Estimate of radiative forcing of Asian biomass-burning aerosols during the period of TRACE-P

Authors


Abstract

[1] The regional radiative impact of biomass-burning aerosols in Asia is estimated using the new and detailed emission data during the experimental period of Transport and Chemical Evolution over the Pacific (TRACE-P) in March 2001. Integration of the USA National Oceanic and Atmospheric Administration (NOAA) Hybrid Single-Particle Lagrangian Integrated Transport model (HYSPLIT) and a solar radiative transfer model (CLIRAD-SW) allow us to simulate the spatial and temporal distributions of black carbon (BC) and organic carbon (OC) aerosols from biomass burning in the South Asian region. It also allows us to estimate further their aerosol optical properties and radiative forcing. We find that an anticyclone over Bay of Bengal dominates the transport of pollutants of the South Asian region. The monthly mean surface concentration of OC and BC is 1.1 μg m−3 in this region. Western Myanmar has the maximum value, with the concentration reaching 12.5 μg m−3. The monthly mean clear-sky direct shortwave radiative forcing ranges from −1.9 to 0.4 W m−2 at the top of the atmosphere and from −0.5 to −12.0 W m−2 at surface, resulting in an increase of the atmospheric heating rate from 0.01 to 0.3°C day−1. Owing to the spatial distributions of the aerosol optical depth ratio (OC/BC) and the surface albedo, there is a strong gradient of heating rate near the source regions, which may modify local circulations.

1. Introduction

[2] Atmospheric aerosols play an important role in atmospheric energy balance and climate change. They affect the climate through direct interaction with solar and terrestrial radiation (direct aerosol effect) and through their effect on the optical properties and life cycle of clouds (indirect aerosol effect) [Charlson et al., 1992; Chylek and Wong, 1995; Levine et al., 1995; Haywood and Boucher, 2000]. By backscattering solar radiation, aerosol particles and clouds exert a cooling (parasol) effect on the climate, dampening the greenhouse gas-induced warming of the Earth [Intergovernmental Panel on Climate Change [IPCC] 2001; Crutzen and Ramanathan, 2003]. Biomass-burning aerosols are a major contributor to the global aerosol loading and radiation budgets [Crutzen and Andreae, 1990; Penner et al., 1992]. Jacobson [2004] shows biomass-burning aerosols are important contributors of global temperature change owing to the cooling effect, as opposed to warming effect due to greenhouse gases. However, the quantitative estimate of aerosol radiation effect is complex because of uncertainties in the emission inventory, chemical component, microphysical properties, optical properties, and, most importantly, their interaction with clouds. Even now, only Jacobson performed an exhaustive simulation.

[3] The major element of biomass-burning aerosols is carbonaceous, which is generally categorized into two types: black carbon (BC) and organic carbon (OC). The BC aerosols absorb strongly sunlight that enhances solar heating of the atmosphere. As a result, there is less penetration of solar energy to the Earth’s surface. It has been suggested [Jacobson, 2000; Hansen et al., 2004] that the radiative forcing of BC contributes substantially to global warming. In contrast to BC, OC aerosols, like sulfate aerosols, mainly scatter solar radiation.

[4] Previous studies applied global models to estimate the direct radiative impact of biomass-burning aerosols. For example, IPCC [2001] estimated the annual global mean radiative forcing of biomass-burning aerosols at the top of the atmosphere (TOA) to be approximately −0.2 W m−2 with a large uncertainty. It is estimated to be in the range of −0.14 to −0.74 W m−2 by Haywood and Boucher [2000], ∼0 and −0.22 W m−2, respectively, for all-and clear-sky conditions [Takemura et al., 2002], and 0.1 and −0.08 W m−2 by Reddy et al. [2005]. More recently, Jacobson [2004] showed that total effect of biomass-burning aerosols (solar plus infrared and direct plus indirect) is about −0.84 W m−2 at the TOA. With the exception of Jacobson, the treatment of aerosol is highly simplified in general circulation models, which tends to underestimate atmospheric solar absorption, particularly in areas experiencing extensive biomass burning [Wild, 1999, 2005]. Furthermore, biomass-burning aerosols have a relatively short lifetime (∼5 days). They have larger spatial and temporal variations, which may substantially affect regional radiative patterns [Ramanathan et al., 2001b] and imply a redistribution of the energy at the regional scale in a climate system [Reddy et al., 2005]. There were a few projects conducted for studying the characteristics of regional biomass-burning aerosols and their climate effect, such as Southern African Regional Science Initiative (SAFARI) [Andreae et al., 1996] and Indian Ocean Experiment (INDOEX) [Ramanathan et al., 2001a]. Neither the observation nor the modeling results from these experiments showed the regional radiative forcing to be one order of magnitude stronger than the global mean values. For example, Myhre et al. [2003] simulated the monthly direct radiative forcing of biomass-burning aerosols in the southern African region during SAFARI-2000 and found that the maximum clear-sky direct radiative forcing is −50 W m−2 and the monthly mean is −4.3 W m−2 for September 2000. In Asia, the complex emission source (industry, dust, and biomass burning) make it difficult to quantify the radiative forcing due to biomass-burning aerosols alone. Nevertheless, Ramanathan et al. [2001a] estimated the clear-sky direct radiative forcing of natural and anthropogenic aerosols to be −7 ± 1 W m−2 over the North Indian Ocean from January to March 1999 (INDOEX).

[5] Southeast and South Asia is one of the major biomass-burning emission source regions in the world [Streets et al., 2004]. The smoke plume from the biomass burning generally spreads downwind thousands of kilometers away and affects air quality, human health, and regional climate. Until now, information on the regional distribution of biomass-burning aerosols from Asia is limited, and their regional radiative impact is not well understood. We have conducted a model simulation to estimate the radiative impact of biomass-burning aerosols in Asia. We integrate a regional transport model with a more accurate solar radiative transfer model based on a detailed aerosol emission inventory obtained during intensive observation period of the Transport and Chemical Evolution over the Pacific (TRACE-P) in March 2001 [Jacob et al., 2003]. We adopt the detailed biomass-burning aerosol (OC and BC) emission and aerosol optical properties data developed for TRACE-P for input to our model. Our results are compared with observations and other modeling studies. Finally, we address the impact of biomass-burning aerosols on the hydrological cycle.

2. Methodology and Data

[6] We integrate the USA National Oceanic and Atmospheric Administration (NOAA) Hybrid Single-Particle Lagrangian Integrated Transport model (HYSPLIT, Draxler and Rolph [2003]) with a solar radiative transfer model (CLIRAD-SW, Chou and Suarez [1999]) and estimate the radiative forcing due to the biomass-burning aerosols in Asia. The domain of the model covers South and East Asia (70°–140°E and 0°–50°N) (Figure 1) with a spatial resolution of 1° × 1° altitude-longitude and a vertical resolution of 10 layers (top height of each layer, 10 m, 1000 m, 2000 m, 3000 m, 4000 m, 5000 m, 6000 m, 7000 m, 8000 m, and 9000 m from ground level). The simulation is performed for March 2001, which is the extensive observation period of TRACE-P. The temporal resolution is 1 hour for solving aerosol transport and 30 min for calculating radiative fluxes. The detailed methodology and data used are described in the following subsections.

