Ambient air quality of a less industrialized region of India (Kerala) during the COVID-19 lockdown

Graphical abstract

Highlights

• •Air quality of major cities of Kerala State (India) during the COVID-19 lockdown.
• •Substantial reduction of NO₂ (-48 %) NO_x (-53 % to -90 %) and CO (-24 % to -67 %).
• •Significant decrease in PM_2.5 (-24 % to -47 %) and PM₁₀ (-17 % to -20 %) levels.
• •Increase of O₃ levels in Northern Kerala during lockdown period.
• •Indications of multiple sources (other than traffic and industry) of air pollution.

Abstract

This study assesses the effect of lockdown, due to the coronavirus disease (COVID-19) pandemic, on the concentration of different air pollutants and overall air quality of a less industrialized region (Kerala) of India. We analysed data from four ambient air quality stations over three years (January to May, 2018–2020) with pairwise comparisons, trend analysis, etc. Results indicated unprecedented reduction in the concentration of the air pollutants: nitrogen dioxide, NO₂ (-48 %), oxides of nitrogen, NO_x (-53 % to -90 %), carbon monoxide, CO (-24 % to -67 %) and the particulate matter (-24 % to -47 % for particulate matter with a diameter of less than 2.5 μm, PM_2.5; -17 % to -20 % for particulate matter with a diameter of less than 10 μm, PM₁₀), as well as the reduction of the National Air Quality Index (NAQI). These reductions indicate an overall improvement of the ambient air quality due to restrictions on transportation, construction, and the industrial sectors during lockdown, even in an area considered less industrial. Despite the general decreasing trend of the concentration of various air pollutants from January to May, suggesting seasonal influences, the trend was intensified in the year 2020 due to the added effect of the lockdown measures. Comparison of the results with those from larger and more industrialized cities suggests that the effects of lockdown are more variable, and focused on the levels of gaseous pollutants. Findings from this study demonstrate the far-reaching effects and implications of the COVID-19 lockdown on ambient air quality, even on less industrialized and less urbanized regions.

1. Introduction

The rapid spread of the coronavirus disease (COVID-19) pandemic caused by the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) induced inenarrable social and economic impacts across the globe. The COVID-19 was first detected in Wuhan (China) and the disease had affected more than 15 million people in 216 countries until 26 July 2020 (https://covid19.who.int). The first COVID-19 case in India was reported on 30 January 2020 in Kerala State, followed by a few more in the first week of February 2020, with escalating cases since the second week of March 2020. As of 26 July 2020, the Government of India had reported more than 1.4 million confirmed cases and 32,771 deaths (https://www.mygov.in/covid-19). Kerala was one of the most affected states of India during the initial stages of the COVID-19 outbreak in the country.

In response to the COVID-19 outbreak, various countries implemented diverse non-pharmaceutical interventions (e.g., personal protection and hygiene, physical distancing, environmental and travel-related measures, etc.) to slow down and to reduce the mortality rates associated with the COVID-19, with the ultimate objective to reach and to maintain the state of low-level or no transmission. In addition to these efforts, many countries imposed large-scale public health and social measures, including restriction of private and public transportation, suspension of educational/commercial/business/religious activities, and geographical area quarantine (often collectively referred to as lockdown) to curtail the spread of the highly transmissible virus within the society (World Health Organisation, 2020). The preventive lockdown was first implemented in Wuhan on 23 January 2020 and subsequently extended for entire China for at least three weeks (Le et al., 2020). Later, several countries across the globe (e.g., Belgium, Italy, Spain, Germany, UK, South Africa, Argentina, Colombia, USA, Israel, New Zealand) implemented the lockdown for varying periods.

Kerala (India) implemented the state-wide lockdown on 24 March 2020, as numbers of COVID-19 cases drastically increased in the State (Table S1; Fig. 1 ). The Government of India also enforced the nation-wide lockdown from 25 March 2020 onwards. Later, the nation-wide lockdown was extended further till the end of May 2020, with varying exemptions in different parts of the country. The lockdown implemented in Kerala was effective against the transmission of the virus, as the COVID-19 cases (registered during March, April and the first week of May) were almost completely recovered by 9 May 2020 (Fig. 1). The rising number of cases since 10 May 2020, however, was attributed to the controlled re-entry of the natives of Kerala stranded in other parts of the country and the globe. As of 26 July 2020, the State Government of Kerala had reported 19,025 confirmed cases and 61 deaths (https://dashboard.kerala.gov.in). Implementation of the lockdown restrictions has resulted in the stagnation of economic activities, such as construction, industrial projects, transportation, etc., leading to rising unemployment, decrease in income generation, and reduced consumer activity. The growth rate of the Gross Domestic Product (GDP) of India in the fourth quarter of the fiscal year 2020 dropped to 3.1 % compared to the previous quarters (4.1–5.2 %), mainly due to the effect of the COVID-19 on the country’s economy (http://mospi.gov.in).

Fig. 1.

History of COVID-19 cases of Kerala (up to 31 May 2020). The period between the dashed vertical lines indicates the state-wide lockdown phase in Kerala.

Despite the infliction on the economic sector, the lockdown measures have unintentionally bring forth benefits for the environment. During the lockdown period, across the globe, emissions from the transportation and industrial sectors were remarkably lower leading to a significant reduction in the levels of environmental pollution (Arora et al., 2020; Muhammad et al., 2020; Paital, 2020). Numerous researchers (e.g., Collivignarelli et al., 2020; Dantas et al., 2020; He et al., 2020; Kanniah et al., 2020; Chin et al., 2020) have analysed data on environmental pollution (mostly air pollution) of different parts of the globe and have reported improvement in the ambient air quality levels during the COVID-19 lockdown. For India, researchers have similarly quantified the impacts of the nation-wide lockdown on the ambient air quality, using observed data from the ambient air quality monitoring stations (Mahato et al., 2020; Sharma et al., 2020) as well as satellite-derived/reanalysis products (Gautam, 2020) or a combination of both (Lokhandwala and Gautam, 2020; Mahato and Ghosh, 2020; Selvam et al., 2020).

