Aerosol opto-physical properties: Temporal variation, aerosol type discrimination and source identification

17 Atmospheric aerosol characterization experiments were conducted, for the first time, over AC- 18 Ahmednagar, a semi-urban location in south-west India, using a multi-spectral solar radiometer 19 from January 2016 to May 2018. The MODIS/Terra retrieved Level-2 daily swath AOD 550 nm 20 data sets during 2011-2021 were also used to infer long-term behaviour of columnar aerosols at 21 AC-Ahmednagar. The daily-averaged, Microtops II Sun photometer measured AOD 500 nm and 22 AOD 1020 nm and MODIS retrieved AOD 550 nm reveal a discernible wide spectrum of variability in 23 their magnitudes. Magnitude-wise, AODs at both the wavelengths depict an increasing trend of 24 about 0.05 year –1 (AOD 500 nm ) and 0.03 year –1 (AOD analysis for AOD performed on a seasonal basis reveals the presence of possible contribution of 31 long-range transported plumes of varied aerosols on different scales and types from a variety of 32 emission sources from different regions. present investigates the prevalent column-integrated aerosol properties using AOD retrieved from the Microtops II photometer measurements for time (ii) To study the spectral variability of II Sun AOD, , to judge different at site. (iii) Seasonal discrimination of dominant aerosol-types based on the constructed contour density map by the 2-dimensional binning of AOD 500 nm versus AE 440-870 nm (iv) Comparison/evaluation of MODIS/Terra algorithm retrieved AOD 550 nm (MOD04_3K_C6.1 (collection 6.1)) against Microtops II Sun photometer measured

regression technique and the statistical metrics of each Langley plot are listed in Table S1.It is seen that at all the wavelength channels the magnitude of lnV 0 is found to be significantly less than 1% which highlights the stability of the instrument for AOD measurements (Table S1).

