Spatiotemporal analysis of the link between VCD_trop NO₂ and population size for provinces in Sumatra Island during 2012-2020

ABSTRACT

Nitrogen dioxide is one of the primary air contaminants that can cause severe ecological and human health effects. This study used the Ozone Monitoring Instrument (OMI) sensor aboard AURA satellite to analyze the spatial and temporal variations of VCD_trop (Tropospheric vertical Columnar density) NO₂ concentration in Sumatra, Indonesia, during 2012–2020. Based on the seasonal characteristics, the VCD_trop NO₂ amounts in the east part of Sumatra were higher in the dry season and slightly lower in the wet season. The increasing trend was recorded in some provinces; North Sumatra, Riau, South Sumatra, and Lampung with the accumulation rate around 8–9 × 10¹⁵ molecules/cm² from 2012 to 2014. Higher VCD_trop NO₂ concentrations were associated with an increase in population size. As a whole, this finding can be useful to arrange air pollution management associated with population growth in urban areas.

KEYWORDS:

• Meteorological factor
• population exposure
• environment
• spatial variation
• VCD_trop NO₂

1. Introduction

Total population in many cities around the world is estimated to increase rapidly every year [1,2]. In order to fulfill the increased population, cities need to generate energies and materials which eventually emit air pollution [3–5]. Annual deaths have also been reported in cities due to the effect of poor air quality [6–8]. The difference air pollution concentration between urban and rural areas or in developed and developing nations could be due to the difference of energy usage and energy production. In this decade, Sumatra region of Indonesia has experienced intensive urbanization, migration, and industrialization processes. These activities have exerted a negative impact on the air quality within that region. NO₂ is one of the key air pollutants in Sumatra and has a substantial effect on human health; thus, it is necessary to study the distribution of VCD_trop NO₂ concentrations and population exposures at the regional scale. There is an absence of studies on the spatial variation of VCD_trop NO₂ pollution related to population exposure in Sumatra over a long-term period.

Nitrogen dioxide is a prominent air quality indicator of burning processes. It leads to the formation of particulate matter and ozone, which are hazardous to human health and surrounding [9,10]. NO₂ is known as a short-lived air pollutant thus it can be used to identify the exposure of air pollutions and their emissions associated with the urban population [11]. The quantitative evaluation of the assoication between NO₂ and population needs a consistent dataset. Therefore, the advanced of recent technology such as the use of satellite data offer high consistency of NO₂ data. Satellite remote sensing is appearing as a promising data source to link the population exposure and air polllution [12]. Mesas-Carrascosa et al. [13] carried out a study on the spatial variation of VCD_trop NO₂ in Spain during the COVID-19 period. The study reported that urbanization aggravated VCD_trop NO₂ pollution in the city. Huang et al. [14] investigated VCD_trop NO₂ exposure in China and found that it was significantly greater in congested areas as compared with other areas. Currently, the use of large-scale population time series data can assist in identifying the effect of population on VCD_trop NO₂ pollution over a long period of time. Most of the above-previous studies associated with VCD_trop NO₂ pollution and population were conducted in a small area over a short period of time.

Various urban factors like energy consumption, infrastructure, innovation, and occupation are associated with total population [12]. In general, urban air polllution emerges due to the resource consumption (such as burning fossil fuel) which relates to population size and various urban factors. Thus, the objective of our study was to conduct a long-range period study on VCD_trop NO₂ pollution related to population and meteorological factors in Sumatra. We applied satellite-based data from VCD_trop NO₂ concentrations that were retrieved from the Ozone Monitoring Instrument (OMI) over Sumatra region from 2012 to 2020. This finding can add new knowledge about long-range periods of spatial variation in VCD_trop NO₂ pollution analysis in the Sumatra region.

2. Materials and methods

2.1. Study area

For this study, the Sumatra Plain was chosen as the study site, with latitudes of 5°0’0’’N to 5°0’0’’S and longitudes of 95°0’0’’E to 105°0’0’’E (Figure 1). The topography of Sumatra western part is undulated with the maximum elevation of about 3,427 m above sea level. While, Sumatra eastern part has a low elevation and a small topography slope with an average elevation around 93 m. The geomorphological characteristic of this area is higher on the western side and lower on the eastern side. Sumatra has a diverse range of land surfaces, including mountains, peatland, basins, hills, plateaus, and plains. In addition, Sumatra has a tropical rainforest climate type with two seasons: dry and wet. Sumatra’s industrialization and economic growth have accelerated in recent years, resulting in a variety of environmental issues. Figure 1 shows the distribution of the main provinces in Sumatra.

Figure 1. Location of the study area. Gray color indicates topographic landscape over the Sumatra Plain.

2.2. VCD_trop NO₂ acquisition

The ozone monitoring instrument (OMI) is a prominent sensor on board the Aura satellite, operating since 2004. This sensor has various benefits for environmental and atmospheric studies. It is measured daily, meaning solar irradiance can be monitored once every 24 h it contains more than 10 years of NO₂ data products such as total vertical column NO₂, tropospheric NO₂, and stratospheric NO₂ columns [15]. This study used the OMI Level-3 daily global gridded VCD_trop NO₂ data product (OMNO₂d) during 2012–2020, with a spatial resolution (0.25 × 0.25), which allows for the observation of finer clusters of atmospheric variables. The data was obtained from the NASA open portal using the Giovanni website (https://giovanni.gsfc.nasa.gov/giovanni/). The OMI passing time for the study period was set at local time 13.00–14.00 h on daily basis. In the OMI Level-3 product, only pixels which were not influenced by row anomaly and also had a minimum cloud contamination or radiative cloud fraction value less than 0.3 for NO₂, were used. Although the value and performance acquired from OMI was lower as compared to TROPOMI. However, the OMI sensor still showed a good agreement with the ground base data. It was due to the lower spatial resolution of OMI as compared to TROPOMI. Numerous studies have revealed the performance of the VCD_trop NO₂ satellite-based dataset, and their accuracy can now be admitted and widely applied in VCD_trop NO₂ studies. For instance, a study by Celarier et al. [16] used ground-based monitoring to validate the VCD_trop NO₂ data and other key parameters. This previous study found the correlation value between both instruments to be around 0.80–0.90. The statistical uncertainty in the VCD_trop NO₂ columns in the study area was conducted using the AMF calculation. The base component was ranging from 0.5 to 1.0 × 10¹⁵ molecules/cm². The relative error varied between 10% and 30% and might have spatial undersampling errors due to coarse resolution of OMI pixels. This cirrcumstance has been revealed by Boersma et al. [17]. The VCD_trop NO₂ data based OMI in areas with extreme aerosol massess like in the Southern Sumatra side might be exposed to above-mean aerosol error contribution.

2.3. Meteorological and population factors

The meteorological data were obtained from the meteorological, climatological, and geophysical agencies of Indonesia. The data collected, such as rainfall, wind speed, and air temperature, covered an 8-year period. Additionally, annual rainfall data were also collected from the CRU TS dataset website (https://crudata.uea.ac.uk/cru/data/hrg/) to obtain spatial information about precipitation variation in the study area. While, total population during an 8-year period in the study area were taken from the central agency statistics of Indonesia (https://www.bps.go.id/)

2.4. Statistical analysis

Pearson correlation analysis was used to examine the relationship between meteorological factors and population data and VCD_trop NO₂ concentration. The IBM SPSS Statistics 21 software was used to evaluate the correlation analysis between these variables.

3. Results and discussion

3.1. Spatial and temporal variation of VCD_trop NO₂ and population size in Sumatra region

In this study, the satellite-based NO₂ data were validated against the NO₂ data from the ground measurement. The validation results showed a good accuracy between the satellite data as shown in Table 1. Figure 2 presents an 8-year mean variation of VCD_trop NO₂ in the entire Sumatra Plain from 2012 to 2020. Clearly, the VCD_trop NO₂ variation was not uniform (Figure 3). Unlike on land, the VCD_trop NO₂ value over adjacent sea was lower than the VCD_trop NO₂ value over the Sumatra land. There was no significant annual variation over the sea. The western coast of China, that was near to the Mallaca Strait, was obtained to have higher NO₂ values than the open sea, could be attributed from vessel emission and also the primary emission sources from the adjacent lands. Kang et al. [18] have reported NO₂ sources were greatly affected by anthropogenic activities, indicating provinces with a large population and intensive industrial activities might contribute to high levels of NO₂. According to Biswal et al. [19], the VCD_trop NO₂ was raised due to the positive fire anomaly, this finding was shown in our study where most of fire cases over the study area (in Riau and South Sumatra Provinces).

