The base stations (Fig. 1), from where the aircraft operations were carried out, represented distinct regions of the IGP. Jodhpur (JDR; 26.25∘ N, 73.04∘ E) in the western IGP is an arid/semiarid region with low urban activities, lying downwind of the Great Indian Desert to its west (JDR has a population density of 161 km-2). Varanasi (VNS; 25.44∘ N, 82.85∘ E) in the central IGP is located downwind of Jodhpur, characterized by extensive anthropogenic activities (automobiles, small and large-scale industries and thermal power plants, and widespread agricultural activities) by its dense population (density 2399 km-2). Bhubaneswar (BBR; 20.25∘ N, 85.81∘ E) is an urban location in the eastern IGP (population density of 2131 km-2) and experiences the influence of the marine aerosol component from the Bay of Bengal (∼50 km away from the base station) in addition to the influence of IGP outflow and local aerosol sources from nearby thermal (coal-based) power plants, mining and fertilizer-based industries, etc. (Panda and Das, 2016). The northwestern part of India has an undulating topography, due to which monsoon currents lose moisture while crossing the western mountain ranges (Aravalli) and resulting in dry arid regions (Moorthy et al., 2007). Strong dust-raising winds are a common feature of the IGP in general and its western parts in particular from April to July (Banerjee et al., 2019). In the central IGP, VNS and its environs hold largely even topography, where the Ganges is the principal river. In the eastern IGP, BBR is topographically made up of western uplands and eastern lowlands, with hillocks in the western and northern parts. These base stations thus provided a west–east cross section of the highly aerosol-laden IGP, where the aerosol characteristics are known to change longitudinally. The spatial map of aerosol optical depth (AOD) at 550 nm (Fig. 1) clearly shows the existence of higher aerosol loading (AOD >0.5) over the observational site during the study period.

Figure 2a shows the actual dates of onset of the monsoon at different parts of India in 2016. As can be seen from the figure, despite a delayed onset at the southern tip of India, the monsoon advanced fast into the central/northern parts of India. Yet, all the flights from the respective base stations were completed ahead of the start of the monsoon to that station. At the eastern IGP, the aircraft flights were made from BBR before the onset of monsoon over India; at VNS, the flights were conducted while monsoon advanced only to the central peninsula. The final set of flights was conducted at JDR when the monsoon covered most of the central and eastern part of India, but had yet to progress towards the northwestern parts.

From each of the base stations, four to five flights were carried out on successive days in different horizontal directions about the station, as shown by the ground projections (horizontal lines in Fig. 2a), with a view to obtain an average subregional representation in the shortest time possible. During each of the flights, measurements were made at six discrete levels following a staircase configuration as shown in Fig. 2 (for BBR, VNS and JDR, respectively). Accordingly, the aircraft initially climbed to the base/ceiling altitude, stabilized and made a horizontal flight along the projected track for about 30 min before climbing up (descending down) to the next higher (lower) levels and stabilizing. This procedure was repeated for all levels (∼0.5, 1, 1.5, 2, 2.5 and 3 km a.g.l.). The ceiling altitude was restricted to 3.5 km based on the unpressurized mode of operation of the aircraft. All the flights were carried out around midday since thorough vertical mixing is established by the daytime convective boundary layer eddies.

The measurements were carried out aboard the instrumented aircraft (Beechcraft 200) fitted with an iso-kinetic inlet, mounted (front facing) at the bottom of the fuselage for aspirating ambient aerosols as detailed in earlier papers (Babu et al., 2016; Vaishya et al., 2018; Gogoi et al., 2019). A constant volumetric flow of 70 L min-1 was maintained using an external pump connected to the main inlet assembly, which provided iso-kinetic flow for the average speed of 300 km h-1 maintained by the aircraft during the entire campaign. The efficiency of this inlet system has already been proven in several previous campaigns (Babu et al., 2016; Nair et al., 2016; Gogoi et al., 2019).

The meteorological conditions across the IGP during the campaign period were generally hot (surface temperature, T∼34.7±2.8 ∘C at JDR, 39±1.9 ∘C at VNS and 32.8±3.6 ∘C at BBR at the time of flight takeoff), with low to moderate relative humidity (RH) at JDR (RH ∼40 %) and VNS (RH ∼60 %). The value of RH at BBR was relatively higher (as high as 80 %) associated with its coastal proximity, in addition to the influence of mild pre-monsoon rainfall during the first (1 June 2016; light rain during noon), third (3 June 2016; heavy rain ∼60 mm in the night) and fourth (4 June 2016; light rain in the morning and during noon) days of observations. The records of T and RH were obtained from the sensors on board the aircraft, while the rainfall data were obtained from the airport meteorological department at BBR.

