Longwave radiative forcing of Indian Ocean tropospheric aerosol

Authors


Abstract

[1] A spectrally resolved discrete-ordinates radiative transfer model is used to calculate the change in downwelling surface and top-of-the-atmosphere (TOA) outgoing longwave (3.9–500 μm) radiative fluxes induced by tropospheric aerosols of the type observed over the Indian Ocean during the Indian Ocean Experiment (INDOEX). Both external and internal aerosol mixtures were considered. Throughout the longwave, the aerosol volume extinction depends more strongly on relative humidity than in most of the shortwave (0.28–3.9 μm), implying that particle growth factors and realistic relative humidity profiles must be taken into account when modeling the longwave radiative effects of aerosols. A typical boundary layer aerosol loading, with a 500-nm optical depth of 0.3, will increase the downwelling longwave flux at the surface by 7.7 W m−2 over the clean air case while decreasing the outgoing longwave radiation by 1.3 W m−2. A more vertically extended aerosol loading, exhibiting a high opacity plume between 2 and 3 km above the surface and having a typical 500-nm optical depth of 0.7, will increase the downwelling longwave flux at the surface by 11.2 W m−2 over the clean air case while decreasing the outgoing longwave radiation by 2.7 W m−2. For a vertically extended aerosol profile, approximately 30% of the TOA radiative forcing comes from sea salt and approximately 60% of the forcing comes from the combination of sea salt and dust. The remaining forcing is from anthropogenic constituents. These results are for the external mixture. For an internal mixture, TOA longwave forcings can be up to a factor of two larger. Therefore, to complete our understanding of this region's longwave aerosol radiative properties, more detailed information is needed about aerosol mixing states. These longwave radiative effects partially offset the large shortwave aerosol radiative forcing and should be included in regional and global climate modeling simulations.

1. Introduction

[2] The Indian Ocean Experiment (INDOEX) has revealed a shortwave surface radiative forcing efficiency (per unit optical depth) of up to 70 W m−2 over the northern Indian Ocean during winter, due to tropospheric aerosols transported from southern Asia [Jayaraman et al., 1998; Satheesh and Ramanathan, 2000]. The major part of the shortwave aerosol opacity is due to anthropogenic sources, and this constitutes a significant perturbation to the tropospheric radiation budget in that it can be observed from satellites over a large geographic area. Potential climatic implications of the direct shortwave aerosol radiative effect include a slowing down of the hydrological cycle [Satheesh and Ramanathan, 2000], and dissipation of low-level cloud cover [Ackerman et al., 2000]. The INDOEX radiation measurement program was concerned mainly with shortwave radiation, as preliminary observations [Jayaraman et al., 1998] and climate modeling studies [e.g., Kiehl and Briegleb, 1993] suggested very large aerosol signatures in and important climatic impacts from the shortwave radiation alone. However, the large shortwave perturbation due to anthropogenic aerosol implies that there should be a noticeable perturbation in the longwave as well.

[3] There have been several studies on the longwave radiative properties of tropospheric aerosols in general. Toon and Pollack [1976] provided the first comprehensive global calculations of aerosol optical properties in the middle infrared, by incorporating several major aerosol types in an external mixture. Ackerman et al. [1976] suggested a decrease in the net surface longwave flux of up to 64 W m−2 in the middle infrared window (8–12 μm), and an increase in boundary layer longwave cooling rates of up to −5 °C/day, due to the addition of a heavy urban aerosol loading. There have been several detailed studies on the longwave radiative effects of mineral dust. Carlson and Benjamin [1980] show that, because of the strong middle infrared absorption by Saharan dust, these particular aerosol particles can increase boundary layer cooling rates by approximately −3 °C/day. Sokolik and Toon [1999] considered in detail the composition of mineral dust, and demonstrated that the total aerosol radiative forcing (shortwave plus longwave) can vary between −20 W m−2 and +11 W m−2 depending on the composition of the dust component alone. Claquin et al. [1998] performed sensitivity studies demonstrating that longwave radiative forcing by dust is nonnegligible, and is sensitive to particle size and to the complex refractive indices of the dust constituents. Quijano et al. [2000] modeled the shortwave and longwave radiative forcing and heating rates due to both Saharan and Afghan dust in clear and cloudy atmospheres, and demonstrate the regional dependence of radiative forcing due to dust.

[4] There have been some measurements demonstrating the surface longwave radiative forcing due to tropospheric aerosols. Lubin and Simpson [1994] measured the spectral longwave radiative signature of anthropogenic aerosols, under typical Los Angeles smog, using a Fourier transform infrared (FTIR) spectroradiometer, and reported an excess downwelling longwave flux of 7.7 W m−2 in the middle infrared window due to the presence of the aerosol. Spankuch et al. [2000], using a similar instrument, found an excess downwelling longwave flux of 1–2 W m−2 in the presence of one type of natural biogenic aerosol (pine pollens).

[5] The abovementioned studies together indicate that aerosol longwave forcing is (1) nonnegligible, and (2) highly variable depending on aerosol composition and hence on region. The INDOEX program has provided a unique database on the aerosol properties of a particular region, in the context of their climatic influence. The longwave radiative properties of the Indian Ocean region need to be understood in detail, so that incorporation of INDOEX results into regional and global climate model simulations can be made with confidence. While INDOEX did not make longwave radiation measurements, this program did garner a thorough enough understanding of the chemical composition and microphysical properties of southern Asian tropospheric aerosol that we can attempt to model this aerosol's longwave radiative signature. While the abovementioned studies that focus on mineral dust provide useful guidance, the Indian Ocean aerosol has a diverse composition that includes sea salt, sulfates, organics, and soot, in addition to mineral dust. We must therefore revisit the early approach of Toon and Pollack [1976] to account for this diverse composition, but with information specific to the Indian Ocean region. We wish to (1) identify which aerosol components, natural or anthropogenic, are most responsible for the longwave radiative forcing, and (2) determine whether the mixing state, external or internal, has an impact on the longwave radiative forcing.