Figure 1.

Model domain used in this study. The dashed box square is defined as the South Asian region in this study.

2.1. Emission and Meteorological Data

[7] Biomass burning is defined as forest burning, savanna/grassland burning, and the burning of crop residues in the field after harvest [Streets et al., 2003b]. Although the composition of biomass-burning aerosols include elements such as OC, BC, K+, Na+, Ca2+, Mg2+, NH4+, H+, Cl, H2SO4, HSO4, SO42−, and NO3[Jacobson, 2004], we focus only on the BC and OC aerosols because of the high emission amount and the tendency to influence and affect the climate [Ferek et al., 1998; Streets et al., 2003b; Reddy et al., 2005]. Streets et al. [2003b] and Woo et al. [2003] provided high spatial (1° × 1°) and temporal (daily) resolution of BC and OC emissions from biomass burning in the Asian region during TRACE-P. In the data set, the satellite data [e.g., advanced very high resolution radar (AVHRR) fire count and Total Ozone Mapping Spectrometer (TOMS) Aerosol Index] were used to determine the spatial/temporal variability. For the model simulation, the hourly emission is obtained simply by dividing the above daily values by daytime hours (local time 8:00–19:00). In addition, the data set also provides information about the estimation on emission uncertainties. We use this information to quantify the upper bound of uncertainties in raidative forcing estimation. Streets et al. estimated that uncertainties (measured as 95% confidence intervals) of the BC and OC emissions for the Asian region are ±420 and ±450%, respectively. These uncertainties are independent for each Asian region, for instance, for South Asian, exceeding ±500%. Streets et al. interpreted that “±500%” means “within a factor of six”, so that the confidence interval would be 17–600% of the mean value. Although the uncertainty of the estimated emission is large, the data we used in this study are currently the most complete and detailed Asian biomass-burning inventory. In comparison, global carbonaceous emission inventories for biomass burning, including those of Penner et al. [1993], Liousse et al. [1996], and Cooke and Wilson [1996], are adopted generally for climate model simulations. Discrepancies among these inventories arise mainly from the temporal resolution, the amount of biomass burnt, and the emission factors used for the various aerosol components of biomass burning.

[8] Figures 2a and 2b show the spatial distributions (1° × 1°) of OC and BC emission, respectively, for March 2001. Both have a similar pattern, and their major source regions are located in India, Indochina (a region that generally includes Thailand, Myanmar, Vietnam, Laos, and Cambodia), and southeastern China. Myanmar has the largest OC and BC emission. For the entire domain, the total OC emission of 0.74 Tg in March 2001 is much larger than the BC emission of 0.10 Tg. Both OC and BC emissions in March 2001 account for about 23% of annual emissions [Streets et al., 2003a; Woo et al., 2003]. Owing to various land use types, OC/BC ratios vary from place to place. It is noticed that the invariant ratio of 7.0 is assumed in a global study [Reddy and Boucher, 2004]. However, in India, Indochina, and other South Asia countries, the OC/BC ratio is about 6.5–8.0, higher than that of southeastern China, which has a mean of about 5.0 (Figure 2c).

Figure 2.

The 1° × 1° spatial distribution of biomass burning aerosol emission for (a) OC, (b) BC, and (c) OC/BC ratio in March 2001.

[9] The meteorological data used in the HYSPLIT are taken from the National Centers for Environmental Prediction (NCEP) reanalysis [Kalnay et al., 1996]. This data set has a spatial resolution of 190.5 km and a temporal resolution of 6 hours on 13 mandatory pressure levels and the surface level (http://www.cdc.noaa.gov/cdc/reanalysis). Additionally, we transfer the pressure coordinate to a σ coordinate in consistence with our modeling system.

2.2. HYSPLIT Model

[10] The HYSPLIT model is used to calculate the air parcel trajectories as well as the advection, dispersion, and deposition of pollutants. The numerical scheme for computing pollutant transport is a hybrid between Eulerian and Lagrangian approaches. Advection and diffusion calculations are made in a Lagrangian framework, while concentrations are calculated on a fixed grid [Draxler and Hess, 1998]. The puff and particle dispersions are used for this model. That is, the transport and dispersion of a pollutant is calculated by assuming the release of a single puff or many particles in each emission grid over the duration of the release. In the puff model, each simulated puff contains an appropriate fraction of pollutant mass and defined with either a Gaussian or top-hat horizontal distribution. The puff is advanced according to the trajectory of its center position, while the size of the puff (both horizontally and vertically) expends in time to account for dispersive nature of turbulent. When a puff expands to cover several meteorological grid points, top-hat puff splits horizontally into four puffs and each with 25% of the mass. In contrast to the top-hat puff, a Gaussian puff splits into five smaller puffs. The center puff contains 60% of the mass while each of the outside 4 puffs gets 10% of the mass. In this study, the dispersion of biomass-burning aerosols is configured as horizontal and vertical top-hat Puff simulation. Air concentrations are then calculated at specific points by assuming that the concentrations within the puff have a defined spatial distribution. A complete and detailed description of the model can be found elsewhere in the literature [Draxler and Hess, 1998; Draxler and Rolph, 2003]. The HYSPLIT model has also been applied to simulations of ozone [Draxler, 2000], dust [Draxler et al., 2001; Zhang et al., 2005], carbonaceous aerosols [Zhang et al., 2005], fire [Sapkota et al., 2005], daily smoke forecast, and contribution of emissions from large forest and wild fires in Indonesia, Mexico, and Florida (NOAA Smoke Forecasting Demo Project, http://www.arl.noaa.gov/smoke/forecast.html).

[11] The mixing state of biomass-burning aerosols is considered as an “external mixture,” meaning that each aerosol component is assumed to be contained in physically separated particles. By contrast, an “internal mixture” means that all aerosols contain a mixture of components from each of the sources. Jacobson [2001b] indicated that when the BC aerosol was associated with a scattering aerosol (either sulfate, nitrate, or organic carbon) by internal mixing, its absorption of sunlight was twice as strong. Also, Liao and Seinfeld [1998] showed that the internal mixing produced a larger heating effect than the external mixing of the same combination of aerosol species. Since the composition and optical properties of internally mixed aerosols are not well understood [Chin et al., 2004], we assume, for simplicity, that biomass-burning aerosols are externally mixed. The average BC and OC aerosol size distributions are based on the aerosol categorization obtained by Hess et al. [1998]. We consider BC as a “soot” aerosol component, and OC as a “water-soluble” aerosol component, in which the aerosol mode radii in the dry state were approximately 0.0118 and 0.0212 μm, respectively. The emission release height is set to be 100 m, the value used in the NOAA Smoke Forecasting Demo Project. Removal by dry deposition and wet scavenging are also factored into the study. We assume a dry deposition velocity of 0.001 m s−1 [Liousse et al., 1996], an in-cloud wet removal rate of 3.2 × 105 L L−1, and a below-cloud wet removal rate of 1 × 10−5 s−1 [Draxler and Rolph, 2003]. The lifetime of biomass-burning aerosols is set to 5 days [Liousse et al., 1996].