Figure 2. (Continued).

Figure 3. Annual mean distributions in VCD_trop NO₂ over Sumatra Island and off-shore seas adjacent to island from 2012 to 2020.

Table 1. The comparison between tropospheric NO₂ column and the ground measurement in eight provinces.

Province	r	slope (mol. cm^−2/ppb)	intercept (mol. com^−2)
Riau	0.67	1.33 ± 0.10	−2.40 ± 0.90
South Sumatra	0.69	1.35 ± 0.12	−2.42 ± 0.92
Aceh	0.68	1.32 ± 0.11	−2.41 ± 0.92
Bengkulu	0.61	0.66 ± 0.14	−5.08 ± 1.89
Lampung	0.63	0.68 ± 0.12	−5.02 ± 1.86
North Sumatra	0.65	1.28 ± 0.15	−3.20 ± 0.89
Jambi	0.66	1.30 ± 0.12	−2.38 ± 1.45
West Sumatra	0.68	1.29 ± 0.14	−2.42 ± 0.89

Economic growth and migration rates were higher in the eastern and southern Sumatra regions than in its western region, resulting in higher VCD_trop NO₂ concentrations on the eastern side. This pattern can be seen in the comparison of Figures 2 and 4. The western side tended to have a lower population, while the eastern side had a higher population, as well as a different climate pattern. The southern area was the most congested area, which was considered based on smaller provinces but had a high population (Figure 4). The northern corner and middle areas showed a low population. The southern part of Sumatra Plain had the highest VCD_trop NO₂ across the study period. Borck and Schrauth [20] have revealed the VCD_trop NO₂ concentration increases with population density with an elasticity value of 0.25. While, particulate matter and O₃ obtained the value of 0.08 and 0.14, respectively. This was concluded that higher population density deteriorate air quality. A study in Germany found the highest contribution of NO₂ originated from road traffic emission which attributed to population exposure through car and bus environment [21]. Therfore, it was even more prominent to calculate for transport system when people were moving inside and outside the city. This condition was shown in the southernmost area of Sumatra where there was a port which connected Sumatra and Java Islands, this port was crowded with car, truck, and buses which queue to enter the boat. It has made a high concentration of VCD_trop NO₂ situated on the south side of Sumatra. Furthermore, capital cities in the east tended to record low VCD_trop NO₂ levels when compared to the southern areas.

Figure 4. Spatial variation of total populaton in Sumatra during (a) 2012–2015 and (b) 2016–2020.

In order to better understand the alterations and variations of the 8-year VCD_trop NO₂ levels in Sumatra comprehensively, we chose eight primary provinces in the Sumatra region, as depicted in Figure 1. The mean of VCD_trop NO₂ in all provinces was fluctuative across the study period (Table 2). Furthermore, an increasing pattern was mostly found in almost all cities during 2012–2016, specifically in Aceh, Bengkulu, and North Sumatra Provinces, where the percentage increase values were above 10%. South Sumatra, on the other hand, decreased slightly during that time period, despite the fact that its total base concentration of VCD_trop NO₂ remained high. During the period 2018–2020, all cities experienced significant change, with Lampung Province experiencing the greatest reduction (17%). This result was consistent with a study by Vadrevu et al. [22]. The decrease in VCD_trop NO₂ was found in 2020, for instance in New Delhi, India, the VCD_trop NO₂ reduced around 61.7%.

Table 2. Changes in mean VCD_trop NO₂ in each primary province during 2012–2020, (unit: 10¹⁵ molecule/cm^2.).

