3.1. Validation of Aqua-AOD with Terra, MISR and SeaWiFS observations.
Recent studies have validated MODIS-AOD with SeaWiFS AOD and MISR AOD to acquire accuracy in the measurement of remote sensing data (Tariq et al., 2018). Figure 3 exhibits the comparison of monthly data of Aqua-AOD at the wavelength of 550nm with observations of Terra-AOD, SeaWiFS and MISR-AOD from 2002 to 2020 over the study region. The highest correlation of 0.97 is obtained between Aqua and Terra AOD due to the same aerosol retrieval algorithm for both sensors. The value of the y-intercept is 0.03 and the slope is 1.11 respectively. A correlation coefficient (R) of 0.93 is obtained among MISR and Aqua-AOD. The slope value of 0.97 and y-intercept of 0.01 exhibits that the MISR satellite data underestimate AOD observations as compared to Aqua. Martonchik et al. (1998) reported that the different aerosol retrieval algorithm and wavelength of MISR and Aqua-MODIS cause this underestimation. The lowest R-value of 0.58 is obtained among Aqua and SeaWiFS AOD. The slope value of 0.85 and y-intercept of 0.07 represent the underestimation of SeaWiFS AOD. This underestimation of AOD is due to the calibration of instruments (Remer et al., 2005).
3.2. Spatial distribution of AOD
Figure 4 represents the distributions of annual mean Aqua-AOD over south Asia during the period September 2002-December 2020. The mean annual AOD values (0-1) retrieved from Aqua-MODIS show significant spatial variability over south Asia. The findings from Figure 4 shows that AOD had a marked impact on countries in South Asia, namely Afghanistan, Nepal, Maldives, Bhutan, Sri Lanka, India, Bangladesh and Pakistan. This study includes South Asian countries due to their geographic locations, vehicular and industrial emissions. High AOD (~0.7) is observed over eastern Pakistan while AOD ˃0.6 exists over southeastern Pakistan due to anthropogenic emissions. This includes an industrial area of Lahore, Gujranwala, Sialkot, Faisalabad, Jacobabad, Larkana, Hyderabad, Nawabshah and Karachi. Sharif et al. (2015) also reported high AOD over the southeastern region of Pakistan. The highest mean annual AOD (˃0.7) has been observed over the IGB region. Recent studies have reported that high AOD over the IGB region is due to the increased industrialization, burning of crop residue, high population density, burning of biomass, heavy urbanization and burning of coal (Jethva et al., 2005; ul-Haq et al. 2016). In addition, low population areas in this study region also contribute to aerosol emissions from garbage and wood burning. Low AOD values (˂0.3) are observed over the northwestern region of Pakistan due to less population and reduced agricultural activity. Mean annual AOD values of ˃0.3 are observed consistently over a wide area of India. In previous studies (Washington et al., 2003), the dominance of heavy aerosol loading over the northern region and IGP of India is observed in Figure 4. Similar to the outcomes acquired by Nizar and Dodamani, (2019), a significant decline in aerosol loading is seen over the Himalayas and Tibetan Plateau as compared to the IGB. Srivastava et al. (2016) reported aerosol loading over the west coast adjoining areas of India. Thus, aerosol loading over ocean regions is due to continental aerosols and marine traffic.
Figure 5 represents the seasonal distribution of averaged Aqua-AOD at a wavelength of 550nm over south Asia from 2002 to 2020. The seasonality of AOD, despite some variations observed from area-to-area, in general, demonstrates lower AOD observed during pre-monsoon and winter season while higher AOD was seen during the period of post-monsoon and monsoon, depending upon the area. During the winter season, the highest AOD (0.8) is observed over the Bihar state of India as shown in Figure 5a. The AOD of ˃0.6 is seen over the IGB region. Similarly, the 0.5 AOD is seen in eastern and southern Pakistan. In agreement with the findings by Tariq and Ali. (2015), higher values of AOD have been observed over the southern region of Pakistan and the main sources of aerosols loading over the region includes desert aerosols transported from Cholistan and Thar deserts. It is evident from Figure 2a that the areas of northwestern and southwestern Pakistan show the minimum AOD of ˃0.2. IGB is often surrounded by haze activity during the winter season which resulted in heavy aerosol loading over the region (Badrinath et al., 2009). In the pre-monsoon, the highest AOD value of ˃0.6 is seen over the IGB region as shown in Figure 2b. The 0.5 AOD is also detected in the eastern region of Pakistan and ˃0.3 is seen over central India. Figure 2b displays that the lowest AOD of 0.2 is observed in northwestern Pakistan.
