Strong radiative heating due to wintertime black carbon aerosols in the Brahmaputra River Valley



[1] The Brahmaputra River Valley (BRV) of Southeast Asia recently has been experiencing extreme regional climate change. A week-long study using a micro-Aethalometer was conducted during January–February 2011 to measure black carbon (BC) aerosol mass concentrations in Guwahati (India), the largest city in the BRV region. Daily median values of BC mass concentration were 9–41μgm−3, with maxima over 50 μgm−3during evenings and early mornings. Median BC concentrations were higher than in mega cities of India and China, and significantly higher than in urban locations of Europe and USA. The corresponding mean cloud-free aerosol radiative forcing is −63.4 Wm−2 at the surface and +11.1 Wm−2at the top of the atmosphere with the difference giving the net atmospheric BC solar absorption, which translates to a lower atmospheric heating rate of ∼2 K/d. Potential regional climatic impacts associated with large surface cooling and high lower-atmospheric heating are discussed.

1. Introduction

[2] Black Carbon (BC) aerosols – a byproduct of the incomplete combustion of fossil fuels and biofuels – are of great concern to the atmospheric sciences community due to their role in climate forcing [Jacobson, 2001; Sato et al., 2003]. They alter the global radiation budget by absorbing and scattering sunlight [Moosmuller et al., 2009]. This reduces the net solar radiative flux to the earth's surface. Light absorption warms the atmosphere in the vicinity of the aerosols, potentially increasing atmospheric stability and changing precipitation patterns [Meehl et al., 2008; Wang, 2007].

[3] A number of studies have identified Southeast (SE) Asia as one of the largest sources of BC aerosols on a global scale [Gustafsson et al., 2009; Menon et al., 2002]. To understand the radiative effects of BC aerosols over this region, several monitoring studies in various locations have been carried out over the past decade. However, there exist very little or no reported data on BC mass concentrations in the Brahmaputra River Valley (BRV) region of SE Asia. The Brahmaputra is the world's fourth largest river. Originating in the Tibetan plateau, the Brahmaputra River flows through the northeastern parts of India and merges with the Ganges River downstream in Bangladesh. The BRV lies between the southeast Tibetan Plateau and the Assam Hills of northeast India. Song and co-workers [Song et al., 2011] recently reported that the surface temperature in the BRV has been increasing at a rate of 0.03 K/year from 1980 to 2005. The magnitude of this temperature change is greater than that over the whole Tibetan Plateau, which has been identified as a major driver of the climate change in the northern hemisphere [Feng et al., 1998; Pan and Li, 1996; Xu et al., 2009]. Precipitation has also increased at a rate of 4 mm/year, and this increase has been attributed primarily to anomalous summer precipitation [Song et al., 2011]. In this paper, we present the results of surface measurements of BC aerosol mass concentrations during the winter of 2011 at Guwahati, India, the largest metropolitan city in the BRV. Diurnal variation of BC mass concentration is discussed along with the associated variations in local meteorological parameters. Finally, the impact of BC aerosols on net radiative forcing and atmospheric warming is assessed and discussed.

2. Experimental Details

[4] A week-long field study was carried out between January 27, 2011 and February 2, 2011 at Guwahati (26°11′N, 91°44′E), which is located in the northeast Indian state of Assam (Figure 1a). The annual mean of the daily maximum temperatures in this state has been rising at the rate of 0.11 K per decade, while the annual mean of the daily temperatures has been rising at a rate of 0.04 K per decade [Assam Science Technology and Environment Council (ASTEC), 2011]. Accompanying this warming trend in recent years are summer months with erratic and unpredictable precipitation including drought-like periods and extreme spells of rain causing flash floods [ASTEC, 2011].

Figure 1.

(a) Location of the sampling site on the map of India with the state of Assam shown in yellow; and (b) MODIS Terra Satellite Retrieval of Aerosol Optical Depth (AOD) over the sampling site during the period of January 20, 2011 to February 3, 2011.

