Radiative Forcing of Carbonaceous Aerosols over Two Urban Environments in Northern India

The radiative forcing of elemental carbon (EC) and organic carbon (OC) has been estimated over two urban environments in Northern India (Jabalpur [JBL] and Udaipur [UDPR]) from November 2011 till November 2012 (till September 2012 over Jabalpur). The elemental carbon concentrations reached 7.36 ± 1.99 μg m over JBL and were as high as 10.78 ± 4.85 μg m over UDPR, whereas the corresponding OC concentrations were much higher in different months (as high as 19.37 ± 12.6 μg m over JBL and 39.71 ± 13.05 μg m over UDPR). The radiative forcing for OC and EC has been estimated using an optical model along with a radiative transfer model. The surface OC radiative forcing was found to range from –2.19 ± 1.93 W m to –3.083 ± 2.29 W m over JBL and –1.97 ± 1.37 to –5.89 ± 2.17 W m over UDPR, whereas the estimated top of the atmosphere (TOA) forcing ranged from –0.87 ± 0.49 to –1.87 ± 0.90 W m over JBL and from –1.23 ± 0.31 to –3.44 ± 1.51 W m over UDPR. However, the effect of EC forcing (as high as –21.75 W m at the surface of and +6.3 W m at TOA over JBL and –38.21 W m at the surface of and +5.05 W m at TOA over UDPR) was found to be more than tenfold higher than OC forcing due to its strong atmospheric absorption, in spite of much lower concentrations compared to OC.


INTRODUCTION
This paper exclusively presents results on the estimation of radiative forcing of two important carbonaceous climateforcing agents, viz., elemental carbon (EC) and organic carbon (OC), over two urban environments in India.EC, having similar absorption characteristics as black carbon (BC) and synonymously used (Chung and Seinfeld, 2002), forms a sizeable fraction of soot.It includes the solid form of pure carbon emitted by incomplete combustion of fossil fuel and biomass.The major anthropogenic activities responsible for its emission are automobile exhaust, industrial processes, coal burning, biofuel and burning of crop residue after harvest (Sahu et al., 2008;Cherian et al., 2009;Kumar et al., 2011;Tiwari et al., 2015), whereas the natural activities responsible for its emission are mainly volcanic eruption and forest fires (Reddy et al., 1999;Martinsson et al., 2009).Global distribution and plausible sources of OC and EC over different environments are explained in et al., 2002).EC is a good absorber of radiation at short wavelengths, and hence, it warms the earth-atmosphere system.The mixing state of EC in its core-shell structure is found to induce different climate forcing.As such, the enhanced atmospheric heating by EC can have a strong impact on the earth's radiation budget and, hence, on the weather and climate (Jacobson, 2000).
Atmospheric OC consists of a complex mixture of various carbon compounds found in the form of primary as well as secondary carbon particles.An appreciable fraction of the atmospheric OCs are volatile at low temperature and coexist with semi-volatile and non-volatile organic compounds in the atmosphere.OC is generated from both anthropogenic (such as the burning of fossil fuel and biomass burning) and natural (such as forest fires and gas-to-particle conversion) sources.OC aerosols are generally of the scattering type and act as good CCN in the atmosphere.Many studies reported that water-soluble organic carbon (WSOC) dominates the OC fraction compared to its insoluble counterpart.WSOC is an important atmospheric constituent, as it is a major scattering species and also can significantly improve the hygroscopicity and CCN activation, controlling the cloud formation (Miyazaki et al., 2006).Secondary organic aerosols are considered to be the major source of WSOC (Miyazaki et al., 2006).Mayol-Bracero et al. (2002) has reported that 45-75% of the total OC is contributed by WSOC.Studies over Po Valley (northern Italy) revealed that the WSOC concentration can reach up to ~58% of the total OC (Stefano et al., 2000).Online measurements of WSOC in total suspended aerosols in St. Louis, Missouri, (in U.S.A.), carried out in 2003 showed ~64% WSOC in June, ~61% in August and ~31% in October (Sullivan et al., 2004).An experiment carried out from September 2006 till August 2007 at the ground location SMEAR III, situated in Helsinki, a metropolitan coastal area in Southern Finland, showed that WSOC accounted for 60% with episodes of biomass burning and ~57% without the biomass burning (Saarikoski et al., 2008).Several studies have reported the measurement of and variation in OC and EC across the globe (Lim et al., 2003;Park et al., 2005;Kondo et al., 2006;Murillo et al., 2013;Yttri et al., 2013).Some of the studies have reported radiative-forcing estimates exclusively of BC aerosols (Hiennola et al., 2016;Peng et al., 2016;Yang et al., 2017).Studies in BC measurements and radiative forcing also have been reported by many investigators over the Indian region (e.g., Sreekanth et al., 2007;Singh et al., 2010;Surendran et al., 2010;Tiwari et al., 2015), which showed a pattern of negative forcing at the surface and positive forcing at TOA due to BC.It is reported that BC forcing can contribute 55-75% of total aerosol atmospheric forcing, even though its mass contribution in total suspended particles is less (Sreekanth et al., 2007;Panicker et al., 2010).However, such studies on OC forcing are sparse.Hence, in this study, we are reporting the important estimates of radiative forcing by OC along with EC over JBL and UDPR.

