GREENHOUSE
VEGETATION
GAS EMISSIONS
FIRES IN SOUTHERN
FROM
AFRICA
R.J. SCHOLES
Division of Forest Science and Technology, CSIR, PO Box 395, Pretoria 0001 South Africa. ([email protected] co.za)
Abstract. Methane (CH4), carbon monoxide(CO), nitrogenoxides (NOx), volatileorganic carbon, and aerosols emitted as a result of the deliberateor accidentalburning of natural vegetationconstitutea large componentof the greenhouse gas emissions of many African countries, but the data needed for calculatingthese emissions by the IPCC methodology is sparse and subject to estimation errors. An improved procedure for estimating emissions from fires in southern Africa has been developed. The proposed procedure involves reclassifying existingvegetationmaps into one of elevenbroad, functionalvegetationclasses. Fuel loads are calculatedwithin each 0.5 x 0.5° cell based on empiricalrelationshipsto climate data for each class. The fractionalarea of each class that burns is estimated by using daily low-resolutionsatellite fire detection, which is calibratedagainst a subsample of pre- and post-fire high-resolutionsatellite images. The emission factors that relate the quantity of gas released to the mass of fuel burned are based on recent field campaigns in Africa and are related to combustion efficiency,which is in turn related to the fuel mix. The emissionsare summed over the 1989 fire season for Africa south of the equator. The estimated emissions from vegetationburning in the subcontinent are 0.5 Tg CH4, 14.9 Tg CO, 1.05 Tg NO, and 1.08 Tg of particles smaller than 2.51xn't The 324 Tg CO: emitted is expected to be reabsorbed in subsequentyears. These estimates are smaller than previous estimates.
10 Introduction Most of Africa has a strongly seasonal climate, with hot, wet summers and warm, dry winters. This kind of climate permits frequent fires, which in turn promote the growth of a fire-tolerant vegetation known as a savanna. Because many savannas are located on extremely infertile soils, frequent application of fire is essential to sustainable low-input pastoralism. Fire has been used for millennia to promote the growth of palatable forage and prevent the encroachment of trees (Goldammer, 1993), and in fact, most fires in Africa are ignited by people rather than by lightning. In the southem hemisphere, savanna fires are concentrated in the dry season, between April and October. Savanna grass is approximately 45% carbon (C) on a dry-mass basis. When it burns, most of this C is released as CO 2. In general, this does not represent a net release of CO 2, however, because the vegetation regrows in subsequent years. On the other hand, it is a net source of a range of other radiatively active and O3-forming trace gases and aerosols (particles), because these are not reabsorbed. If the fire frequency or fuel load are increased systematically, vegetation fires are a net source of CO v as well; conversely, if fires are suppressed, savannas become a major C sink as tree biomass and organic soil matter increase. The degree to which colonial and post-colonial fire regulations have altered the long-term frequency of fire is unknown. In some places fire may now be more frequent than it was previously, whereas in others, the reverse may be true. African savanna fires
Environmental Monitoring and Assessment 38:169-179, 1995. © 1995 Kluwer Academic Publishers. Printed in the Netherlands,
170
R.J. SCHOLES
have a profound effect on the regional and global atmosphere (Crutzen and Andreae, 1990), but there is little data to suggest that the magnitude of the impact has changed in recent times. The emissions of partial combustion products (CO, CH 4, and aerosols) are inversely proportional to the degree of oxygenation during combustion (Ward and Radke, 1993). The fine, dry grass fuels that predominate in savannas are highly oxygenated during combustion and therefore release relatively small amounts of these gases. Emissions of nitrogenous gases (predominantly NO x, but also small amounts of nitrous oxide (N20) and several other gases) depend on the N content of the fuel (Lobert et al., 1991), which is typically low (0.5-1%). Because of the vast area of Africa that burns, the gross trace gas emissions are globally significant (Crutzen and Andreae, 1990; Levine, 1990; Lacaux, Cachier, and Delmas, 1993) and could dominate the greenhouse gas inventories of large but lightly industrialized African countries, such as Angola, Tanzania, and Zambia. The general procedure for estimating emissions from biomass burning (which should