In the monsoon season, high AOD ranging from ˃0.6 to 0.9 has been observed over the IGB region and eastern Pakistan. The high AOD during monsoon over IGB and eastern region of Pakistan are due to biomass burning and emissions from vehicles. The 0.8 AOD value is also observed in the southeastern region of Pakistan. Similarly, the ˃0.6 AOD value is noticed in the eastern region of India and the southwestern region of Pakistan. It is evident from Figure 5c that the minimum AOD of ˃0.3 is seen in northwestern Pakistan and the southern region of India. In the post-monsoon season, the ˃0.7 AOD value is detected over few areas of the IGB region as shown in Figure 4d. The ~0.5 AOD value is noticed in central India. Similarly, the ˃0.3 AOD value is observed in the southwestern region of Pakistan. Finally, it is noticed from Figure 2d that the northern region of Pakistan shows a minimum AOD of 0.2. Chung et al. (2005) analyzed that anthropogenic aerosols contribute about 70% to the overall AOD over south Asia.
Seasonal distributions of AOD anomalies over south Asia from 2002 to 2020 are shown in Figure 6. During the winter season, a high anomaly value of 0.2 is observed over few areas of eastern Pakistan and western India as shown in Figure 6a. Major sources of high AOD anomaly values over these regions are due to excessive crop residue burning, urbanization and industrialization. In all the seasons, regions with low AOD anomaly values were situated in less-developed and rural regions of southwestern and northeastern Pakistan. In the pre-monsoon season, the anomaly of 0.1 is seen over the IGB region and eastern Pakistan. The most noticeable feature is that (only during monsoon) negative AOD anomaly values were seen over southeastern and eastern Pakistan and eastern India. The positive AOD anomaly values were dominated in western and southern parts of Pakistan and India respectively as shown in Figure 6c. In terms of the magnitude of AOD anomalies and area coverage, it is worth presenting that positive AOD anomaly centers were observed during winter, followed by post-monsoon, pre-monsoon, and least being monsoon. In post-monsoon, the highest AOD anomaly is observed over southeastern Pakistan and western areas of India as shown in Figure 6d.
Figure 7 represents spatial correlation maps of AOD with meteorological variables (TEMP, RH and WS) and MODIS-Aqua retrieved (EVI) over south Asia during the period September 2002-December 2020. A correlation of ˃0.26 is observed over the eastern region of Pakistan and the western region of India as shown in Figure 7a. The highest positive correlation of ˃0.54 is obtained among AOD and RH over the northeastern region of Pakistan and the central part of India while a negative correlation is observed over a southwestern region of Pakistan as shown in Figure 7b. Xueliang et al. (2012) reported that meteorological variables such as pressure, WS, RH, TEMP and wind direction change the chemical composition and path of aerosols. The increase in RH values on the western side of the Indian region accompanied by a decline in RH over eastern regions of India. The highest correlation among AOD and WS is observed over the southeastern region of Pakistan and the eastern region of India. The maximum negative correlation of -0.58 is observed among the AOD and EVI over the western region of India as shown in Figure 7d. The negative correlation shows that an increase (decrease) in AOD results in a decrease (increase) in EVI over the area. In northeastern and southeastern Pakistan, a negative R (~-0.3) among the AOD and EVI is seen due to the presence of high aerosols loading and low EVI. Tariq et al. (2021) also indicated a negative correlation among AOD and EVI over Lahore, Multan, Faisalabad, Jacobabad and Karachi. The highest positive R of 0.54 is observed over Madhya Pradesh and Maharashtra, India.
3.3. Temporal variations in AOD
Figure 8 exhibits the time series of monthly Aqua-AOD spatially averaged over the study region during the period September 2002- December 2020. The highest AOD (0.58) is found in June 2018 while the lowest AOD (0.17) is observed in December 2005. Anthropogenic emissions due to rapid urbanization, more fuel consumption and biomass burning are responsible for high aerosol loadings over south Asia. Prasad et al. (2007) reported high AOD over the IGB region in the summer as compared to other seasons. The slope value of 0.0002 represents an increasing trend of monthly mean AOD and the y-intercept is found to be 0.29. Ali et al. (2014) investigated aerosol loading over Lahore and obtained high AOD in the summer season due to high temperatures and dust activity.