Data Sampling and Extraction
The sampling of OC and EC over JBL and UDPR was performed by using a fine-particulate-matter sampler (APM 550 MFC, Envirotech Instrument Pvt.Ltd.) operated at a flow rate of 16.7 L min -1 using quartz filter paper (2500 QAO-UP, Palls Life Science Inc., with a diameter of 47 mm).Details of the meteorological conditions at the sites are available in Panicker et al. (2015).The Palls quartz filter papers were pre-treated for 4 hrs in the muffle furnace at 900°C to remove any contamination through human error or carbon residue (Ali et al., 2016).The sampling was done at six-hour intervals.The filter papers were analyzed with a DRI Thermal/Optical Carbon Analyzer (Model 2001, Atmoslytic Inc., USA) for obtaining OC and EC fractions.The IMPROVE_A protocol (Chow et al., 2011) was utilized by the instrument to detect EC and OC concentrations.A small sample punch (0.496 cm 2 ) was taken from a quartzfiber filter, and the carbonaceous compounds adsorbed onto it were converted to carbon dioxide (CO 2 ) by passing it through an oxidizer (heated manganese dioxide [MnO 2 ]).A methanator (hydrogen-enriched nickel catalyst) further reduced this CO 2 to methane (CH 4 ), and CH 4 equivalents of OC and EC were quantified with a flame ionization detector (FID).OC subfractions, viz., OC 1 , OC 2 , OC 3 , OC 4 , were obtained, respectively, at 140, 280, 480 and 580°C temperatures (in a non-oxidizing helium [He] atmosphere), and EC subtractions, viz., EC 1 , EC 2 and EC 3 , were obtained, respectively, at 580, 740 and 840°C temperatures (in an oxidizing atmosphere of 10% oxygen in a balance of helium).A detailed description of data extraction using a DRI analyzer is available elsewhere (Panicker et al., 2015;Ali et al., 2016).