[5] Measurements of BC mass concentration were carried out at ∼8 m above ground level using a micro-Aethalometer (Magee Scientific, Model AE51). This model is lightweight (250 g), small (117 × 66 × 38 mm3), and provides automatic, near real-time, filter-based measurement of BC mass concentration [Ferrero et al., 2011]. It measures the attenuation of light from an 880-nm wavelength LED source by a PTFE-coated borosilicate glass fiber filter (Fiberfilm™ Filters, T60 material, Pall Corporation) upon which aerosol is being deposited. The change in optical filter transmittance is calculated and converted to BC mass concentration [Ferrero et al., 2011]. During our study, the instrument was operating continuously with a PM2.5 sampling inlet and the sampling period was 1 min. The data were subsequently averaged over 30 min periods.

3. Results and Discussion

3.1. Black Carbon Concentration

[6] Median values of the diurnal cycle of BC mass concentration are shown in Figure 2 and range from 9 to 41 μg m−3 with two major maxima in the morning (i.e., 7:30–9:00 local time) and the evening (i.e., 21:00–23:00 local time). Variability in BC mass concentration is greatest during the evening and early morning hours with values over 50 μg m−3 on multiple occasions. Sunrise was at ∼6:10 and sunset was at ∼17:05 local time during the sampling period. The diurnal variation in BC mass concentration coincides with the diurnal variability of local weather parameters, which are plotted in Figure 3. BC mass concentrations rise from sunrise until they peak between 8:00 and 8:30 after which they fall until reaching a relatively steady minimum in the mid-afternoon from about 14:00 to 16:30. High BC mass concentrations in the morning hours are due to increased BC emissions compounded with pollutants being trapped near the surface in a residual shallow nocturnal boundary layer [Stull, 1988]. Increasing BC emissions in the morning are from anthropogenic activities including transportation emissions from the morning commute and biomass burning associated with domestic heating. Temperature and wind speed maxima occur in the mid-afternoon and coincide with the minimum of BC mass concentration. During the mid-afternoon, surface BC concentrations are reduced as BC is mixed vertically by the deepening boundary layer and dispersed horizontally by higher wind speeds (between 1.7 and 3.5 m/s). BC mass concentrations begin to rise after 16:30. After sunset, reduced vertical mixing results from subsidence of the boundary layer, and consequently BC mass concentrations increase until 20:30 to 23:00 as emissions are trapped near the surface. Anthropogenic emissions from industrial activities and transportation sources are reduced in the evening while contributions due to biomass burning for domestic heating increase. All anthropogenic emissions are reduced as the night progresses, and as a result BC concentrations decrease slowly.

Figure 2.

Median values of the diurnal cycle of BC mass concentration. Error bars represent upper and lower quartiles. Plotted points are the median values for aggregated data from the subsequent half hour for available days (e.g., 1:00 represents the time period from 1:00 to 1:29).

Figure 3.

Median values of the diurnal cycle of various meteorological parameters: (a) temperature (in Kelvin), (b) relative humidity (in %), (c) air pressure (in kiloPascal), and (d) surface wind speed (in meters per second) during the measurement time period in Guwahati.

[7] A global survey of BC mass concentrations at various urban locations is given in Table 1 for comparison with values measured in this study. The range of median BC values measured in Guwahati, India are much higher than values measured elsewhere. Overall values are higher than those from even the most polluted cities in this list, which include Kolkata, India, Xi'an, China, and Hyderabad, India. Values in Guwahati are much higher (by a factor of 3 to 10 times) than those at urban locations in the USA and Europe. However, the time period during which measurements were taken is different for different locations, significantly changing BC emission rates and atmospheric dilution through mixing, both important to BC mass concentrations. Aerosol optical depth (AOD) retrievals from the Moderate Resolution Imaging Spectroradiometer (MODIS) on the Terra satellite are shown in Figure 1b with elevated aerosol optical depth (AOD) values (i.e., 0.66–0.82) at and near the sampling site, confirming elevated aerosol concentrations in the region.

