Inuence of COVID-19 Lockdown on Aerosol Optical Depth over Pokhara and Kyanjin Gompa in Nepal

The outbreak of the COVID-19 pandemic and the subsequent global economic shutdown provided an opportunity to conduct a real-time experiment assessing the inuence of global emission reductions in the Aerosol Optical Depth (AOD) level, an indicator of air pollution over Nepal. Nepal's government imposed a lockdown on the country for approximately three months (from 24 March onwards) in 2020. The purpose of this study is to examine the temporal uctuation in Aerosol Optical Depth (AOD) caused by the COVID-19 shutdown by comparing its value during the same time period of the past year over two sites: Pokhara and Kyanjin Gompa. We comparatively analyzed the variation of diurnal mean and monthly average AOD of two selected sites, from the month of January to May 2020 and January to May 2018. By examining the time-series graph of daily average AOD prior to and during the lockdown period, our study showed an apparent uctuation in AOD throughout the studied areas. The major ndings of the research revealed that after the lockdown, a signicant variation in monthly averaged AOD was observed, ranging from 20–60% deviation over Pokhara and 25–50% deviation in Kyanjin Gompa at different wavelengths. This conrms previous studies on aerosols and other particulate matter during COVID-19 lockdown, as well as theoretical assumptions. In addition, we performed the heatmap correlation analysis among AOD, Total precipitable water (Tpw), Angstrom exponent (α), Turbidity coecient (β), and Visibility (V) during the studied period with possible explanations. We believe this research work serves as a crucial reference for our government to implement appropriate policies for pollution control over the studied areas in the future.


Introduction
The minute solid particles or liquid droplets suspended in the atmosphere are known as an aerosol.
Atmospheric aerosols are mainly produced by the mechanical disintegration processes occurring over land (e.g., lift of dust) and ocean (e.g., sea-spray) and by chemical reactions occurring in the atmosphere [1,2]. Natural aerosols such as dust, y ash, soot, pollen grains, and carbon particle occur naturally as they mix in the atmosphere by different natural phenomena, thereby covering the majority of the aerosol in the atmosphere, on the other hand, anthropogenic aerosols are located considerably in a polluted area due to human activities, including fossil fuel combustion, biomass burning, industrial smoke, power plants, coal mining, and other several activities, overlaying approximately 10% of the entire aerosols [3].
These different aerosol groups often clump together to form a hybrid of both natural and anthropogenic. Aerosols can make their way into the atmosphere almost everywhere globally, depending on the season and weather conditions [2].
Aerosol scatters and absorb sunlight (size range 0.1-1 micrometres are most effective) [4,5] and change the cloud's characteristics [6]. They directly affect limiting visibility and causing a risk to human health, including premature deaths, allergies, respiratory problems such as In uenza, Pneumonia, and harmful cardiovascular effects such as heart attacks and strokes [3,7,8]. On examining the long-term records of aerosol particles and lung cancer incidence, Tie et al. [9] reported that the mortality due to lung cancer is closely correlated to the levels of aerosol particles present in the atmosphere near the surface. High Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js aerosol pollution causes wide-ranging consequences for human health, natural ecosystems, visibility, weather, radiative forcing, and tropospheric oxidation (self-cleaning) capacity [10]. In addition, aerosols in uence the melting of snow and ice in the Himalayas and around the Tibetan Plateau, which ultimately plays a crucial role in climate change [11].
Aerosol Optical Depth (AOD) is an optical measure of light extinction (absorption and scattering) of the solar beam due to the number of total aerosols in the entire atmospheric column and provides an indirect measure of aerosols in the atmosphere [3]. As aerosol deposition is continually rising due to industrialization and urbanization, high population density, or biomass burning in different regions of Nepal, aerosols studies, particularly AOD, has risen rapidly in recent years [12]. In the past, several works (e.g., Acharya et al. [3], Bhattarai [20], over the different regions of the globe; India, Spain, Malaysia, Europe, USA, China, has claimed that lockdown had contributed signi cantly in improving air quality due to decrement in emission. Ranjan et al. [21] aimed to study the effect of the Covid-19 pandemic on aerosol optical depth, concluding a huge reduction in AOD, about 6% to 45 %, over the different regions of India. Acharya et al.
[18] revealed that lockdown reduced AOD by ~ 20% at maximum places in Europe and USA. Raza et al. [22] reported that AOD was reduced in Pakistan during the lockdown period in 2020 in comparison to the previous years. From the worldwide study of AOD during the pandemic period, Sanap [23] recorded a decrease in AOD from mid-March to April 2020; however, a drastic reduction in AOD was observed in May.
Furthermore, despite many impacts of lockdown on people's social life and the national economy, several ndings [e.g., Shrestha et al. [24], Gautam et al. [25], Baral & Thapa [26] observed that lockdown had temporarily better environmental conditions in Nepal.
In this particular paper, we present results on the implications of observed diurnal and monthly averaged variation of aerosol optical depth due to anthropogenic aerosols over the two sites of Nepal by dividing the study into two scenarios, i.e., before the lockdown (1 January 2020-23 March 2020) and during the lockdown (24 March 2020-31 May 2020). We emphasized the variation of anthropogenic aerosol pollution in the most polluted and the least polluted regions. This work is also based on the observational ndings utilizing Total Precipitable Water (Tpw) and visibility correlation with their respective AOD wavelength during the study period. This paper is organized as follows: in Sect. 2, we describe the methodology, in Sect. 3, we discuss the results, and nally, we conclude our study in Sect. 4. The required data for this research work has been taken from AERONET provided by GSFC NASA (https://aeronet.gsfc.nasa.gov/). The Aerosol Robotic Network (AERONET), a ground-based remote sensing aerosol network of well-calibrated Sun/sky radiometers established in the early 1990s, is a wellestablished and productive facility for passive aerosol measurement [27]. The study site was chosen based on available data for Nepal in order to maximize the use of available databases. Daily averages of all AOD data points were calculated and then examined on a monthly average basis. Figure 1 shows a network of the stations in the map of Nepal used in this study for AOD observation.