Province	2012	2014	2016	2018	2020	Changes% (2012–2018)	Changes% (2018–2020)
Riau	6.86	6.87	6.14	6.96	5.99	+1.5	−13.9
South Sumatra	6.44	5.97	5.59	5.82	5.75	−9.6	−1.7
Aceh	4.19	4.39	4.43	4.66	4.41	+11.2	−5.4
Bengkulu	4.17	4.22	4.70	4.63	4.44	+11.0	−4.1
Lampung	8.04	7.31	6.99	8.34	6.92	+3.7	−17.0
North Sumatra	4.87	5.26	5.24	5.39	4.78	+10.7	−11.3
Jambi	5.29	4.72	4.58	4.82	4.75	−8.9	−1.5
West Sumatra	4.21	3.96	4.25	4.53	4.25	+7.6	−6.2

Based on Figure 2, the variation of VCD_trop NO₂ in Sumatra sustained a notable reduction in most areas from 2018 to 2020. The provinces that maintained a decrease in VCD_trop NO₂ levels from 2018 to 2020 were particularly similar to those that maintained an increase in VCD_trop NO₂ levels from 2012 to 2018. In addition, the concentration in major cities like South Sumatra and Jambi was still dropping. But, the reduction in South Sumatra was the lowest as compared with other provinces during 2018–2020, with a value of 1.7%. As a whole, the VCD_trop NO₂ levels in the main provinces have reduced significantly, especially in Lampung Province in the southern Sumatra region. This reduction could be due to the tight policy implemented during the COVID-19 lockdown in 2020, which also affected other provinces. In addition, there was also a notable reduction in Riau Province in the eastern part of Sumatra. while other provinces only showed a slight reduction below 7%.

Figure 2. The VCD_trop NO₂concentration over Sumatra during 2012–2020 (unit: 10¹⁵ molecule/cm²).

Most areas were relatively stable, with only a slight fluctuation across the study period. Yearly distributions and change rates in the eight main provinces are shown in Table 2. Lampung (8.0 × 10¹⁵ molecules/cm²), Riau (6.8 × 10¹⁵ molecules/cm²), South Sumatra (6.4 × 10¹⁵ molecules/cm²), Jambi (5.2 × 10¹⁵ molecules/cm²), North Sumatra (4.8 × 10¹⁵ molecules/cm²), West Sumatra (4.3 × 10¹⁵ molecules/cm²), Aceh (4.2 × 10¹⁵ molecules/cm²), and Bengkulu (4.1 × 10¹⁵ molecules/cm²), respectively, had the highest average VCD_trop NO₂. Only Aceh and North Sumatra Provinces showed a continuous increase from 2012 to 2018. Prior to 2020, all provinces declined dramatically due to the impact of the COVID-19 control measure. In particular, Lampung Province had the highest VCD_trop NO₂ across the study period, although it seemed to fluctuate throughout the years, but this province had a much higher level than other areas. The high concentration in that province is linked to high population density and intense migration via a harbor connecting Sumatra and Java. The congested traffic when vehicles queued to enter ferryboats produced a large amount of VCD_trop NO₂ emissions from this source. A study by Merico et al. [23] noted the impact of NO2 pollution was higher in harbour area than urban areas and it became a main pollutant concern in that area.

In addition, the yearly change in the Aceh, West Sumatra, and Bengkulu provinces showed a similar trend across the study period that was stable at less than 4.0 × 10¹⁵ molecules/cm². In contrast, most provinces increased significantly from 2012 to 2018, then decreased significantly from 1% to 17% in 2020. This decline could be due to the implementation of the COVID-19 lockdown policy during 2020. The reduction of VCD_trop NO₂ was also observed in some cities around the world such as cities in China [24], India [25], London, Milan, and Paris [26]. This policy has led to the closure of industrial activities and the restriction of public transportation and social activities [27]. As a whole, in normal condition, the influence of VCD_trop NO₂ in each province was affected by the population size, road traffic, and biomass burning activities. These three factors governed greatly the VCD_trop NO₂ in Sumatra region.