Methodology
The aerosol optical model OPAC (Optical Properties of Aerosols and Clouds) (Hess et al., 1998) and radiative transfer model SBDART (Santa Barbara Discrete ordinate Atmospheric Radiative Transfer) (Ricchiazzi et al., 1998) were used for obtaining the radiative forcing of OC and EC.The OPAC model derives aerosol optical properties, such as the aerosol optical depth (AOD), single-scattering albedo (SSA) and asymmetric parameter (ASP), and Angstrom exponent (ANG), from observations of chemicalcomposition datasets.This model uses a default database and has provisions to include user-specified inputs.The model includes different environments, such as urban, continental polluted and continental average, which can be chosen according to the prevailing environmental conditions.The OPAC model constrains the surface number concentration provided to obtain a vertical profile using equation N = N(o) e (-h/z) , where the N(o) is surface number concentration of individual aerosol species and Z is scale height.More details of the OPAC model are available elsewhere (Hess et al., 1998).The EC data obtained from the DRI Thermal/Optical Analyzer was converted into number density using the methodology proposed for the urban model in Hess et al. (1998) for BC and then used in an urban environment of the OPAC model to obtain AOD, SSA and ASP and ANG for EC.
As there is no predefined OC component in the OPAC model, the OC mass has been bifurcated into water-soluble OC (WSOC) and water-insoluble OC (WIOC) using the methodology reported by Ram et al. (2010) in a similar environment.The default refractive index of water-soluble and insoluble components of the OPAC model has been replaced with the refractive index of WSOC and WIOC (Arola et al., 2011) at 450 and 650 nm in order to create an OC environment in the model.Furthermore, the observed OC mass fractions have been converted into insoluble and water-soluble number density (Hess et al., 1998) and used in the OPAC model to get AOD, SSA, ASP and ANG for OC at different humidity conditions.
The optical properties obtained for OC and EC, i.e., AOD, SSA and ASP, were separately included in the SBDART model (Ricchiazzi et al., 1998) to obtain shortwave fluxes at the surface and the top of the atmosphere (TOA).This model uses the assumption of plane-parallel atmosphere.It is provided with default and user-specified options for various atmospheric parameters.We used the OC and EC optical parameters as mentioned above.Apart from that, we used surface albedo obtained from MODIS satellite observations (https://ladsweb.modaps.eosdis.nasa.gov/api/v1/productPage/product=MCD43A1C) for this study.Tropical model profiles of temperature and humidity (McClatchy et al., 1972) also were utilized for this study.Aerosol profiles in the model (McClatchy et al., 1972) are constrained with the visibility at 5 and 23 km with respective scale heights of 0.99 and 1.45 km.For an intermediate value of visibility, it uses a weighted average of scale heights (Panicker et al., 2008).Ozone concentrations were obtained from OMI (Ozone monitoring instrument), and the monthly mean of total atmospheric water-vapor column was acquired from MODIS level 2 products (https://ladsweb.modaps.eosdis.nasa.gov/api/v1/productPage/product=MYD07_L2).
Shortwave fluxes were derived at the surface and TOA separately by using the optical properties of OC and EC; in both cases, common inputs were ozone, water vapor and atmospheric profiles.Also, fluxes were derived for "no aerosol" conditions by switching off the aerosol components.The difference between fluxes for aerosol and no-aerosol conditions was used to obtain the forcing at the surface and at TOA separately for OC and EC over JBL and UDPR.Forcing efficiency (F eff ), which is the forcing per unit optical depth (Conant, 2000;Pandithurai et al., 2004;Panicker et al., 2014a, b), was also estimated to standardize forcing for different AOD conditions.

Radiative Forcing of Elemental Carbon
A detailed discussion of measurement methods, variations and basic characteristics for OC and EC over JBL and UDPR is provided in Panicker et al. (2015).The major sources of OC and EC over JBL were biomass burning and vehicular exhaust.However, a wide spectrum of sources, such as road transport, fossil-fuel emissions, road dust and biomass burning, contributed to OC and EC over UDPR (Panicker et al., 2015).The monthly variations in OC and EC mass are presented in Panicker et al. (2015) and also given in Table 1.Aerosol direct radiative forcing is a quantitative estimate of the amount of radiation lost by the surface and gained by the atmosphere due to aerosol scattering/absorption.The radiative forcing for EC and OC over JBL and UDPR has been estimated as explained in section 2. Such estimated forcing over JBL and UDPR during different seasons, viz., pre-monsoon (March-May), monsoon (June-September), post-monsoon (October-December) and winter (January-February), and plausible reasons for its variation are explained below.
The direct-radiative-forcing estimates of EC over JBL (at the surface and TOA) are depicted in Fig. 1.The average EC forcing over JBL varied from -16.99 ± 12.68 to -21.75 ± 10.24 W m -2 at the surface and +3.87 ± 2.3 to +6.31 ± 2.51 W m -2 at TOA.A minus sign for surface forcing denotes the loss of radiation at the surface, whereas a positive sign for TOA forcing denotes strong radiative absorption in the atmosphere.The net atmospheric forcing by EC varied from 21.2 ± 15.38 to 28.03 ± 12.04 W m -2 over JBL (Fig. 3).The EC radiative forcing over UDPR was found to Table 1.Monthly concentration of OC and EC over JBL and UDPR (Panicker et al., 2015).

