IPCC Fourth Assessment Report: Climate Change 2007
Climate Change 2007: Working Group I: The Physical Science Basis Biomass Burning Aerosol

The TAR reported a contribution to the RF of roughly –0.4 W m–2 from the scattering components (mainly organic carbon and inorganic compounds) and +0.2 W m–2 from the absorbing components (BC) leading to an estimate of the RF of biomass burning aerosols of –0.20 W m–2 with a factor of three uncertainty. Note that the estimates of the BC RF from Hansen and Sato (2001), Hansen et al. (2002), Hansen and Nazarenko (2004) and Jacobson (2001a) include the RF component of BC from biomass burning aerosol. Radiative forcing due to biomass burning (primarily organic carbon, BC and inorganic compounds such as nitrate and sulphate) is grouped into a single RF, because biomass burning emissions are essentially uncontrolled. Emission inventories show more significant differences for biomass burning aerosols than for aerosols of fossil fuel origin (Kasischke and Penner, 2004). Furthermore, the pre-industrial levels of biomass burning aerosols are also difficult to quantify (Ito and Penner, 2005; Mouillot et al., 2006).

The Southern African Regional Science Initiative (SAFARI 2000: see Swap et al., 2002, 2003) took place in 2000 and 2001. The main objectives of the aerosol research were to investigate pyrogenic and biogenic emissions of aerosol in southern Africa (Eatough et al., 2003; Formenti et al., 2003; Hély et al., 2003), validate the remote sensing retrievals (Haywood et al., 2003b; Ichoku et al., 2003) and to study the influence of aerosols on the radiation budget via the direct and indirect effects (e.g., Bergstrom et al., 2003; Keil and Haywood, 2003; Myhre et al., 2003; Ross et al., 2003). The physical and optical properties of fresh and aged biomass burning aerosol were characterised by making intensive observations of aerosol size distributions, optical properties, and DRE through in situ aircraft measurements (e.g., Abel et al., 2003; Formenti et al., 2003; Haywood et al., 2003b; Magi and Hobbs, 2003; Kirchstetter et al., 2004) and radiometric measurements (e.g., Bergstrom et al., 2003; Eck et al., 2003). The ωo at 0.55 µm derived from near-source AERONET sites ranged from 0.85 to 0.89 (Eck et al., 2003), while ωo at 0.55 µm for aged aerosol was less absorbing at approximately 0.91 (Haywood et al., 2003b). Abel et al. (2003) showed evidence that ωo at 0.55 µm increased from approximately 0.85 to 0.90 over a time period of approximately two hours subsequent to emission, and attributed the result to the condensation of essentially non-absorbing organic gases onto existing aerosol particles. Fresh biomass burning aerosols produced by boreal forest fires appear to have weaker absorption than those from tropical fires, with ωo at 0.55 µm greater than 0.9 (Wong and Li 2002). Boreal fires may not exert a significant direct RF because a large proportion of the fires are of natural origin and no significant change over the industrial era is expected. However, Westerling et al. (2006) showed that earlier spring and higher temperatures in USA have increased wildfire activity and duration. The partially absorbing nature of biomass burning aerosol means it exerts an RF that is larger at the surface and in the atmospheric column than at the TOA (see Figure 2.12).


Figure 2.12. Characteristic aerosol properties related to their radiative effects, derived as the mean of the results from the nine AeroCom models listed in Table 2.5. All panels except (b) relate to the combined anthropogenic aerosol effect. Panel (b) considers the total (natural plus anthropogenic) aerosol optical depth from the models. (a) Aerosol optical depth. (b) Difference in total aerosol optical depth between model and MODIS data. (c) Shortwave RF. (d) Standard deviation of RF from the model results. (e) Shortwave forcing of the atmosphere. (f) Shortwave surface forcing.

Modelling efforts have used data from measurement campaigns to improve the representation of the physical and optical properties as well as the vertical profile of biomass burning aerosol (Myhre et al., 2003; Penner et al., 2003; Section 2.4.5). These modifications have had important consequences for estimates of the RF due to biomass burning aerosols because the RF is significantly more positive when biomass burning aerosol overlies cloud than previously estimated (Keil and Haywood, 2003; Myhre et al., 2003; Abel et al., 2005). While the RF due to biomass burning aerosol in clear skies is certainly negative, the overall RF of biomass burning aerosol may be positive. In addition to modelling studies, observations of this effect have been made with satellite instruments. Hsu et al. (2003) used SeaWiFs, TOMS and CERES data to show that biomass burning aerosol emitted from Southeast Asia is frequently lifted above the clouds, leading to a reduction in reflected solar radiation over cloudy areas by up to 100 W m–2, and pointed out that this effect could be due to a combination of direct and indirect effects. Similarly, Haywood et al. (2003a) showed that remotely sensed cloud liquid water and effective radius underlying biomass burning aerosol off the coast of Africa are subject to potentially large systematic biases. This may have important consequences for studies that use correlations of τaer and cloud effective radius in estimating the indirect radiative effect of aerosols.

Since the biomass burning aerosols can exert a significant positive RF when above clouds, the aerosol vertical profile is critical in assessing the magnitude and even the sign of the direct RF in cloudy areas. Textor et al. (2006) showed that there are significant differences in aerosol vertical profiles between global aerosol models. These differences are evident in the results from the recently published studies and AeroCom models in Table 2.5. The most negative RF of –0.05 W m–2 is from the model of Koch (2001) and from the Myhre et al. (2003) AeroCom submission, while several models have RFs that are slightly positive. Hence, even the sign of the RF due to biomass burning aerosols is in question.

