Savanna burning in Africa occurs over a wide range of environmental, vegetation, and land use conditions. The emission factors for trace emissions from these fires can vary by a factor of 6 to 8, depending on whether the fires burn in miombo woodlands or in ecosystems where grass vegetation dominates. Ground‐based measurements of smoke emissions and aboveground biomass were made for fires in grassland and woodland savanna ecosystems in South Africa and Zambia. A high combustion efficiency ( trueη⌢ ) was measured for the pure grassland; i.e., a high proportion of carbon was released as CO2. The trueη⌢ was lower for woodland savanna ecosystems with variable amounts of grass and with a more compact layer of leaf material and litter lying near the ground. The trueη⌢ was found to be dependent on the ratio of grass to the sum of grass and litter. Models developed for estimating emissions were integrated in a nomogram for estimating total emissions of CO2, CO, CH4, nonmethane hydrocarbons, and particles of less than 2.5 μm diameter per unit area.

Fires are dominant factors in shaping the structure and composition of vegetation in African savanna ecosystems. Emissions such as CO2, NOx, CH4, and other compounds originating from these fires are suspected to contribute substantially to changes in global biogeochemical processes. Limited quantitative data exist detailing characteristics of biomass, burning conditions, and the postfire environment in African savannas. Fourteen test sites, differentiated by distinct burn frequency histories and land‐use patterns, were established and burned during August and September 1992 in savanna parklands of South Africa and savanna woodlands of Zambia. Vegetation physiognomy, available fuel loads, the levels of biomass consumed by fire, environmental conditions, and fire behavior are described. In the South African sites, total aboveground fuel loads ranged from 2218 to 5492 kg ha−1 where fire return intervals were 1–4 years and exceeded 7000 kg ha−1 at a site subjected to 38 years of fire exclusion. However, fireline intensity was only 1419 kW m−1 at the fire exclusion site, while ranging from 480 to 6130 kW m−1 among the frequent fire sites. In Zambia, total aboveground fuel loads ranged from 3164 kg ha−1 in a hydromorphic grassland to 7343 kg ha−1 in a fallow shifting cultivation site. Dormant grass and litter constituted 70–98% of the total fuel load among all sites. Although downed woody debris was a relatively minor fuel component at most sites, it constituted 43–57% of the total fuel load in the fire exclusion and shifting cultivation sites. Fire line intensity ranged between 1734 and 4061 kW m−1 among all Zambian sites. Mean grass consumption generally exceeded 95%, while downed woody debris consumption ranged from 3 to 73% at all sites. In tropical savannas and savanna woodlands of southern Africa, differences in environmental conditions, land‐ use patterns, and fire regimes influence vegetation characteristics and thus influence fire behavior and biomass consumption.

During a field study in southern Africa (Southern African Fire‐Atmosphere Research Initiative (SAFARI‐92)), black carbon formation was quantified in the residues of savanna fires. The volatilization ratios of C, H, N, and S were determined by measuring their contents in the fuel and residue loads on six experimental sites. The volatilization of sulfur (86 ± 8%) was significantly higher than previously reported. Volatilization of H, N, and S was significantly correlated with that of carbon, enabling us to estimate their volatilization during savanna fires by extrapolation from those of carbon. By partitioning the residues in various fractions (unburned, partially burned, and ash), a strong correlation between the H/C ratio in the residue and the formation of black carbon was obtained. The ratio of carbon contained in ash to carbon contained in the unburned and partially burned fraction is introduced as an indicator of the degree of charring. As nitrogen was enriched in the residue, especially in the ash fraction of >0.63 mm, this indicator may be useful for an assessment of nutrient cycling. We show that the formation of black carbon is dependent on the volatilization of carbon as well as the degree of charring. The ratio of black carbon produced to the carbon exposed to the fire in this field study (0.6–1.5%) was somewhat lower than in experimental fires under laboratory conditions (1.0–1.8%) which may be due to less complete combustion. The average ratio of black carbon in the residue to carbon emitted as CO2 ranged from 0.7 to 2.0%. Using these ratios together with various estimates of carbon exposed or emitted by savanna fires, the worldwide black carbon formation was estimated to be 10–26 Tg C yr−1 with more than 90% of the black carbon remaining on the ground. The formation of this black carbon is a net sink of biospheric carbon and thus of atmospheric CO2 as well as a source of O2.

No abstract