more properly be called vegetation burning, because biomass refers only to the living material, and fires mostly consume dead material) is to divide the study region into vegetation types and calculate the emission per type (OECD, 1991; UNEP et al., 1995). In each type, the area (A:m 2) mean fuel load (L:g m-2), combustion completeness (the fraction of the biomass exposed to the fire that actually burns--C:nondimensional fraction), and average time between fires (R:years) is estimated, usually based on expert opinion, because few detailed measurements exist. The amount of biomass consumed per annum (B: g yT-1) in each vegetation class is then calculated from B = A R -1 L C. The emission factors (F:g emission product/g fuel) are typically estimated by measuring the increase in a given gas relative to the increase in CO 2 in a smoke plume. This measurement is then converted to a dry-mass basis by assuming a C content in the fuel (41-45%) and an efficiency of conversion of fuel C to CO 2 (80-98%). The emission of combustion products is then given for each gas by E a~ = B Fg~s, which is summed over the various vegetation classes. The method proposed by the Intergovernmental Panel on Climate Change for calculating emissions from savanna burning follows this simple approach (UNEP et al., 1995). The disadvantage of the method is that the range of possible values for R, L, C, and F is very large, and the procedures for validating them in the field are demanding. This means that the accuracy of the estimate is generally very low. This paper briefly describes a method (documented in more detail in Scholes, Kendall, and Justice (1995) and Scholes, Ward, and Justice (1995) that greatly reduces the uncertainties by applying the following strategies: 1. The area burned is measured using a combination of low- and high-resolution satellite images. 2. The fuel load is modeled in four categories, constrained by climate. 3. The emission factors are related to the fuel mixture. The method has been applied to Africa south of the equator. It could probably also be applied to West Africa and other tropical areas that are subject to frequent and extensive fires, but the predictive equations would have to be recalibrated for areas outside of southern Africa.
64
GREENHOUSE GAS EMISSIONS FROM VEGETATION FIRES IN SOUTHERN AFRICA
171
The objectives of this paper are to describe a procedure for estimating regional trace gas emissions from vegetation fires and to provide data that will assist in the preparation of national greenhouse gas emission inventories in southern Africa.
2. M e t h o d s The best available vegetation maps (Table I) of southern Africa at an appropriate scale (1:1 M to 1:5 M) were digitized and edge-matched using geographic information system (GIS) technology to form a detailed, composite vegetation map of the subcontinent. The mapped vegetation classes were reclassified into eleven broad types (Table II), each of which has distinctive fuel types and relationships between fuel production and climate. The maps were converted to raster form at a scale matching the available long-term monthly
TABLE I The best availablebroad-scale vegetation maps for Africa south of the equator. Where no suitable map is available,the all-Africamap by White (1983) can be used
Country
Africa South Africa, Lesotho, and Swaziland Botswana
Namibia Zimbabwe and Mozambique Angola Zambia Malawi Tanzania Zaire
Kenya Uganda
Source
White, F: 1984, Vegetationof Africa, Unesco/AETFAT,Pads. Acocks, J.P.H.: 1955, Veld Types of South Africa. Memoirs of the Botanical Survey of South Africa 28 (revised 1975 as Memoir 40). Ministry of Agriculture, Botswana: 1991, Vegetationmap of the Republic of Botswana, Land UtilisationDivision, Private Bag 003, Gaborone. Giess, W.: 1971, VegetationMap of South West Africa. Dinteria: 4:5-114 Granvaux-Barbosa,L.A.: 1967, Vegetationmap of the flora Zambeziaca area, Collins, Harare. Granvaux-Barbosa, L.A.: 1970, Fitograficade Angola, histituto de InvesticaoCientificade Angola. Edwards, A.C.R.: 1976, Republic of Zambia Vegetation Map, Institut fur AngewandteGeodesia, Frankfort. 1980, National Atlas of Malawi, Department of Survey, Blantyre. Ministry of Lands: 1984, TanzaniaVegetationCover Types, Forestry Division, Ministry of Lands, Governmentof Tanzania. 1932, Atlas of Southern Katanga, VegetationMap. NASA-GIMMS: In press, PathfinderProject on Central African Forests. Ministry of Wildlife: 1986, Biotic Communitiesof Kenya. 1964 (reissued 1972), Vegetationof Uganda, Department of Land Surveys, Entebbe.