2.3. Aerosol Optical Properties and Optical Depth

[12] The wavelength-dependent aerosol single scattering albedo (ω), asymmetry parameter (g), and extinction coefficient (σe) and aerosol optical depth (hereafter referred to as AOD) are calculated using the Optical Properties of Aerosols and Clouds (OPAC) software package of Hess et al. [1998]. In the OPAC, the Mie theory is applied to compute the single scattering albedo, asymmetry parameter, and extinction coefficient at 61 wavelengths between 250 and 40,000 nm. The OPAC has been widely used to estimate the aerosol optical properties in the study of aerosol radiative effect. For instance, OPAC was used during the Bay of Bengal experiment [Ramachandran and Jayaraman, 2003], INDOEX experiment [Collins et al., 2002], and Asian Pacific Regional Aerosol Characterization Experiment (ACE-Asia) [Conant et al., 2003].

[13] Because the computational cost of directly including OPAC in our modeling system is too high, we adopt a treatment similar to Grant et al. [1999], in which we parameterize the optical properties using results of OPAC calculations. Table 1 lists the computed ω, g, and σe of hydrophobic BC aerosols at eight wavelengths (250, 300, 350, 500, 550, 1000, 1750, and 5000 nm). These values are saved for later use in the radiative transport model (as described in section 2.4). We consider the OC aerosol to be water-soluble with its size growing with the relative humidity. Figures 3a and 3b show that the ω and g of the OC aerosol increase with relative humidity. The relationship between the optical properties (ω and g) and relative humidity for the eight wavelengths can be fitted by the exponential growth equation:

equation image

where y is the optical properties (ω or g), and y0 is the optical properties at the relative humidity of 0%, A and t are the regression coefficients, and RH is the relative humidity. The fitting for each wavelength has a R2 greater than 0.95. Figure 3c shows the variation of σe with relative humidity for the OC aerosol at eight wavelengths. In order to obtain a better regression relationship, we further categorize relative humidity into three classes (0–50, 50–90, and 90–99%). Class 1 (RH = 0–50%) is a linear regression, and the other two classes are the fitting using equation (1). The R2 is 0.98 for all cases of fitting. The σe is normalized at a number density of 1 particle cm−3.

Figure 3.

(a) Single scattering albedo, (b) asymmetry parameter, and (c) extinction coefficient of OC aerosols against relative humidity at eight wavelengths. The lines represent the regression fit.

Table 1. Optical Properties (ω, g, and σe) of BC Aerosol Used in This Studya
WavelengthsAerosol Optical Properties
ωgσe
  • a

    The σe is normalized at a number density of 1 particle cm−3.

UV-250 nm3.08 × 10−15.02 × 10−11.36 × 10−6
UV-300 nm3.13 × 10−14.53 × 10−11.21 × 10−6
UV-350 nm2.91 × 10−14.22 × 10−11.02 × 10−6
VIS-500 nm2.26 × 10−13.53 × 10−16.39 × 10−7
VIS-550 nm2.09 × 10−13.36 × 10−15.54 × 10−7
IR-1000 nm9.65 × 10−22.25 × 10−12.47 × 10−7
IR-1750 nm3.18 × 10−21.35 × 10−11.31 × 10−7
IR-5000 nm2.40 × 10−33.66 × 10−24.49 × 10−8

[14] The total AOD, τ, for both BC and OC is calculated from

equation image

where σe,j is the extinction coefficient of the aerosols in layer j, normalized to 1 particle cm−3. Nj(h) is the total aerosol number density in the layer j, which is obtained from the HYSPLIT model simulation. In the range of RH= 90–99%, the σe increases rapidly (Figure 3c), and the AOD is very sensitive to RH. Using the regression relationship, the aerosol optical properties can be effectively computed as a function of RH.

2.4. Solar Radiative Transfer Model and Radiative Forcing

[15] The radiative forcing of aerosols is determined by taking together its optical properties with its temporal and spatial distribution in the atmosphere. The aerosol radiative forcing at TOA, in the atmosphere, and at the surface are computed using the solar radiative transfer model (CLIRAD-SW) developed by Chou and Suarez [1999] and Chou et al. [2002]. This model includes the absorption of solar radiation by water vapor, O3, O2, CO2, scattering due to gases (Rayleigh scattering), and the absorption and scattering due to aerosols. Interactions among the gaseous absorption and surface reflection are explicitly included using the δ–Eddington approximation [Joseph et al., 1976]. Fluxes are then computed using the two-stream adding approximation and virtually integrated over the entire spectrum from 175 to 10,000 nm, which is divided into 11 bands in the model.

[16] This model needs the values of the vertical profiles of temperature, humidity, ozone, and the aerosol optical properties in spatial and temporal variations. These values are obtained as follows. The temperature and humidity fields are based on the NCEP reanalysis data set as described in section 2.1. The vertical ozone distribution is the same as the global background values used in the Regional Acid Deposition Model (RADM) [Chang et al., 1987]. It is noticed that the clear-sky fluxes in the troposphere is not sensitive to the ozone amount [Chou et al., 2002], and the use of the climatological values of ozone does not have any significant effects on the results of this study. The spectral variation of the aerosol optical properties (ω, g, TOA) is derived using the OPAC outputs (see section 2.3). The surface albedo data for March 2001 is taken from the Moderate Resolution Imaging Spectroradiometer (MODIS) 16-day Albedo Products [Lucht et al., 2000; Schaaf et al., 2002].

[17] Radiative forcing has been used as an indicator of potential climatic importance and has often been linked to climate change [for example, Liao and Seinfeld, 1998; Haywood and Boucher, 2000]. Aerosol radiative forcing is defined as the change in net irradiance at a certain atmospheric level with and without aerosol effects. In this study, we focus on clear-sky direct radiative forcing of biomass-burning aerosols at the TOA (ΔFTOA) and at the surface (ΔFSFC) given by

equation image
equation image

where BBA stands for biomass-burning aerosols. The FTOA and FSFC represent the net downward fluxes at the TOA and surface, respectively. The radiative forcing in the atmosphere (ΔFATM) can also be defined as (ΔFTOA − ΔFSFC), representing the solar energy absorption by aerosols in the atmosphere.