While many studies have documented improvements in air quality in major cities, the effects of lockdown measures on ambient air quality of less urbanized and less industrialized regions are not investigated comprehensively at the global level. Whereas urban and industrialized areas generally show higher levels of pollution compared to less urbanized and less industrialized areas, with air quality improving in less urban (or rural) areas (Strosnider et al., 2017), contrasting patterns in some air pollutants (e.g., SO₂) between urban and rural areas during the COVID-19 lockdown have also been reported (Wang et al., 2020). Moreover, research has also not addressed the effect of short-term fluctuations and seasonal trends in the concentrations of air pollutants due to differences in seasons between the lockdown (i.e., summer) and pre-lockdown (i.e., winter) periods in a broader regional context. Hence, this study investigated the ambient air quality in a less industrialized region (the maritime State of Kerala) of India.

The study sought to answer the following research questions. (1) To what extent did the lockdown during the COVID-19 improve the ambient air quality of Kerala State? (2) How significant are the seasonal trends in the reduction of air pollutant levels, and how different are their influences in 2020 compared to the previous years? (3) How do changes in the ambient air quality of Kerala State differ from more urbanized and industrialized areas? We hypothesized that significant reductions in air pollution would emerge from the lockdown measures, even in the less industrialized state of Kerala, though the detailed trends may differ from those of the more urbanized and industrialized areas.

2. The study area of Kerala

The Kerala State is an elongated strip of land, along the windward side of the Western Ghats, in the southwest tip of Peninsular India (Fig. 2 ). The State covers an area of 38,863 km² (roughly the size of Switzerland), which accounts for about 1.2 % of the total geographical area of India. The physiography of the State is classified into the highland (in the east), the lowland (in the west) and the midland in between. The majority of the area of Kerala experiences tropical humid climate, while the eastern slopes of the Western Ghats fall under tropical semi-arid climate. Kerala receives a mean annual rainfall of about 2800 mm, of which, 70 % occurs as the Indian summer monsoon rainfall (Thomas and Prasannakumar, 2016). The State, though small in size, has varying topographical and climatic conditions leading to a wide spectrum of land use/ land cover and cropping pattern. However, roughly 60 % of the land area is utilized for agricultural purposes, whereas nearly 30 % of the State is under forest cover (http://www.forest.kerala.gov.in/index.php/forest/forest-area). Kerala State witnessed recurrent floods in 2018, 2019 and 2020, where the August 2018 flood occurred in Kerala was the worst flood since 1924. The August 2018 flood primarily affected the central parts of the State and inundated about 1100 km² in Palakkad, Thrissur, Ernakulam, Idukki, Kottayam, Alappuzha, and Pathanamthitta districts. The flood had severely affected the croplands (∼46 %) and settlement areas (∼21 %) of these districts (Lal et al., 2020a). The census 2011 data (https://censusindia.gov.in) indicate that the population of Kerala as 33,387,677 (population density = 859 persons km⁻²), which forms roughly 3 % of the population of India. The major urban areas of the State include Thiruvananthapuram, Ernakulam and Kozhikode, with a population of 0.97, 0.60 and 0.61 million, respectively.

Fig. 2.

Map of Kerala State (India) including the locations of the ambient air quality monitoring stations.

In general, the levels of air pollutants (except particulate matter) in Kerala are within the permissible limit prescribed by the national ambient air quality standards (http://kerenvis.nic.in). The concentration of various air pollutants, especially particulate matter, however, in Ernakulam and Kozhikode districts was reported nearing the alarming level in the recent years (Government of Kerala, 2020a). Tobollik et al. (2015) observed a considerable health burden for the population living in urban Kerala mainly due to particulate matter with a diameter of less than 2.5 μm and 10 μm (PM_2.5 and PM₁₀) concentrations, albeit local air quality standards being met, and reported that a decrease of 10 % in particulate matter concentrations may save 15,904 (95 % confidence interval: 11,090-19,806) life years. The industries and vehicular traffic are the major drivers for the deterioration of air quality of the State. The industrial sector of Kerala is not well-developed compared to other states of India, which is evinced by the substantially lower figures of the per-capita domestic product and per-capita manufacturing value added in the State (Thomas, 2005). Among the different states of India, Kerala ranks 12^th position in terms of the total number of industries and 18^th place in terms of the fixed capital (http://mospi.nic.in). Based on the annual survey of industries of 2016-17, 6507 industries were operating in the manufacturing sector of Kerala, where more than 60 % of the industries is related to food, non-metallic minerals, tobacco, wood, rubber and plastic products (Government of Kerala, 2020b). Among the different districts of the State, Ernakulam has the greatest number of industries (20 %), where the two dominant industrial clusters (consisting mostly of chemical industries) of the district are Eloor and Ambalamugal.

The major transport infrastructure of the State includes 0.27 million km of road, 1588 km of railway, 1687 km of inland waterway, 585 km coastal route with 18 ports and four international airports. The road density of Kerala is 3.9 km km⁻², which is roughly three times the national average. The primary road network of the State is the National highways, which handle about 40 % of the total traffic, while the State highways and the major district roads carry another 40 %. The traffic on the roads of the State shows a steadily increasing rate of 12–14% per year. As of March 2019, Kerala had more than 13 million registered motor vehicles and the last two decades witnessed a compounded annual growth rate of above 10 % (Government of Kerala, 2020a).

3. Materials and methods

The ambient air quality data of four air quality monitoring stations in Kerala, viz., Plamood, Thiruvananthapuram (K1), MG Road, Ernakulam (K2), Eloor, Ernakulam (K3) and Palayam, Kozhikode (K4) were collected from the Kerala State Pollution Control Board (Fig. 2). The ambient air quality data analyzed include gaseous pollutants, such as nitrogen monoxide (NO), nitrogen dioxide (NO₂), oxides of nitrogen (NO_x), ammonia (NH₃), sulphur dioxide (SO₂), carbon monoxide (CO), and ozone (O₃) as well as particulate matter (PM_2.5 and PM₁₀). Measurements of all air pollutants were not available, however, from all the monitoring stations (Table 1 ). Data for daily ambient air quality of all the stations were pre-processed to remove the outliers. We computed the National Air Quality Index (NAQI) (after Central Pollution Control Board, 2014) using the processed air quality data. The NAQI uses NO₂, NH₃, SO₂, CO, O₃, PM_2.5, PM₁₀ and lead (Pb) for computing the index. The index requires the concentration of a minimum of three pollutants, including at least PM_2.5 or PM₁₀. We converted the concentration of each air pollutant (C_p) to a normalized number (i.e., sub-index, I_p) using segmented linear functions (Eq. 1).

$$I_p=\left[\left\{\frac{\left(I_{HI}-I_{LO}\right)}{\left(B_{HI}-B_{LO}\right)}\right\}\times \left(C_p-B_{LO}\right)\right]+I_{LO}$$ (1)

where, B_HI is the breakpoint concentration greater or equal to C_p, B_LO is the breakpoint concentration smaller or equal to C_p, I_HI and I_LO are the air quality index values corresponding to B_HI and B_LO, respectively. Finally, Eq. 2 provides an estimation of the NAQI (Central Pollution Control Board, 2014):

$$NAQI=Max\left(I_p\right)$$ (2)

Table 1.