Table 1. Mean/Median BC Values in Urban Locations Worldwide
LocationPeriodBC (μg m−3)
Delhi, IndiaaDec16.7
Delhi, IndiabFeb19
Hyderabad, IndiabJan21
Kanpur, IndiacDec6–20
Kolkata, IndiadAnnual26.5
Mumbai, IndiaeJan–Mar7.5–17.5
Bangalore, IndiafNov0.4–10.2
Xi'an, ChinagJan21.6
Urban, EuropehDec–Feb3.5–4.2
Maryland, USA (suburban)iAnnual0.25–3
Guwahati, India (present study)Jan–Feb9–41

3.2. Estimation of Aerosol Radiative Forcing

[8] The primary input parameters required for calculating aerosol radiative forcing are AOD, aerosol single scattering albedo (SSA) and asymmetry parameter (g) and surface albedo. We have used the Optical Properties of Aerosols and Clouds (OPAC) model developed by Hess et al. [1998]to obtain these parameters. OPAC estimates wavelength-dependent AOD, SSA, and g for aerosols which are spherical and externally mixed, at eight values of relative humidity (RH) (0%, 50%, 70%, 80%, 90%, 95%, 98%, and 99%). It also facilitates user-defined mixtures of aerosol components to best fit the measured aerosol parameters such as BC mass concentration (in our case). For this study, the urban aerosol mode of OPAC – consisting of water-soluble, insoluble, and black carbon aerosol components – was used at 80% RH to match the urban landscape and meteorological conditions of the sampling site. The measured mean BC mass concentration was input into the urban aerosol mode of OPAC to estimate the AOD, SSA, and g in the 0.2 to 5.5 μm wavelength range. The number density of BC aerosols in the urban mode was varied to match the measured monthly mean BC mass concentration, while the number densities of insoluble and water-soluble aerosols were kept unchanged. This iterative procedure is described in earlier papers [Babu et al., 2002]. The daily mean BC mass concentration measured during our sampling period was 23.2 μg m−3 with a standard deviation of 8.7 μg m−3. A number concentration of 390,000 particles cm−3 of BC in the urban aerosol mode of OPAC corresponded to our measured mean BC mass concentration. Similar high number concentrations of BC aerosols estimated using OPAC have been previously reported for different regions of India [Pant et al., 2006; Ramachandran and Kedia, 2010].

[9] Spectrally-varying AOD provides qualitative information pertaining to aerosol size distribution and loading. The variation of AOD as a function of wavelength (λ) is best described using the Ångström power law expression [Schuster et al., 2006]:

display math

where α is the Ångström exponent and β the turbidity coefficient. α is a qualitative indicator of the aerosol size in the atmosphere. β is a measure of total particulate load in a vertical column and is equal to AOD at 1 μm wavelength. A value of α ≤ 1 implies the aerosol size distribution is dominated by coarse-mode aerosols such as dust and sea salt aerosols, whereas a value ofα ≈ 2 indicates that the size distribution is dominated by fine-mode aerosols such as freshly generated combustion aerosols [Schuster et al., 2006]. For this study, by evolving a least squares fit between the OPAC output values of AOD and λon a log-log scale,α and βwere evaluated to be 1.3 and 0.4 respectively. These values indicate that aerosol loading over our sampling site was significant and its size distribution was dominated by fine-mode aerosols (i.e., sub-micron or less in size), which typically are associated with vehicular emissions and open biomass burning. The AOD at a wavelength of 0.55 μm estimated by OPAC was ∼0.9, which is in agreement with the mean AOD measured by MODIS Terra satellite over Guwahati, India during January (see Figure 1b).