[18] Moreover, the atmospheric heating rate (equation image) due to the absorption of solar radiation by various gases and aerosols can be calculated as follows [Liou, 2002]:

equation image

where ga is the gravitational acceleration, cp is the specific heat at constant pressure, and ΔF(p) is the net flux divergence of a pressure layer p. The change in atmospheric heating rate (ΔQ) with and without the contribution of biomass-burning aerosols is then given by

equation image

3. Results and Discussion

[19] In the following subsections, we present the results obtained from our modeling system. These results include the biomass-burning aerosol concentration, AOD, radiative forcing (ΔFTOA, ΔFSFC, ΔFATM), and ΔQ. Our discussions focus on the South Asian region between 70°–110°E, 5°–30°N, including Indochina, southern China, India, and Bay of Bengal, as shown in Figure 1. It is a major source/sink region of the biomass-burning aerosols.

3.1. Overall Meteorological Conditions and Transport of Biomass Burning Aerosols

3.1.1. Mean Meteorological Conditions

[20] The meteorological conditions and transport pathways of TRACE-P have been described by Fuelberg et al. [2003]. Fuelberg’s study, however, does not focus on the South Asian region. Here we present the mean meteorological fields in March 2001, which allow us to determine the general meteorological conditions over our modeling domain. However, the mean results may filter out some information such as the seasonal transition from winter to spring and the decay of prolonged El Niño Southern Oscillation (ENSO) cold phase conditions [Fuelberg et al., 2003]. Figure 4a depicts the mean sea level pressure field for a Siberian anticyclone (∼1028 hPa) centered at 50°N, 90°E and an Aleutian cyclone (∼1006 hPa) centered at ∼50°N, 160°E. These anticyclone and cyclone dominate the entire domain and drive the transport of Asian atmospheric pollutants, especially for the formation and intensification of Aleutian cyclone [Fuelberg et al., 2003]. Figure 4b shows that the mean surface flow north of ∼35°N is offshore. South of this region, the flow is more parallel to the coastline, with some onshore flow over Indochina. In the South Asian region, there is a weak anticyclone over the Bay of Bengal that dominates the transport of pollutants of this region substantially. That is, the anticyclonic flow over the Bay of Bengal leads to onshore flow over Indochina and India. Figure 4c shows the mean surface relative humidity, which is an important parameter affecting the optical properties of OC aerosols (see section 2.3). India and Myanmar exhibit low humidity conditions (<50%) in the March dry season.

Figure 4.

(a) Mean sea level pressure (hPa), (b) mean surface streamlines, and (c) mean relative humidity (%) in March 2001.

[21] Figures 5a, 5b, and 5c depict the mean geopotential height and wind field at 500, 700, and 850 hPa, respectively. In March, the West Pacific subtropical high centered at approximately 20°N, 140°E is well established. Its ridge axis extends toward Indochina [Chen and Chen, 2003]. Driven by the West Pacific subtropical high and middle latitude low (near 50°N, 140°E), the time-averaged wind field patterns reveal the presence of westerly flow over the midlatitude North Pacific Basin at each pressure level. At the 500-hPa level (∼5000 gpm), the westerly flow dominates most of the middle latitude areas. At the 700-hPa level (∼3000 gpm), the northwesterly offshore flow prevails north of ∼30°N and turns to westerly offshore flow around 25°N. In addition, there is an anticyclone centered at ∼15°N, 80°E, extending from 700 hPa to the surface (see Figure 4b). At 700 hPa, the anticyclonic flow transports the air mass from the Indian subcontinent to the Bay of Bengal and the Indian Ocean. At 850 hPa (∼1500 gpm), the anticyclone moves eastward into the Bay of Bengal. The southerly flow over Indochina extends northward to southeastern China. Figures 5d, 5e, and 5f depict the mean relative humidity at 500, 700, and 850 hPa, respectively. Compared to the spatial distribution of relative humidity over the Indian subcontinent and the land surrounding Bay of Bengal (Figure 4c), reversed relative humidity distribution is observed at 850 and 700 hPa. The difference in distribution of relative humidity at certain altitudes over this region is related to the vertical structure of anticyclonic flow.

Figure 5.

Mean geopotential height (gpm) and wind flows (m s−1) at (a) 500 hPa, (b) 700 hPa, and (c) 850 hPa in March 2001. Geopotential height intervals for Figures 5a–5c are 30, 20, and 10 gpm, respectively. Major cyclonic and anticyclonic circulation centers are labeled “L” and “H”, respectively. Mean relative humidity (%) at (d) 500 hPa, (e) 700 hPa, and (f) 850 hPa in March 2001.

3.1.2. Trajectory Analysis

[22] Trajectory analysis is used to trace the convective sources of air. In the study, the 5-day forward trajectories at 00 and 12 Coordinated Universal Time (UTC) during March 2001 are calculated using the HYSPLIT model (Figure 6). We conduct a study of the forward trajectory starting at SE China (25°N, 115°E), Thailand (15°N, 102°E), Myanmar (15°N, 102°E), and East India (18°N, 82°E), where the major source regions of the biomass-burning emission are located. The color code indicates trajectory altitudes during a 5-day computational period. Figure 6a indicates that the air mass, originated from SE China, travels along highly disordered routes and remains, for the most part, below the boundary layer, a behavior that is also revealed by the complex wind flow patterns at the surface (Figure 4b) and in the low troposphere (Figure 5c). In Figure 6b, most trajectories show northeastward routes over Indochina. A few trajectories reach southeastern China and the West Pacific Ocean. These trajectories are above 1000 m in altitude, corresponding to the convergence and orographic lifting at surface and southerly flow at 850 hPa. Figure 6c shows relatively stable eastward trajectories and higher trajectory altitudes. This is because a surface convergence and orographic lifting over Myanmar results in the uplift of an air parcel released from ground sources to the lower troposphere (∼2000 m). In the lower troposphere, the westerly flow prevails, forming a transport belt (around 25°N) for long-range transport. Two main trajectory groups form in East India’s source region, as shown in Figure 6d. One group presents clockwise trajectories over Bay of Bengal. The other group forms clusters of the trajectories traveling in a northeasterly direction to southern China. Their routes are initially driven by low-level convergence over the Indian subcontinent and subsequently dominated by anticyclonic flow and westerly flow at the lower troposphere.

Figure 6.

Five days forward trajectories of major biomass-burning emission areas: (a) SE China (25°N, 115°E), (b) Thailand (15°N, 102°E), (c) Myanmar (20°N, 95°E), and (d) East India (18°N, 82°E). Each trajectory starts at 00 and 12 UTC during March 2001, and trajectory altitudes (as mean sea level, meter) are denoted by the color scale.