65
Scale
1:5 000 000
1:1 500 000
1:2 000 000 1:3 000 000
1:2 500 000 1:2 500 000 1:1 500 000 1:1 000 000
1:2 000 000 1:500 000 1:250 000 1:1 000 000
1:500 000
172
R.J. SCHOLES
0
0
d
d
0
~
d
0
d
0
~
0
d
0
d
0
0
d
d
~
d
×
deo~de~edd~
,
VEGETATION
GAS EMISSIONS
FIRES IN SOUTHERN
FROM
AFRICA
R.J. SCHOLES
Division of Forest Science and Technology, CSIR, PO Box 395, Pretoria 0001 South Africa. ([email protected] co.za)
Abstract. Methane (CH4), carbon monoxide(CO), nitrogenoxides (NOx), volatileorganic carbon, and aerosols emitted as a result of the deliberateor accidentalburning of natural vegetationconstitutea large componentof the greenhouse gas emissions of many African countries, but the data needed for calculatingthese emissions by the IPCC methodology is sparse and subject to estimation errors. An improved procedure for estimating emissions from fires in southern Africa has been developed. The proposed procedure involves reclassifying existingvegetationmaps into one of elevenbroad, functionalvegetationclasses. Fuel loads are calculatedwithin each 0.5 x 0.5° cell based on empiricalrelationshipsto climate data for each class. The fractionalarea of each class that burns is estimated by using daily low-resolutionsatellite fire detection, which is calibratedagainst a subsample of pre- and post-fire high-resolutionsatellite images. The emission factors that relate the quantity of gas released to the mass of fuel burned are based on recent field campaigns in Africa and are related to combustion efficiency,which is in turn related to the fuel mix. The emissionsare summed over the 1989 fire season for Africa south of the equator. The estimated emissions from vegetationburning in the subcontinent are 0.5 Tg CH4, 14.9 Tg CO, 1.05 Tg NO, and 1.08 Tg of particles smaller than 2.51xn't The 324 Tg CO: emitted is expected to be reabsorbed in subsequentyears. These estimates are smaller than previous estimates.
10 Introduction Most of Africa has a strongly seasonal climate, with hot, wet summers and warm, dry winters. This kind of climate permits frequent fires, which in turn promote the growth of a fire-tolerant vegetation known as a savanna. Because many savannas are located on extremely infertile soils, frequent application of fire is essential to sustainable low-input pastoralism. Fire has been used for millennia to promote the growth of palatable forage and prevent the encroachment of trees (Goldammer, 1993), and in fact, most fires in Africa are ignited by people rather than by lightning. In the southem hemisphere, savanna fires are concentrated in the dry season, between April and October. Savanna grass is approximately 45% carbon (C) on a dry-mass basis. When it burns, most of this C is released as CO 2. In general, this does not represent a net release of CO 2, however, because the vegetation regrows in subsequent years. On the other hand, it is a net source of a range of other radiatively active and O3-forming trace gases and aerosols (particles), because these are not reabsorbed. If the fire frequency or fuel load are increased systematically, vegetation fires are a net source of CO v as well; conversely, if fires are suppressed, savannas become a major C sink as tree biomass and organic soil matter increase. The degree to which colonial and post-colonial fire regulations have altered the long-term frequency of fire is unknown. In some places fire may now be more frequent than it was previously, whereas in others, the reverse may be true. African savanna fires
Environmental Monitoring and Assessment 38:169-179, 1995. © 1995 Kluwer Academic Publishers. Printed in the Netherlands,
170
R.J. SCHOLES
have a profound effect on the regional and global atmosphere (Crutzen and Andreae, 1990), but there is little data to suggest that the magnitude of the impact has changed in recent times. The emissions of partial combustion products (CO, CH 4, and aerosols) are inversely proportional to the degree of oxygenation during combustion (Ward and Radke, 1993). The fine, dry grass fuels that predominate in savannas are highly oxygenated during combustion and therefore release relatively small amounts of these gases. Emissions of nitrogenous gases (predominantly NO x, but also small amounts of nitrous oxide (N20) and several other gases) depend on the N content of the fuel (Lobert et al., 1991), which is typically low (0.5-1%). Because of the vast area of Africa that burns, the gross trace gas emissions are globally significant (Crutzen and Andreae, 1990; Levine, 1990; Lacaux, Cachier, and Delmas, 1993) and could dominate the greenhouse gas inventories of large but lightly industrialized African countries, such as Angola, Tanzania, and Zambia. The general procedure for estimating emissions from biomass burning (which should more