Month
Jabalpur (JBL) Udaipur (UDPR) EC (µgm -3 ) OC (µgm -3 ) EC (µgm -3 ) OC (µgm  be higher than that over JBL.It ranged between -18.38 ± 11.84 and -38.21 ± 16.24 W m -2 at the surface and between +2.34 and +5.05 W m -2 at TOA during the period of observation.The net atmospheric forcing over UDPR varied from 20.72 ± 12.68 to 43.22 ± 18.99 W m -2 (Fig. 3).Seasonal variation in the EC forcing over JBL and UDPR is shown in Fig. 1(b), which indicates that the highest surface forcing occurs in the post-monsoon season (-19.43 ± 13.21 W m -2 ) over JBL, propotional to the higher mass concentration during this season, which could be associated with more biomass burning at this time.The lowest forcing was found in the monsoon season and is associated with the rain and wash out of EC.By contrast, the EC forcing over UDPR was highest in the pre-monsoon season (-34.32 ± 16.81 W m -2 ) and showed lower values in winter.The higher EC forcing in the pre-monsoon season could be associated with large concentrations of EC over the region due to long-range transport.
The EC forcing at the surface was more or less propotional to the EC mass over both stations.For example, the EC concentration over JBL was high during January 2012 (7.36 µg m -3 ), which induced high forcing (-21.75W m -2 ) (Fig. 1) at the surface.However, the propotional difference in EC forcing with respect to the difference in EC mass (for example, over JBL, the EC mass was lowest in September 2012 (Table 1), whereas its magnitude of surface forcing was higher than during February 2012, when the EC mass was higher) is associated with the inclination of solar radiation in different months.The magnitude of EC forcing obtained here is comparable to those extracted using BC fractions reported over different regions in India (Panicker et al., 2010;Tiwari et al., 2015;Sreekanth et al., 2007).

Radiative Forcing of Organic Carbon
Very few studies have reported exclusive radiative forcing of the OC component.Most of the results on OC forcing are model-based (e.g., Haywood and Boucher, 2000;Yin et al., 2015), and observational results are sparse.The estimated OC forcing at the surface and TOA are shown in Fig. 2. OC is a combination of water-soluble and water-insoluble fractions.The forcing has been estimated by bifurcating the OC into WSOC and WIOC in different months according to Ram et al. (2010).The forcing by OC was found to be negative at the surface as well as TOA.The surface-forcing values were found to vary from -2.19 ± 1.93 W m -2 to -3.083 ± 2.29 W m -2 over JBL and -1.97 ± 1.37 to -5.89 ± 2.17 W m -2 over UDPR.However the TOA values varied from -0.87 ± 0.49 to -1.87 ± 0.90 W m -2 over JBL and -1.23 ± 0.31 to -3.44 ± 1.51 W m -2 over UDPR.Seasonal variation in OC forcing is summarized in Fig. 2(b); it showed higher forcing in the post-monsoon season (-3.083 ± 2.29 W m -2 ) and lower forcing in the pre-monsoon season (-2.19 ± 1.93 W m -2 ) over JBL.The estimated OC forcing was found to be high in the pre-monsoon season (-4.88 ± 2.53 W m -2 ) and low in the post-monsoon season (-1.97 ± 0.8 W m -2 ) over UDPR.The forcing due to OC was also found to be merely propotional to the OC mass.However, the slight disproportionate variation in OC forcing could be associated with the difference in the combination of WSOC and WIOC during different months apart from the solar inclination.