2. Model and Input Data

2.1. Radiative Transfer Model

[6] We calculate the downwelling longwave flux at the surface, and the upwelling longwave flux at the top-of-the-atmosphere (TOA), using the discrete-ordinates radiative transfer formulation of Stamnes et al. [1988]. This algorithm is used in four-stream mode, and absorption by trace gases is specified in 106 bands covering the wave number range 20–2560 cm−1 (3.9–500-μm wavelength) using the method of exponential sum fitting of transmissions (ESFT) [Wiscombe and Evans, 1977]. The tables of exponents and weights for these bands were kindly provided by Dr. Si-Chee Tsay (NASA Goddard Space Flight Center, personal communication, 1993). Between 20 and 2000 cm−1, the bands are 20 cm−1 wide, while at larger wave numbers they are 80 cm−1 wide. These ESFT tables were originally computed from the LOWTRAN 7 radiative transfer model [Kneizys et al., 1988], and therefore the water vapor continuum used in these calculations is from this absorption model.

[7] Because the surface and TOA longwave fluxes depend as much on the thermal structure of the ocean-atmosphere system as on the optical properties of the scattering and absorbing media, we must make certain that the thermal structure of the model atmosphere is representative of the region. We constructed a 56-layer model atmosphere, in which the temperature, number density, and ozone abundance above 14 km is taken from the U.S. Standard Atmosphere (1976) tropical annual model. Below 14 km, we make use of INDOEX rawinsonde data from the Kashidhoo Climate Observatory (KCO) (4.965°N, 73.466°E). Between 13 January and 29 March 1999, 190 sondes were launched from KCO. We take the average and standard deviation of temperature and relative humidity over all of the sondes, on a grid of 250 meter resolution between sea level and 10 km above sea level, and on a grid of 1 km resolution between 10 and 14 km. The average temperature profile is shown in Figure 1, and the average relative humidity profile is shown in Figure 2, for the altitude range 0–10 km. The tropospheric temperature profile remains steady throughout the INDOEX 1999 field phase; standard deviations in temperature are uniformly around 2 K at all altitudes. The relative humidity profile exhibits considerably more variability (standard deviations of 8–10% within 1 km of the surface, increasing to 20% or more above 2 km).

Figure 1.

Temperature and aerosol profiles based on INDOEX rawinsonde and lidar data, respectively. Filled circles joined by the heavy solid line depict the average vertical temperature profile from 190 sondes launched from the Kaashidoo Climate Observatory during January–March 1999. The error bars represent 1σ in these temperature measurements. The dotted and light solid curves depict the vertical profiles in relative 500-nm aerosol opacity for the boundary layer and extended aerosol profiles (labeled 1 and 2), respectively.

Figure 2.

Average vertical profile in relative humidity from 190 sondes launched from the Kashidhoo Climate Observatory during January–March 1999. The error bars represent 1σ in these measurements.

[8] Throughout the middle infrared, and irrespective of aerosol optical properties, the upwelling and downwelling fluxes will depend significantly on both temperature and relative humidity. The former determines the overall magnitude of emission, while both specify the column abundance of the most important “greenhouse” gas. We therefore adopt three model atmospheres, to consider a large range of variability in atmospheric thermodynamics over the Indian Ocean during winter. A typical, or “average” model atmosphere uses the mean temperature and relative humidity of all 190 sondes on the tropospheric altitude grid specified above. A cooler and drier, or “low relative humidity (RH)” model atmosphere subtracts one standard deviation from the “average” temperature and relative humidity profiles at all altitude grid points. A warmer and wetter, or “high RH” model atmosphere adds one standard deviation to the average temperature and relative humidity profiles at all altitude grid points.

[9] For the vertical profile of aerosol burden, we adopt the model of Satheesh et al. [1999] and Lelieveld et al. [2000], which are based on INDOEX lidar and C-130 observations. Approximately one third of the time during the northern Indian Ocean winter, the aerosol was well mixed in the first 1 km above the surface, decreasing in abundance exponentially above 1 km with a scale height of 0.8 km. For the remaining two thirds of the time, the aerosol was well mixed in the first 2 km above the surface, with an additional plume between 2 and 3 km in which the optical depth at 500 nm was three times larger than in a layer near the surface. These boundary layer and higher altitude aerosol profiles are shown in Figure 1, and are referred to as profiles 1 and 2, respectively.

[10] When calculating radiative fluxes due to aerosols, diurnal variability needs to be considered because transport processes on short timescales can exert considerable influence on aerosol optical properties [Sokolik and Toon, 1999]. This is particularly true in the shortwave, and for the longwave diurnal effects can also matter if there is significant diurnal variability in temperature or relative humidity. We examined the variability in the rawinsonde data from KCO, and found a diurnal variation in near-surface temperature of ±2 °C, with most of this variation occurring between the hours 0000 and 0600 UTC (highest daily temperatures at 0600). No diurnal temperature variability was found in the lower troposphere, even when limiting the altitude range to 0–1 km. We noticed a diurnal variation of ±10% in relative humidity near the surface, again with most of this variation occurring between 0000 and 0600 UTC (higher values at 0000). No diurnal variability in relative humidity was noticed in the lower troposphere, even when limiting the altitude range to 0–1 km. With only this small variability present, we can consider longwave radiative flux calculations made using the average temperature and relative humidity profiles of Figures 1 and 2 to be a representative diurnal average. This diurnal average should be suitable for a discussion of the first-order aerosol radiative effects considered in this study, such as the effect of the aerosol vertical profile and extremes in mixing state (external versus internal mixture). However, the reader is cautioned that if high-time-resolution longwave fluxes are required for the radiative interactions in a given process study, the longwave flux calculations should be time-dependent.

