全球变化知识资源中心

Over half of all wood harvested worldwide is used as fuel, supplying ~9% of global primary energy. By depleting stocks of woody biomass, unsustainable harvesting can contribute to forest degradation, deforestation and climate change. However, past efforts to quantify woodfuel sustainability failed to provide credible results. We present a spatially explicit assessment of pan-tropical woodfuel supply and demand, calculate the degree to which woodfuel demand exceeds regrowth, and estimate woodfuel-related greenhouse-gas emissions for the year 2009. We estimate 27–34% of woodfuel harvested was unsustainable, with large geographic variations. Our estimates are lower than estimates from carbon offset projects, which are probably overstating the climate benefits of improved stoves. Approximately 275 million people live in woodfuel depletion ‘hotspots’—concentrated in South Asia and East Africa—where most demand is unsustainable. Emissions from woodfuels are 1.0–1.2 Gt CO₂e yr⁻¹ (1.9–2.3% of global emissions). Successful deployment and utilization of 100 million improved stoves could reduce this by 11–17%. At US$11 per tCO₂e, these reductions would be worth over US$1 billion yr⁻¹ in avoided greenhouse-gas emissions if black carbon were integrated into carbon markets. By identifying potential areas of woodfuel-driven degradation or deforestation, we inform the ongoing discussion about REDD-based approaches to climate change mitigation.

Traditional woodfuels, which include both firewood and charcoal used for cooking and heating, represent approximately 55% of global wood harvest and 9% of primary energy supply^{1, 2}. The current extent and future evolution of traditional woodfuel consumption is closely related to several key challenges to sustainable development. Roughly 2.8 billion people worldwide³, including the world’s poorest and most marginalized, burn wood to satisfy their basic energy needs. Woodfuels can impact public health⁴, cause deforestation or forest degradation⁵, and contribute to climate change^{6, 7, 8}. Climate impacts arise from two pollutant flows: CO₂ is emitted because a fraction of woodfuel is harvested unsustainably; methane (CH₄), black carbon and other short-lived climate forcers (SLCFs) are emitted because of incomplete combustion, which also emits health-damaging pollutants. Thus, woodfuels present society with two important links between local and global impacts; incomplete combustion releases pollutants that damage health and warm the atmosphere, and unsustainable harvesting drives both forest degradation and climate change.

Risks to public health are increasingly well characterized⁴, whereas impacts on deforestation, degradation and global climate remain highly uncertain. Historically, woodfuel demand was considered a major driver of land cover change^{9, 10} (LCC). However, early research failed to account for regrowth, consumers’ response to scarcity, and use of trees outside forests^{11, 12}. More recent local or regional assessments find conflicting results^{13, 14, 15, 16, 17}, suggesting that geography is an important determinant of woodfuel sustainability. However, few systematic studies of woodfuel sustainability and greenhouse gas (GHG) emissions have been conducted¹⁸. The Intergovernmental Panel on Climate Change’s Fourth Assessment claimed that 10% of global woodfuel is harvested unsustainably^{19, 20}, and the Fifth Assessment stresses that net emissions from woodfuels are unknown¹⁷. Better understanding of the contribution of woodfuels to deforestation, forest degradation and climate change is needed to evaluate the impact of the growing wave of household energy interventions and inform emerging REDD (Reducing Emissions from Deforestation and Forest Degradation) methodologies^{21, 22}.

Here we present a spatially explicit snapshot of woodfuel supply and demand (Supplementary Section 1) throughout tropical regions where traditional woodfuel consumption is concentrated. Using 2009 as a base year, we quantify the extent to which woodfuel demand exceeds supply, identify specific ‘hotspots’ where harvesting rates are likely to cause degradation or deforestation, quantify the carbon emissions that result from current woodfuel exploitation, and estimate the emission reductions that could be achieved from large-scale interventions²³.

Nearly all landscapes produce a measurable increment of woody biomass either as new growth or as regrowth from previous disturbances. This assessment considers supply/demand balance over one year. If an area is harvested for woodfuel below the annual growth rate, then woody biomass stocks are not depleted and harvesting is sustainable. However, if annual harvesting exceeds incremental growth, it is unsustainable, leading to a decline of woody biomass, forest degradation and net carbon emissions. In this assessment, we define the wood harvested in excess of the incremental growth rate as non-renewable biomass²⁴ (NRB).

We treat woodfuel demand as an exogenous factor derived from a mix of national and sub-national studies supplemented by data from the Food and Agriculture Organization (FAO), International Energy Agency (IEA), and United Nations^{1, 25, 26} (UN). Woodfuel demand has subsistence and commercial components. Subsistence demand occurs primarily in rural areas, where people collect their own fuel using simple non-motorized forms of transportation from within a few hours of their homes. Commercial demand originates in urban and some densely populated rural locations and is typically supplied by motorized transport over much longer distances.