[10] To estimate the short-wave clear sky aerosol radiative forcing at the surface and top of the atmosphere (TOA), the estimated values of AOD, SSA, and g by OPAC were input into thelibrary for radiative transfer (libRadtran) – a collection of algorithms for calculation of solar and thermal radiation in the Earth's atmosphere [Mayer and Kylling, 2005]. Aerosol radiative forcing is sensitive to the surface albedo. Surface reflectance measured over Guwahati by MODIS Terra and Aqua satellites (16-Day, Level 3 Global 500 m 1 km SIN Grid) at seven wavelength bands centered at 0.645, 0.859, 0.469, 0.555, 1.24, 1.64, and 2.13 μm were used. The radiative forcing calculations were performed using eight radiation streams at 1 h intervals and 24 h averages were obtained. The solar zenith angles were computed at 5° interval. Aerosol radiative forcing at the top of the atmosphere (TOA) and surface (SFC) is defined as the difference between the net flux with and without aerosols. For this study period, the estimated mean SFC forcing (clear-sky and diurnally averaged) was −63.4 Wm−2and mean TOA forcing (clear-sky and diurnally averaged) was +11.1 Wm−2. The positive TOA forcing indicates that the color and optical thickness of the aerosols yield a scene that is darker when viewed from TOA than that of the bright surface in the absence of aerosols. The difference between the radiative forcing at the TOA and the SFC is designated as the net atmospheric forcing ΔF, which represents the amount of solar energy absorbed in the atmosphere by aerosols. The estimated mean value of ΔF over Guwahati during this study's period was +74.5 Wm−2. The heating rate (Kelvin per day) due to ΔF is calculated from the first law of thermodynamics and hydrostatic equilibrium as follows [Ramachandran and Kedia, 2010]:

display math

where ∂T/∂t is the heating rate in K d−1, g/Cp is the lapse rate (taking g as the acceleration due to gravity and Cp the specific heat capacity of air at constant pressure = 1006 J kg−1 K−1) and ΔP is atmospheric pressure difference (taking ΔPas 300 hPa, mid-latitude pressure width of troposphere) in the first 3 km above ground. The clear-sky ∂T/∂t for this study was calculated to be ∼2 K/d.

[11] To examine the possible outflow trajectories of aerosols from Guwahati, 7-day isentropic back trajectories, computed using the Hybrid Single Particle Lagrangian 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 msl at every km were computed and analyzed. The trajectories at all the different heights suggested outflow of pollutants to continental China and Tibet. Such outflow has been suggested to be a major cause of the rapid melting of glaciers and permafrost [Hansen and Nazarenko, 2004].

4. Conclusions

[12] Winter-time BC mass concentrations observed in Guwahati – a major city in the BRV region of SE Asia – are higher than those measured in mega cities of India and China, and much higher than in urban locations of Europe and USA. Median values of the diurnal cycle of the BC mass concentration over the study period in January–February 2011 are in the range of 9–41 μg m−3, with individual values as high as 50 μg m−3during evening and early morning times. Analysis of spectrally-varying AOD suggests that sub-micron sized aerosols, likely emitted from combustion processes, dominated the aerosol pollution during the sampling days. The high level of BC mass concentrations reduces the shortwave radiation reaching the surface by 63.4 Wm−2 and reduces the outgoing shortwave radiation at the top of the atmosphere by 11.1 Wm−2under cloud-free conditions. The shortwave atmospheric absorption translates to a clear-sky lower atmospheric heating rate of ∼2 K/d. This large surface cooling accompanied with significant atmospheric heating could qualitatively explain the regional climate change in the BRV region. Such a situation could intensify low-level inversion, which slows down convection and in turn inhibits cloud formation. Additionally, indirect effects associated with BC aerosols such as the cloud ‘burn off’ effect [Ackerman et al., 2000] could affect the normal precipitation pattern over this region. Speculations based on limited observation from one single station in the BRV region such as ours are not enough to adequately address this region's climate change. Nonetheless, this study emphasizes the influence of large BC emissions on the climate of the BRV region and the pressing need for future studies.


[13] This material is based upon work supported by NASA EPSCoR under Cooperative Agreement NNX10AR89A, by NASA ROSES under grant NNX11AB79G, and by the Desert Research Institute. We acknowledge Tony Hansen of Magee Scientific Corp, California for loaning a micro-Aethalometer to conduct this study. His valuable suggestions on careful interpretation of the data are much appreciated. We also acknowledge thepro bono help from Mr. and Mrs. Jagadish Chakrabarty of Guwahati in setting up the sampling platform and performing timely filter change during sampling.

[14] The Editor thanks two anonymous reviewers for assisting with the evaluation of this paper.