2.2. Aerosol Optical Properties

[11] The volume extinctions, single scattering albedos, and scattering asymmetry factors of aerosol particles of most size ranges and compositions depend significantly on relative humidity. This is especially true for a relative humidity greater than 80%, in which the size distribution and refractive index of a water soluble aerosol constituent will be modified by both particle growth and transformation to an aqueous solution, and in which the size distribution and refractive index of some water insoluble constituents may be modified as the particles may become coated with either liquid water or an aqueous solution of any soluble constituents. For drier atmospheres, the relative humidity dependence in aerosol optical properties can sometimes be neglected in the shortwave [e.g., Satheesh et al., 1999]. However, this relative humidity dependence cannot be neglected in the longwave. Blanchet and List [1983] considered the composition of Arctic haze, with both soluble and insoluble constituents, and showed that while the spectrally resolved shortwave volume absorption coefficient varies by up to a factor of two as relative humidity increases from 0 to 99%, the volume absorption coefficient in the longwave can vary by up to two orders of magnitude.

[12] Based on INDOEX chemical and microphysical data from both 1998 and 1999, Indian Ocean tropospheric aerosol was found to consist of six major constituents: sea salt, sulfates, soot, dust, organics, and fly ash. In an external mixture, the dry size distribution of each constituent was found to be well represented by a standard lognormal distribution:

equation image

where J is the number of components, and with mode radii rmj and standard deviations σj given in Table 1. For five of the constituents J = 1, except for sea salt, which has an accumulation mode and a coarse mode. Mineral dust will generally exhibit more than one mode [Sokolik and Toon, 1999], and in fact Satheesh et al. [1999] first noticed a bimodal distribution in the mineral dust. However, they subsequently identified the smaller particle mode as ash. In an external mixture, the radiative properties of each aerosol constituent are calculated separately, and these radiative properties are then mixed in the radiative transfer model. In an internal mixture, the refractive indices must somehow be combined to estimate the refractive index of an aggregate particle. The size distribution of these aggregate particles will generally be represented by a different lognormal distribution than that for the external mixture. Both cases are considered in this study.

Table 1. Size Distribution Parameters for the Aerosol Constituents Used in This Study
ConstituentMode radius rmStandard deviation σ
sea salt-accumulation mode0.416 μm2.03
sea salt-coarse mode3.42.0
sulfates0.03062.24
soot0.01182.0
dust0.52.20
organics0.02122.24
ash0.392.0

[13] Given the complex refractive index of a constituent dielectric medium at wavelength λ, m = nik, the single scattering albedo ωjλ, asymmetry factor gjλ, phase function moments Pmjλ, and volume extinction coefficient βext, λ, are calculated from Mie theory [Wiscombe, 1980]. In the Mie calculations, the lognormal size distribution of (1) is integrated over the particle size range 0.001 to 20 microns. Table 2 gives the complex refractive indices for each of the individual aerosol constituents, and for the internal mixture (the computation of the latter is discussed in detail below). In addition, because the constituent with the largest optical depth contribution is the sulfates, we allow for this medium a transformation from a coated particle to a pure sulfuric acid solution at and above this medium's critical relative humidity of 80% [Nilsson, 1979], and use refractive indices for a sulfuric acid solution given by Palmer and Williams [1975].

Table 2. Complex Refractive Indices (Real and Imaginary Components) for the Individual Indian Ocean Aerosol Constituents Considered in This Studya
Wavelength, μmSea saltSulfates (dry)Sulfuric acidDust and fly ash
  • a

    Most of these refractive indices are from the study of Hess et al. [1998]. The refractive indices for the 25% sulfuric acid solution are from the study of Palmer and Williams [1975]. The refractive indices for the internal mixture are calculated from those of the individual constituents, as discussed in the text.