properly be called vegetation burning, because biomass refers only to the living material, and fires mostly consume dead material) is to divide the study region into vegetation types and calculate the emission per type (OECD, 1991; UNEP et al., 1995). In each type, the area (A:m 2) mean fuel load (L:g m-2), combustion completeness (the fraction of the biomass exposed to the fire that actually burns--C:nondimensional fraction), and average time between fires (R:years) is estimated, usually based on expert opinion, because few detailed measurements exist. The amount of biomass consumed per annum (B: g yT-1) in each vegetation class is then calculated from B = A R -1 L C. The emission factors (F:g emission product/g fuel) are typically estimated by measuring the increase in a given gas relative to the increase in CO 2 in a smoke plume. This measurement is then converted to a dry-mass basis by assuming a C content in the fuel (41-45%) and an efficiency of conversion of fuel C to CO 2 (80-98%). The emission of combustion products is then given for each gas by E a~ = B Fg~s, which is summed over the various vegetation classes. The method proposed by the Intergovernmental Panel on Climate Change for calculating emissions from savanna burning follows this simple approach (UNEP et al., 1995). The disadvantage of the method is that the range of possible values for R, L, C, and F is very large, and the procedures for validating them in the field are demanding. This means that the accuracy of the estimate is generally very low. This paper briefly describes a method (documented in more detail in Scholes, Kendall, and Justice (1995) and Scholes, Ward, and Justice (1995) that greatly reduces the uncertainties by applying the following strategies: 1. The area burned is measured using a combination of low- and high-resolution satellite images. 2. The fuel load is modeled in four categories, constrained by climate. 3. The emission factors are related to the fuel mixture. The method has been applied to Africa south of the equator. It could probably also be applied to West Africa and other tropical areas that are subject to frequent and extensive fires, but the predictive equations would have to be recalibrated for areas outside of southern Africa.
64
GREENHOUSE GAS EMISSIONS FROM VEGETATION FIRES IN SOUTHERN AFRICA
171
The objectives of this paper are to describe a procedure for estimating regional trace gas emissions from vegetation fires and to provide data that will assist in the preparation of national greenhouse gas emission inventories in southern Africa.
2. M e t h o d s The best available vegetation maps (Table I) of southern Africa at an appropriate scale (1:1 M to 1:5 M) were digitized and edge-matched using geographic information system (GIS) technology to form a detailed, composite vegetation map of the subcontinent. The mapped vegetation classes were reclassified into eleven broad types (Table II), each of which has distinctive fuel types and relationships between fuel production and climate. The maps were converted to raster form at a scale matching the available long-term monthly
TABLE I The best availablebroad-scale vegetation maps for Africa south of the equator. Where no suitable map is available,the all-Africamap by White (1983) can be used
Country
Africa South Africa, Lesotho, and Swaziland Botswana
Namibia Zimbabwe and Mozambique Angola Zambia Malawi Tanzania Zaire
Kenya Uganda
Source
White, F: 1984, Vegetationof Africa, Unesco/AETFAT,Pads. Acocks, J.P.H.: 1955, Veld Types of South Africa. Memoirs of the Botanical Survey of South Africa 28 (revised 1975 as Memoir 40). Ministry of Agriculture, Botswana: 1991, Vegetationmap of the Republic of Botswana, Land UtilisationDivision, Private Bag 003, Gaborone. Giess, W.: 1971, VegetationMap of South West Africa. Dinteria: 4:5-114 Granvaux-Barbosa,L.A.: 1967, Vegetationmap of the flora Zambeziaca area, Collins, Harare. Granvaux-Barbosa, L.A.: 1970, Fitograficade Angola, histituto de InvesticaoCientificade Angola. Edwards, A.C.R.: 1976, Republic of Zambia Vegetation Map, Institut fur AngewandteGeodesia, Frankfort. 1980, National Atlas of Malawi, Department of Survey, Blantyre. Ministry of Lands: 1984, TanzaniaVegetationCover Types, Forestry Division, Ministry of Lands, Governmentof Tanzania. 1932, Atlas of Southern Katanga, VegetationMap. NASA-GIMMS: In press, PathfinderProject on Central African Forests. Ministry of Wildlife: 1986, Biotic Communitiesof Kenya. 1964 (reissued 1972), Vegetationof Uganda, Department of Land Surveys, Entebbe.
65
Scale
1:5 000 000
1:1 500 000
1:2 000 000 1:3 000 000
1:2 500 000 1:2 500 000 1:1 500 000 1:1 000 000
1:2 000 000 1:500 000 1:250 000 1:1 000 000
1:500 000
172
R.J. SCHOLES
0
0
d
d
0
~
d
0
d
0
~
0
d
0
d
0
0
d
d
~
d
×
deo~de~edd~
,
