Satellite-based monitoring of aerosols is useful to understand their spatial and temporal variability. Aerosol distribution shows large variations on daily, seasonal and inter-annual scales specifically in a country like India, owing to the short lifetimes, heterogeneity in the sources, and dependency of aerosols on the meteorological conditions. In addition, the Indian sub-continent exhibits large variability in vegetation cover and topography that might affect the aerosol distribution. Further, the aerosol retrievals from different satellite sensors may vary at regional and seasonal scales because of differences and uncertainties in calibration, sampling, cloud screening, treatment of the surface reflectivity and aerosol microphysical properties. To this end, a combination of satellite measurements can be used to examine the changing levels of aerosols with a greater reliability.

In this paper, we have analyzed the trends of Aerosol Optical Depth (AOD) over the Indian region during pre-monsoon season using datasets from two satellite sensors, i.e., Moderate Resolution Imaging Spectroradiometer (MODIS) onboard Aqua and Ozone Monitoring Instrument (OMI) onboard Aura platforms. We have used the combined Dark Target Deep Blue (DTDB) AOD estimates at 550 nm from MODIS; and the Ultra-Violet (UV) algorithm (OMAERUV) AOD retrievals at 500 nm from OMI. Though the standard quality assured Level 3 data products have been utilized from both the sensors, however, those might be affected by the uncertainties due to issues in the instrument calibration, data sampling, cloud screening and algorithm accuracy. Following this, it can be noted that the expected error (EE) of AOD from MODIS over land is within ± (0.05 + 0.15 * AOD) for Dark Target retrievals and ± (0.03 + 0.20 * AOD) for Deep Blue algorithm; while OMI AOD estimates are associated with retrieval uncertainty corresponding to (largest of) 0.1 or 30% (Anh et al., 2014; Bilal & Nichol, 2017; Kaufman et al., 1997; Remer et al., 2005; Torres et al., 2007, 2013; Wei et al., 2019). The AOD variability is studied for the last 17 years (2005–2021), with specific reference to three zones, i.e., Zone 1 (lying in Indo-Gangetic Plains (IGP)), Zone 2 (lying in Desert region) and Zone 3 (lying in South Indian region) (Fig. 1a). Zone 1 lies in one of the major hotspots of increasing atmospheric pollution due to rapid urbanization/industrialization and growing energy demands. This region demonstrates a significant variability in aerosol loading based on the complex combination of anthropogenic factors (biomass burning, vehicular emissions and industrial activities) mixed with the contribution from the natural sources, particularly in the pre-monsoon season when there is frequent transportation of mineral dust from the Thar Desert region (Mehta, 2015). Selection of Zone 2 is based on the dominance of natural aerosols as it lies in the desert dust dominated region in the Indian sub-continent. During the pre-monsoon period, there is transport of mineral dust from the Thar Desert and the Middle-East to the northern and north-western India by south-westerly summer winds (Dey et al., 2004). Over Zone 3, cleaner air is generally expected when compared with Zone 1, but rapid industrialization and urbanization is leading to large amount of anthropogenic aerosols in this region. In addition, higher temperatures during pre-monsoon months play an important role in heating and lifting the loose soil with association of wind speed.

Fig. 1
figure 1

A Study area with three highlighted zones; B averaged AOD variation in (a) March (b) April (c) May as seen through MODIS, same for (a′), (b′) and (c′) but from OMI; C Correlation between MODIS and OMI AOD datasets for Indian region over the years 2005–2021 during (a) MAM (b) March (c) April (d) May; D Time series of Seasonal (MAM) mean AOD over Indian landmass during 2005–2021 using AOD from (a) MODIS and (b) OMI; E Time series of Seasonal (MAM) mean AOD over (a) Zone 1 (b) Zone 2 (c) Zone 3 during 2005–2021 using AOD from MODIS and OMI; for D and E, trends where ever statistically significant have been indicated in the graphs, along with the level of significance