Details of the ambient air quality monitoring stations, Kerala*.

Station Code	Location	Type	Pollutants**	Data availability	Remarks
K1	Plamood (Thiruvananthapuram)	Urban/Residential	NO, NH3, SO2, CO, O3, PM2.5, PM10	2018-2020
K2	MG Road (Ernakulam)	Urban/Residential	NOx, NH3, SO2, CO, PM2.5, PM10	2018-2020
K3	Eloor (Ernakulam)	Industrial	NO, NO2, NOx, NH3, SO2, CO, O3, PM2.5, PM10	2018-2020	PM2.5 is not available in 2019.
K4	Palayam (Kozhikode)	Urban/Residential	NOx, NH3, SO2, CO, O3, PM2.5, PM10	2018-2020

*Details belong to the ambient air quality monitoring stations under the scope of this study; ** Data available in 2020.

The NAQI contains six classes: good (0–50), satisfactory (51–100), moderate (101–200), poor (201–300), very poor (301–400) and severe (401–500).

The analysis utilized daily data (24 h) from 1 January to 31 May of 2018, 2019 and 2020. We considered the period between 1 January and 23 March as the pre-lockdown period, and the duration from 24 March to 17 May as the lockdown period. Following this classification, the data between 1 January 2020 and 23 March 2020 represent the pre-lockdown period (PLD) 2020, while the data from 24 March 2020 to 17 May 2020 are termed as the lockdown (LD) 2020. Similarly, we considered the daily average of the air quality data between 1 January and 23 March of 2018 and 2019 as the PLD_mean, and we treated the daily average of the data between 24 March and 17 May as the LD_mean. We computed descriptive statistics of the different air pollutants for the PLD 2020, LD 2020, PLD_mean and LD_mean for pairwise comparisons. Since the air quality data rarely follow the normal distribution, we applied the Mann-Whitney test for analysis of statistical significances between groups, as the Mann-Whitney test is applicable for differences in medians as well as in the shape and spread of the distributions.

We tested the daily time series of each pollutant for the entire period (i.e., 1 January to 31 May) of all years (i.e., 2018, 2019 and 2020) for presence of monotonic trends using the Mann-Kendall test (Kendall, 1975; Mann, 1945). Since the Mann-Kendall test assumes that the data are independent and identically distributed, we tested the data series for serial correlation. We applied the Mann-Kendall test to the original time series, when a significant autocorrelation lacks at the 5 % level, and to the pre-whitened series, when a significant serial correlation occurrs (von Storch, 1999).

We compared changes in the ambient air quality data during the LD 2020 in Kerala State with the changes in the levels of various air pollutants recorded in the urbanized and industrialized regions of India as well as across the globe during the lockdown period. This comparison evaluated differences in the magnitude of changes between the urbanized/industrialized and less urbanized/industrialized regions.

4. Results

The various air pollutants measured at the different ambient air quality monitoring stations of Kerala show notable temporal variability during January 2020 to May 2020 (Fig. 3 ). The wider temporal variations are noted in PM_2.5 and PM₁₀ at K1, NO_x and NH₃ at K2, NO, NO₂, NO_x and NH₃ at K3 and CO at K4. In general, all the air quality monitoring stations of Kerala witnessed a gradual reduction in the concentration of most of the air pollutants during the lockdown window.

Fig. 3.

Daily time series (1 January 2020 - 31 May 2020) of the concentration of different air pollutants measured at (a) K1, (b) K2, (c) K3 and (d) K4. The period between the dashed vertical lines indicates the state-wide lockdown phase in Kerala.

4.1. Comparison of ambient air quality during pre-lockdown and lockdown phases (PLD 2020 vs. LD 2020)

Table 2 provides the summary statistics of the ambient air quality data of the air quality monitoring stations (K1, K2, K3 and K4) of Kerala. A comparison of the air quality data during the PLD 2020 and LD 2020 (Fig. S1) shows significant variability in the concentration of most of the air pollutants between the periods (p ≤ 0.001). The average of the particulate matter shows considerable differences between the PLD 2020 and LD 2020 at K1 (PM_2.5 = -71 % and PM₁₀ = -45 %) and K2 (PM_2.5 = -58 % and PM₁₀ = -49 %), whereas the differences at K3 are comparatively lower in magnitude (Table 2). The gaseous pollutants show noticeable changes at K3 and K4 compared to K1 and K2 (except for NO_x) (Fig. S1). The average concentration of the air pollutants, such as NH₃, NO₂, NO_x and CO measured at K3 during the LD 2020 is lesser than 50 % of their concentration during the PLD 2020. A similar change is visible in the concentration of NO_x, CO and SO₂ at K4. The significant changes in the concentration of the air pollutants, such as NO (K1 = -21 %, K3 = -28 %), NO₂ (K3 = -70 %), NO_x (K2= -90 %, K3 = -65 %, K4 = -71 %) and CO (K1 = -24 %, K2 = -10 %, K3 = -50 %, K4 = -75 %) imply reduced emissions from the transportation and industrial sectors due to the lockdown.

Table 2.

Summary statistics of various air pollutants of different stations (2a to 2d) during the pre-lockdown and lockdown periods.