(B)
RESULTS AND DISCUSSION (i) Spectral variation of AOD Spectral AOD is an essential parameter in determining the extent to which aerosols directly perturb the Earth-atmosphere radiation budget.It is found to depend on their columnar distribution, size spectrum, and refractive index (Pawar et al., 2012;Varpe et al., 2018).Fig. S1 shows the average AOD spectra over the site AC-Ahmednagar in winter and pre-monsoon seasons.As seen in Fig. S1, the length of vertical bars shows larger to smaller variation in AODs from shorter to longer wavelengths respectively during both seasons.The relatively strong wavelength dependence of AOD is evident in Fig. S1, which illustrates that AOD is higher at shorter wavelengths (λ < 0.6 μm).AOD gradually decreases towards longer wavelengths (λ > 0.6 μm) despite seasonal change attributed to the dominance of the fine-mode aerosols over coarsemode aerosols at the observing site.Moreover, the spectral dependence of AOD is found to be decay type indicating the possibility of a similar kind of aerosol particle size distribution albeit variation in their columnar content based on AOD magnitude at each wavelength (More et al., 2013).According to Mie scattering theory (Quenzel et al., 1970), the occurrence of a higher columnar concentration of fine-mode aerosols in the atmosphere over the observing site cause wavelength-selective scattering of the solar irradiance producing higher AODs at the shorter wavelengths.Similarly, the coarse-mode aerosols provide likewise contributions to the AOD at both wavelengths (Schuster et al., 2006).In the multi-wavelength radiometer (MWR) studies over Dehradun (30°00' N -30°30' N and 78°18' E -78°36' E), Rana et al. (2009)  The study also reveals more AOD decrease at higher wavelengths as a consequence of the reduction in coarse-mode aerosols in the winter season while the AOD amplitude at longer wavelengths is seen to be higher during pre-monsoon season showing the substantial contribution of coarse-mode aerosols over fine-mode aerosols.The occurrence of high AODs at larger wavelengths during pre-monsoon season could be assigned to the dominance of coarsemode particles due to prominent surface wind activity which causes lifting-up of the soil-dust from the dry surface into the atmosphere and the long-range transported mineral-dust from arid regions (Aher et al., 2014;Pawar et al., 2015).During the winter season, the frequency of the strong nocturnal inversions is normally high (Vernekar et al., 1993).Aerosols produced due to anthropogenic activities (e.g., vehicular emissions, domestic cooking, industrial emissions, etc.) are released into the surface layers and get trapped in the ABL due to less atmospheric ventilation (Kolhe et al., 2018).Further, the concentration of the soil dust aerosols is less during winter on account of calm weather conditions.As a result of this, AODs are comparatively higher at shorter wavelengths during winter as compared to those observed in the pre-monsoon season (Dani et al., 2012).equator crossing time of about 10:30/13:30 hr (IST).Recently, many researchers have performed clear and extensive studies on the validation of MODIS aerosol products (Chu et al., 2002;Ichokuet al., 2002;Levy et al., 2005Levy et al., , 2010;;Bennouna et al., 2011;Bibi et al., 2015;Ma et. al., 2016;Vijayakumar et al., 2016;Ali et al., 2017;Wang et al., 2017aWang et al., , 2017b;;Boiyo et al., 2017;Gupta et al., 2018;Tian and Gao, 2019;Sharma et al., 2021) nm has calculated by using the Ångström empirical formula (Eq.S1) where, AE is the Ångström exponent estimated by employing linear regression technique to spectral AOD curve in the 440-870 nm spectral range (Pawar et al., 2015).The validation process consists of plotting of the collocated MODIS/Terra and Microtops II AODs on a 2-D scatter diagram and performing linear regression fit to the plotted data (More et al., 2013;Sayer et al., 2014;Wang et al., 2017aWang et al., , 2017b;;Boiyo et al., 2017;Sharma et al., 2021).Here, the scatter diagram of the Microtops II calculated AOD 550 nm against MODIS/Terra Dark Target-Deep Blue combined (DT-DB) retrieved AOD550 nm is constructed and is shown in Fig. S3.The expected error (EE) envelope, determined from the MODIS/Terra 3-km DT/DB combined product over Land and Ocean, representing the percentages of the AOD retrievals falling within/above/below (PWE/PAE/PBE) the custom error range, is determined by employing Eq. ( S2), The regression line is represented by the red solid line, the blue and green solid lines are above and below EE bounds respectively, and the black dashed line indicates the X = Y line.The estimated y-intercepts and slopes for linear regression analysis are given in Eq. ( S3), In this study, the following statistical parameters are used to quantitatively evaluate the consistency and uncertainty that may be prevalent in the Microtops-II and MODIS/Terra AOD retrievals: the Pearson's correlation coefficient (r), the coefficient of determination (R 2 ), the mean absolute/absolute percentage error (MAE, Eq. ( S4)), the root-mean-square error (RMSE, Eq. ( S5)), and the relative mean bias (RMB, Eq. ( S6)).
Significant overestimation in MODIS/Terra AOD retrievals in the presence of low and high aerosol loadings may be produced due to the errors in instrument calibration, surface reflectance assumptions, and aerosol models used for aerosol inversion algorithm (More et al., 2013 Beegum et al. (2008) have identified the impact of long-range advection of aerosols from the adjoining areas resulting in the significant variations in the AOD magnitudes and their wavelength dependence measured during March to May 2006 in the Integrated Campaigns for Aerosols, gases, and Radiation Budget (ICARB).Also, picking of dust aerosols by the prominent convective eddies caused by the strong heating of the land surface (a meso-scale mechanism) could be the plausible reason behind the modulation of the observed AOD and its spectral variability(Moorthy et al., 2007).(ii)Comparison of Microtops II Sunphotometer measured and MODIS retrievedAODsThe comparison of the ground-based AOD measurements with those retrieved from the satellite-based observations and thereby the assessment of satellite aerosol product retrievals has major theoretical as well as practical implications in the up-gradation of the retrieval algorithms and in climate research.The MODerate resolution Imaging Spectroradiometer (MODIS) onboard Terra/Aqua satellite provides spectral information of aerosol parameters in the mid-visible wavelength region.Terra/Aqua satellites are operated at 705 km altitude with an . It is crucial for the individual researcher to choose MODIS retrieved aerosol products comprising of the realistic results based on the quantitative analysis of aerosol data and the comparison of different MODIS retrieved aerosol products.For comparison, MODIS retrieved AODs require to be spatio-temporally collocated with Microtops II AOD measurements.The collocation is needed since the MODIS sensor scan encompasses spatial coverage of the study area from the space whereas Microtops II is a ground-based point measurement.The objective of the spatiotemporal collocation process is to co-locate the air mass perceived by the MODIS sensor from the space over the study location, which is concurrently observed by the ground-based Microtops II sensor.In the present study, the comparison of the Terra platform MODIS sensor DT-DB combined algorithm retrieved AOD (MOD04_3K_C6.1 (collection 6.1) retrieved from https://ladsweb.modaps.eosdis.nasa.gov/) is attempted by comparing it with ground-based Microtops II Sun photometer derived AOD data at the same wavelength.For achieving the realistic comparison between Microtops II and MODIS AODs, AODs for Microtops II at 550

.
Fig. S1.The Langley plots, a plot of signal voltage lnV(λ) against relative optical air mass, at 440, 500, 675, 870, and 1020 nm wavelengths displaying linear regression equations obtained at each wavelength and the respective Pearson's correlation coefficients (r).

Fig
Fig. S3.2-D scatter diagram of the collocated daily MODIS AOD 500 nm (calculated) against Microtops-II Sunphotometer measured AOD 500 nm at site AC-Ahmednagar for the period January 2016-May 2018.

Table S1 .
; Sharma Values of the factory generated and observed calibration constants (lnV0) at each Microtops II channel with corresponding % deviation.