0.251.386.63 × 10−71.429.98 × 10−31.380.001.533.00 × 10−2
0.301.372.68 × 10−71.412.66 × 10−31.380.001.532.50 × 10−2
0.351.364.68 × 10−81.411.66 × 10−31.380.001.531.70 × 10−2
0.401.365.43 × 10−91.401.66 × 10−31.380.001.531.30 × 10−2
0.451.363.97 × 10−91.401.66 × 10−31.370.001.538.50 × 10−3
0.501.362.84 × 10−91.401.66 × 10−31.370.001.537.80 × 10−3
0.551.352.98 × 10−91.402.00 × 10−31.370.001.535.50 × 10−3
0.601.351.16 × 10−81.402.00 × 10−31.370.001.534.50 × 10−3
0.651.351.97 × 10−81.402.33 × 10−31.360.001.534.50 × 10−3
0.701.355.46 × 10−81.402.33 × 10−31.360.001.534.00 × 10−3
0.751.352.73 × 10−71.402.83 × 10−31.361.48 × 10−71.534.00 × 10−3
0.801.353.56 × 10−71.393.33 × 10−31.361.21 × 10−71.534.00 × 10−3
0.901.355.80 × 10−61.394.32 × 10−31.364.53 × 10−71.534.00 × 10−3
1.001.352.04 × 10−51.395.16 × 10−31.362.75 × 10−61.534.00 × 10−3
1.251.345.30 × 10−51.396.32 × 10−31.351.00 × 10−51.535.00 × 10−3
1.501.342.47 × 10−41.387.62 × 10−31.352.08 × 10−41.535.70 × 10−3
1.751.331.84 × 10−41.375.89 × 10−31.342.06 × 10−41.536.40 × 10−3
2.001.321.09 × 10−31.343.40 × 10−31.332.50 × 10−41.537.60 × 10−3
2.501.282.03 × 10−31.315.15 × 10−31.293.10 × 10−21.521.40 × 10−2
3.001.402.39 × 10−11.391.89 × 10−11.372.01 × 10−11.523.90 × 10−2
3.201.488.11 × 10−21.466.43 × 10−21.431.08 × 10−11.512.40 × 10−2
3.391.432.05 × 10−21.431.78 × 10−21.415.10 × 10−21.511.93 × 10−2
3.501.418.41 × 10−31.427.94 × 10−31.403.80 × 10−21.511.80 × 10−2
3.751.383.23 × 10−31.403.67 × 10−31.372.90 × 10−21.501.20 × 10−2
4.001.374.19 × 10−31.394.73 × 10−31.363.20 × 10−21.506.70 × 10−3
4.501.351.19 × 10−21.381.33 × 10−21.344.20 × 10−21.508.70 × 10−3
5.001.341.11 × 10−21.371.23 × 10−21.325.10 × 10−21.481.80 × 10−2
5.501.311.06 × 10−21.351.37 × 10−21.307.60 × 10−21.463.60 × 10−2
6.001.289.48 × 10−21.317.91 × 10−21.311.42 × 10−11.445.50 × 10−2
6.201.397.96 × 10−21.396.77 × 10−21.381.29 × 10−11.436.30 × 10−2
6.501.353.49 × 10−21.383.71 × 10−21.358.50 × 10−21.425.20 × 10−2
7.201.332.89 × 10−21.344.47 × 10−21.301.02 × 10−11.461.30 × 10−1
7.901.313.13 × 10−21.264.42 × 10−21.251.49 × 10−11.228.90 × 10−2
8.201.303.32 × 10−21.195.67 × 10−21.262.50 × 10−11.121.20 × 10−1
8.501.303.53 × 10−21.299.60 × 10−21.372.74 × 10−11.062.10 × 10−1
8.701.313.69 × 10−21.651.22 × 10−11.402.29 × 10−11.192.90 × 10−1
9.001.313.84 × 10−21.691.50 × 10−11.392.02 × 10−11.854.40 × 10−1
9.201.303.96 × 10−21.571.67 × 10−11.371.93 × 10−12.225.40 × 10−1
9.501.294.11 × 10−21.488.28 × 10−21.392.66 × 10−12.946.50 × 10−1
9.801.274.39 × 10−21.446.36 × 10−21.442.10 × 10−12.916.50 × 10−1
10.01.264.63 × 10−21.426.38 × 10−21.441.84 × 10−12.575.00 × 10−1
10.61.226.06 × 10−21.376.83 × 10−21.391.51 × 10−11.912.50 × 10−1
11.01.198.63 × 10−21.348.12 × 10−21.361.89 × 10−11.832.00 × 10−1
11.51.171.26 × 10−11.311.10 × 10−11.392.05 × 10−11.813.50 × 10−1
12.51.162.28 × 10−11.291.91 × 10−11.352.24 × 10−11.745.00 × 10−1
13.01.182.69 × 10−11.302.22 × 10−11.362.53 × 10−12.003.50 × 10−1
14.01.243.26 × 10−11.332.71 × 10−11.382.98 × 10−11.632.20 × 10−1
14.81.283.50 × 10−11.322.98 × 10−11.403.25 × 10−11.542.40 × 10−1
15.01.293.56 × 10−11.323.35 × 10−11.413.31 × 10−11.512.06 × 10−1
16.41.373.84 × 10−11.483.38 × 10−11.453.95 × 10−11.473.20 × 10−1
17.21.433.90 × 10−11.623.67 × 10−11.544.13 × 10−11.493.70 × 10−1
18.01.473.89 × 10−11.613.44 × 10−11.573.64 × 10−11.774.70 × 10−1
18.51.483.85 × 10−11.583.38 × 10−11.583.55 × 10−12.055.70 × 10−1
20.01.523.62 × 10−11.693.36 × 10−11.603.49 × 10−12.208.20 × 10−1
21.31.533.52 × 10−11.683.30 × 10−11.633.54 × 10−12.399.40 × 10−1
22.51.543.46 × 10−11.673.27 × 10−11.673.47 × 10−12.691.00 × 100