2a. Station: K1
Period		NO	NH3	SO2	CO	O3	PM2.5	PM10	NAQI
		μg m−3	μg m−3	μg m−3	mg m−3	μg m−3	μg m−3	μg m−3	–
PLD 2020	Median	9.07	11.34	17.29	1.38	42.85	35.14	57.89	72
	Q1	8.32	10.67	16.42	1.27	38.70	29.71	52.69	66
	Q3	9.90	13.07	18.69	1.58	48.06	43.09	66.13	90
	Skewness	0.57	0.68	−0.33	2.93	0.20	1.23	1.59	1.43
LD 2020	Median	7.20	6.93	16.08	1.05	32.43	10.17	31.76	53
	Q1	6.75	5.85	15.61	1.01	28.26	7.38	17.69	51
	Q3	7.59	7.46	16.63	1.10	37.25	14.70	43.13	55
	Skewness	−0.89	1.88	−0.38	−0.69	0.85	0.91	−0.36	−0.18
PLDmean	Median	1.39	6.20	6.86	0.98	64.39	45.76	79.54	81
	Q1	1.34	5.21	5.70	0.91	57.97	38.22	69.65	72
	Q3	1.64	7.56	10.07	1.17	69.16	57.75	90.36	96
	Skewness	1.15	0.59	1.53	5.10	0.08	0.59	0.82	1.45
LDmean	Median	1.90	3.61	4.68	1.01	47.48	24.79	52.77	57
	Q1	1.39	3.24	4.28	0.95	42.00	21.66	46.66	52
	Q3	1.97	4.00	5.14	1.09	51.37	30.45	68.95	71
	Skewness	−0.35	0.35	1.38	3.32	0.10	0.19	0.22	1.00

2b. Station: K2
Period		NOx	NH3	SO2	CO	PM2.5	PM10	NAQI
		μg m−3	μg m−3	μg m−3	mg m−3	μg m−3	μg m−3	–
PLD 2020	Median	62.88	19.96	3.00	1.11	41.32	73.90	87
	Q1	23.48	9.92	3.00	0.89	34.11	65.00	72
	Q3	112.31	26.66	3.10	1.52	50.85	86.29	132
	Skewness	0.14	1.86	1.91	0.44	−0.21	−0.40	0.61
LD 2020	Median	6.18	20.26	2.98	1.00	17.41	37.58	50
	Q1	5.05	3.02	2.06	1.00	15.01	31.35	50
	Q3	7.54	30.30	3.00	1.01	21.22	43.67	54
	Skewness	3.37	0.80	0.05	1.70	0.35	0.20	2.00
PLDmean	Median	46.50	1.90	2.95	1.60	53.00	98.20	103
	Q1	35.71	1.60	2.40	1.45	40.55	80.96	86
	Q3	60.43	2.50	4.20	1.70	70.75	111.24	136
	Skewness	1.28	2.13	2.79	0.23	0.19	0.07	0.75
LDmean	Median	47.35	1.75	18.50	1.90	29.40	62.50	95
	Q1	40.25	1.60	15.00	1.80	24.63	53.40	90
	Q3	57.66	2.50	19.70	2.30	34.08	71.73	104
	Skewness	1.67	1.06	−0.10	0.42	−0.02	0.02	0.48

2c. Station: K3
Period		NO	NO2	NOx	NH3	SO2	CO	O3	PM2.5	PM10	NAQI
		μg m−3	μg m−3	μg m−3	μg m−3	μg m−3	mg m−3	μg m−3	μg m−3	μg m−3	–
PLD 2020	Median	12.05	5.22	15.01	17.32	6.41	0.66	35.50	15.00	35.02	37
	Q1	8.80	3.10	10.93	17.07	5.53	0.51	27.35	10.58	34.19	35
	Q3	12.68	7.19	18.16	20.02	7.86	0.82	40.28	16.17	38.87	47
	Skewness	1.12	4.37	3.77	1.85	2.12	0.30	−0.35	0.00	1.09	4.16
LD 2020	Median	8.63	1.56	5.23	4.89	5.44	0.33	24.61	14.84	35.35	35
	Q1	8.59	1.52	5.17	4.84	5.17	0.30	21.87	14.57	33.14	35
	Q3	8.72	1.71	5.35	4.94	5.85	0.35	27.62	15.13	35.55	36
	Skewness	−2.05	−1.19	2.85	2.34	2.12	−0.64	0.66	0.60	0.76	3.16
PLDmean	Median	8.04	24.68	23.00	51.90	11.93	0.75	13.21	50.05	46.58	83
	Q1	6.29	16.73	19.36	49.39	11.10	0.69	11.90	41.44	42.18	69
	Q3	11.2	29.37	26.07	56.06	12.87	0.86	14.11	60.92	53.48	103
	Skewness	2.71	0.87	2.46	−0.48	−0.04	1.37	−0.45	−0.15	0.36	0.57
LDmean	Median	5.69	14.24	16.91	45.03	5.46	0.73	12.83	32.34	33.65	55
	Q1	5.46	12.92	15.86	44.84	5.11	0.69	12.23	1.10	30.69	37
	Q3	6.03	15.31	17.61	47.35	6.03	0.81	13.42	37.52	36.76	64
	Skewness	1.43	0.97	−1.07	1.03	2.34	3.83	3.97	1.15	0.55	3.85

2d. Station: K4
Period		NOx	NH3	SO2	CO	O3	PM2.5	PM10	NAQI
		μg m−3	μg m−3	μg m−3	mg m−3	μg m−3	μg m−3	μg m−3	–
PLD 2020	Median	45.38	5.07	4.58	1.03	6.51	30.62	78.15	79
	Q1	39.03	3.10	4.09	0.76	5.04	25.11	64.38	66
	Q3	57.44	5.64	9.53	1.23	8.05	41.45	85.32	86
	Skewness	−0.15	−0.40	1.06	−0.06	0.23	0.30	−0.10	0.02
LD 2020	Median	13.29	3.38	2.30	0.26	9.92	26.50	43.28	45
	Q1	11.97	3.25	1.97	0.16	8.32	22.35	36.49	41
	Q3	14.47	3.71	2.65	0.42	10.63	30.46	47.54	51
	Skewness	0.59	0.90	1.28	0.24	0.34	−0.01	−0.40	−0.28
PLDmean	Median	79.28	5.27	3.99	0.97	7.31	70.59	107.35	135
	Q1	71.15	4.02	3.71	0.82	5.94	59.97	92.83	103
	Q3	95.11	6.52	7.96	1.34	8.77	86.13	125.94	187
	Skewness	0.16	−0.30	1.38	0.16	1.15	−0.22	−0.01	0.18
LDmean	Median	71.64	1.76	4.79	0.75	3.44	39.72	73.82	90
	Q1	65.42	1.55	4.12	0.64	2.96	33.77	63.90	83
	Q3	77.16	2.00	5.14	0.83	3.78	45.11	82.33	96
	Skewness	−0.10	0.18	1.02	0.41	2.92	−0.24	−0.43	0.16