3.1.3. Transport of Biomass Burning Aerosols

[23] Figure 7 shows the monthly mean concentrations of biomass-burning aerosols (OC + BC) and wind fields from the surface layer to an altitude of 3 km. In the South Asian region (dashed box in Figure 1), the monthly mean and standard deviation of the hourly biomass-burning aerosol concentrations are 1.1 μg m−3 and 3.6 μg m−3, respectively. The maximum mean concentration reaches 12.5 μg m−3 in western Myanmar. The aerosol concentration in the downwind areas such as the Indian Ocean and West Pacific Ocean ranges from 0.1 to 2.0 μg m−3. The 1-km layer (Figure 7b) has similar spatial distributions of wind field and biomass-burning aerosol concentration as the surface. This similarity indicates a vertically well-mixed layer from the surface to a height of 1 km. However, because of a strengthened northerly flow from the westward-tilted anticyclone, the plume of biomass-burning aerosols disperses from East India over the Bay of Bengal. In addition, the increasing southeasterly flow over Indochina carries more biomass-burning aerosols to southern China. At altitudes above 1000 m, the concentrations evidently decrease with height near the source regions, eventually dropping below 1.0 μg m−3 at altitudes above 3000 m (not shown in figure). Figure 7d shows the westerly flow forming a belt shape of biomass-burning aerosols around 25°N. This belt of aerosols is then transported downwind to East Asia at altitudes above 2 km, a result identical to that found in the study of Carmichael et al. [2003]. Overall, our results show a good agreement with those of INDOEX campaign [Rajeev et al., 2000; Lelieveld et al., 2001; Ramanathan et al., 2001a]. (The INDOEX study examined a 3-km-thick layer of the South and Southeast Asia biomass-burning aerosol. Its outflow extends to the Indian Ocean and the western Pacific.) However, it is difficult to quantify the concentration of biomass-burning aerosols directly through measurement alone.

Figure 7.

Simulated mean mass concentration of biomass burning aerosols and the wind field of NCEP reanalysis in March 2001 at four levels: (a) surface (10 m), (b) 1000 m, (c) 2000 m, and (d) 3000 m concentration. (Unit: G μg−3).

3.2. Optical Depth of Biomass Burning Aerosols

[24] As described in section 2.3, the AOD of each grid column is calculated on the basis of equation (2). Figure 8a shows the spatial distribution of mean biomass burning AOD (OC + BC) at a wavelength of 550 nm for March 2001. The spatial distribution of AOD is correlated with the columnar concentration. Figure 8a reveals a pattern similar to that shown in Figure 7. The higher AOD is located near the source areas, and the plumes are elongated to downwind areas such as southern China and the Bay of Bengal. The South Asian region has an AOD of 0.015 ± 0.038. The maximum AOD of about 0.107 is located in western Myanmar. The second highest AOD appears around eastern India, corresponding to the higher columnar aerosol concentration in this region (Figure 7).

Figure 8.

(a) The mean aerosol optical depth (AOD) of biomass burning aerosol in column; (b) The MODIS-observed AOD of 550 nm in March 2001 (source: http://g0dup05u.ecs.nasa.gov/OPS/Giovanni/modis.MOD08_M3.shtml); (c) TOMS aerosol index in March 2001 (source: http://reason.gsfc.nasa.gov/OPS/Giovanni_toms/toms.ep.shtml).

[25] We further compare our results with the AOD observed by MODIS (Figure 8b) and the aerosol index (hereafter referred to as AI) derived from TOMS (Figure 8c) for the same period of time. Both MODIS and TOMS provide aerosol information contributed by total aerosols, whereas our study focuses on biomass-burning aerosols alone. Streets et al. [2003a] reported that the OC + BC aerosols from biomass-burning emissions were attributed to 51 and 22% of the total anthropogenic OC + BC aerosols (biomass burning is considered to be an anthropogenic source) for Southeast and South Asia, respectively. On the basis of the INDOEX results, the total anthropogenic OC + BC aerosols contributed as much as 40% to the total anthropogenic aerosols by mass [Ramanathan et al., 2001b]. Anthropogenic sources contributed to the observed AOD about 60 [Satheesh et al., 1999] to 75% [Lelieveld et al., 2001]. Additionally, the emission data we used in this study contain large uncertainties, as described in section 2.1. The MODIS AOD and TOMS AI also have their inherent uncertainties. For example, the retrieved AOD over land has larger uncertainties than that retrieved over sea [Chin et al., 2004]. The TOMS AI data are particularly suitable for detecting the presence of absorbing aerosols such as smoke, soot, and dust over a high reflecting surface [Hsu et al., 1999]. However, absorbing aerosols contained in the boundary layer are not readily seen by TOMS because the signal is weak relative to apparent noise from the ground [Herman et al., 1997]. Therefore we only compare the observed MODIS AOD and TOMS AI with our simulated AOD in spatial distribution for the South Asian region. Figure 8c shows that mineral dust in northwest China (desert area) and the anthropogenic sources in east China (highly populated area) enhanced values from TOMS AI. By contrast, no mineral dust is identified over inland Asia by MODIS because of a retrieval problem. The main locations of large biomass-burning AOD in Figure 8a (eastern India, western Bay of Bengal, Myanmar, and Thailand) correspond with relatively higher AOD or AI in Figures 8b and 8c, respectively, in comparison with the surrounding areas. However, the former is about one order in magnitude lower than the latter, owing to the coexistence of other anthropogenic aerosols in the same region.

[26] Figure 9 further depicts ratio of the AOD of OC to that of BC (OC/BC). The ratio is less than 6 in most areas, especially in India and Myanmar (approximate to 3). Compared with the source emission and concentration ratios of OC/BC (∼7), the AOD ratio of OC/BC is lower. It implies that BC aerosol has a relatively higher contribution to the total AOD than the OC aerosol in unity mass concentration. We note that in the South Asian region, the OC/BC is higher over the ocean than over the land. By contrast, India and Myanmar, which are the major source areas of biomass burning, have a low ratio of about 3. However, the emission (Figure 2c) and concentration (not shown here) ratio of OC/BC have similar values over the South Asian region. We suggest that the above contrast could be related to local relative humidity. On the basis of the discussion in section 2.3, the AOD of OC aerosols is very sensitive to RH at high RH value. Figures 4c, 5d, 5e, and 5f show that the areas which have a lower AOD ratio of OC/BC also experience lower relative humidity (below 50%). This is especially so in the major source area of the South Asian region.

Figure 9.

The ratio of the aerosol optical depth (AOD) of OC to that of BC (OC/BC).

3.3. Direct Radiative Forcing and Heating Rate

[27] In order to understand the radiative impact from different aerosol components, we consider three cases (OC, BC, and OC + BC aerosols) to simulate direct radiative forcing. In the South Asian region, lower relative humidity (see section 3.1) and cloudless conditions frequently prevail [Takemura et al., 2003] during the dry season. It is therefore reasonable to consider only a clear-sky condition when calculating the aerosol radiative forcing.