3.2. Seasonal variations of VCD_trop NO₂

One of the largest proportions of global tropical peatland is located in Indonesia, specifically in Sumatra and Kalimantan. Peatland fires not only ignited surface vegetation, but also the peat in the subsurface, that then emitted large number of carbon dioxide to the air. The forest fires in Sumatra were highly associated with the distribution of meteorological conditions affected by El Niño Southern Oscillation (ENSO) event. Several studies have found the El Niño event related to remarkable high temperature and low rainfall in this region. The high anomaly was very high in the eastern and southern sides of Sumatra that was in line to areas of forest fires in 2015–2018. The VCD_trop NO₂ concentrations over the study area were 5.7 and 5.8 × 10¹⁵ molecules/cm² in 2015 and 2018, respectively. If we compared to other countries like Middle East, a study by Barkley et al. [28] revealed there was a linear increase of NO₂ trend by up to 12% per year. In contrast, in China, a significant reduction of NO₂ was observed since 2011 due to effective air quality measures [29].

Seasonal variation of VCD_trop NO₂ over Sumatra region over 8 years is depicted in Figure 5. In the Sumatra region, the dry season started from April to September and wet season started from October to March. Because the effect of meteorology varied highly in different regions, the seasonal distribution properties was detected in all over the region, with the eastern and southern parts of Sumatra having the highest levels in the dry season. This was consistent with another study by Swartz et al. [30] who assumed higher VCD_trop NO₂ levels were induced by the impact of biomass burning during the wet season. Because Sumatra is mostly covered by peatland areas that are prone to land fires [31], this has contributed to higher VCD_trop NO₂ emissions for several years. Contrarily, western Sumatra had the lowest temperatures during the two seasons. This might be due to the distinct formation of NO_x in certain areas and also the VCD_trop NO₂ accumulation because its major sink such as the photolysis rate was low. This result was in line to glyoxal recorded over China from OMI sensor [32]. The increment of VCD_trop NO₂ amount could be associated with the intense fossil fuel combustion in the winter period.

Figure 5. Seasonal VCD_trop NO₂ concentration over Sumatra during 2012–2020, (a) dry season and (b) wet season (unit: 10¹⁵ molecule/cm²).

We believed that the primary origins of NO_x in western Sumatra were natural emissions, closely related to temperatures, soil properties, and rainfall. Meanwhile, areas on the east side of Sumatra were mostly associated with anthropogenic activities like industrial emissions. Specifically, microbiological activity from peatland areas was one of the main contributors to high NO_x emissions during the wet season. Swartz et al. [30] revealed the NO_x time span is generally longer in the wet season, resulting in slight air pollution during that period. Furthermore, the highest VCD_trop NO₂ concentration was recorded during the dry season in the southern corner of Sumatra, where there was a dense harbor. Many people traveled to other provinces during the holiday season, using this location as a border crossing point.

3.3. Effect of meteorological factors

Table 3 shows the correlation coefficients between meteorological factors and the VCD_trop NO₂ concentration at the study site. In general, several factors, such as total NO_x emission, NO_x life time, and NO₂ transport from different areas, needed to be studied in order to understand VCD_trop NO₂ in a specific area (Wang et al. 2019a) [15]. Based on our findings, we discovered that VCD_trop NO₂ had a significant seasonal trait, which meant that it was higher during the dry season and lower during the wet season. Therefore, according to the VCD_trop NO₂ emission aspect, a great increase in fossil fuel burning is the main driver of high NO_x emissions during wet periods. Particularly during that period, the time span of NO_x was greater as compared with other periods due to the poor solar radiation and air temperature. These factors would contribute to the deferment of atmospheric reaction processes. In contrast, increased solar radiation occurred during the dry period, making the chemical reaction more active in reducing VCD_trop NO₂ pollutants. The solar radiation and humidity had a prominent role in shaping the VCD_trop NO₂ in atmosphere according to Voiculescu et al. [33]. Falocchi et al. [34] explained the VCD_trop NO₂ concentration normally reduced when the solar radiation raised up to 250 W.m⁻².

Table 3. Pearson correlation coefficients between meteorological variabels and VCD_trop NO₂ in the study site.

Variable	VCDtrop NO2
Temperature	−0.880**
Wind speed	0.742*
Rainfall	−0.617*
Population	0.686*

**Correlation is significant at the 0.01 level (2-tailed).

*Correlation is significant at the 0.05 level (2-tailed).