The air pollutants, such as NO₂, NH₃, SO₂, CO, O₃, PM_2.5, PM₁₀, and Pb, are considered for the calculation of the NAQI (Central Pollution Control Board, 2014). In this study, however, all the air pollutants except Pb were used for the computation of the NAQI. The temporal variability of the NAQI of all the stations shows significant differences between the PLD 2020 and LD 2020 (p ≤ 0.001) (Fig. 4 ). On average, the NAQI of K2 and K4 during the LD 2020 shows a reduction by 43 %, whereas the NAQI of K1 is decreased by 26 % compared to the NAQI of the PLD 2020 (Table 2). The reduction in the NAQI implies an improvement in the overall air quality. Following the classification of the NAQI, the PLD 2020 is characterized by moderate to satisfactory air quality in all the stations, whereas the LD 2020 is dominated by good to satisfactory air quality. All the monitoring stations do follow this trend, whereas difference in the NAQI between the periods is less at K3 compared to other stations. Although most of the air pollutants show a significant decrease in their concentration during the LD 2020 (compared to the PLD 2020), O₃ at K4 exhibits a significantly higher concentration during the LD 2020 (p ≤ 0.001) (Fig. S1d). Comparison of the average concentration of the air pollutants, as well as the average NAQI, indicates that the air quality of the major cities of Kerala during the LD 2020 is significantly different compared to the PLD 2020.

Fig. 4.

Daily NAQI between 1 January 2020 and 31 May 2020 computed at (a) K1, (b) K2, (c) K3 and (d) K4. The horizontal lines represent the threshold of different NAQI classes.

4.2. Trend of air pollutants between January and May (2018, 2019 and 2020)

Although the LD 2020 showed a decrease of the concentration of the air pollutants in Kerala State, the reduction in the concentration (due to lockdown) cannot be determined without estimating the role of short-term fluctuations or the seasonal trends in the concentration of the air pollutants during the analysis period. The fluctuations can relate to meteorological, seasonal or even anthropogenic and cultural factors (Chen et al., 2015; Cichowicz et al., 2017; Mohtar et al., 2018). Zangari et al. (2020) observed hardly any significant changes in the concentration of PM_2.5 and NO₂ in New York City during the lockdown period compared to the same period in 2015-2019. Hence, we tested the daily time series of the ambient air quality data of the stations from January to May of 2018, 2019 and 2020 for the presence of trends.

Results of the trend analysis of the concentration of the air pollutants of the various stations indicate a significant decreasing trend over the analysis period in 2020 (Table 3 ). The concentration of all the air pollutants at K1, K2 (except CO), K3 (except PM_2.5 and PM₁₀) and K4 (except O₃, NH₃ and SO₂) shows significant negative trends (95 % confidence levels), implying a gradual decrease of the concentration from January to May. On the other hand, the concentration of O₃ at K4 shows a significant increasing trend. A comparison of the trends of the concentration of the air pollutants in 2020 with the trends in 2018 and 2019 indicates that most of the air pollutants follow a decreasing trend in all the years (e.g., PM_2.5, PM₁₀ at K1, K2 and K4, NH₃ at K1 and K4), whereas a few exhibit alternating trends (e.g., NO_x at K2 and K3, CO at K1 and K3), and a few show reversal of the trends (SO₂ and CO at K2, O₃ at K4) in 2020.

Table 3.

Results of trend analysis of pollutant levels of various stations during January 1 to May 31 of 2018, 2019 and 2020.

Station K1
Pollutant	Test Statistic (S)			Z Score
	2018	2019	2020	2018	2019	2020
NO	1903	−855	−2775	3.12**	−1.39	−4.46***
NH3	−1018	−991	−2321	−1.67	−1.61	−3.73***
SO2	−834	−895	−1881	−1.37	−1.45	−3.02**
CO	−1752	653	−1229	−2.87**	1.06	−1.98*
O3	−1460	−2599	−2303	−2.39*	−4.22***	−3.70***
PM2.5	−1414	−1631	−2033	−2.32*	−2.65**	−3.30***
PM10	−1804	−1933	−2295	−2.96**	−3.14**	−3.69***
Station K2
Pollutant	Test Statistic (S)			Z- Score
	2018	2019	2020	2018	2019	2020
NOx	−414	938	−1918	−1.04	1.54	−3.18**
NH3	ND	−36	−990	ND	−0.11	−1.71*
SO2	1495	3190	−3197	5.29***	5.51***	−5.36***
CO	839	2416	−664	2.14*	3.96***	−1.08
PM2.5	−1168	−1779	−2149	−2.93**	−2.92**	−3.49***
PM10	−782	−2272	−2527	−1.96*	−3.73***	−4.10***
Station K3
Pollutant	Test Statistic (S)			Z Score
	2018	2019	2020	2018	2019	2020
NO	−1132	−776	−1907	−2.10*	−1.26	−3.07**
NO2	−1457	−2345	−2620	−2.68**	−3.81***	−4.26***
NOx	−1718	1215	−3718	−3.20**	1.97*	−6.35***
NH3	1503	−2023	−3371	2.76**	−3.29**	−5.42***
SO2	−416	−2695	−1287	−0.71	−4.38***	−2.07*
CO	2375	−2252	−1374	4.67***	−3.69***	−2.21*
O3	−1433	2565	−3031	−2.42*	4.17***	−4.88***
PM2.5	−2752	ND	545	−5.60***	ND	0.91
PM10	−1511	1592	9	−2.55*	2.59**	0.01
Station K4
Pollutant	Test Statistic (S)			Z-Score
	2018	2019	2020	2018	2019	2020
NOx	−2021	−2088	−1873	−3.28**	−3.43***	−3.01**
NH3	−1885	−1945	−191	−3.06**	−3.16**	−0.31
SO2	−454	1713	−328	−0.77	2.81**	−0.53
CO	−2007	−333	−2259	−3.26**	−0.54	−3.74***
O3	−2126	−1372	1497	−3.45***	−2.23*	2.41*
PM2.5	−1358	−1359	−2439	−2.21*	−2.21*	−3.92***
PM10	−1766	−2045	−1825	−2.90**	−3.32***	−2.93**

ND-Not tested due to multiple data gaps.