Atmospheric Forcing of OC/EC and Forcing Efficiency
The atmospheric forcing of OC and EC, i.e., the difference between TOA and surface forcing, has been computed and depicted in Fig. 3. EC-induced atmospheric absorption showed an enhancement of up to 28 W m -2 (January 2012) over JBL.However, it was up to 43 W m -2 (May 2012) over UDPR.The OC atmosperic absorption was negligible compared to EC forcing.It may be noted that the OC mass is much higher than the EC mass; however, the EC atmospheric forcing was up to 10 times higher than that of OC forcing.This is because of the fact that EC is a strong absorber of radiation and warms the earth-atmosphere system.However, OC is a weak absorber and a good scatterer and therefore scatters radiation back and forth.Hence, the net forcing exerted by OC in spite of its high concentration is less significant compared to EC.
The forcing efficiency (F eff ) (forcing per unit optical depth) for OC and EC over JBL and UDPR are depicted in Figs.4(a)-4(b).A unit increase in AOD (at 500 nm) was found to increase EC forcing up to -295 W m -2 AOD -1 over UDPR and -254 W m -2 AOD -1 over JBL at the surface.However, EC-TOA F eff has shown a projection of up to +74.33 W m -2 over JBL and +42.4 W m -2 over UDPR.The OC F eff (at 450 nm) at the surface reached -77.8 W m -2 AOD -1 over UDPR and -48.55 W m -2 AOD -1 over JBL.The corresponding OC-TOA F eff was as high as -25.42 W m -2 AOD -1 over JBL and -45.46 W m -2 AOD -1 over UDPR.The seasonal mean surface F eff for EC and OC was found to be higher in the pre-monsoon season (-241.89± 168.24 W m -2 AOD -1 for EC and -48.56 ± 42.85 W m -2 AOD -1 for OC) over JBL.Lower surface F eff was found in the post-monsoon season for EC (-210.1 ± 142.77 W m -2 AOD -1 ) and during the monsoon season for OC (-38.48 ± 22.84 W m -2 AOD -1 ) over JBL.However EC-F eff over UDPR was higher in winter (-267.28 ± 177.56 W m -2 AOD -1 ), while OC-F eff was higher in the pre-monsoon season (-65.70 ± 34.28 W m -2 AOD -1 ).Similarly, the lower surface F eff over UDPR was in the post-monsoon season (-220.13 ± 201.70 W m -2 AOD -1 for EC and -20.99 ± 14.64 W m -2 AOD -1 for OC).The forcing efficiency, which is a projection of radiative forcing per unit increase in optical depth, necessarily is a function of AOD and radiative forcing.It may be noted from Table 2 that for a forcing of -19.43 W m -2 at the surface for an EC-AOD of 0.092 in November 2011, the F eff was -210.1 W m -2 .By contrast, for a lower forcing of -17.93 W m -2 in March 2012, the forcing efficiency was -241.89W m -2 AOD -1 at the surface (for an AOD of 0.074) over JBL.This illustrates that even though higher forcing is obtained for higher AOD, the F eff may be less and vice versa.However, when different forcings arise for the same AOD due to changes in available radiation associated with solar inclination, the higher forcing case will induce more F eff .E.g., the same AOD, ~0.074, in February and June 2012 over Jabalpur induced respective forcings of -16.99 W m -2 and -18.81 W m -2 at the surface due to differences in available radiation associated with solar inclination.Here, the respective F eff were ~229 W m -2 AOD -1 and ~254 W m -2 AOD -1 -i.e., the higher the forcing, the higher the F eff for the same AOD.Hence, this illustrates that forcing efficiency is strongly dependent on AOD, radiative forcing and solar inclination, which in turn depend on the individual seasons.
The net atmospheric radiative absorption was converted to atmospheric heating, and heating rates were computed, as explained in Panicker et al. (2014a) for OC and EC.Monthly variations in heating rates due to EC and OC are depicted in Fig. 5.The heating rates for EC ranged between 0.71 ± 0.63 and 0.93 ± 0.25 K day -1 over JBL and 0.70 ± 0.44 to 1.4 ± 0.82 K day -1 over UDPR.However, OC showed negligible heating rates compared to EC (0.035 to 0.045 K day -1 over JBL and 0.016 to 0.076 K day -1 over UDPR).
The sensitivity of radiative forcing to different parameters, such as albedo, and changes in mass concentration has been studied and depicted in Tables 2 and 3 for two representative months (November and March).It is seen that in spite of higher concentrations of EC and OC, the lower TOA forcing over UDPR is associated with lower albedo over the region.Hence, detailed analysis has been done with default vegetation albedo and different combinations of albedo surfaces (such as 50-50% vegetation-sand, 70-30% sand/vegetation and vice versa).It can be seen that changes in albedo induce changes of 1-13% in EC surface forcing and larger changes in TOA forcing.Similarly, larger changes were observed in TOA forcing for OC (Table 3).This study suggests that the albedo of the underlying  surface is a key parameter in extracting radiative forcing of carbonaceous aerosols.The sensitivity to mass concentration also has been examined and depicted in Table 2.A change of 10% in EC and OC mass was found to produce a 1-9% change in AOD and another 3-11% change in radiative forcing over the sites.