We develop a map of supply–demand balance by estimating harvesting pressure, first from subsistence and then commercial harvesters (Fig. 1a, b). Areas exploited to satisfy commercial demand form a ‘woodshed’, which represents the region that would satisfy demand if the full mean annual increment (MAI) is used²⁷ (Fig. 1c shows commercial woodsheds for a high-demand area of East Africa; Supplementary Fig. 5 shows the entire pan-tropics).

a, Pixel-level supply-demand balance. b, Local-level balance. c, Commercial balance. odt; oven-dry tonnes of woody biomass.

Many woodfuel-dependent regions are characterized by high rates of deforestation. Others, particularly parts of China and India, have experienced recent afforestation. Although not directly linked to woodfuel demand, these processes, which we define collectively as LCC, impact woodfuel supplies. Deforestation creates large volumes of non-renewable woodfuel^{28, 29}, and afforestation augments renewable woodfuel supplies by adding to the growing stock of ‘dendro-energy biomass’ (DEB). Neither process has been explicitly accounted for in previous woodfuel assessments. When deforestation occurs in regions accessible to woodfuel users, the cleared woody biomass may be used as timber and woodfuel. Similarly, afforestation adds DEB equivalent to the MAI of the surrounding land class. However, the degree to which LCC by-products are actually used as woodfuel is unknown. To accommodate this uncertainty, we explore two scenarios, described in Table 1. In Scenario A, we assume LCC by-products are not used. In Scenario B, we assume they are used, yielding two NRB components (NRB_B1 and NRB_B2): NRB_B1 indicates the use of LCC by-products; NRB_B2 indicates the wood harvested in excess of MAI to satisfy the demand that remains after accounting for the use of those by-products. In populated regions experiencing high rates of deforestation, large volumes of DEB are accessible, and NRB_B2 may be zero (Supplementary Section 5).

Full table

Woodfuel demand in 2009 was ~1.36 Gt. If by-products of LCC were not used (Scenario A), pan-tropical expected fNRB_A was 27–30% (367–413 Mt). If by-products of LCC were used (Scenario B), we estimate they contributed 8.3% (113 Mt) of pan-tropical woodfuel supply (fNRB_B1). We also find that 22–25% (296–340 Mt) of the remaining demand was harvested unsustainably (fNRB_B2). Adding fNRB_B1 and fNRB_B2, the total fraction of NRB using LCC by-products is 30–34%. The uncertainty results from uncertain productivity and contribution of plantations (Supplementary Section 6). This is largest in Asia, where forest plantations may be a substantial source of supply, and smallest in sub-Saharan Africa, which has few plantations³⁰. Figure 2 shows a global map of fNRB_B2 (maps of fNRB_A and fNRB_B1+B2 are shown in Supplementary Fig. 7).

Shading indicates the percentage fNRB estimated in sub-national units resulting from direct woodfuel harvesting (Scenario B2). The rectangle shows the region illustrated in Fig. 1.

Climate impacts arise from emissions of well-mixed GHGs, which include CO₂ and CH₄, and SLCFs, which include black and organic carbon aerosols, CO and volatile organic compounds (VOCs). Emissions of well-mixed GHGs and SLCFs as a result of unsustainable harvesting and incomplete combustion from traditional woodfuels (Methods) were 1.0–1.2 Gt of CO₂ equivalent (CO₂e) in 2009: 1.9–2.3% of global emissions and 3.5–4.3% of emissions in the pan-tropical region³². National emissions vary widely (Supplementary Table 2). India and China have the largest populations of traditional woodfuel users and highest overall emissions, but relatively low per capita emissions. In contrast, Kenya, Ethiopia and Uganda, which constitute part of the East African hotspot, rank among the highest emitters in absolute and per capita terms.

There is geographic variation in the mix of pollutants emitted by traditional woodfuels because of variations in fNRB and in the extent of charcoal use, which has different emission characteristics from fuelwood (Methods). Globally, after accounting for uptake by the fraction of woody biomass that is sustainably harvested, CO₂ contributes 34–45% of total climate forcing. Black carbon has a similar impact, contributing 35–42%, and CH₄, CO and VOCs account for the remaining 31–37%. This variation has policy implications; at present, carbon markets value reductions of CO₂, CH₄ and N₂O, but do not value black carbon abatement, which favours interventions in regions with high fNRB.

Interventions in household energy have been implemented for decades with multiple objectives³³: including forest conservation; health improvements; and climate change mitigation, as well as poverty alleviation and economic development. The Global Alliance for Clean Cookstoves (GACC), the largest stove programme so far, proposes to deploy 100 million improved stoves by 2020 (ref. 23). With large spatial variation in fNRB, impacts of interventions vary with geographic patterns of stove uptake. We examine this variation with four intervention scenarios (Methods and Supplementary Section 7).