Confidence levels: 95 % (*), 99 % (**) and 99.9 % (***).

The continuously decreasing trend of the concentration of the air pollutants (e.g., CO, PM_2.5 and PM₁₀) over the period (i.e., from January to May) is likely due to the reduction of the number of vehicles during April and May, as a result of the closure of the educational institutes and related activities for summer holidays. Besides, the winter months (January and February) typically have higher levels of air pollution compared to the summer season due to the decreased levels of solar flux and photochemical activity and fewer instances of long-range pollutant transport (Zangari et al., 2020). Since there a significant decreasing trend exists for most of the pollutants in all the years, it could mean that any changes recorded in the concentration of the pollutants between the PLD 2020 and LD 2020 could have resulted from this general trend, as well as the effect of lockdown. Hence, reversal of the positive trend, as well as the intensification of the general negative trend (manifested in Mann Kendall test statistic) in 2020 compared to the previous years, may be treated as the effect of lockdown.

An estimation of the number of days in 2018–2020, in which the concentration of the different air pollutants exceeding the national ambient air quality standards, provides additional details. The gaseous pollutants showed very few exceedance during the years (except NO₂ for a few days at K3), whereas the particulate matter exceeded the limits for a significantly greater number of days. The national ambient air quality standards for PM_2.5 and PM₁₀ (24 h) are 60 and 100 μg m⁻³, respectively. The number of days (during the pre-lockdown time frame) exceeding the national standards in 2018 and 2019 is remarkably higher than that in 2020 for both PM_2.5 and PM₁₀ (Fig. S2). At K1, for example, 32 % of the total days between 1 January 2018 and 23 March 2018 (PLD 2018) and 22 % of the total days between 1 January 2019 and 23 March 2019 (PLD 2019) exceeded the national standard for PM_2.5, while during the PLD 2020, only 7 % of the total days surpassed the standard. During the LD 2018 (24 March 2018 to 17 May 2018), LD 2019 (24 March 2019 to 17 May 2019) and LD 2020, however, none of the days exceeded the national standard for PM_2.5. We note such a pattern in all the years and all the stations for both PM_2.5 and PM₁₀. Figure S2 indicates that the number of days exceeding the national standards for PM_2.5 and PM₁₀ also shows a declining pattern from 2018 to 2020 for both the pre-lockdown and lockdown phases, but in differing magnitudes. As definite trends are notable within the concentration of the air pollutants during the analysis period, the estimation of the change in the concentration of the pollutants due to the effect of lockdown should consider this trend also.

4.3. Relative changes in the ambient air quality in Kerala during lockdown

Since the changes in the concentration of the various air pollutants during the LD 2020 result from the combined effect of the seasonal trends as well as the lockdown, we estimated the effect of lockdown on the ambient air quality as the relative change in the concentration of the air pollutant (ΔP_x) between the PLD 2020 and LD 2020 with respect to the corresponding average change of the previous years. This expression is (Eq. 3):

$$\Delta P_x\left(in\%\right)=\left(\frac{{PLD}_{norm}-{LD}_{norm}}{{PLD}_{norm}}\right)\times 100$$ (3)

where, ΔP_x is the relative change in the concentration of any given air pollutant during the LD 2020 with respect to the PLD 2020, ${PLD}_{norm}=\frac{{PLD}_{2020}}{{PLD}_{mean}}$ and ${LD}_{norm}=\frac{{LD}_{2020}}{{LD}_{mean}}$.

The ΔP_x of most of the air pollutants in all the stations implies a remarkable reduction of the concentration in the LD 2020 with respect to the PLD 2020 (Fig. 5 ). The significant reduction in the concentration of the pollutants, such as NO₂ (-48 %) NO_x (-53 % to -90 %), CO (-24 % to -67 %) as well as the particulate matter (-24 % to -47 % for PM_2.5, -17 % to -20 % for PM₁₀) is correlated to the decreased emissions from the transportation and industrial sources during the lockdown period. One may notice that the rate of reduction in the concentration of the air pollutants is non-uniform across the State, implying the role of regional socio-economic, meteorological and anthropogenic factors controlling air quality.

Fig. 5.

Relative change (ΔP_x) in the concentration of the air pollutants during the LD 2020: (a) K1, (b) K2, (c) K3 and (d) K4. No data (or gap) indicates unavailability of pollutant data.

Among the different stations, K1 exhibits a comparatively lower percentage of reduction of the concentration of the various air pollutants (Fig. 5a). NO_x and SO₂ concentrations at K2, NO_x and NH₃ concentrations at K3, NO_x, SO₂, and CO levels at K4 show significant reduction (> 50 %) in their concentration during the LD 2020. The concentration of the particulate matter is reduced only at K1 and K2 (Fig. 5a, b), however, the concentration of the particulate matter in a few stations (K3 and PM_2.5 at K4) shows an increase during the LD 2020 (Fig. 5c, d). The relative change of the NAQI of the different stations suggests that the improvement in the air quality is significant at K2 (-38 %) and K4 (-15 %), whereas the change in the NAQI of K1 is insignificant. The relative change at K3 shows an increase (+43 %), however, which is probably due to the higher background and residual pollution.

5. Discussion

The improvement of the air quality of Kerala State during the LD 2020, though varying in magnitude, results from the combined effect of the lockdown measures (i.e., restrictions on the transportation sector and cessation of construction and industrial activities) and the seasonal trends. Among the different air pollutants, the most significant and widespread reduction is observed for CO (-24 to -67 %), whereas the reduction in the concentration of NO_x (-53 to -90 %) and SO₂ (-58 to -84 %) is also apparent in some of the stations (Fig. 5). Since the air quality in India has been arguably improving consistently since 2018 (Narain et al., 2020), the improvements in air quality in 2020 could have also been a continuation of the trend. Kerimray et al. (2020) and Zangari et al. (2020) discussed the significance of this synergism on the improvement of air quality. As a less industrialized and less urbanized state of India, Table S2 shows a comparison of changes in the air quality of Kerala during the lockdown with various studies from more urbanized and industrialized areas worldwide.