Supporting data used in this study include aerosol backscattering coefficients and depolarization ratio measured by the Cloud-Aerosol Transport System (CATS) aboard the International Space Station (ISS). The CATS is comprised of an elastic backscatter lidar consisting of two high-repetition-rate (4–5 kHz), low-energy (1–2 mJ) Nd:YVO4 lasers operating at three wavelengths (1064, 532 and 355 nm). The receiver subsystem consists of a 60 cm telescope having a 110 µrad field of view, photon-counting detectors and associated control electronics (Yorks et al., 2014, 2016). As the altitude of ISS orbit is about 405 km (51∘ inclination), CATS provides a comprehensive coverage of the tropics and midlatitudes, with nearly a 3 d repeat cycle. Level 2 data of CATS (https://cats.gsfc.nasa.gov/data/, last access: 9 July 2020) are used (Lee et al., 2019) in the present study, which provide the geophysical parameters, such as the vertical feature mask, profiles of cloud and aerosol properties (i.e., extinction, particle backscatter), and layer-integrated parameters (i.e., lidar ratio, optical depth). In addition, types of aerosols are also derived based on CATS aerosol typing algorithms where eight aerosol types (in CATS mode 7.1) are identified: volcanic, dust, dust mixture, clean/background, polluted marine, marine, polluted continental and smoke. Incorporating the information of backscatter color ratio (1064/532 nm) and spectral depolarization (ratio of perpendicular to parallel backscatter) ratio (1064/532 nm), CATS operating mode 7.1 provides the characteristics of aerosol regimes (Yorks et al., 2016) as given in Table 1.

Aerosol number size distributions [d(N)/d(log⁡Dp)], representative of each of the three subregions of IGP are presented in Fig. 3; the panels from left to right represent the subregions JDR, VNS and BBR, from the west to east IGP. Three distributions are shown for each station, representative of (i) near the surface with proximity to emission sources (600 m a.g.l.), (ii) in the upper atmospheric boundary layer (ABL, 2000 m a.g.l.) and (iii) in the free troposphere (3100 m a.g.l.) following the mean ABL heights (1.3±0.5, 2.3±0.5 and 1.4±0.2 km for JDR, VNS and BBR respectively; Vaishya et al., 2018) at local noon time. Aerosol number concentrations below 0.542 µm are not size-classified and represented as a single count (between 0.3 and 0.542 µm); they are shown as a function of altitude in Fig. 4a. The figures clearly reveal that at all altitudes above different stations, the size distributions are consistently bimodal, with a prominent accumulation mode (<1 µm) and a weaker secondary mode (>1 µm). The concentration of particles in the unclassified size regime (below 0.542 µm) showed a gradual decrease with increase in altitude at all stations and a spatial distinctiveness with the highest near-surface concentration in the central IGP (most anthropogenically impacted subregion of the IGP) depicting sharper altitude variation compared with the other two subregions.

As is well-established that during the pre-monsoon and prior to the onset of the monsoon, both the natural and anthropogenic aerosol species coexist in large abundance over the IGP. We examined in Fig. 4b, the altitude profiles of accumulation mode aerosols (NAC, concentration below 1 µm), which are mostly attributed to anthropogenic origin and coarse mode aerosols (above 1 µm), which are mostly of natural origin. NAC showed only weak altitudinal dependence above 1 km at all the subregions; though at VNS, there was a sharp increase in the concentration below 1 km, obviously due to source proximity. This feature is seen in Fig. 4a also. This observation is supported by the collocated measurements of aerosol total number concentrations (NT) as measured by a condensation nuclei (CN) counter aboard the aircraft (Jayachandran et al., 2020) in the size range above 2.5 nm, showing the highest values of NT in the entire altitude range of measurements over VNS. On the other hand, the vertical profiles of coarse mode aerosol concentrations (NC) showed significantly large abundance over the western IGP (arid/semiarid regions) represented by JDR, similar to the springtime observations reported by Gogoi et al. (2019).

These observations are also in line with the reported values of dust fractions (Vaishya et al., 2018) during the same campaign, showing the enhancement of dust fraction from 10 % to 20 % at 300 m to more than 90 % above 2 km altitude at JDR, while the smallest dust faction (<10 %) was observed at BBR in the entire altitude range. Over the central IGP, synoptic wind-driven desert dust aerosols lead to elevated layers of aerosols having a higher dust fraction (>50 %). However, it should be noted that dust over the central IGP is more absorbing in nature because of its mixing with other anthropogenic emissions (such as BC; Vaishya et al., 2018), while that over western IGP is rather pristine in nature. Thus, quantification of the absolute magnitude of coarse mode aerosol concentrations is very important to understand the significance of elevated aerosol load on radiative perturbations. The increasing concentration of coarse mode particles with the increase in altitude across the entire IGP is another interesting feature in the present study, which is most conspicuous at the central IGP and least at the west, implying their increasing role at higher altitude, probably due to the lofted regional dust and advected mineral dust from west Asian regions.

With the goal of quantifying this, the size distribution spectra are averaged for each altitude level and for each station. From these spectra, the geometrical mean diameter (Dg) is estimated as a function of altitude, using the following equation: 2Dg=exp⁡∑luniln⁡(Dpi)N, where Dpi=√Di×Di+1 denotes the geometric midpoint of each channel of the APS, ni is the particle concentration in the ith channel and N=∑luni is the total concentration. Accordingly, Dg of a spectrum of particles is the 50 % probability point of an equivalent diameter having half of the particle concentrations larger than this size and the remaining half below. The vertical profiles of Dg and mode (=Dp(nmax⁡)) of the distributions are shown in Fig. 5. It clearly shows the increase in the coarse mode fraction in the size distribution, with both the mode and Dg showing a steady increase with altitude, especially Dg. The rate of increase in Dg with altitude increases from west to east across the IGP, with the highest values in BBR (Fig. 5b). In the central IGP where the mixed aerosol type prevails, the increase in Dg within the ABL is rather weak, but in the free troposphere it increases more sharply probably due to the faster decrease in the accumulation mode concentration (Fig. 4) or the prevalence of advected dust at higher altitudes or both. The previous observations reveal the nonuniform distribution of dust and anthropogenic sources of aerosols. Nearly steady values of NC in the entire column at JDR are attributed to the strong convective mixing of coarse mode dust aerosols up to the lower free-tropospheric region. On the other hand, altitude variation in accumulation and coarse mode aerosols is relatively more variable at BBR and VNS, compared to that at JDR (Fig. 4b) as indicated by the profiles of Dg.

Apart from the number-weighted expression of aerosol size distributions, the mass-weighted distributions carry useful information for quantifying regional distinctiveness of the dominance of coarse mode particles. Even though the fine mode aerosols are extremely numerous in the atmosphere and important for microphysical processes, they represent only a very small proportion of total particle mass, whereas coarse mode particles, even though far less numerous, have significant volume. In simple terms, particle number concentrations are dominant in the fine mode (<0.1 µm), the surface area is predominantly in the accumulation mode (0.1 to 1 µm) and the volume, and hence mass, is divided between the accumulation mode and coarse particle mode. In the present study, since the size range of particle counts is confined in the accumulation and coarse mode regimes (between 0.5 and 20 µm), a quantitative picture of aerosol mass concentrations is obtained by assuming a uniform density equal to 2 g cm-3 following Moorthy et al. (1998) and Pillai and Moorthy (2001). Since the size-resolved particle densities are not known, we did not use effective density (mass–mobility relationship defined as the mass of the particle divided by its mobility equivalent volume) of particles to calculate the mean particle mass size distributions.

Figure 6a shows the altitudinal variation in coarse mode aerosol mass concentrations over all the observational sites, along with the values of coarse mode mass fractions (FMC). Over VNS and JDR, consistently higher values of MC were seen in the entire altitude range. This is in line with the higher values of coarse mode aerosol concentrations (NC) at these sites, JDR being the highest. On the other hand, the values of MC at BBR decreased significantly from the surface to the lower free-tropospheric region. The higher values MC observed near the surface at BBR can be attributed to the influence of local sea-salt aerosols, however not affecting the values of Dg, due to the abundance of accumulation mode aerosols over this site.

Similar to that of NC, FMC showed (Fig. 6b) gradually increasing values with altitude at all the locations. The high values of coarse mode mass fraction and an increasing trend with altitude are indicative of the role of upper-level transport of dust from the western desert region, in addition to those contributed locally due to thermal convective processes. Compared to the other two stations, the highest value of FMC (∼70 %) near the surface was seen at JDR indicating the role of the arid nature of the region. This exercise clearly explains the abundance coarse mode dust decreasing from west to east, along with an increase in the contribution of anthropogenic fine/accumulation mode aerosols.

With the goal of examining the transport of mineral dust (by the synoptic winds), the spatial distributions of UV aerosol index, aerosol types and aerosol absorption optical depth (AAOD), all derived from the Level-3 OMAERUVd data product (daily, 1.0∘×1.0∘) from the Ozone Monitoring Instrument (OMI, on board the Aura satellite; Levelt et al., 2006), are examined. OMAERUV uses the pixel-level Level-2 aerosol data product of OMI at three wavelengths (355, 388 and 500 nm) to derive AAOD. Higher values of AAOD at 388 nm are indicative of the presence of dust or biomass burning aerosols. This is because absorption by dust and organic carbon from biomass burning sources has strong wavelength dependency, with higher absorption at near-UV wavelengths. As the period of this campaign was devoid of major fire activities over the study region (northern India) which normally peaks in April to May and October to November, corresponding to burning after the wheat and rice harvests (Vadrevu et al., 2011; Venkataraman et al., 2006), the AAOD values would be representative of dust loading. This aspect is confirmed in the subsequent section using lidar depolarization ratio.

Figure 7a–d show the spatial distributions of UV aerosol index, aerosol type and AAOD at 388 and 500 nm, while the synoptic winds are shown in Fig. 7e. A very good association between the westerly advection and dust loading extending from west to central IGP is noticeable from the figure. This lends further support to the role of advected dust leading to higher MC and FMC at higher altitudes, seen in Fig. 6. In this context, it is also worth noticing that based on observational data and regional climate modeling, Banerjee et al. (2019) have clearly shown (in their Fig. 7) the significant vertical extent of dust loading, of both local and remote origin, during the pre-monsoon and summer across the IGP reaching altitudes as high as 600 hPa.

The volume size distribution of aerosols (shown in Fig. 8) at three distinct altitude regions of the atmosphere also clearly shows the altitudinal change in the pattern of distribution, changing from coarse mode dominance near the surface to accumulation mode dominance at the ceiling altitude over BBR. Those at JDR, the pattern of distributions remains the same in the entire column. Similar to JDR, VNS also depicted a significant enhancement in coarse mode aerosols in the upper levels (at 2 and 3 km altitudes) of the atmosphere. Similar to these observations, based on the collocated spectral scattering properties of aerosols obtained during the same experiment, Vaishya et al. (2018) have reported that the aerosol population changes from supermicron-mode-dominant natural aerosols to submicron-mode-dominant anthropogenic aerosols, as we move from west to east in the IGP. Moreover, the large abundance of coarse particles (>2 µm) along with significant fine/accumulation mode aerosols in the column highlights the complex mixture of dust with other anthropogenic components in all three regions, making a complex scenario for aerosol radiation and aerosol cloud interaction processes. Based on the combination of satellite remote sensing and regional climate model simulations, Banerjee et al. (2019) have also shown the presence of a dry elevated layer of dust (at altitudes between 850 and 700 hPa, taking place in multiple layers) during June across the IGP, transported from the Thar Desert to the northern Bay of Bengal. To ascertain this further, we have examined the data from CATS aboard ISS.

Geophysical parameters derived from CATS on board the ISS are very useful to infer aerosol features in the atmospheric column, especially at altitudes above the ceiling altitude of the aircraft (3.1 km). In the present study, we have considered three products from CATS for the campaign period, viz. (i) depolarization ratio, (ii) attenuated backscatter coefficients and (iii) aerosol types.

Figure 9 shows the vertical cross section of depolarization ratio for three passes during the campaign period and close to the three subregions (identified by the rectangular boxes in the figure). Higher values (∼0.3) of depolarization ratios are seen in the western IGP (JDR, bottom panel), suggesting the dominance of nonspherical (dust) particles. The depolarization ratio decreases towards the east across the IGP with values equal to 0.1 at the central IGP and ∼0.2 in the eastern BBR. These lend additional support to the inference on the influence of dust aerosols during the campaign period. Supporting the patterns of depolarization ratio, aerosol types (from CATS operating mode 7.1) in Fig. 10a indicate the significant presence of dust at JDR, while the aerosol types over VNS and BBR are a mixture of dust and polluted continental and carbonaceous aerosols. Vertical profiles of total attenuated backscatter coefficients show the vertical extent of the aerosol layer to be as high as 5 km (as has been shown by Banerjee et al., 2019) over all the sites (Fig. 10b).

BC is the chief anthropogenic absorbing aerosol species, and the IGP is known to be among the global hot spots of it (Govardhan et al., 2019). The height resolved information on FBC is important not only in radiative forcing but also in CCN activation as well (Bhattu et al., 2016). Collocated measurements of BC during SWAAMI–RAWEX have been used to examine the vertical profiles of BC and its variation across the IGP prior to the onset of the Indian summer monsoon. Figure 11a shows the vertical profiles of BC for the three subregions. Each profile is the average of all the profiles obtained from measurements made from each of the base stations. It is seen that BC remained low (∼1 µg m-3) and depicted very weak altitude variations at the western and eastern IGP regions, while in the central IGP there is a rapid decrease in BC from the high value (∼3 µg m-3) near the surface. Above 2 km, all the profiles overlap, though a weak increase is indicated over BBR, which is examined later. The very high values of BC close to the surface at VNS are attributed to the wide-spread anthropogenic activities in the central IGP including the cluster of thermal power plants in that region. Consequently, the columnar concentration of BC (integrated up to 3.1 km) is also the highest at VNS.

However, the vertical profiles of the fractional contribution of BC (FBC) to the total composite aerosol mass (estimated from the volume size distribution, considering a uniform density of 2 g cm-3, especially in view of the abundance of dust) shows (Fig. 11b) subregional distinctiveness. It remains the lowest (∼6 %) in the western IGP, with very little altitude variation. In the central IGP, FBC is quite high (∼15 % to 20 %) within the ABL and drops off fast above 2 km approaching the values seen for the western IGP. FBC depicts an elevated peak at around 1 km above ground level at VNS, while at BBR, higher FBC values occur at still higher altitudes, where the near-surface values are much lower and comparable to those at JDR. There is a steady increase in FBC from near the surface to higher altitudes, and above 2 km the values are comparable to the peak values seen at VNS (at ∼1 km altitude). Despite this, the integrated BC concentration at BBR falls in between that of JDR and VNS, mainly because of the large values occurring in the lower atmosphere at VNS. It may be recalled that based on SWAAMI–RAWEX aircraft measurements, Vaishya et al. (2018) have reported uniform scattering characteristics across the IGP, while the absorption coefficients showed subregional distinctiveness leading to a west-to-east gradient (decrease) in the vertical structure of single-scattering albedo (SSA).

Investigation of the vertical profiles of BC mass concentrations on individual days (Fig. S2 in the Supplement) helps to see the distinctiveness in each subregion, resulting from the spatially heterogeneous nature of emission sources and advection, especially at BBR where the inland profiles, made during flights perpendicular to the coastline (on 2 and 3 June), show significantly higher values of BC at higher altitudes than those along the coastline. At BBR, this arises mainly because of spatially heterogeneous source impacts. The regions towards the northwest of BBR are characterized by large-scale urban and industrial activities (“Ambient air quality status and trends in Odisha: 2006–2014”). Similarly, near-surface BC concentrations at VNS were higher when the flights were confined to the NE, NW and SW of the city center, while the values in the SE sector were lower. On the other hand, at JDR, the profiles revealed a better spatial homogeneity.

To quantify the climatic implications of BC, the heating rate profiles of BC are examined based on the estimation of shortwave aerosol direct radiative forcing (DRF) due to BC alone. DRF due to BC represents the difference between the DRF for aerosols with and without the BC component. The in situ values of scattering (σsca) and absorption (σabs) coefficients measured on board the aircraft were used to estimate spectral values of AOD (layer-integrated σsca+σabs), single-scattering albedo (SSA) and asymmetry parameter (g) for each level, assuming a well-mixed layer of 200 m above and below the measurement altitude (details are available in Vaishya et al., 2018). The layer mean values of AOD, SSA and Legendre moments of the aerosol phase function (derived from the Henyey–Greenstein approximation) are used as input in the Santa Barbara DISORT Atmospheric Radiative Transfer (SBDART; Ricchiazzi et al., 1998) model to estimate diurnally averaged DRF (net flux with and without aerosols) at the top (DRFTOA) and bottom (DRFSUR) of each of the layers. The atmospheric forcing (DRFATM) for each of the levels is then estimated as 3DRFATM=DRFTOA-DRFSUR. In order to estimate the forcing due to BC alone, optical parameters for aerosols were deduced again. For this, values of σabs were segregated into the contributions by BC (σBC) and organic carbon (OC) (σOC), where σBC was estimated following inverse wavelength dependence of BC (e.g., Vaishya et al., 2017). Based on this, a new set of AOD and SSA for BC-free atmosphere is calculated and fed into SBDART for estimating DRFALL-BC without the BC component. Thus, DRF due to BC is 4DRFBC=DRFALL-DRFALL-BC. Here, DRFALL represents forcing due to all the aerosol components, including BC.