Furthermore, VCD_trop NO₂ transport was linked to meteorological factors. During the wet season, the mean wind speed was greater, which contributed to the higher VCD_trop NO₂ emissions. Furthermore, the atmospheric layer tended to be low during the wet season, which was simply due to inversion layers, resulting in more consistent VCD_trop NO₂ levels in the lower troposphere zone. In order to examine the VCD_trop NO₂ variation in distinct areas, this study analyzed the meteorological factors in these eight provinces. Our result obviously found that VCD_trop NO₂ variation was related to the seasonal changes of meteorological factors. Rainfall and air temperature effects, which created a cooler condition during the wet season, were able to attenuate the oxidation and wet deposition rates. Additionally, compared with the eastern provinces (Aceh, North Sumatra, Riau, Jambi, South Sumatra, and Lampung), the western provinces (Bengkulu and West Sumatra) showed different meteorological circumstances (Figure 6). When compared to other regions, provinces in western Sumatra with a lower latitude zone and close to the Indian Ocean had a higher mean temperature. Furthermore, due to the monsoon effect, there was a greater amount of rainfall in the western areas (Figure 7), which was helpful for VCD_trop NO₂ removal. This finding was in line with the seasonal variation in VCD_trop NO₂ levels in the provinces with high VCD_trop NO₂ levels (Figure 5).

Figure 6. Yearly mean variations of tropospheric NO₂ column, air temperature, and wind speed in eight provinces during 2012–2020.

Figure 7. Spatial distribution of total average precipitation over Sumatra during (a) 2012–2015 and (b) 2016–2020.

Our study had some limitations with the lack of ground measurement NO₂ data presented in this study. The use of large coverage dataset included information on actual NO₂ pollution could improve the retrieval of VCD_trop NO₂ from satellite. Additional work also could analyze other air pollutants such as SO₂, CO, PM and O₃. Thus, planned future study will include the use of this technique to greater spatial scale for instance analyzing the association for distinct provinces in Indonesia.

4. Conclusion

Therefore, to improve air quality and reduce air pollution, a new measure needs to be applied in Sumatra. Understanding the VCD_trop NO₂ concentration is a key to controlling air pollution because NO₂ is a benchmark for the air quality status in a certain area. This policy can also be useful to arrange effective action for air pollution management in the future. That is a reason why the study about the spatial and temporal distribution of VCD_trop NO₂ from 2011 to 2020 over Sumatra is prominent. The VCD_trop NO₂ concentration in Sumatra is substantial, and the variation of VCD_trop NO₂ is irregular. The concentration was higher in eastern Sumatra and lower in the western part. The VCD_trop NO₂ pollution revealed long-range changeability that varied provincially. Firstly, the VCD_trop NO₂ concentrations in most provinces rose remarkably until 2018, except in 2020 because of a tight policy implemented for the COVID-19 lockdown. Then, the variation of VCD_trop NO₂ has dropped in the current year. The seasonality of VCD_trop NO₂ concentrations is clearly visible in Sumatra. The concentration of VCD_trop NO₂ in eastern Sumatra was slightly higher during the dry season and slightly lower during the wet season. The higher temperature, greater amount of rainfall, and stronger wind speed were helpful to VCD_trop NO₂ elimination. According to the spatio-temporal distribution analysis of VCD_trop NO₂ during the 8-year study, the efficacy of environmental regulations applied in Sumatra was less evident. Thus, for future tasks, a study regarding the relationship between other urban air pollutants such as CO, SO₂, O₃, and PM with population size should be conducted. Also, the use of finer resolution product such as TROPOMI can improve global analysis of VCD_trop NO₂, that will also useful for studies on detailed spatiotemporal distribution in nitrate aerosols. This output will help to give a new insight about urban air pollution studies

Author contributions statement

Conceptualization and supervision: MR; writing review and editing: MR, WMRI, THA, TEA, SN, and ND; data curation and formal analysis: MR, WMRI and MTL; evidence collection, review, and editing: MR, WMRI, HGA, HA and MTL.

Availability of Data and material

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Disclosure statement

On behalf of all authors, the corresponding author states that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.