The lockdown over the Yangtze River Delta Region (China) lowered NO_x, SO₂, and PM_2.5 emissions by approximately 29–47 %, 16–26 %, and 27–46 % during the Level I and Level II response periods, respectively (Li et al., 2020). In the urban areas of São Paulo (Brazil), the partial lockdown witnessed a drastic reduction of NO (up to -77.3 %), NO₂ (up to -54.3 %), and CO (up to -64.8 %) concentrations compared to the five-year monthly mean. Otmani et al. (2020) reported significant reduction of NO₂ (-96 %), SO₂ (-49 %) and PM₁₀ (-75 %) concentrations in Sale city (Morocco). The urban areas of Europe also evinced reduced levels of air pollutants (Sicard et al., 2020): NO, NO₂, PM_2.5 and PM₁₀ show a reduction by 56–82%, 29–70%, -8 (i.e., increased by 8 %) to 13 %, and -1 (i.e., increased by 1 %) to 40 %, respectively. Chin et al. (2020) observed 40–93 % reduction in the concentration of SO₂, CO, NO₂ and particulate matter at Denver, Colorado, USA during the “stay at home” period compared to the same period in 2018 and 2019. In the Indian context, Sharma et al. (2020) observed an overall decrease of NO₂ (-18 %), CO (-10 %), PM_2.5 (-43 %) and PM₁₀ (-31 %) compared to the previous years. Significant reduction in the air pollutant levels is also reported in Gujarat State (India): NO₂ (-30 to -84 %), CO (-3 to -55 %), SO₂ (-22 to -58 %), PM_2.5 (-38 to -78 %) and PM₁₀ (-32 to -80 %) (Selvam et al., 2020). Mahato et al. (2020) noted that NO₂, CO, SO₂, PM_2.5 and PM₁₀ levels were reduced by 52.68 %, 30.35 %, 17.97 %, 53.11 %, and 51.85 %, respectively compared to pre-lockdown period. The reduction in the particulate matter during the lockdown was also significant compared to the same period of 2017–2019 (i.e., -34.19 % for PM_2.5 and -58.12 % for PM₁₀).

The reduction of the air pollutant levels in the lockdown period significantly varies among the urban, industrial, semi-urban and rural regions. Kanniah et al. (2020) observed that the reduction of the air pollutant levels varies among the industrial (NO₂ = -33 to -46 %; PM_2.5 = -19 to -42 %; PM₁₀ = -28 to -39 %), urban (NO₂ = -63 to -64 %; CO = -25 to -32 %; PM_2.5 = -23 to -32 %; PM₁₀ = -26 to -31 %), semi-urban (NO₂ = -55 to -56 %; CO = -25 to -27 %; PM_2.5 = -15 to -28 %; PM₁₀ = -22 to -27 %) and rural regions (NO₂ = -26 to -34 %; CO = -6 to -7 %; PM_2.5 = -4 to -27 %; PM₁₀ = -10 to -24 %). Wang et al. (2020) also reported that NO₂, CO, PM_2.5 and PM₁₀ concentrations of Hangzhou city during lockdown were reduced (compared to the same period of 2019) by 58.4 %, 22.3 %, 42.7 % and 47.9 %, respectively in urban areas, whereas it was 48.0 %, 20.8 %, 18.5 % and 39.6 %, respectively in rural areas. In general, comparison of the reduction of the air pollutant levels in Kerala State during the LD 2020 with these studies suggests a lesser degree of reduction of the gaseous pollutants (e.g., NO and NO₂) and the particulate matter in the State, compared to the highly urbanized and industrialized regions. Although the reduction in the gaseous air pollutant concentrations of Kerala State is comparable with that of the suburban and rural areas (Kanniah et al., 2020; Wang et al., 2020), reduction of the particulate matter shows considerable variability among the regions.

Contrary to the general reductions in the air pollutant levels across Kerala, a few stations exhibit a substantial increase in the concentration of NH₃ (5–100 % at K1, K2 and K4), SO₂ (36–85 % at K1 and K3), O₃ (3 % at K1 and 224 % at K4), PM_2.5 (> 50 % at K3 and K4) and PM₁₀ (40 % at K4) during the LD 2020 (Fig. 5). The major source of NH₃ emissions is agriculture, while minor sources include industrial processes, vehicular emissions and volatilization from soils and oceans (Behera et al., 2013). Hence, the increase in the concentration of NH₃ in Kerala State during the LD 2020 indicates the contributions from the sources other than industrial and transportation sectors. On the other hand, the major urbanized/industrialized areas of India (e.g., Kolkata, Delhi, Chennai) observed a considerable reduction in NH₃ concentration due to the shutdown of transportation sector (Bedi et al., 2020; Mahato et al., 2020). One may also note that K3 showed a significant reduction of NH₃ levels (-67 %) during the LD 2020, which could be attributed to the cessation of the industrial activities. The SO₂ levels at K1 and K3 showed a remarkable increase during the LD 2020 (Fig. 5a, c). Although the lockdown exerted cessation of the industrial operations, the relative increase in SO₂ levels could be due to the high background/residual pollution and/or due to regional factors (e.g., Kerimray et al., 2020; Wang et al., 2020). Shehzad et al. (2020) observed a faint trail of NO₂ emission along the major maritime routes of the Indian Ocean, implying the presence of active marine traffic during the period. Hence, the onshore transport of SO₂ due to marine traffic may also influence the concentration of SO₂ of Kerala State.