The vertical profiles of atmospheric heating rate (HR, estimated based on the atmospheric pressure difference between the top and bottom of each layer and aerosol-induced forcing in that layer) due to BC alone show (Fig. 12) maximum influence of BC in trapping the SW radiation at VNS, followed by BBR and JDR. Interestingly, the altitudinal profiles of heating rate are distinctly different over the regions, with BBR showing an increase with altitude, while VNS shows the opposite pattern with maximum heating (∼0.81 K d-1) at 500 m above ground. Enhanced heating at 500–2000 m altitude is seen at JDR. These results indicate the dominant role of absorbing aerosols near the surface at VNS, while the atmospheric perturbation due to elevated layers of absorbing aerosols is conspicuous at BBR (HR ∼0.35 K d-1 at the ceiling altitude). The column-integrated values of atmospheric forcing due to BC alone are 7.9, 14.3 and 8.4 W m-2 at JDR, VNS and BBR, respectively.

In this context, we have examined the possible role of the large network of thermal power plants (TPPs) over the northern part of India, which is reported to have a significant contribution to regional emissions (Singh et al., 2019). These include the emissions of SO2, NOx, CO2, CO, volatile organic compounds (VOCs), suspended particulate matter (PM2.5 and PM10, including BC and OC) and other trace metals like mercury (Guttikanda and Jawahar, 2014; Sahu et al., 2017), dispersing over large areas through stacks. Fly ash from coal-fired power plants causes severe environmental degradation in the nearby environments (5–10 km) of TPPs (Tiwari et al., 2019). Over the IGP, since more than 70 % of the thermal power plants are coal based, emissions of CO2 and SO2 make up more than 47 % of the total emission share, while the relative share of PM2.5 and NOx is ∼15 % and 30 % (GAINS, 2010). Based on the in situ measurement of BC in fixed and transit areas in close proximity of seven coal-fired TPPs in Singrauli (located ∼700 km northwest of BBR), Singh et al. (2019) have reported that BC concentration reached as high as 200 µg m-3 in the transit measurements. The Energy and Resources Institute, India, has also reported that emission levels of the carbonaceous (soot or BC) particles are estimated to be around 0.061 gm kW h-1 per unit of electricity from Indian thermal power plants (Vipradas et al., 2004). Based on emission pathways and ambient PM2.5 pollution over India, Venkataraman et al. (2018) have reported that the types of aerosols emitted from coal burning in thermal power plants and industry in eastern and peninsular India are similar to those of residential biomass combustion. The ongoing discussion thus clearly indicates that TPPs are major sources of BC in the atmosphere.

As it is not possible to measure BC from space, to infer the role of these emissions from thermal power plants in causing the higher BC fraction at higher altitude over BBR, we have examined the spatial distribution of the concentrations of the co-emitted NO2 and SO2 in Fig. 13, in which the locations of major coal-based TPPs (https://www.ntpc.co.in/en/power-generation/coal-based-power-stations, last access: 9 July 2020) are also marked. The data are obtained from OMI on board the Aura satellite. Higher concentrations of NO2 and SO2 are readily discernible from the figure around the regions (marked in the figure) during the period of flight experiments where there are clusters of TPPs. As the energy consumption is the highest during summer and most dependent on TPPs, these TPPs should be operating to near full capacity. This provides indirect support to the high concentrations of BC (co-emitted) at higher levels. In general, these TPPs have tall stacks (heights in the range 200 to 400 m) and aid easy ventilation to the lower free-tropospheric altitudes.

To further ascertain this, the spectral properties of aerosol absorption are examined. First, we have examined the frequency distribution of absorption Ångström exponent (αabs, derived from the linear fit on log-log scale between corresponding absorption coefficients to Aethalometer wavelengths) in Fig. 14, separately for the mixed layer (ML, below 1.5 km) and above (≥2 km). The frequency distribution of αabs reveals a clear shift towards lower values as we move from JDR to BBR, both within the ML and above, even though the values of αabs are mostly between 1 and 1.5. Based on laboratory studies and field investigations, it has been shown that the higher values of αabs (∼2) are representative of biomass burning emissions, while the values of ∼1 are indicative of fossil fuel combustions (Kirchstetter et al., 2004). The values of αabs>1 are indicative of the presence of biomass burning, whose relative abundance increases with the steepness of the absorption spectra, as has been reported elsewhere from the laboratory experiments (Hopkins et al., 2007).

Examining Fig. 14 in the above light, it emerges that a significant contribution of BC from fossil fuel combustions mixed with biomass burning origin prevails at higher altitudes over BBR, while the association between the two decreases abruptly from ML to a higher height at VNS. The consistent higher values of BC in the column associated with the values of αabs lying between 1 and 1.5 can also be due to the aging of BC at higher heights, during which BC mixes with other species and its Ångström exponent increases. This is because the spectral dependence of absorption steepens when BC (even though its source could be fossil fuel) is coated with a concentric shell of weakly absorbing material (Gogoi et al., 2017). Further investigations are needed in this direction.

The spatial variation in the altitude profiles of NAC, Dg, FMC and FBC across the IGP hints to several possible implications of their direct and indirect effects. Altitudinal increases in the values of Dg and FMC along with depolarization ratios are indicative of the presence of dust (>4 µm) in the lower free troposphere, which is known to produce a long-wave (warming) radiative effect (Miller et al., 2006; Tegen and Lacis, 1996). Conversely, significant abundance of accumulation mode aerosols, in general, might contribute significantly to scattering. For example, a clear seasonal change in the vertical profiles of aerosol concentrations (NT-APS) as measured by APS in the size (diameter) regime of 0.5 to 20 µm is noticeable at JDR, changing from the much steeper variation (vertically) in winter (as reported by Gogoi et al., 2019) to a near-steady one during prior to the onset of the monsoon (Fig. 15). Based on airborne measurements during SWAAMI–RAWEX, Vaishya et al. (2018) have reported that the values of SSA at west IGP varied between 0.935 (at 530) in spring and 0.84 (at 530 nm) prior to the onset of the monsoon, indicating a seasonal change in the aerosol type and consequently their optical properties.

To examine the role of the dynamical processes in different seasons, we have shown the profiles of vertical velocity (in pressure coordinates from 1000 to 100 hPa) in Fig. 16. These are obtained from ERA-Interim reanalysis data sets. Here, the positive and negative signs of vertical velocity (ω) are indicative of updraft (as indicated by -ve values of ω) and downdraft (as indicated by +ve values of ω). A clear seasonal transformation is seen, with increasingly stronger updrafts dominating over the IGP from December to June, with the intensity increasing from west to east. In the western IGP regions, the sign of vertical velocity seems to change from December to June, progressively enhancing the magnitude of deep convection towards the onset of the monsoon, imparting stronger vertical dispersion and more homogeneous distribution of aerosols in the column.

Regionally, the seasonal transformation of vertical velocity is more prominent over the eastern IGP BBR, where the magnitude of vertical velocity is consistently higher from surface to upper-tropospheric regions prior to the onset of the monsoon. During this period, the head of the Bay of Bengal is known to be one of the regions where deep convection exists (Bhat et al., 2001). Since size distribution is a dominant factor in determining the direct radiative forcing (Tegen and Lacis, 1996; Liao and Seinfeld, 1998; Seinfeld et al., 2016), a clear seasonal change in the altitudinal variations in aerosol type and size distributions associated with distinct transport and convective processes will have a strong radiative impact. In particular the columnar distribution of coarse mode dust and highly absorbing BC needs explicit representations in climate models for accurate understanding of the net top-of-atmosphere (TOA) direct radiative forcing. Apart from the direct radiative implications, abundance of coarse mode dust particles (having sizes larger than critical diameter) and aged BC (coated with hygroscopic materials) in the lower free troposphere can act as CCN in a supersaturated environment. Recent studies suggest that mineral aerosols are the dominant IN for cirrus clouds (Storelvmo and Herger, 2014).