SUMMARY
1.The radiative forcing for EC and OC was estimated using the OPAC model in conjunction with the SBDART model.2. The average EC forcing at the surface of JBL varied from -16.99 ± 12.68 W m -2 to -21.75 ± 10.24 W m -2 and from +3.87 ± 2.34 to +6.31 ± 2.51 W m -2 at TOA in different months.The forcing values for OC were found to be significantly less in spite of its higher concentration than EC, indicating that it is a small climate-forcing agent compared to its counterpart.6.The atmospheric heating due to EC was found to reach 0.93 ± 0.25 K day -1 over JBL and 1.4 ± 0.82 K day -1 over UDPR, whereas it reached 0.045 ± 0.039 K day -1 over JBL and 0.076 ± 0.035 K day -1 over UDPR for OC. 7. The forcing efficiency of EC was found to be as high as -295 W m -2 over UDPR and -254 W m -2 over JBL at the surface.However, the OC forcing efficiency showed an enhancement as high as -77.81W m -2 over UDPR and -48.56 W m -2 over JBL. 8.The sensitivity of albedo to radiative forcing showed that changes in albedo induce changes of 1-13% in the EC surface forcing and larger changes in TOA forcing, indicating that the albedo of the underlying surface is a key factor in forcing estimates.A change of 10% in EC and OC mass produces a 1-9% change in AOD and a further 3-11% change in the radiative forcing.

Fig. 1 .
Fig. 1.(a) The Monthly and (b) Seasonal variation of EC forcing at surface and TOA.

Fig. 2 .
Fig. 2. (a) The Monthly and (b) Seasonal variation of OC forcing at surface and TOA.

Fig. 5 .
Fig. 5.The Monthly variation of (a) EC and (b) OC Atmospheric Heating rates.

3.
The surface EC forcing ranged between -18.38 ± 11.84 to -38.21 ± 16.24 W m -2 over UDPR.The TOA forcing ranged between +2.34 ± 0.847 W m -2 to +5.01 ± 2.75 W m -2 during the period.4. The negative forcing values at the surface show the reduction in radiation at the surface, and positive TOA forcing values indicate strong raduative absorption by EC aerosols.5.

Table 2 .
Sensitivity of AOD and forcing for change in mass of EC and OC.

Table 3 .
Sensitivity of OC and EC radiative forcing to different albedo conditions.