Full size image

Spatial variation of averaged AOD (2005–2021) over the Indian region using MODIS and OMI data during March, April and May months is shown in Fig. 1b. It can be seen that during pre-monsoon season, there is a high aerosol loading over the northern part of the county due to meteorological factors explained earlier. North-east region, however, shows high aerosol loading which might be attributed to the frequent forest fires during the pre-monsoon months along with the cloud contamination due to cloud cover which mostly prevail over the region during pre-monsoons (Pathak et al., 2012; Sahu et al., 2015). Southern peninsular region shows low aerosol loadings during the pre-monsoon season. A general trend of increasing AOD from March to May can be attributed to the fact that as the heat increases from March to May, lifting of loose soil owing to strong winds becomes more prominent along with the mixing of aerosols from anthropogenic activities (Mehta, 2015). The values of mean AOD (2005–2021) over the Indian region from (MODIS, OMI) is found to be (0.40, 35) during pre-monsoon period with the monthly averages of (0.37, 0.30) for March, (0.39, 0.37) for April and (0.43, 0.37) for May, respectively. Figure 1c shows the scatter plot for spatial correlation between the AOD values from MODIS and OMI. The values of correlation (R) were found to be 0.70, 0.55 and 0.64 for the months of March, April and May, respectively. It is worth noting here that highly correlated pixels were found to be associated with low surface reflectivity areas and vice versa.

The linear trend in AOD over the considered time frame was observed in two equal halves of 8 years each, one between 2005–2012 and the other between 2013–2021 (excluding 2020) to specifically see how the AOD trends have changed during recent years as compared to 2005–2012. Year 2020 was excluded as there was a nationwide lockdown over the Indian region [during March, April and May (MAM)] due to Covid-19 and there was a sudden halt of anthropogenic activities. This low AOD loading over the Indian region observed in 2020 might affect the trends and hence, needs to be studied as a special case (Mehta et al., 2021). Time series of the pre-monsoonal mean AOD using both MODIS and OMI data over Indian region is shown in Fig. 1d. The trends were tested for statistical significance using the Student’s t-test following the approach by Weatherhead et al. (1998). An increasing trend is observed from data of both the sensors in both the halves (2005–2012 and 2013–2021). Increasing trend from MODIS is much more prominent than the trend from OMI. Also, increasing trend in the second half (2013–2021) is more than that of first half (2005–2012) for both MODIS and OMI, and the trend is statistically significant. Trend from MODIS is 0.011 and 0.014 year−1 for first and second half, respectively, whereas trend from OMI is 0.002 and 0.006 year−1. Further, time series of the seasonal (MAM) mean AOD using both MODIS and OMI data over the three zones is shown in Fig. 1e. For Zone 1, higher positive AOD trends are found in the latter half of the time period compared to the first half from both MODIS and OMI. In Zone 2, though the overall AOD trends for the entire time frame are negative, the nature of trend has changed from negative to positive in the latter half. The negative AOD trends in the period 2005–2012 are in line with some previous studies done on desert dust loading during pre-monsoon season associated with the increasing rainfall (Pandey et al., 2017). However, in the second half (2013–2021), before the year 2019, there was a regular increase in the AOD values followed by a dip in year 2019 and then again an increase in 2021. In this context, some other studies (Ross et al., 2018; Sarkar et al., 2019; Sharma and Majumdar, 2017) have indicated that there has been an increase in the frequency of meteorological droughts, heat waves, dust storms and maximum surface temperatures over India, especially during the recent years. As a large fraction of aerosols over zone 2 is due to transported dust from the Arabian Sea (AS) (Gautam et al., 2010); changes in wind speed over AS might also influence the changes in the aerosol loading over Zone 2. Interestingly, there has been a recent reversal in the global terrestrial stilling of winds (Zeng et al., 2019); especially over Asia, contrary to the findings in the previous decades. These factors might have been responsible for the recent change in the nature of aerosol trend over the western Indian region; though exact reason could call for in-depth analysis of changes in natural and anthropogenic emissions; meteorological conditions and transport mechanism. On the other hand, trends are positive over both periods of 2005–2012 and 2013–2021 from both MODIS and OMI data over Zone 3.

We found that over the entire Indian region, in the last eight years, the trends have been increasing at a faster pace than the previous decade. In fact, for the desert region, where there had been a fall in the AOD values during the past years, recent years have witnessed an increasing trend. This increasing trend over the Indian mainland at a faster pace could be a matter of worry given the anthropogenic load along with the associated population of the country and its ever increasing demands. The lockdown period has set a classic example of how the aerosol burden can be reduced in light of decreasing emissions and could help the policy makers provide mitigation solutions in the coming years to control the increasing aerosol load.