Site Selection
More detailed information about the aeronet sites is shown in table 1.

Instrumentation
The CIMEL Electronic CE-318 sun/sky radiometer, which is part of the AERONET worldwide network, was used for all of the observations reported in this paper. This sensor operates by measuring the intensity of solar radiation at any speci ed wavelength and converting it to optical depth using the corresponding concentrations at the top of the atmosphere [27]. More information about the AERONET equipment utilized for optical depth retrieval can be found in Schmid's work [28]. The AERONET archive is classi ed into three quality levels: raw data at level 1.0, cloud-screened data at level 1.5, and quality-assured data at level 2.0. The AERONET data used in this work are version 3.0 at level 1.5, which includes cloud exclusion and quality controls, but may lack nal calibration.
In this study, AODs of corresponding wavelengths 1640 nm, 1020 nm, 870 nm, 675 nm and 440 nm have been used to analyze diurnal and monthly time series analysis. The spectral AOD () data obtained from the AERONET using the wavelength of 440 -870 nm is used to calculate the different optical properties of the aerosols as wavelength exponent (α), turbidity coe cient (β), visibility (V) using Angstrom's turbidity law, given by Iqbal et al. [29]: See formula 1 in the supplementary les.
Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js β represents the amount of aerosol presents in the air in the vertical column of the atmosphere, while α is related to the size distribution of aerosol in the atmosphere, having the value ranging from 0 to 4. We calculate α from the wavelength of 440 -870 nm by using the formula provided by Sayer et al. [30]: See formula 2 in the supplementary les.
where r1 and r2 are the AOD at wavelengths λ 1 and λ 2 .
The higher value of α indicates the higher percentage of smaller sized aerosol particles in the air and vice versa [31,32].
Additionally, visibility was investigated, which is de ned as the degree of clarity or the farthest distance through the atmosphere to the horizon at which prominent objects may be distinguished with the naked eye. Its value in km can be computed by using the following formula [29].
See formula 3 in the supplementary les.

Result And Discussion
Based on the methodology and datasets discussed above, AOD data were analyzed on two sites: Kyanjin The above gures illustrate that both studied sites had the most aerosol loading at wavelength 440 nm with the variant of time followed by other wavelengths in increasing order. This means that the submicron aerosol particle is dominant in this vicinity than that of coarse mode aerosol. The dominance of shorter wavelength AOD was also reported in different literatures [7,33,34]. Minimum aerosols loading in the atmosphere was observed during the month of January, which may be due to the weak generation mechanisms as well as the colder ground surface. The higher AOD over Pokhara (above 0.3 for most cases) signi es that Pokhara and the surrounding around the valley are heavily polluted [35,36]. On the other hand, the AOD was recorded below 0.4 for all studied periods over the site Kyanjin_Gompa.  The comparative analysis of these two sites, with the same period of 2018, depicts that AOD was decreased drastically in the month of March, April, and May in both stations, which clearly shows the effects due to lockdown. Overall, the monthly deviation of AOD (at different wavelengths) during the lockdown period ranges between ~ 25-50% for Kyanjin_Gompa and ~ 20-60% for Pokhara, with a maximum decrement of AOD in the month of March. This result is consistent with the global study of AOD during the lockdown period by Sanap [23]. However, comparatively less deviation in AOD was observed in the month of April and May. The decrement in AOD in our study regions during lockdown is more than in the different areas of India (6 to 45% as recorded by Ranjan et al. [21]) and Europe (~ 20% observed by Acharya et al. [18]). During the lockdown, the production of the anthropogenic aerosol has decreased, which resulted in the lowering of the monthly average AOD, thus enhancing the fact that human activities that help in the production of anthropogenic aerosols are signi cant factors affecting AOD. weather and climate change for studying the association between two parameters, revealing physics behind them, and forecasting its impact on the local and global scale.