[28] On the basis of three-dimensional radiation fluxes calculated by the solar radiative transfer model (CLIRAD-SW), the aerosol radiative forcing can be further estimated from equations (3) and (4). Figures 10a–10f show the mean clear-sky shortwave radiative forcing at the TOA, atmosphere (ATM), and surface (SFC) with respect to the OC and BC biomass-burning aerosols for March 2001. The spatial distribution patterns of the radiative forcing are similar to that of the AOD (see Figure 8a). Maximum values of both parameters are located near the source regions. The OC aerosols result in negative ΔFTOA in the entire domain, particularly in the South Asian region. The forcing ranges from 0 to −3.1 W m−2, which is nearly identical to the region of maximum reduction in solar radiation at the surface (−0.5 to −4.8 W m−2, see Figure 10c). The difference between ΔFTOA and ΔFSFC is attributed to the absorption by atmosphere (0 to 1.7 W m−2, see Figure 10b). The OC aerosols enhance the reflected solar radiation at the TOA, leading to a cooling effect due to less incoming solar radiation to the atmosphere and surface. Inversely, the BC aerosols absorb downward and upward solar radiation, therefore substantially reducing any solar radiation reaching the surface and reflecting back to space. This absorption results in a positive radiative forcing at the TOA (ΔFTOA). Figure 10d shows that BC aerosols cause positive ΔFTOA within a range of 0.0 to 1.6 Wm−2, i.e., a warming effect, in the South Asian region. Meanwhile, the solar absorption by BC aerosols within the atmosphere (see Figure 10e) leads to a significant reduction in solar radiation reaching the Earth’s surface, and to a higher negative radiation forcing of up to −7.4 W m−2. Overall, from a radiation budget viewpoint, the BC and OC aerosols create warming and cooling effects on the atmosphere over the South Asian region, respectively. The maximum radiative forcing at the TOA, ATM, and SFC appears in western Myanmar for both BC and OC aerosols.

Figure 10.

The mean clear-sky direct radiative forcing in March 2001 due to the OC aerosols at (a) TOA, (b) ATM, and (c) SFC; due to the BC aerosols at (d) TOA, (e) ATM, and (f) SFC; due to the BC + OC aerosols at (g) TOA, (h) ATM, and (i) SFC.

[29] Figures 10g–10i further illustrate the mean clear-sky direct radiative forcing of the total biomass-burning carbon aerosols (OC + BC) at the TOA, ATM, and SFC. For the TOA (Figure 10g), the ΔFTOA ranges from −1.9 to 0.4 W m−2. The biomass-burning carbon aerosols have a net effect of radiative cooling of the earth-atmospheric system. However, Jacobson [2004] indicated the other chemistry components have a cooling effect, primarily due to their effect on the swelling of aerosol particles. Therefore the biomass-burning aerosols create cooling effect on the atmosphere over the South Asian region. For the surface (Figure 10i), it ranges from −0.5 to −12.0 W m−2 in the South Asian region, with a minimum value located in western Myanmar. The ratio ΔFSFCFTOA has been calculated for each grid point in the domain. This ratio for the sea is found to be approximately 3, which is consistent with those observed over the tropical northern Indian Ocean by Satheesh and Ramanathan [2000] and over the Bay of Bengal by Kaufman et al. [2002].

[30] The radiative forcing efficiency (ΔF/AOD) [Satheesh and Ramanathan, 2000; Ramanathan et al., 2001a; Tahnk and Coakley, 2002] is used for normalizing radiative forcing in unity AOD. It is useful for quantifying and comparing aerosol radiative effects with respect to different places and aerosol component conditions. On the basis of Figure 10, the ΔFTOA and ΔFSFC for each of the 1° × 1° latitude-longitude grid boxes over the South Asian region are plotted against the corresponding AOD at a wavelength of 550 nm. The results are shown in Figures 11a–11c for the OC, BC, and OC + BC aerosols, respectively. Evidently, the aerosol direct radiative forcing is almost linearly correlated with the AOD, giving an independent relationship in four categories: ΔFTOA over land, ΔFTOA over sea, ΔFSFC over land, and ΔFSFC over sea. Table 2 further lists the slopes and R2 values of the linear fits in Figure 11. Here the slopes are expressed as radiative forcing efficiency. At the TOA, Table 2 shows that the OC aerosols have a negative ΔFTOA/AOD, and its value over sea is larger than that over land (i.e., −31.6 and −39.7 W m−2 for the land and sea, respectively). On the other hand, the BC aerosols have a positive ΔFTOA/AOD, and the ΔFTOA/AOD over land is about 2 times larger than that over sea (i.e., 48.7 and 25.7 W m−2 for the land and sea, respectively). The result that a lower surface albedo causes the OC aerosol to enhance negative forcing over the sea, whereas a higher surface albedo causes the BC aerosol to enhance positive forcing over land, can be explained by the radiative forcing parameterization schemes given by Charlson et al. [1992, equation [3]] and Chylek and Wong [1995, equation [8]]. Furthermore, the ΔFTOA/AOD of OC + BC aerosols is −15.0 and −28.9 W m−2 over land and sea, respectively. However, the ΔFTOA/AOD over land has a lower R2 of 0.70 than that over sea (R2 = 0.97) because of the opposite signs of radiative forcing and the perturbations of the land surface albedo over the land (Figure 10g).

Figure 11.

The relationship of clear-sky direct radiative forcing and AOD at 550 nm. (a) OC aerosol, (b) BC aerosol, and (c)BC + OC aerosol.

Table 2. Aerosol Direct Radiative Forcing Efficiency (ΔF/AOD, Unit in W m−2) and the R2 of Linear Fits Over the South Asian Region (in Parenthesis)
SpeciesTOASurface
LandSeaLandSea
OC−31.6 (0.95)−39.7 (0.99)−51.0 (0.97)−59.7 (0.99)
BC+48.2 (0.78)+25.7 (0.81)−254.9 (0.98)−277.3 (0.99)
OC + BC−15.0 (0.70)−28.9 (0.97)−89.4 (0.98)−93.3 (0.99)

[31] At the surface, Table 2 shows that the ΔFSFC/AOD is larger over the sea than over the land for all cases. The ΔFSFC/AOD of BC aerosols either over land or sea is about five times higher than that of OC aerosols. The surface radiation is very sensitive to the BC aerosols. This result is consistent with the work of Jacobson [2001a], which shows global BC direct forcing at the surface from combined biomass-burning and fossil-fuel sources to be just over four times stronger than OC direct forcing. For OC + BC aerosols, the ΔFSFC/AOD is −89.4 and −93.3 W m−2 over the land and sea, respectively.