Table 2.5. Estimates of anthropogenic carbonaceous aerosol forcing derived from models published since the TAR and from the AeroCom simulations where different models used identical emissions. POM: particulate organic matter; BC: black carbon; BCPOM: BC and POM; FFBC: fossil fuel black carbon; FFPOM: fossil fuel particulate organic matter; BB: biomass burning sources included.

No Modela LOAD POM (mgPOM m–2τaer POM  τaer POMant (%) LOAD BC (mg m–2RF BCPOM (W m–2RF POM (W m–2RF BC (W m–2RF FFPOM (W m–2RF FFBC (W m–2RF BB (W m–2Reference 
Published since IPCC, 2001 
SPRINT         0.12 –0.24 0.36 –0.05 0.15 –0.01 (Takemura et al., 2001) 
LOA 2.33 0.016   0.37 0.30 –0.25 0.55 –0.02 0.19 0.14 (Reddy et al., 2005b) 
GISS 1.86 0.017   0.29 0.35 –0.26 0.61 –0.13 0.49 0.065 (Hansen et al., 2005) 
GISS 1.86 0.015   0.29 0.05 –0.30 0.35 –0.08b 0.18b –0.05b (Koch, 2001) 
GISS 2.39     0.39 0.32 –0.18 0.50 –0.05b 0.25b 0.12b (Chung and Seinfeld., 2002) 
GISS 2.49     0.43 0.30 –0.23 0.53 –0.06b 0.27b 0.09b (Liao and Seinfeld, 2005) 
SPRINTARS 2.67 0.029 82 0.53 0.15 –0.27 0.42 –0.07b 0.21b 0.01b (Takemura et al., 2005) 
GATORG 2.55     0.39 0.47 –0.06 0.55 –0.01b 0.27b 0.22b (Jacobson, 2001b) 
MOZGN 3.03 0.018       –0.34         (Ming et al., 2005a) 
CCM       0.33     0.34       (Wang, 2004) 
UIO-GCM       0.30     0.19       (Kirkevag and Iversen, 2002) 
AeroCom: identical emissions used for year 1750 and 2000 (Schulz et al., 2006) 
UMI 1.16 0.0060 53 0.19 0.02 –0.23 0.25 –0.06b 0.12b –0.01 (Liu and Penner, 2002) 
UIO_CTM 1.12 0.0058 55 0.19 0.02 –0.16b 0.22b –0.04 0.11 –0.05 (Myhre et al., 2003) 
LOA 1.41 0.0085 52 0.25 0.14 –0.16c 0.32c –0.04b 0.16b 0.02b (Reddy and Boucher, 2004) 
LSCE 1.50 0.0079 46 0.25 0.13 –0.17 0.30 –0.04b 0.15b 0.02b (Schulz et al., 2006) 
ECHAM5-HAM 1.00 0.0077   0.16 0.09 –0.10c 0.20c –0.03b 0.10b 0.01 (Stier et al., 2005) 
GISS 1.22 0.0060 51 0.24 0.08 –0.14 0.22 –0.03b 0.11b 0.01b (Koch, 2001) 
UIO_GCM 0.88 0.0046 59 0.19 0.24 –0.06 0.36 –0.02b 0.18b 0.08b (Iversen and Seland, 2002) 
SPRINTARS 1.84 0.0200 49 0.37 0.22 –0.10 0.32 –0.01 0.13 0.06 (Takemura et al., 2005) 
ULAQ 1.71 0.0075 58 0.38 –0.01 –0.09 0.08 –0.02b 0.04b –0.03b (Pitari et al., 2002) 
Average A–K 2.38 0.019   0.38 0.26 –0.24 0.44 –0.06 0.25 0.07   
Average L–T 1.32 0.008 53 0.25 0.10 –0.13 0.25 –0.03 0.12 0.01   
Stddev A–K 0.42 0.006   0.08 0.14 0.08 0.13 0.04 0.11 0.09   
Stddev L–T 0.32 0.005 0.08 0.09 0.05 0.08 0.01 0.04 0.04   


a MOZGN: MOZART (Model for OZone and Related chemical Tracers-GFDL(Geophysical Fluid Dynamics Laboratory)-NCAR (National Center for Atmospheric Research); for other models see Note (a) in Table 2.4.

b Models A to C are used to provide a split in sources derived from total POM and total BC: FFPOM = POM × 0.25; FFBC = BC × 0.5; BB = (BCPOM) – (FFPOM + FFBC); BC = 2 × FFBC; POM = 4 × FFPOM.

c Models L, O and Q to T are used to provide a split in components: POM = BCPOM × (–1.16); BC = BCPOM × 2.25.

The mean and median of the direct RF for biomass burning aerosol from grouping all these studies together are similar at +0.04 and +0.02 W m–2, respectively, with a standard deviation of 0.07 W m–2. The standard deviation is multiplied by 1.645 to approximate the 90% confidence interval, leading to a direct RF estimate of +0.03 ± 0.12 W m–2. This estimate of the direct RF is more positive than that of the TAR owing to improvements in the models in representing the absorption properties of the aerosol and the effects of biomass burning aerosol overlying clouds.