25.01.563.37 × 10−11.653.31 × 10−11.703.03 × 10−12.998.00 × 10−1
27.91.583.31 × 10−11.653.23 × 10−11.703.03 × 10−12.577.80 × 10−1
30.01.583.24 × 10−11.643.19 × 10−11.703.03 × 10−12.426.70 × 10−1
35.01.563.57 × 10−11.663.57 × 10−11.703.03 × 10−12.426.20 × 10−1
40.01.554.63 × 10−11.634.23 × 10−11.703.03 × 10−12.347.00 × 10−1
Wavelength, μmSootOrganicsInternal mixture 
0.251.624.50 × 10−11.533.00 × 10−21.485.59 × 10−2  
0.301.744.70 × 10−11.538.00 × 10−31.485.00 × 10−2  
0.351.754.65 × 10−11.535.00 × 10−31.484.67 × 10−2  
0.401.754.60 × 10−11.535.00 × 10−31.484.52 × 10−2  
0.451.754.55 × 10−11.535.00 × 10−31.484.37 × 10−2  
0.501.754.50 × 10−11.535.00 × 10−31.484.30 × 10−2  
0.551.754.40 × 10−11.536.00 × 10−31.484.19 × 10−2  
0.601.754.35 × 10−11.536.00 × 10−31.484.12 × 10−2  
0.651.754.35 × 10−11.537.00 × 10−31.484.15 × 10−2  
0.701.754.30 × 10−11.537.00 × 10−31.484.09 × 10−2  
0.751.754.30 × 10−11.538.50 × 10−31.484.13 × 10−2  
0.801.754.30 × 10−11.521.00 × 10−21.474.18 × 10−2  
0.901.754.35 × 10−11.521.30 × 10−21.474.31 × 10−2  
1.001.764.40 × 10−11.521.55 × 10−21.474.43 × 10−2  
1.251.764.50 × 10−11.511.90 × 10−21.474.64 × 10−2  
1.501.774.60 × 10−11.512.25 × 10−21.474.86 × 10−2  
1.751.794.80 × 10−11.471.75 × 10−21.464.90 × 10−2  
2.001.804.90 × 10−11.428.00 × 10−31.444.78 × 10−2  
2.501.825.10 × 10−11.421.20 × 10−21.425.26 × 10−2  
3.001.845.40 × 10−11.422.20 × 10−21.471.62 × 10−1  
3.201.865.40 × 10−11.438.00 × 10−31.518.93 × 10−2  
3.391.875.50 × 10−11.437.05 × 10−31.496.37 × 10−2  
3.501.885.60 × 10−11.455.00 × 10−31.495.87 × 10−2  
3.751.905.70 × 10−11.454.00 × 10−31.485.57 × 10−2  
4.001.925.80 × 10−11.465.00 × 10−31.475.59 × 10−2  
4.501.945.90 × 10−11.461.30 × 10−21.476.28 × 10−2  
5.001.976.00 × 10−11.451.20 × 10−21.466.54 × 10−2  
5.501.996.10 × 10−11.441.80 × 10−21.447.22 × 10−2  
6.002.026.20 × 10−11.412.30 × 10−21.421.14 × 10−1  
6.202.036.25 × 10−11.432.70 × 10−21.461.11 × 10−1  
6.502.046.30 × 10−11.463.30 × 10−21.469.24 × 10−2  
7.202.066.50 × 10−11.407.00 × 10−21.441.22 × 10−1  
7.902.126.70 × 10−11.206.50 × 10−21.321.13 × 10−1  
8.202.136.80 × 10−11.011.00 × 10−11.241.32 × 10−1  
8.502.156.90 × 10−11.302.15 × 10−11.311.90 × 10−1  
8.702.166.90 × 10−12.402.90 × 10−11.662.32 × 10−1  
9.002.177.00 × 10−12.563.70 × 10−11.872.95 × 10−1  
9.202.187.00 × 10−12.204.20 × 10−11.863.35 × 10−1  
9.502.197.10 × 10−11.951.60 × 10−11.962.89 × 10−1  
9.802.207.15 × 10−11.879.50 × 10−21.932.72 × 10−1  
10.02.217.20 × 10−11.829.00 × 10−21.832.34 × 10−1  
10.62.227.30 × 10−11.767.00 × 10−21.631.72 × 10−1  
11.02.237.30 × 10−11.725.00 × 10−21.591.64 × 10−1  
11.52.247.40 × 10−11.674.70 × 10−21.562.19 × 10−1  
12.52.277.50 × 10−11.625.30 × 10−21.533.02 × 10−1  
13.02.287.60 × 10−11.625.50 × 10−21.602.82 × 10−1  
14.02.317.75 × 10−11.567.30 × 10−21.522.79 × 10−1  
14.82.337.90 × 10−11.441.00 × 10−11.483.03 × 10−1  
15.02.337.90 × 10−11.422.00 × 10−11.473.39 × 10−1  
16.42.368.10 × 10−11.751.61 × 10−11.593.54 × 10−1  
17.22.388.20 × 10−12.082.42 × 10−11.713.92 × 10−1  
18.02.408.25 × 10−11.981.80 × 10−11.763.99 × 10−1  
18.52.418.30 × 10−11.851.70 × 10−11.804.20 × 10−1  
20.02.458.50 × 10−12.122.20 × 10−11.934.89 × 10−1  
21.32.468.60 × 10−12.062.30 × 10−11.975.19 × 10−1  
22.52.488.70 × 10−12.002.40 × 10−12.045.34 × 10−1  
25.02.518.90 × 10−11.882.80 × 10−12.094.92 × 10−1  
27.92.549.10 × 10−11.842.90 × 10−11.984.86 × 10−1  
30.02.579.30 × 10−11.823.00 × 10−11.944.59 × 10−1  
35.02.639.70 × 10−11.924.00 × 10−11.974.83 × 10−1  
40.02.691.00 × 10−01.865.00 × 10−11.935.56 × 10−1  

[14] To account for relative humidity, the medium's complex refractive index must be modified by mixing with that of water, to estimate an equivalent complex refractive index [e.g., Blanchet and List, 1983]:

equation image

where (r/ro)RH is the particle growth factor at relative humidity RH. The growth factor is unique to a given medium, and depends on many variables including molecular weight of the liquid, particle radius, temperature, and surface tension. Generally speaking, the growth factor increases monotonically with relative humidity until the RH reaches 70–80%, and then increases exponentially for higher relative humidities. In addition to the refractive index, the size distribution of the component must be modified using the growth factor. Following the approximation used by Nilsson [1979], we select one growth factor for the mode radius of a given component, and modify the size distribution by multiplying the mode radius by this representative growth factor. The growth factors used for each of the constituents are given in Table 3. Dust is considered nonhygroscopic and in our model does not grow with relative humidity. For the other constituents, representative growth factors were chosen from Table VII of the study of Nilsson [1979] or from the work of d'Almeida et al. [1991]. For use in the radiative transfer model, the optical properties of each aerosol component are calculated for the relative humidity (rounded to the nearest 10%) of each layer.

Table 3. Growth Factors Used to Adjust the Particle Size Distributions for the Mie Scattering Calculations
ConstituentRelative humidity
30%40%50%60%70%80%90%
sea salta1.0001.0001.5731.6201.7901.9652.345
sulfatesb1.1401.1801.3051.3501.4351.5391.746
sootc1.0001.0001.0001.0161.0331.1861.407
dust1.0001.0001.0001.0001.0001.0001.000
organicsd1.0311.0551.0901.1501.2601.5541.851
ashe1.0191.0341.0571.0981.1761.3981.645
internal mixtured1.0311.0551.0901.1501.2601.5541.851

[15] For our external mixture, we must vary the fractional contributions of the constituents with total (500 nm) optical depth. This variation in fractional contribution is necessary to differentiate between background and anthropogenic constituents. As is standard in studies of aerosol optical properties [Jayaraman et al., 1998; Satheesh et al., 1999], we use 500 nm as a reference wavelength, and calculate ωjλ, gjλ, Pmjλ, and βext for 500 nm as well as for all middle infrared wavelengths. The sea salt is considered a background constituent, and its 500-nm optical depth is held constant at 0.04. As total optical depth increases, the optical depths of all other constituents increase in the proportions reported by Satheesh et al. [1999]. The exception is the dust, which is also considered a background constituent. Once the 500-nm optical depth of dust reaches 0.05 (which occurs at total 500-nm aerosol optical depth 0.4), the optical depth of dust is then held constant at 0.05 while the optical depths of the remaining constituents (ash, sulfates, organics, soot) increase in the proportions given by Satheesh et al. [1999]. For the sea salt, a Mie scattering calculation using the size distributions of Satheesh et al. [1999] reveals that 24% of the 500-nm optical depth is due to the coarse mode, with the remaining 76% due to the accumulation mode. For all total aerosol optical depth increments considered in this study, the fractional contributions of each constituent to the total 500-nm optical depth are given in Table 4. For all infrared wavelengths, the optical depth contributions from the various constituents are weighted according to these fractions.

Table 4. Fractional Contribution to the Total 500 nm Aerosol Optical Depth from Each Constituent in the External Mixture
Total aerosol optical depthSea saltDustNitrateSootOrganicsAsh
0.10.4000.0840.2520.1020.1140.048
0.20.2000.1120.3360.1360.1520.064
0.30.1330.1220.3640.1470.1650.069
0.40.1000.1260.3780.1530.1710.072
0.50.0800.1000.4020.1640.1800.074
0.60.0670.0830.4170.1700.1870.076
0.70.0570.0720.4270.1740.1920.078
0.80.0500.0630.4350.1780.1940.080
0.90.0440.0560.4410.1800.1980.081
1.00.0400.0500.4460.1820.2000.082

[16] Using this external mixture, we wish to calculate the longwave fluxes as a function of total column aerosol optical depth at the standard visible reference wavelength 500 nm. We denote this reference total column optical depth as τc, and we want τc to vary over the range 0–1 as observed during the INDOEX field program. Let s(z) denote a vertical aerosol opacity profile as shown in Figure 1. To scale the spectral aerosol optical depth as a function of relative humidity, and hence altitude z, we define for the jth aerosol constituent:

equation image

The total 500-nm aerosol optical depth for the given model atmosphere is

equation image

where fjc) is the fractional contribution of the jth aerosol constituent to the 500-nm aerosol optical depth as given in Table 4. The spectral optical depth of the jth aerosol constituent, as a function of altitude, is therefore:

equation image

The single scattering albedo ωλ(z) and phase function moments Pmλ(z) for the external aerosol mixture are calculated from the constituent quantities ωjλ(z) and Pmjλ (z) by:

equation image

where τ(z) is the total optical depth for the layer at altitude z (gaseous absorption plus aerosol extinction).

3. Longwave Radiative Fluxes—External Mixture

[17] Figure 3 shows the downwelling longwave flux at the ocean surface as a function of visible wavelength (500 nm) aerosol optical depth, for all three model atmospheres and both aerosol profiles. During the 1999 INDOEX field program, aerosol optical depths (500 nm) were observed in the range 0–1, with typical values of 0.3 for the boundary layer aerosol profile and 0.7 for the aerosol profile extending to 3 km [Satheesh and Ramanathan, 2000]. With no aerosol, the downwelling longwave flux increases in increments of 15 W m−2 as we switch from the low RH to the average and then to the high RH model atmospheres, due to increasing temperature and water vapor abundance. In this figure, the 500-nm aerosol optical depth is that of the particular aerosol column in the model atmosphere, including relative humidity effects. Therefore, we expect the downwelling fluxes for both aerosol profiles to be nearly identical for a given model atmosphere, as plotted in Figure 3. For a given 500-nm aerosol optical depth, the downwelling surface fluxes for the boundary layer aerosol profile (profile 1) are slightly larger than those for the deeper profile (profile 2); in the former, the entire emitting aerosol layer lies more closely confined near the surface, in a warmer temperature range (Figure 1). However, we should bear in mind that for a given aerosol number density, the total optical depth of aerosol profile 2 will be larger than that of profile 1. For a typical aerosol optical depth of profile 1 (0.3 at 500 nm), the downwelling longwave surface flux is increased by 7.7 W m−2 over the clean air case, for our average model atmosphere. For a typical aerosol optical depth of profile 2 (0.7 at 500 nm), the downwelling longwave surface flux is increased by 11.2 W m−2 over the clean air case, for our average model atmosphere.

Figure 3.

Downwelling longwave flux at the Indian Ocean surface, as a function of total aerosol column optical depth at 500 nm, for all three model atmospheres. For a particular model atmosphere, the dotted curve depicts the flux for the boundary layer aerosol profile (profile 1) and the solid curve depicts the flux for the vertically extended aerosol profile (profile 2).