As a departure from the general behaviour of O₃ across Kerala State, K1 and K4 recorded an increase in O₃ concentration during the LD 2020. Similar observations were reported by various researchers (e.g., Mahato et al., 2020; Sicard et al., 2020; Siciliano et al., 2020; Tobías et al., 2020) and attributed to the decrease of NO_x concentration along with the increase in the reactivity of the volatile organic compounds (VOC) mixture, the decrease of NO and corresponding reduction of the O₃ consumption (i.e., titration) and the higher rate of insolation and increased temperatures (Tobías et al., 2020). In general, the bivariate relationship between NO_x and O₃ at K4 is negatively correlated (Fig. S3) implying that the increase in O₃ levels (224 %) at K4 could be associated with the decrease in the NO_x concentration (i.e., -68 %), while K1 and K3 show hardly any definite relationships. Although the general pattern of PM_2.5 and PM₁₀ can presumably reflect the reduction of transportation and industrial activities, K3 and K4 recorded a significant increase of particulate matter during the LD 2020 (Fig. 5c, d). The anomaly at K3 and K4 could relate to various reasons, such as emissions from the sources other than transportation and industrial sectors (e.g., domestic/residential sectors, burning of biomass, log-range transport, etc.) and a higher degree of background/residual pollution, and the contributions from these sources might have offset the reduction of the particulate matter due to the lockdown (Li et al., 2020; Otmani et al., 2020; Sicard et al., 2020). The concentration of the particulate matter would have significant implications in the quality of life of Kerala State as a reduction in PM_2.5 concentration by 10 % would save 15,904 life years (Tobollik et al., 2015). However, the majority of premature deaths in the Indian context occur in the less-urbanized/rural areas as compared to the heavily urbanized/industrialized regions, where the major factors are PM_2.5 and O₃ from the domestic, agricultural and residential sectors (Karambelas et al., 2018). Therefore, the results of this study have extended implications in the context of health risks associated with air pollution of Kerala State.

Since the transboundary transport is postulated for the increased levels of air pollutants of Kerala State, the role of long-range transport on the air quality was investigated by analysing the air mass trajectories for the LD 2020 by HYbrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) transport model (Rolph et al., 2017; Stein et al., 2015). The back-trajectory analysis at K3 and K4 indicates movement of air mass mostly from the Arabian Sea and the eastern parts of peninsular India (Fig. 6 ). Contrastingly, K1 receives air parcels from multiple regions including the Arabian Sea, the Bay of Bengal, the Indian Ocean and the northern parts of Sri Lanka. However, the seasonal effects on the transport of the pollutants are not addressed in this study. Although the effect of lockdown in Kerala State is manifested as the reduction in the concentration of most of the pollutants, the long-range transport of the pollutants may act as a limiting factor for further improvement of air quality.

Fig. 6.

Air mass trajectories reached at the air quality monitoring stations (K1, K3 and K4) of Kerala during 1–30 April 2020. The hourly endpoints indicate back trajectories for 24 -h period.

Results of this study help understand the differences in the response of the ambient air quality to short-term human interventions in less industrialized and less urbanized regions to recognize, monitor, and prioritize potential public health concerns and opportunities for action, which are beneficial to develop appropriate policy measures considering reductions in the concentration of the primary and secondary pollutants, as well as the background/residual pollution levels. Although the unprecedented restrictions on the transportation, construction and industrial sectors caused serious negative effects on the economy, the lockdown offered an incomparable opportunity to investigate the role of emissions from various sectors controlling the ambient air quality. The improvement of the ambient air quality due to the lockdown measures seems to help develop better ambient air quality management programmes, especially in the urbanized/industrialized areas, but has differing implications in the less-industrialized/less-urbanized regions.

5.1. Limitations of study

This study addressed the effect of lockdown on the ambient air quality of Kerala (India) considering the temporal variability and trends in the air quality data. This study has three major limitations. The first limitation is a focus of the analysis on the major pollutants, such as NO, NO₂, NO_x, NH₃, SO₂, CO, O₃, PM_2.5, PM₁₀, but all the stations (except K3) did not have the comprehensive data of the pollutants. The second limitation is the lack of long-term air quality data at the monitoring stations. We analysed the trends and estimated the averages of the pollutants based on the data of the previous two years (2018 and 2019). If data were available for a quite long period, however, the estimates of the average concentration of the pollutants, as well as the seasonal trends would have been more reliable. Thirdly, unavailability of meteorological data at the ambient air quality stations limited the present study to understand the role of local meteorological variables on the ambient air quality. We analysed rainfall, minimum and maximum temperature, and relative humidity of January to May (2018–2020) recorded by India Meteorological Department at Thiruvananthapuram, however, indicates hardly any significant differences in the meteorological conditions between 2020 and the previous years.

6. Summary and conclusions

In summary, examination of air pollutants in Kerala State between the pre-lockdown and lockdown periods, seasonal trends in air pollutants, along with comparison with data from more industrialized/urbanized areas, furnished the following answers to the research questions posed in this study. First, the levels of different air pollutants showed a significant reduction during the lockdown period compared to the period before lockdown. The sizeable reductions in concentrations of the pollutants, including NO₂ (-48 %) NO_x (-53 % to -90 %), CO (-24 % to -67 %) as well as the particulate matter (-24 % to -47 % for PM_2.5, -17 % to -20 % for PM₁₀), correlates with decreased emissions from transportation and industrial sources during the lockdown period. Second, a significant decreasing trend in air pollutant levels of Kerala State from January to May over three years (2018, 2019 and 2020) does imply seasonal fluctuations as a cause of the improved air quality. In 2020 affected by the lockdown measures, however, the general decreasing trend in air pollutants was intensified (manifested as the change in Mann Kendall S), or, the positive trend was reversed. Third, the effect of lockdown on the air quality of Kerala State differs from that reported for highly urbanized and industrialized regions, as well as the typical rural regions. Although the reduction of the concentration of the gaseous pollutants of Kerala is comparable with highly urbanized and industrialized regions, the reduction of particulate matter shows considerable variability among the regions. In general, the effects of lockdown were more variable and focused, in that the reduction in the concentration of air pollutants was jointly controlled by lockdown measures as well as seasonal effects. Long-range transport of pollutants originating from the Arabian Sea, as well as higher background and residual pollution, also accounted for the variability.

In conclusion, details of improvements in the ambient air quality of Kerala State can help to understand interactions between short-term human interventions and environmental quality during a global pandemic. Findings from this study provide an example from a less urbanized and less industrialized region in how to recognize, monitor, and prioritize potential public health concerns and opportunities for action, which may be beneficial for developing appropriate policy measures. Although the improvement in the ambient air quality due to the COVID-19 pandemic has been considered an experiment for developing better air quality management programmes across more polluted/urbanized/industrialized regions, this study affirms that such management measures of air quality could have divergent implications in less industrialized and less urbanized regions as well.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is Supplementary data to this article: