First ever in-situ measurements of black carbon (BC) aerosols in the troposphere (up to 9 km) made over central India and the resulting atmospheric impact as revealed by the environment lapse rate are presented. The altitude distribution of BC showed multiple peaks; two surprisingly large peaks, one at ∼4.5 km, and another above 8 km. Associated with these, rapid decrease in the environmental lapse rate and a sharp increase in the atmosphere stability were observed, probably caused by the atmospheric warming by the BC layers. This important observation calls for extensive high altitude profiling of BC to quantify the resultant warming, increase in stability and consequent increase in BC lifetime.
 The atmospheric warming by aerosol black carbon (BC) and its consequence on climate forcing remains one of the important issues related to atmospheric aerosols. The strong absorption over a wide range of wavelengths, coupled with its smaller size (sub micron) and longer atmospheric lifetime makes BC all the more important for climate impact. Over the tropics the strong convection supports deeper vertical lofting of BC in the atmosphere. In addition, direct injection of aerosol at high altitudes by ever increasing air traffic [e.g., Hendricks et al., 2005] is a potential source of BC in the upper troposphere (and even in lower stratosphere at high latitudes). Presence of BC aerosols in the upper troposphere and lower stratosphere has been identified and even quantified though in a limited way [Pusechel et al., 1992; Blake and Kato, 1995; Strawa et al., 1999]. Such information is non-existent over Indian region.
 In aerosol radiative forcing, the vertical distribution of BC has immense importance, especially in the presence of clouds [Satheesh et al., 2008; Chand et al., 2009]. The warming potential of BC strongly increases when located above highly reflecting/scattering clouds [Podgorny and Ramanthan, 2001], and leads to local warming, reduction in the local relative humidity and increase in stability [Ackerman et al., 2000]. Despite, available information is limited and direct experimental measurements are virtually non-existent over tropics were strong convection and cloudiness coexist. The very few recent efforts over India [Moorthy et al., 2004; Babu et al., 2008], were confined only to the first two to three kilometers of the atmosphere. In the backdrop of the above, in situ measurements of the vertical distribution of BC along with concurrent measurements of atmospheric temperature were undertaken as a part of the RAWEX (Regional Aerosol Warming EXperiment) field experiment using a high altitude balloon from the tropical station (Hyderabad, HYD, 17.48 N, 78.40 E, 557 m msl), India, during the spring of 2010; a season when elevated aerosols layers have been reported to occur over these regions [Satheesh et al., 2008]. Important findings from this first ever attempt over India are reported, and possible impacts are discussed.
2. Measurement Details
 In situ measurements of BC mass concentrations (MB) were carried out using an Aethalometer (Model AE-42, of Magee Scientific, USA) installed in a gondola attached to hydrogen inflated zero pressure balloon designed and fabricated at the National Balloon Facility of the Tata Institute of Fundamental Research (TIFR), Hyderabad (∼20 km off the urban centre of Hyderabad). The 109755 m3 balloon, made of 25 μm thick, linear low-density polyethylene film, has been designed to lift a payload of ∼350 kg up to a ceiling altitude of ∼35 km. It carried several other payloads such as GPS receiver, sensors for temperature and RH, telemetry and tele-command systems all fitted to the gondola, which was attached to the balloon using a parachute (21 m diameter), that deploys on the descent enabling profiling during the descent and a safe recovery of the payloads after the flight.
 The Aethalometer estimated the BC mass concentration by measuring the attenuation of light at 880 nm while the aerosol particles deposited on the quartz filter tape. The instrument was specially configured for volumetric flow with external pump operation, and supplemented with DC pumps capable of providing a flow rate of 14 liters per minute (LPM) at ground level (950 hPa ambient pressure). The assembly has been tested in a thermo vacuum chamber, and consistent operations were ensured down to an ambient pressure level of 250 hPa (corresponding to an altitude of ∼9 km) and temperature of −40°C. The Aethalometer was operated at a time base of 5 sec and the data has been telemetered down, where it was received and recorded along with the GPS coordinates (latitude, longitude, altitude, and time) besides being stored in its flash memory onboard. An onboard telecommand system was used to turn off/on the Aethalometer at the desired altitudes (when the pressure, monitored continuously during the flight, dropped to ∼250 hPa level). Ballast cans with a ∼50 kg of ballast powder were attached to the gondola to ensure a slow and steady ascent rate (∼2.6 ± 0.7 m s−1) to enable high resolution vertical profiles during the ascent. The descent, however, was faster (∼4 m s−1) and was essentially controlled by the parachute.
 The principle of operation, data deduction and error budgeting of the Aethalometer are described in detail in several earlier papers [Babu et al., 2008, and references therein] and are not repeated. Necessary corrections for the amplification and shadowing effects were carried out following Weingartner et al.  and Arnott et al.  during analysis of the data. For the present configuration, the measured BC had an uncertainty of <15% or 25 ng m−3 whichever was higher.
Figure 1 (left) shows the photograph of the balloon with the payload just after the launch and the flight path. The balloon ascent started at 07:30 local time, well after sunrise, on 17th March 2010, and after the measurements landed at a location ∼40 km north–east of the launch site at 11:30 LT (trajectory shown in Figure 1 (middle panel)) and the payloads were recovered in safe working condition. Since the time difference and spatial separation between the ascent and descent profiles was not significant, the BC values in both legs were grouped together in altitude bins of 100 m and averaged so that each altitude bin contained a minimum of 12 data points to have a reliable statistics.
3. Results and Discussion
 The mean altitude distribution of BC thus obtained is shown in Figure 1 (right), where the filled circles represent the mean values at the mid-point of each bin and the horizontal lines through the points indicate the standard deviations. The BC profile shows several important and surprising features.
 1. BC concentration remained nearly steady in the lower atmosphere, up to an altitude of ∼2 km above ground, above which it gradually increased superposed with weak fluctuations.
 2. A sharp thin layer (P1) of very high BC (concentrations going as high as 12 μg m−3) is noted in the altitude region from 4.4 km to 4.8 km, which contained more than 50 independent measurements.
 3. Above this, the values remained moderate to low, superposed with several smaller and narrower peaks, before coming down to very low values at ∼6 km and remaining so until ∼8 km.
 4. A second layer of enhanced BC is indicated above 8 km; however this peak (P2) is not fully resolved as the Aethalometer was commanded off because the flow rate decreased drastically above 8.5 km.
 Integrating the vertical profile up to 8.5 km, the columnar loading of BC is calculated as ∼23 mg m−2, and the peaks at 4.5 km and at 8.3 km contributed respectively ∼12% and ∼10% to this loading.
 Examining the possible mechanisms leading to the formation of these surprising elevated layers of enhanced BC, we considered the following (a) Vertical lofting from the surface by the strong thermal convection over the land (b) Local confinement by convectively stable layers, trapped between unstable layers and inversions (c) Long-range transport by change in advection patterns and (d) Other possible local source of elevated BC.
 During the experiment, another Aethalometer, inter-compared with the one, flown on the balloon, was in continuous operation at the launch site. The temporal variation of BC mass concentration, obtained using this Aethalometer on the launch day is shown in Figure 2 (top). The start and end of the balloon flights are also marked on the X-axis. The diurnal variation of the BC shown by the surface aethalometer, is typical to continental sites, and basically arises out of the boundary layer dynamics (fumigation) and would be conspicuous in seasons when the day-night temperature differences are larger, as has been discussed in several earlier papers [e.g., Babu and Moorthy, 2002]. It is clearly seen that the small dilution in the surface concentration due to increased convective mixing and deepening of the atmospheric boundary layer, though might explain the near steady value of BC in the lower atmosphere, is totally inadequate to account for the large concentration seen at 4.5 km (P1) and 8.5 km (P2).
 To examine the role of atmospheric thermodynamics, we have plotted in Figure 2 (middle), the vertical profiles of wind speed, wind direction and potential temperature, all derived from the concurrent measurements using the radiosonde, attached to the gondola. The wind direction remained steady northerly–northeasterly up to 5.5 km, more than 1 km above the first and strongest layer, while the speed did not reveal rapid changes; being between 4 to 7 m s−1. Thus there was no drastic change in the airmass type and advection strength that could have led to the formation of P1. From 5.6 to 6.4 km, though the wind drastically shifted to westerly and remained so up to 9 km and above, BC concentration did not reveal any impact; it remained very low until 8 km. Between 8 and 8.5 km BC increased significantly to reach a peak at 8.3 km. Interestingly, the wind speed sharply increased above 8 km, from ∼6 m s−1 to very high values of ∼15 to 20 m s−1, associated with tropical jet stream and this might have led to confinement of BC at this level leading to the high BC. In the lower levels, the altitude distribution of θ (potential temperature) showed several structures with strong inversions at ∼1 km, 1.6 km, and 4.5 km. From the ground to 1 km, potential temperature remained constant showing an adiabatic condition and the prevalence of convective turbulence. Above 1 km, θ increased with altitude showing a convectively stable sub-adiabatic atmospheric region up to ∼2 km. Above 2 km, θ remained steady with altitude again, showing a convectively unstable region extending up to ∼4 km. In this region, the wind speeds were also higher (∼7 m s−1) and directed from north–easterly region. Above 4.7 km, θ showed a sharp inversion and changed back to sub-adiabatic, stable condition that extended up to 7 km. It is quite possible that this strong inversion might be responsible for confining BC at 4.5 km leading to the sharp peak (P1). It is, however, also possible that this strong inversion has resulted from a local warming caused by enhanced absorption (of solar radiation) by the BC in strong layer P1. Earlier studies have shown that the local pollution of the previous day might be trapped in the entrainment zone and would be observed during the next morning. We have examined the surface aethalometer data for a week prior to the launch and found them to be quite similar. The PBL height attains the altitude of 3 to 4 km over Hyderabad typically late in summer (from mid-April onwards) when the sunrise is early and the ambient temperatures go well above 40°C. Thus, though the PBL is important in deciding the BC concentration above surface, it cannot totally explain the observed layer above 4 km especially in the month of March. We shall examine this later in section 4.
 To examine the role of long-range transport, 7-day isentropic back trajectories, computed using the Hybrid Single Particle Legrangian Integrated Trajectory (HYSPLIT) model of the National Oceanic and Atmospheric Administration (NOAA) at different altitude levels from the surface (550 m msl) to 8 km at every km are shown in Figure 2 (bottom). At the lower levels, below 4 km, the horizontal extent of the trajectories are short and are confined to the Indian landmass; originating from the east coast of peninsular India (below 3 km) while at 4 km, it arrived from the northwestern part of India. At higher altitudes, above 5 km, all the trajectories have very long spatial extent and originate from the African regions or even beyond that. However BC remained very low in those altitudes up to 8 km; before increasing again. Between 8 and 9 km, where the peak P2 occurred, the trajectories did not show any conspicuous shift and thus do not indicate the possibility of long-range transport.
 This leads to the possibility of BC occurring locally in the upper altitude levels. Hyderabad, a large urban area in central India, is a major national hub for air transport too. Checking with the Air Traffic Control (ATC), it is learnt that about 200 long range aircrafts, that overpass Hyderabad without landing, passed through the corridor between ∼8 and 10 km altitude of our flight domain; while about 250 to 300 aircrafts, that land at Hyderabad used a lower corridor of 4 to 5 km. It is quite possible that the exhaust from these aircrafts, coupled with the longer residence time in these regions would have led to the observed layers.
4. Possible Impacts on Atmospheric Stability
 The layers of enhanced BC, observed at 4.5 km and above at 8 km are rather surprising primarily because they were observed for the first time, and also because of their large magnitudes and their significant implications to aerosol radiative forcing (being located above the clouds). In this context, the layer at 8.3 km, though totally unexpected, assumes immense importance. Based on model simulation and observation during INDOEX, Ackerman et al.  reported that atmospheric warming by the layer of enhanced BC reduced the cloud cover and could offset the aerosol induced radiative cooling at the top of the atmosphere on a regional scale. Recently, Zarzycki and Bond  have reported that the BC above low level clouds account for about 20% of the global burden and 50% of the forcing. Based on satellite observations, Chand et al.  have shown that while absorbing aerosols above reflecting clouds darkens the scene by enhanced absorption, over dark surface they tend to brighten the scene by increased backscatter.
 Thus the occurrence of high, elevated BC layers in the present study has implications to the local atmospheric stability. In Figure 3 (left) we have shown the vertical profile of mean ambient temperature derived from concurrent measurements using balloon borne sensors during the ascent and descent legs. In the same panel is plotted the environmental lapse rate (dT/dz). To our surprise, we notice a sharp decrease in the environmental lapse rate from ∼−9 K km−1 in the region below 3 km to reach values as low as ∼−0.8 K km−1 at 4.5 km at the peak of the first layer P1. It gradually recovered to near normal values at higher altitudes before started falling down again around the second peak at 8 km to ∼−6 K km−1. These sharp changes in lapse rate appear to be caused by absorption of radiation by the BC layer. To examine this, we estimated the heating rate profile using the altitude distribution of BC and the vertical profile of extinction coefficient from the CALIOP (Cloud Aerosol Lidar with Orthogonal Polarization) onboard CALIPSO (Cloud Aerosol Lidar Pathfinder Satellite Observation) satellite (http://www-calipso.larc.nasa.gov) which had an over pass over the study region on the day of BC profiling. The vertical profile of percentage contribution of BC extinction to the layer extinction derived using CALIPSO data showed large variability with values ranging from ∼10% to ∼ 80% with most of the high values above 3 km. The layer extinction and the contribution of BC to the layer extinction are incorporated into the urban aerosol model [Hess et al., 1998] and the corresponding profile of single scattering albedo (ω) and phase functions are estimated. The diurnally averaged values of the net short wave radiation absorbed in each layer and the heating rates are estimated using SBDART [Ricchiazzi et al., 1998] and as described by Moorthy et al. . Since the extinction profile over the study location from CALIPSO was available only up to 5.5 km on the day of the balloon flight, the heating rate estimation is also restricted to 5.5 km. The heating rate profile thus obtained, shown in Figure 3 (right), depicted significant heating of ∼2.8 K day−1 at ∼4.5 km, which might have definitely contributed to the observed decrease in the lapse rate around the same altitude.
 Thus, it appears that the large, elevated BC layer absorbed solar radiation leading to warming of the local ambient, which increased the stability (decreased the lapse rate) and led to the sharp inversion in potential temperature. This stable layer, thus created, is conducive for maintaining the BC layer longer, without dissipation (by inhibiting turbulent mixing) and thus increases its lifetime leading to further enhanced absorption. This effect would be significantly enhanced by the local BC aerosols, lofted by the dynamics of the planetary boundary layer. Thus, a combination of the PBL evolution and warming by the BC layer would be acting together, leading to the formation of the large BC layer at ∼4.5 km. However, the PBL dynamics becomes quite insignificant at altitudes as high as 8 km or above, where the peak P2 was observed. Our observations, thus raises several questions including: “Do BC layers build ‘their own homes’ up in the atmosphere?” More focused investigations are necessary to examine this issue further.
 We could not estimate the aerosol heating at higher levels, due to non-availability of aerosol extinction profiles. Nevertheless, it is pertinent to note that the warming at P2 would be much higher (as the air is thinner). It is very important because these are the altitudes (>8 km) where thin cirrus clouds form over the tropics and the strong BC-induced warming would significantly influence these clouds and their forcing. To examine these, more high altitude flights are planned during the coming spring, when the prevailing winds are favorable for high altitude ballooning and safe recovery of the payloads.
 The first-ever high altitude, in-situ measurements of BC in the troposphere (up to 9 km) made over central India revealed the following.
 1. The altitude distribution showed multiple peaks in BC concentration, with two large peaks the one at ∼4.5 km, and another above 8 km, probably associated with high-altitude aircraft emissions.
 2. Associated with the peak in BC concentration at ∼4.5 km, a rapid decrease in the environmental lapse rate and a sharp increase in the atmosphere stability were observed, apparently resulting from warming by the BC layers.
 3. This raises new issues on the lifetime of elevated BC layers, their sources, radiative forcing and probable impacts on cirrus clouds.
 The study was carried as part of the Regional Aerosol Warming Experiment (RAWEX) under Aerosol Radiative Forcing over India (ARFI) project of Indian Space Research Organization (ISRO). We acknowledge NOAA Air Resources Laboratory for the provision of the HYSPLIT transport and dispersion model and READY website (http://www.arl.noaa.gov/ready.html) used in this publication. One of the authors thank Department of Science and Technology (DST), New Delhi for Swarna Jayanti Fellowship.
 The Editor thanks B. V. Krishna Murthy and an anonymous reviewer for their assistance in evaluating this paper.