[18] Figure 4 shows the upwelling longwave flux at the TOA, for all three model atmospheres and both aerosol profiles. Due to water vapor abundance and temperature structure, the clean air TOA longwave fluxes for the low RH and high RH model atmospheres are different from that of the average model atmosphere by +7.8 and −3.4 W m−2, respectively. For a given 500-nm aerosol optical depth, the difference in TOA longwave flux between the two aerosol profiles is slightly greater than that for the surface downwelling longwave flux. In aerosol profile 2, the extension of the aerosol plume to an altitude of 3 km results in a noticeable increase in “greenhouse” trapping of longwave radiation. For a 500-nm aerosol optical depth of 0.3 in the average model atmosphere, the TOA longwave fluxes are 292.7 and 292.1 W m−2, for profiles 1 and 2, respectively. For a 500-nm aerosol optical depth of 0.7 in the average model atmosphere, the TOA longwave fluxes are 292.1 and 291.2 W m−2, for profiles 1 and 2, respectively. For a typical aerosol optical depth of profile 1 (0.3 at 500 nm), the upwelling TOA flux is reduced by 1.3 W m−2 from the clean air case, for our average model atmosphere. For a typical aerosol optical depth of profile 2 (0.7 at 500 nm), the upwelling TOA flux is reduced by 2.7 W m−2 from the clean air case, for our average model atmosphere.

Figure 4.

Upwelling longwave flux at the top-of-the-atmosphere, as a function of total aerosol column optical depth at 500 nm, for all three model atmospheres. For a particular model atmosphere, the dotted curve depicts the flux for the boundary layer aerosol profile (profile 1) and the solid curve depicts the flux for the vertically extended aerosol profile (profile 2).

[19] Figures 5 and 6show the downwelling surface and upwelling TOA fluxes in the middle infrared window (8–12 μm). Comparison of the vertical axes of these figures with those of Figures 3 and 4 shows that the majority of the total longwave flux lies outside the middle infrared window (approximately 92% and 65% for the downwelling surface and upwelling TOA fluxes, respectively). However, the major part of the aerosol forcing lies inside this window. For the typical model atmosphere, between 72% and 80% of the TOA aerosol forcing lies inside the middle infrared window, depending on aerosol profile, and also increasing with total aerosol optical depth. For the dry model atmosphere, this range decreases slightly to 68–73%, and for the wet model atmosphere, the range is 71–77%. Thus, although an increase in relative humidity will make water vapor emission more prominent within the middle infrared window, the increase in relative humidity also increases the aerosol opacity by particle growth among most of the constituents.

Figure 5.

As in Figure 4, but for the downwelling flux in the middle infrared window. The labels a, b, and c refer to the low RH, average, and high RH model atmospheres, respectively.

Figure 6.

As in Figure 5, but for the upwelling flux in the middle infrared window. The labels a, b, and c refer to the low RH, average, and high RH model atmospheres, respectively.

[20] In Figure 7, we illustrate the relative contributions to the downwelling surface and outgoing TOA longwave flux from the background and anthropogenic components, for a typical model atmosphere and aerosol profile 2. Once the total (500 nm) aerosol optical depth increases above 0.1, the contribution from sea salt alone remains constant at 3.4 W m−2 and −0.8 W m−2 for the downwelling surface and TOA outgoing flux, respectively. Once the total (500 nm) aerosol optical depth increases above 0.4, the contribution from both the sea salt and dust remains constant at 5.4 W m−2 and −1.5 W m−2 for the downwelling surface and TOA outgoing flux, respectively. Further increases in aerosol radiative forcing with increasing optical depth are due to the anthropogenic components.

Figure 7.

Downwelling surface (a) and upwelling top-of-atmosphere (b) longwave fluxes, as a function of 500-nm aerosol optical depth, calculated using the average model atmosphere and the vertically extended aerosol profile (profile 2). The solid curve represents the flux from all aerosol constituents. The dashed and dotted curves represent the flux from just the sea salt and the sea salt plus dust, respectively. For these latter curves, the 500-nm optical depth refers to that of an aerosol with all constituents added.

4. Longwave Radiative Fluxes—Internal Mixture

[21] To model the optical properties of an internal mixture, we consider an aggregate particle composed of all six aerosol constituent materials which are mixed according to the mass fractions δj of the total dry aerosol layer observed in the field. The volume-weighted aggregate refractive index (real or imaginary part) is:

equation image

where ρj are the densities of the constituents. The mass fractions and densities are for each of the constituents are given in Table 5. The complex refractive index resulting from (1) is given in Table 2. This is the simplest way to model an internal mixture, and has often been used in climate-related studies [e.g., Haywood and Shine, 1995]. Sokolik and Toon [1999] considered more sophisticated effective medium approximations, and found that for dust mixtures, the simple volume-weighted mean of (7) compares well with them in the middle infrared. The particle size distribution for our internal mixture is that of the complete bimodal distribution observed by Satheesh et al. [1999], given by (1) with J = 2, mode radii rm1 = 0.135 and rm2 = 0.955 μm, and standard deviations σ1 = 2.477 and σ2 = 2.051. The smaller particle mode is dominant, and the relative mode amplitudes are N1 = 104 × N2. The relative humidity dependence in the optical properties of this single-component aerosol layer is determined using (2) and the growth factors listed in Table 3, with the size distribution adjusted as was done for the external mixture.

Table 5. Densities and Mass Fractions for the Six Individual Constituents Making up the Internal Mixture
ConstituentDensity ρj, g cm−3Mass fraction δj
sea salt2.2a0.06b
sulfates1.769a0.41
soot2.3a0.10
dust2.5a0.21
organics1.8c0.12
ash2.45d0.10

[22] The longwave forcing of the internal mixture is found to be larger than that of the external mixture, as shown in Figure 8. For the typical model atmosphere, a typical low aerosol profile (500-nm optical depth 0.3) will exhibit a change in the downwelling surface flux of 11.1 and 7.7 W m−2 for the internal and external mixture, respectively. The corresponding changes in upwelling TOA longwave flux are −1.6 and −1.2 W m−2 for the internal and external mixture, respectively. A typical high aerosol profile (500-nm optical depth 0.7) will exhibit a change in the downwelling surface flux of 21.5 and 11.2 W m−2 for the internal and external mixture, respectively. The corresponding changes in upwelling TOA longwave flux are −4.9 and −2.7 W m−2 for the internal and external mixture, respectively.

Figure 8.

Downwelling surface (a) and upwelling top-of-atmosphere (b) longwave fluxes, as a function of 500-nm aerosol optical depth, calculated using the average model atmosphere. The upper two curves (dotted and solid) pertain to the internal mixture and the lower two curves (dotted and solid) pertain to the external mixture. The dotted curves depict the flux for the boundary layer aerosol profile (profile 1) and the solid curves depict the flux for the vertically extended aerosol profile (profile 2).

[23] These relatively large increases in longwave radiative forcing from the internal mixture can be understood by examining the volume extinctions as shown in Figure 9. Here, the spectral volume extinctions are scaled using the dry aerosol value at 500 nm, i.e., the quantity bλ of (3). For the internal mixture (Figure 9a), the values of bλ vary by nearly an order of magnitude as a function of relative humidity, and in the middle infrared window are between 15% and 80% of the 500-nm volume extinction. This is approximately true for only two of the other constituents, dust, and ash, which together are approximately 25% of the external mixture, by volume. The sulfates are approximately 47% of the internal mixture by volume, and at the highest relative humidities (where we use sulfuric acid refractive indices), the volume extinctions in the middle infrared window are between 15% and 40% of the 500-nm volume extinction (i.e., smaller volume extinctions than for the internal mixture). For sulfates at lower relative humidities (solid curves of Figure 9b), the volume extinctions are an order of magnitude smaller than those of the internal mixture. The organics and soot together are approximately 22% of the external mixture, by volume, and their middle infrared volume extinctions are also an order of magnitude smaller than those of the internal mixture. Only for the sea salt (a mere ∼6% by volume of the external mixture) do we see middle infrared volume extinctions larger than those of the internal mixture. Thus, the radiative properties of the external mixture are dominated by particles having smaller middle infrared volume extinctions than those that result when the constituents are internally mixed. This is the result of the strong dependence in middle infrared volume extinction on particle size, in the size range typical of tropospheric aerosols [e.g., Blanchet and List, 1983]. Note that this radiative contrast between the internal and external mixture depends on the assumption that the particle size distribution for the internal mixture can be represented by the bimodal distribution for the entire aerosol layer, given by Satheesh et al. [1999].

Figure 9.

(opposite) The volume extinction coefficients calculated from Mie theory for the internal mixture (a) and the individual components of the external mixture (b–h) as a function of wavelength and relative humidity. The volume extinction coefficients are scaled to a reference value of unity at 500 nm for the dry particle size distribution, as per (3) and related text.

5. Discussion

[24] Satheesh et al. [1999] also considered the question of internal versus external mixing, in the shortwave, and concluded that there was little difference between the two cases. Examining Figure 9, we see that there is much less variability in the shortwave than in the longwave between the volume extinction for the internal mixture and those of the individual constituents. This explains the conclusion of Satheesh et al. [1999], which is not valid for the longwave. In fact, if we examine the longwave cooling rates in the lower troposphere as a function of altitude (Figure 10), we notice that the cooling rates can be up to a factor of two larger in the internally mixed case. Thus, a complete description of Indian Ocean aerosol radiative properties will require additional information from the field about mixing states.

Figure 10.

Vertical profiles of the longwave cooling rate in the lower troposphere, plotted for aerosol optical depth (500 nm) increasing from right to left in increments of 0.1, (a) for the internal mixture in the boundary layer aerosol profile, (b) for the external mixture in the boundary layer profile, (c) for the internal mixture in the vertically extended profile, and (d) for the external mixture in the vertically extended aerosol profile.

[25] When externally mixed, the Indian Ocean tropospheric aerosol can reduce the outgoing planetary longwave radiation flux by up to 3.2 W m−2, while increasing the surface downwelling longwave flux by up to 15.7 W m−2. For an internal mixture, these radiative perturbations are even larger as shown in Figure 8. These radiative signatures are considerably smaller than their counterparts in the shortwave as measured during INDOEX, but they should be detectable by satellite and ground-based instruments, particularly spectroradiometers such as the Atmospheric Infrared Sounder (AIRS) scheduled as part of NASA's Aqua platform (launch in spring, 2002), or FTIR systems [e.g., Lubin and Simpson, 1994; Spankuch et al., 2000]. These longwave radiation changes due to aerosol are approximately of the same order as variability in the longwave flux due to changing tropospheric temperature and water vapor abundance alone, as can be seen in the clear-sky limits of Figures 3, 4, 5, and 6. However, if the Indian Ocean aerosol is a persistent and widespread phenomenon, the aerosol-induced radiative fluxes will establish a background longwave radiation environment that is different from that of clean air, and thermodynamically induced longwave radiation variability will occur with reference to this altered background level. The longwave radiative effects of the tropospheric aerosols partially offset the shortwave effects and also have implications for the thermodynamics of the lower troposphere. If an increasing portion of the longwave radiation emitted by the surface is absorbed and reemitted by a scattering and absorbing layer near the surface, the result will be a reduction in the transfer of heat between the surface and the entire troposphere. This may serve to stabilize surface temperatures (e.g., reducing the diurnal cycle) and to reduce convection [Ackerman, 1977]. This would complement the aerosol shortwave effect in possibly slowing down the region's hydrological cycle [Satheesh and Ramanathan, 2000]. This study suggests that longwave radiative flux changes due to Indian Ocean tropospheric aerosols are large enough and have potentially enough significance that they should be included in simulations of regional and global climate.

Acknowledgments

[26] This research was supported by the National Science Foundation under NSF ATM-9730276.

Ancillary