We optimistically assume that 100 million state-of-the-art improved cookstoves are successfully disseminated according to different scenarios. Resulting emission reductions range from 98–161 MtCO₂e yr⁻¹. The largest reductions result from targeting the highest per capita woodfuel consumers. This is followed by reductions achieved by targeting consumers in regions with the highest rates of NRB, although uncertainties in emission reductions from individual stoves make the difference insignificant. The smallest reductions result from dissemination in the most business-friendly countries. The emission reductions achieved by prioritizing health improvements fall between these extremes (Fig. 4).

Bars indicate GHG emissions/uptake, data points show net emissions, error bars indicate standard deviations, and numbers indicate annual reductions achieved by shifting from baseline to intervention.

One limitation of the study is a lack of reliable woodfuel consumption data. When possible, we used national and sub-national data sets. However, for most countries, we relied on data compiled by international organizations containing unknown uncertainties that make it difficult to communicate the uncertainty in these results. A second limitation is that the analysis considers a single year and does not account for potential behavioural changes among woodfuel users in response to scarcity. Potential responses include decreasing consumption, switching to non-woody fuels, or taking measures to increase woody biomass supply. Such responses are site-specific and difficult to model globally, but they could be incorporated in national and sub-national dynamic models.

Using the best available data, we estimate that unsustainable harvesting and incomplete combustion contributed 1.9–2.3% of global emissions of well-mixed GHGs and SLCFs in 2009. Globally, emissions were split evenly between CO₂, black carbon and other SLCFs. In 12 nations, emissions from woodfuels were 50% or more of the country’s total emissions, demonstrating the dominant role that traditional woodfuels have in places with few industrial emissions (Supplementary Table 2).

Our estimates of fNRB are considerably lower than estimates used by woodfuel projects in the carbon market. Project revenues depend directly on fNRB. A review of 305 carbon projects in 45 countries reveals a median fNRB of 90% with minimal regional variation (Supplementary Section 6). We identified only four countries in which sub-national fNRB exceeds 80% as a result of woodfuel demand. Just 8% of existing projects fall within these areas. Thus, project developers are very likely overstating the emission reduction potential of improved stoves.

Household energy forms a major component of the United Nations’ promotion of ‘Sustainable Energy for All’³⁴. However, high upfront costs are a barrier to implementing sustainable solutions. Despite finding lower fRNB values than market actors assume, successfully disseminating 100 million state-of-the-art cookstoves would reduce traditional woodfuel emissions by 98–161 MtCO₂e yr⁻¹. At US$11 per tCO₂e, the average price of offsets from stove projects in 2012 (ref. 35), these reductions would be valued at US$1.1–1.8 billion if black carbon can be integrated into carbon markets. This far exceeds current investments in household energy in the Global South, which do not garner the same level of finance as other major health impacts such as malaria, tuberculosis and HIV. In addition, we find that policy objectives are important determinants of emission reductions, introducing variation of 60%. Countries with high per capita woodfuel use or high NRB rates yield the largest emissions reductions. However, neither group overlaps completely with countries experiencing the highest disease burden from woodsmoke exposure (Fig. 5). Thus, improved stove dissemination among populations suffering from the largest disease burden results in fewer emission reductions than dissemination in regions with high rates of woodfuel consumption or unsustainable harvesting. However, we identified a small group of countries that rank poorly in all categories (red text in Fig. 5). Others rank poorly in two out of three categories (blue text in Fig. 5). These countries deserve clear prioritization. The sub-national data set generated by this research can be used to more accurately identify high-priority areas and pinpoint locations where interventions would have the greatest impact. Moreover, by identifying areas where woodfuel-driven degradation or deforestation is likely to occur, our assessment fills a critical gap in knowledge about the extent to which woodfuel demand may contribute deforestation or forest degradation and informs emerging REDD-based approaches to climate change mitigation.

We use the WISDOM model³⁶ (Supplementary Section 1) to characterize sustainability and net carbon emissions of traditional woodfuels in 90 developing countries located primarily in tropical regions, using 2009 as a base year. Woodfuel demand was derived from national and sub-national studies (Supplementary Section 1) supplemented by data from the FAO, IEA and UN (refs 1, 25, 26). From these data, we mapped subsistence and commercial components of traditional woodfuel demand. Subsistence demand occurs in rural areas, where people use woodfuels they collect themselves or purchase locally. This wood is harvested within a few hours’ walking distance. Commercial demand originates in urban and some densely populated rural locations and is carried using motorized transport over longer distances (Supplementary Section 1).

Woodfuel supply is defined by the productivity of woody biomass, which we model as a function of above-ground biomass (AGB) stock. We use recent maps of land cover and ecological zones^{37, 38} to define a broad system of land units, including cropland and crop mosaic (often neglected in assessments of woodfuel supply). Each land unit is assigned an AGB stock using three types of source: AGB distribution maps; geo-referenced field plots; and forest inventories from known locations for specific forest types (Supplementary Section 1). AGB distribution was derived from two recent data sets^39,