standard) applies to nature reserves, scenic spots, and other areas of special protection; Grade II (second level standard) applies to residential areas; commercial, transportation, and mixed-residential areas; cultural areas; and general industrial areas specified in urban planning, as well as rural areas; and Grade III (third level standard) apply to industrial zones (Ministry of Environmental Protection of the People’s Republic of China 1996). Grade I standards are the most stringent and Grade III the least stringent. Although we present the standards for all grades, the study region is Grade II. Chinese standards are set for both daily (24-h) and annual periods, whereas the US NAAQS and WHO guidelines are for a range of time frames including 1-h, 24-h, and annual values. The primary US NAAQS standards presented here are intended to protect public health, with an adequate margin of safety to be protective of the health of sensitive subpopulations, such as children, the elderly, and people with asthma (US Environmental Protection Agency 2012). Guidelines provided by the WHO are developed with the goal of protecting human health from adverse impacts of air pollution and are intended to inform policy makers from different countries (WHO 2011).

Table 2 Comparison of daily averages to ambient air quality standards and guidelines
Full size table
Fig. 6
figure 6

Pollutant average pollutant concentrations versus standards and guidelines, by station. a PM10; b NO2; c SO2. Note: Chinese Grade III PM10 standard, 250 μg/m3 (not shown). Chinese Grade III NO2 standard, 120 μg/m3 (not shown); US NAAQS and WHO NO2 guidelines (not shown) are not for 24-h averages. Chinese Grade II SO2 standard, 150 μg/m3; Chinese Grade III SO2 standard, 250 μg/m3 (not shown); US SO2 NAAQS (not shown) are for 1-h concentrations, not for 24-h averages

Full size image

In the case of PM10, the average concentration over the study period was higher than the Chinese Grade I standard and WHO guidelines for all stations; for three of four stations, the average concentration over the study period was lower than the US standard and the Chinese Grade II standard (Fig. 6a). Average NO2 concentrations across the study period did not exceed Chinese ambient air quality standards, while US standards and WHO guidelines are not applicable with this data set because they both refer to 1-h concentrations instead of 24-h averages (Fig. 6b). For SO2, the average concentration over the study period was lower than the Chinese Grade I standard at three of the four stations (Fig. 6c). The US NAAQS for SO2 applies to a 1-h concentration and thus is not applicable to the 24-h averages available in this data set. Average SO2 concentrations at all four stations were higher than the WHO guidelines for SO2 of 20 μg/m3.

3.3 Summary of Studies of Air Pollution and Health in Lanzhou

The city of Lanzhou, with about three million people, has been the subject of very few studies of how air pollution affects health compared to other cities of similar size. For example, our literature search identified 13 such studies. In contrast, Chicago, IL, in the US, also with approximately three million residents, has hundreds of such studies. Table 3 summarizes the identified studies, including the pollutants considered, time frame, health outcome, methods, and key results.

Table 3 Summary of air pollution and health studies for Lanzhou, China
Full size table

Although we searched for non-English studies, all identified studies were in English, with the exception of one study (Wang et al. 2012) in Chinese with English abstract. Of the 13 studies, four used questionnaires to assess health status, mostly for respiratory symptoms. Nine used data on hospital admissions, two for cardiovascular and cerebrovascular disease and seven for respiratory disease. The findings generally indicate that higher levels of air pollution were associated with higher risk of health outcome. However, a wide range of health outcomes have not been studied in Lanzhou, including mortality and adverse birth outcomes. The time frame of the studies was generally a few years, with the longest study at 5 years.

4 Discussion

We found high levels of air pollution in the city of Lanzhou, with an average PM10 concentration (141.5 μg/m3) across the four monitors and duration of the study period more than six times greater than the WHO annual average guideline of 20 μg/m3. NO2 and SO2 concentrations averaged across all monitors and over the duration of the study period also exceeded WHO guidelines. Previous work in Lanzhou has noted that the area’s topography (Fig. 1) of increased susceptibility to inversion layers, creating a stable atmosphere that allows pollution to accumulate near ground level. In addition, local coal-fueled industries, household use of coal for cooking, industry including smelters, rapidly increasing vehicle ownership and use, freight transfer stations, and transport of dust from the Hexi region of Gansu Province contribute to the air quality issues in Lanzhou (Chen and Sun 2010; Feng and Wang 2012; Niu et al. 2011; Qi et al. 2001; Ta et al. 2004; Xie et al. 2010). In fact, air pollution in Lanzhou has been so severe that civil servants were asked to walk to work, rather than use vehicles, on high pollution days (Lie 2007). A previous study found higher levels of TSP in Lanzhou compared to three other Chinese cities (Wuhan, Chongqing, Guangzhou), but the lowest levels of NOx, and second highest levels of SO2 (Qian et al. 2001). That study also measured several size fractions of particulate matter in schoolyards in the four cities, finding Lanzhou to have the highest coarse particulate matter and TSP. Results from our study suggest that concentrations of pollutants in Lanzhou remain a cause for concern.