Correlation Analysis: Heat map
In this study, we examined the relationship among aerosol parameters. Results in Fig. 6 represent correlation analysis for AOD at wavelengths 870 nm, 440 nm, Total Precipitable Water (Tpw), Angstrom Exponent (α), Turbidity Coe cient (β) through heat map. Here, we compared the correlation analysis of different parameters during January-May of 2020, prior and after lockdown with January-May of 2018 to investigate the effect of lockdown on their correlation coe cient value at Aeronet sites viz. Kyanjin_Gompa and Pokhara. Each square shows the correlation coe cients between the different variables on each axis. The correlation coe cient is scaled from + 1 to -1, and the value close to 0 indicates there is less or no linear trend between the two variables. The value closer to ± 1 indicates the strong correlation between them. The gure shows that the principal diagonal coe cients are all 1 (dark red) because those squares correlate each variable to itself, so it is a perfect correlation. A consistent reduction in correlation coe cient was observed among the observational data from 25 March to 31 May It is worth mentioning that the sensitivity of the Pearson coe cients depends upon the chosen time frame. Correlations can be highly sensitive to the time frame of observation [44]. In particular, when short time frames with few data are analyzed, the correlation parameters may be misleading. Therefore, researchers should carefully scrutinize different correlation regimes before drawing any general conclusions. Principally, the more data analyzed and the longer the time frame, the more meaningful the Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js correlation results. Therefore, our interest is to compare the long-term mean AOD level of the same periods in the near future.

Conclusion
This study aims to investigate the aerosol records, AOD variability, from two stations, including one less and one high polluted region of Nepal prior to and after the lockdown period. From the diurnal & monthly variation, and heatmap correlation analysis, we have concluded the following results.
Studies had shown that the time series of AOD for urban city Pokhara had a comparatively large value than the less polluted site Kyanjin_Gompa. By investigating the diurnal mean AOD of Pokhara and Kyanjin_Gompa, it was observed that the maximum AOD values at different wavelengths were recorded up to 1.55 at Pokhara and 0.13 at Kyanjin_Gompa before lockdown. However, there was a signi cant decrement in AOD values during the lockdown, which is visible from the comparison of AOD with the same period of the past year.
During the studied period over Pokhara and Kyanjin_Gompa, monthly variation in Aerosol Optical Depth showed a drastic fall in AOD compared to the past year. Decrement in AOD was recorded up to 20-60% for Pokhara and 25-50% for Kyanjin_Gompa. The maximum reduction occurred in March, and a slightly less drop in AOD was observed in other months.
Restrictions in emissions of various anthropogenic aerosols due to the shutdown of the novel coronavirus pandemic could be the possible reason behind this improvement on aerosols. Other various studies of AOD also supported our results during Covid-19 lockdown at different parts of the globe.
Correlation analysis among AOD, water precipitation (Tpw), angstrom exponent (α), turbidity coe cient (β), and visibility was performed during the study period. We found quite less value of correlation coe cient for short term analysis; as climate data follows a periodic cycle in the long term and due to the effect of seasonal variation, a higher value of correlation coe cient can be expected for long-term aerosols study. However, our study shows a greater association of AOD with visibility and water precipitation at the least polluted Kyanjin_Gompa site than at Pokhara.
Considering the threat to the environment and health of the population, the control technologies on the production and emission of a large number of aerosols must be adopted in the polluted regions. We believe that the study of such a measure of aerosols helps well enough to build a solid understanding and a correct illustration of the essential aerosol-cloud-precipitation methods that drive precipitation and the Earth's radiative balance. This could successfully incorporate the understanding of aerosols into regional and global climate models.