[32] The radiative forcing of the atmosphere (ΔFATM) can be defined as ΔFTOA − ΔFSFC. In our study, the ΔFATM of OC + BC aerosols presented positive values in the entire domain. The maximum value of 10.0 W m−2 appeared in western Myanmar (see Figure 10h). The change in the atmospheric heating rate (ΔQ) due to the existence of aerosol can be expressed as equation (6). Figure 12a shows the spatial distribution of the average ΔQ from the surface to the level of 3 km, which is the major aerosol-loading layer. We find that the ΔQ has a strong gradient, in the range of 0.01–0.3°C day−1 surrounding the source regions. Its maximum is located in western Myanmar. The perturbations of the ΔQ are mainly attributed to the higher ΔFATM of BC aerosols over land. Figure 12b further shows the vertical cross-sectional ΔQ along 20°N. The low atmosphere of India and Myanmar had a larger ΔQ, particularly for the level of 1000–3000 m. By contrast, between the above two regions, the Bay of Bengal has a relatively weaker ΔQ. We suggest that the sea-land gradients in ΔQ over the low troposphere may drive or modify the local circulation. Further studies should be conducted.

Figure 12.

Change in heating rate (a) average of 0–3 km and (b) the vertical cross section along 20°N. (Unit: °C day−1).

[33] The above-presented ΔFTOA, ΔFATM, ΔFSFC, ΔF/AOD, and ΔQ are diurnal averaged values. However, incoming shortwave radiation is the function of the solar zenith angle, so the zero value will occur at night. Therefore the actual values for daytime should be considered as approximately double the diurnally average value.

3.4. Comparisons With Other Studies

[34] In global modeling studies, IPCC [2001] suggested value of −0.2 W m−2 (a range of −0.07 to −0.6 W m−2) for the direct ΔFTOA contributed by biomass-burning aerosols. Haywood and Boucher [2000] reported that the global average direct ΔFTOA has an estimated range of −0.14 and −0.74 W m−2, while the absorbed radiation by BC aerosols ranged from 0.16 to 0.2 W m–2. More recently, several global modeling studies have taken steps to improve the treatment of aerosol concentration, optical properties [Kirkevåg and Iversen, 2002; Takemura et al., 2002; Chin et al., 2004; Reddy and Boucher, 2004], chemical components, and microphysical properties [Jacobson, 2001a]. Nevertheless, the results still showed discrepancies for ΔFTOA, even for clear-sky conditions. For example, Takemura et al. [2002] estimated the global average clear-sky direct ΔFTOA to be −0.22 W m−2 for biomass-burning aerosols (OC + BC) and 0.07 W m−2 for BC aerosols, while Reddy et al. [2005] indicated that these two values are −0.06 W m−2 and 0.29 W m−2, respectively. These discrepancies are primarily due to the uncertainties of absorbing BC aerosols. In this study, we use a finer spatial and temporal resolution model and a more realistic emission data set to compute ΔFTOA for biomass-burning aerosols. Averaged over the South Asian region (70°–110°E, 5°–30°N), the ΔFTOA in March 2001 is −0.27 W m−2 (−1.9 to 0.4 W m−2) for OC + BC and 0.13 W m−2 (0.0–1.6 W m−2) for BC aerosols. Although our results cannot compare with the global results because of different time and spatial conditions, the estimated ΔFTOA is still within the range of earlier global studies.

[35] We further compare our results with other campaigns as listed in Table 3. These campaigns estimated the clear-sky direct radiative forcing of biomass-burning aerosols using modeling, observation, and hybrid methods. The observation methods include satellite [Kaufman et al., 2002; Tahnk and Coakley, 2002], ground-based [Ramanathan et al., 2001a], and aircraft [Anderson et al., 1996; Keil and Haywood, 2003] approaches to calculate aerosol optical properties and radiative forcing. The use of both model simulations and field observations to estimate aerosol radiative forcing is influenced by many factors, such as methodology, hypothetical assumptions, and the background situation. The magnitudes of our ΔFTOA and ΔFSFC are lower than but close to those observed by Ramanathan et al. [2001b] over the tropical Indian Ocean, which are much lower than those observed by Tahnk and Coakley [2002] and Kaufman et al. [2002] over the Bay of Bengal. The results are reasonable given that these observations include the effects of anthropogenic aerosols, which result in higher forcing. Other studies in the South African and tropical South Atlantic region revealed that forcing was stronger because of a stronger emission [Streets et al., 2004] and relatively higher UV-absorbing aerosol amounts as observed by TOMS [Herman et al., 1997].

Table 3. Comparison of Clear-Sky Direct Radiative Forcing of Biomass Burning Aerosols (BBA) at the TOA and Surface
Field CampaignStudiesAerosol SpeciesModeling or ObservationTime PeriodAreaΔFTOA (W m−2)ΔFSFC (W m−2)
TRACE AAnderson et al. [1996]BBAObservation21 Sep. to 26 Oct. 1992Tropical South Atlantic region −2.4 (sea) to −25.0 (land)
SAFARI-2000Myhre et al. [2003]BBAModelingSep. 2000South Africa−4.3 
INDOEXRamanathan et al. [2001b]BBA + anthropogenic aerosolObservationJan.–Apr. 1996–1999Tropical India Ocean0.0 ± 2.0−14.0 ± 3.0
Collins et al. [2002]BBA + anthropogenic aerosolModelingJan.–Mar. 1999Indian Ocean and continental regions−9.0 (sea) to 4.0 (land)−5.0 to −40.0
Tahnk and Coakley [2002]BBA + anthropogenicObservationJan.–Mar. 1996–2000Bay of Bengal−8.3−18.7
Kaufman et al. [2002]BBA + anthropogenic aerosolObservation + modelingMar. 1997Bay of Bengal−5.0 to −15.0−20.0 to −40.0
TRACE-PThis studyBBAModelingMar. 2001South Asian region−0.3 (−1.9 to 0.4)−1.4 (−0.5 to −12.0)

[36] In addition, the emission uncertainty mentioned in section 2.1 should be further considered to quantity the uncertainty of radiative forcing. An additional simulation using the upper bound on OC and BC emission data sets is performed. The upper bound emission data are described in section 2.1. As a result, the average clear-sky direct ΔFTOA and ΔFSFC with upper bound emission are −1.1 W m−2 (a range of −6.7 to 2.9 W m−2) and −7.1 W m−2 (a range of −0.5 to −58.2 W m−2) over the South Asian region, respectively. The average values are four to five times larger than those of original simulation (−0.3 and −1.4 W m−2, as listed in Table 3). Besides, the ranges of ΔFTOA and ΔFSFC with upper bound emission are larger than original ones. In comparison, the results are quantitatively consistent with some pervious studies as listed in Table 3. The magnitudes of ΔFTOA fall within the ranges of the studies [see Ramanathan et al., 2001b; Collins et al., 2002] including biomass-burning and anthropogenic aerosols, indicating that upper bound BC/OC emission can account for the contribution of anthropogenic aerosols to ΔFTOA. In contrast, it is only one fourth of the modeled ΔFTOA by Myhre et al. [2003] for considering biomass-burning aerosols alone over S. Africa. At surface, our average ΔFSFC is only half of the results of Ramanathan et al. [2001b] and Tahnk and Coakley [2002]. However, its high bound value of −58.2 W m−2 is higher than those listed in Table 3, again showing the strong response of ΔFSFC to biomass-burning aerosols. In summary, the numerical experiment for upper bound emission indicates that biomass-burning aerosols alone potentially contribute to radiative forcing at the same level as anthropogenic aerosols. Obviously, the radiative forcing of biomass-burning aerosol can not be ignored in South Asian region.

[37] The ΔFTOA and ΔFSFC listed in Table 3 show a significant fluctuation in various regions. Although we have used a more detailed emission data set and applied a detailed regional modeling system to simulate the radiative impacts, the uncertainties in estimating the radiative forcing of biomass-burning aerosols in this study still remain owing to unknown variables such as (1) the size distribution of aerosol, (2) the aerosol mixture, and (3) the aerosol chemistry. We need more reliable measurements of biomass-burning aerosol characteristics over the source regions and a more complete scheme of aerosol processes [e.g., Jacobson, 2002] in order to improve our model simulations.

3.5. Impact on the Hydrological Cycle

[38] Overall, biomass-burning aerosols in clear-sky conditions result in less solar irradiance reaching the Earth’s surface but greater heat in the lower atmosphere. The regional cooling at the surface accompanied with the warming of the lower atmosphere was also shown in the global climate model (GCM) study as reported by Ramanathan et al. [2001a]. Such an effect on the temperature structure and energy balance of the lower troposphere leads to more stable meteorological conditions. These perturbations will affect the dynamic and thermodynamic processes in the atmosphere and further impact the regional hydrological cycle and rainfall [Chung et al., 2002; Menon et al., 2002]. Here we attempt to quantify the radiative impact on the hydrological cycle according to the balance between radiation and evaporation at the surface [Ramanathan et al., 2001b]. The decrease in solar radiation at the Earth’s surface is expected to cause a decline in evaporation [Roderick and Farquhar, 2002]. A reduction in the evaporation will have to be balanced by a reduction in rainfall, which will effectively spin down the hydrological cycle [Ramanathan et al., 2001b]. Hence we estimate the change in pan evaporation resulting from a change in solar irradiance using the following equation (7) given by Roderick and Farquhar [2002]:

equation image

where λ (∼2.4 MJ kg−1) is the latent heat of vaporization of water; δEpan (kg m−2 s−1) is the change in pan evaporation; δRs (J m−2 s−1) is the change in solar irradiance at the surface; γ (∼67 Pa K−1) is the psychrometric constant; and s is the slope of the saturation vapor pressure-temperature relationship at the temperature of air. The ratio s/(s + γ) varies from 0.48 at 5°C to 0.82 at 35°C [Roderick and Farquhar, 2002]. In this study, the maximum value of δRs is −32.1 MJ m−2 per month (i. e., −0.012 kw□ h m−2 × 3.6 MJ kw−1 × 24 × 31 = −32.1 MJ m−2) in western Myanmar. With s/(s + γ) of 0.73 at an average temperature of 27°C in March 2001 (obtained from NCEP reanalysis), the reduction in monthly pan evaporation can be estimated from equation (7) to be 14.1 mm m−2 per month. It is noted that the accumulated precipitation was about 48.8 mm m−2 in this area during March 2001 [Climate Prediction Center (CPC) Merged Analysis of Precipitation (CMAP)] [Xie and Arkin, 1996]. The significance of evaporation reduction caused by biomass-burning aerosols and its subsequent impact on the regional hydrological cycle is shown. Further model study of the hydrological cycle response due to biomass-burning aerosol forcing is needed.

4. Concluding Remarks

[39] Using the detailed aerosol emission data measured during TRACE-P experiment in March 2001, we have integrated a three-dimensional Lagrangian model (HYSPLIT) with a solar radiative transfer model (CLIRAD-SW) to estimate the spatial distributions of BC and OC aerosols and the radiative forcing of the Asian biomass burning. The monthly mean concentration of biomass-burning carbon aerosols is well-mixed in the 1-km layer near the surface with a value ranging from 1.0 to 12.5 μg m−3 surrounding the source regions. There is a persistent aerosol layer with a thickness of 3 km over most of the South Asian region, and the plumes of biomass-burning aerosols extend far to the Indian Ocean and the western Pacific Ocean. The aerosol optical depth (AOD) of the biomass-burning carbon aerosols reaches a maximum value of 0.11 over western Myanmar. Compared with the OC aerosol, the BC aerosol makes a remarkable contribution to the AOD, especially in the source region.

[40] Because of the strong absorption of solar radiation, the BC aerosol has a positive clear-sky shortwave radiative forcing at the top of the atmosphere. However, this positive radiative forcing is overshadowed by the negative radiative forcing due to the reflection of shortwave radiation by the OC aerosol. As a result, the biomass-burning carbon aerosols have a net effect of radiative cooling of the earth-atmospheric system. At the surface, the negative radiative forcing due to the BC aerosol is five times stronger than that due to the OC aerosol. Averaged over the Asian region (as summarized in Figure 13), the clear-sky direct shortwave radiative forcing of carbon aerosols ranges from −1.9 to 0.4 W m−2 at the top of the atmosphere and from −0.5 to −12.0 W m−2 at the surface. In the atmosphere, the enhanced shortwave heating due to the carbon aerosols reaches a maximum of 0.3°C day−1.

Figure 13.

The clear-sky direct radiative forcing of biomass burning aerosols at TOA and surface and the change in heating rate at low atmosphere over the South Asian region.

[41] The strong heating of the atmosphere and cooling of the surface due to the BC aerosol reduce significantly the atmosphere lapse rate. It is expected that the negative radiative forcing at the top of the atmosphere and the enhanced atmospheric stability due to the biomass burning in the South Asian region have the effects of weakening the atmospheric circulation and hydrological cycle.

Acknowledgments

[42] This work was supported by the National Science Council, Taiwan under the contracts NSC92-2111-M-008-018-AGC, NSC93-2111-M-008-016-AGC, NSC94-2111-M-008-018-AGC, and NSC94-2752-M-008-006-PAE. We appreciate the Center for Global and Regional Environmental Research, University of Iowa, for providing us biomass-burning aerosols emission data. Special thanks are also given to the Air Resources Laboratory, NOAA, and the Climate and Radiation Branch of Goddard Space Flight Center, NASA for technical support.

Ancillary