In general, the measured pollutants did not exhibit spatial heterogeneity, with less heterogeneity in PM10 than the other pollutants. This result suggests that issues of spatial misalignment, where health and pollution data are on different spatial scales (e.g., city-level health data vs point-level exposure data) may be less problematic for epidemiological research in Lanzhou than in other areas with more spatially heterogeneous pollution patterns, and underscore the need to investigate spatial heterogeneity of the locations and pollutants of interest (Peng and Bell 2010).

Our results indicate a lack of strong day-of-the-week patterns and some seasonal patterns for pollution levels, with lowest PM10 in summer and higher SO2 in November through February. This indicates that day of the week may not be an important confounder in studies of short-term exposure to air pollution in this area. However, sensitivity analysis (e.g., including and then omitting day of the week as a variable in separate models) could further investigate this issue. Further, our finding of generally higher pollutant concentrations in winter compared to summer months is consistent with the sparse previous studies of ambient air pollution in Lanzhou (Chu et al. 2008a, b; Costabile et al. 2010; Ma 2013; Qian et al. 2001; She et al. 2011; Ta et al. 2004; Wang et al. 2009). Winter in this area is characterized by a stable atmosphere, particularly during the nighttime (Chu et al. 2008a).

Data availability remains an issue in this region of the world. Additional monitor locations, hourly concentration data, information on other pollutants (e.g., ozone, size distributions of particulate matter), and a longer historical record would allow for a more detailed analysis of long-term trends in pollution levels over multiple years, examination of diurnal trends in pollution levels, and a more powerful analysis of spatial heterogeneity or spatial patterns in pollutant distribution. The current monitoring network in Lanzhou includes measurements of PM2.5 and carbon monoxide (CO). Such information and research would contribute greatly to a more complete understanding and characterization of air pollution in Lanzhou.

The high concentrations of air pollutants in Lanzhou imply substantial health consequences to Lanzhou residents. Our findings of seasonal variability in air pollution concentrations have implications for future studies that examine how health is affected by air pollution in the region. Studies identified through our systematic review indicate that air pollution is associated with human health in Lanzhou, although a small number of health responses were investigated, typically for short time frames. These few health studies combined with the high levels of pollution observed in this study and earlier work suggest a potentially substantial health burden for residents of this city and suggest the need for future epidemiological studies of health risk.

Our findings have implications for other major industrial cities in China and other growing urban centers, which may have either no or recent air quality monitoring programs.

According to Ministry of Environmental Protection of the People’s Republic of China, Lanzhou is similar to most Chinese cities suffering from heavy air pollution (Ministry of Environmental Protection of the People’s Republic of China 2014). These cities share major air pollution sources, including industrial emissions, coal combustion, fugitive dusts, and traffic emissions. Lanzhou has consistently ranked as one of the ten worst cities for air pollution since the Chinese government began releasing the air pollution index for major cities in 2003.

These results demonstrate levels of air pollutants that are likely to be associated with a strong public health burden, although little scientific research has been conducted to date to link pollution to health specifically in this city. Expanding industrial centers in China can be accompanied by air pollution levels that relate to the type, size, and ownership of industry (Wang and Deng 2013). Decisions made in the next few decades regarding the rapid urbanization of Chinese cities, with a doubling of the population from 2000 to 2030 (Cao et al. 2012; Chiu 2012) will have a profound effect on air quality and public health, especially in the context of a changing climate (Gu and Han 2010). Achieving healthy levels of air quality may be more difficult under a changing climate, for example, as higher temperatures promote tropospheric ozone formation. In fact, rapid urbanization of Lanzhou has increased the urban heat island effect in recent decades (Li et al. 2010; Pan and Yang 2013; Qin et al. 2011).

Our data includes information through 2012; however, since the end of 2011, the Gansu Provincial government has implemented a series of environmental protection initiatives, and the Ministry of Environmental Protection has stated that the air quality in Lanzhou has improved substantially in recent years (Ministry of Environmental Protection of the People’s Republic of China 2014). Therefore, the Ministry sets Lanzhou as an example for other major cities in the Beijing-Tianjin-Hebei region and Yangtze River delta region. In order to inform air quality control in other cities, understanding is needed of the spatial and temporal patterns of air pollution in Lanzhou before and during the implementation of environmental initiatives for air quality. Future work on air pollutant levels after these initiatives, combined with this work of air quality conditions up through 2012, can support investigation of the accountability of these environmental policies and initiatives.

5 Conclusions

We investigated air quality in Lanzhou and identified high levels of air pollution, with PM10 concentrations exceeding health-based regulations and guidelines. Our systematic review of studies on outdoor air pollution and health identified 13 studies, which is an order of magnitude lower than the number of similar studies for many other cities. However, the existing studies generally found that air pollution in Lanzhou is associated with increased risk of health outcomes such as hospital admissions for respiratory disease. The findings on the spatial and temporal patterns of air pollutants can inform future epidemiological studies, as well as aid the design of networks for air pollution monitors. The high levels of air pollution combined with the review of the scientific literature indicate that ambient air pollution presents a substantial public health concern in Lanzhou.