Figure 9 demonstrate the monthly average AOD and their standard deviations over the study area from 2002 to 2020. The highest AOD (~0.44) is detected in June while the second-highest AOD (~0.43) is seen in July. The minimum AOD (~0.22) is obtained in December. The relative change between January and June AOD is ~76%. This depicts the impact of the high temperature during the summer months over the study region. The interannual variability of AOD is higher in the summer season than in winter as noticeable from the standard deviation of AOD as shown in Figure 9. Tariq et al. (2021) also reported high AOD over Pakistan in summer as compared to the winter season.
Table 1. Descriptive statistics of AOD over south Asian countries from September 2002 to December 2020.
Countries
|
Mean AOD
|
Maximum AOD
|
Minimum AOD
|
Pakistan
|
0.34±0.16
|
0.66
|
0.02
|
India
|
0.36±0.13
|
0.72
|
0.08
|
Bangladesh
|
0.45±0.11
|
0.57
|
0.26
|
Nepal
|
0.25±0.15
|
0.55
|
0.02
|
Bhutan
|
0.06±0.006
|
0.07
|
0.05
|
Sri Lanka
|
0.27±0.03
|
0.33
|
0.24
|
Afghanistan
|
0.20±0.08
|
0.39
|
0.06
|
The Aqua-AOD measured at 550 nm ranging from 0.02 to 0.66 with a mean value of 0.34 ± 0.16 over Pakistan, 0.08 to 0.72 with a mean of 0.36 ± 0.13 over India, 0.26 to 0.57 with a mean of 0.45±0.11 over Bangladesh, 0.02 to 0.55 with mean of 0.25 ± 0.15 over Nepal, 0.05 to 0.07 with mean of 0.06 ± 0.006 over Bhutan and 0.24 to 0.33 with mean of 0.27 ± 0.03 over Sir-Lanka as shown in Table 1. The descriptive statistics show that the aerosol loading was relatively higher over Pakistan, India and Bangladesh than Nepal, Afghanistan, Bhutan and Sri-Lanka.
Figure 10 depicts the time series of MODIS-Aqua retrieved AOD over Kabul, Lahore, Karachi, New Delhi, Bangalore, Varanasi, Kathmandu and Dhaka from September 2002 to December 2020. The slope value of -2E-06 depicts the decreasing trend of AOD over Kabul with an intercept value of 0.5915 as shown in figure 10a. The highest AOD of ~0.28 over Kabul was observed in March 2012. Figure 10b shows the increasing trend of Aqua-AOD over Lahore with a slope of 2E-05 and intercept of 0.5915 while the highest AOD of ~1.36 was observed in November 2014. The slope value of 5E-06 represents an increasing trend of AOD over Karachi as shown in Figure10c. The highest AOD of ~1.0 over Karachi was observed in August 2010. In New Delhi, the positive slope value of 0.0006 shows the increasing trend of AOD while the highest AOD (~1.4) was observed in July 2019. Figure 10d shows slightly high AOD over New Delhi throughout the study period due to the high aerosol loading over the city from the IGB region. This increase in AOD affects the climate of New Delhi badly for the last few years. Panda et al. (2009) found increasing trends of AOD over New Delhi during the pre-monsoon and winter because of dust activity and burning of biomass.
The highest AOD of ~0.51 over Bangalore was observed in May 2018 as shown in Figure 10e. The slope value of 3E-05 and intercept of 0.119 represents rising trend of AOD over Bangalore. Ramachandran et al. (2012) found an increasing trend of AOD over Hyderabad and Bangalore due to the rapid increase in urbanization. Figure 10f shows the increasing trend of AOD over Varanasi, India. The slope and y-intercept were 4E-05 and 0.4788 respectively while the maximum AOD of ~1.5 was observed in December 2019. The positive slope value of 2E-05 depicts rising trend of AOD over Katmandu with a maximum AOD of ~1.1 in June 2018 as shown in Figure 10g. Previous studies showed that the katabatic winds and topography of Katmandu cause pollutants to be trapped under inversion layers near the valley (Panday and Prinn, 2009). Figure 10h shows the increasing trend of AOD over Dhaka with the highest AOD of ~1.2 in June 2018. The y-intercept and slope from the linear model are 0.4204 and 4E-05 respectively.