Estimating carbon footprint from biofuel production from oil palm: methodology and results from 2 pilot areas in Indonesia Meine Van Noordwijk*, Sonya Dewi*, Ni’matul Khasanah*, Andree Ekadinata*, Subekti Rahayu*, Jean Pierre Caliman**, Mukesh Sharma†, Rosediana Suharto‡ *World Agroforestry Centre, ICRAF, Southeast Asia ** SMARTRI †AsianAgri ‡Indonesia Palm Oil Commission ABSTRACT In the last five years Indonesian palm oil production grew by 13.4% per year, with growth in export at 16.2% per year and slow growth in domestic consumption. Oil palm production in Indonesia and Malaysia is now in the focus of the debates on Biofuel and Carbondioxide (CO2) and other greenhouse gas (GHG) emissions, through its association in the public debate with deforestation and (over)use of peatland. The potential use of palm oil as biodiesel to reduce dependency on, and emissions from, the use of fossil fuel has focused debate on the emissions caused by the conversion of land to oil palm and subsequent steps in the production. Carbondioxide (CO2) and other greenhouse gas emissions due to the production of palm oil can be attributed to three phases of the production process: A. the initial conversion of preceding vegetation into a palm oil plantation, usually based on 'land clearing', leading to a 'C debt' B. the balance of emission and absorption during the growth cycle of the oil palms, depending on growth rate, green manure and organic waste management and fertilizer practices, leading to a time‐averaged C‐stock that influences 'C debt' and repay time, C. transport to the refinery followed by CPO (Crude Palm Oil) and kernel production, transesterification into biofuel and further transport to the end users. A comprehensive accounting system on carbon and other GHG emissions of biofuel production of oil palm has to include the whole life cycle. We develop a tool called Biofuel Emission Reduction Estimator Scheme (BERES) to calculate emission reduction factors, using a life cycle approach. The number of years for payback time from carbon debt, the emissions saving from fossil fuel substitution and fossil fuel substitution efficiency can then be calculated from the balance between total sequestration and emissions. Under a pilot study in 2 areas in Indonesia we conducted a subset of BERES, which embraced components A and B within the overall life cycle of biofuel production from oil palm. We estimated carbon emissions using different definition of organizational and operational boundaries. In Indonesia, the schemes of oil palm production are varied from purely large‐scale company operations, cooperation between large‐scale company (nucleus) and smallholders (plasma), and independent smallholders. The net result is very sensitive to the preceding vegetation. A minimum emission reduction efficiency of 35% can only be reached in a second production cycle of oil palm, or when oil palm replaced vegetation of less than 40 t C/ha. Investment in CH4 capture at the mill can improve the situation. Where peat soils are used, the effects of drainage on emissions usually means the target efficiency cannot be met. A third factor with considerable influence is the use of N fertilizer in relation to yield. Increase in N use efficiency can lower costs as well as help reaching the fossil fuel substitution efficiency. 1. Introduction Biofuels appeared to be such a nice way of reducing the climate change challenge: it reduces political dependence on fossil fuel supply, can be done with minimal change to existing engines and modes of transport, and provides new sources of income for rural economies. The potential use of palm oil as biodiesel to reduce dependency on and emissions from the use of fossil fuel (so far involving small volumes only) has focused debate on the emissions caused by the conversion of land to oil palm and subsequent steps in the production. European countries have been leading in the development of environmental standards for biofuel and developed a ‘typical’ emission scenario for palm oil production. Oil palm production in Indonesia and Malaysia has become focus of debate on greenhouse gas (GHG) emissions, through its association in the public debate with deforestation and (over)use of peatland. In the last five years Indonesian palm oil production grew by 13.41%/year, with growth in export at 16.24%/year and slow growth in domestic consumption. The main use of palm oil in Indonesia is in the food industry, with limited other uses (including biodiesel) due to the increasing price of feed stock (CPO, olein). The dependency on export is very high; Indonesia biggest buyers are India (28.5%), China (25.4%) and Netherlands (12.3%); Pakistan is becoming an important market partner. Production situations of oil palm in Indonesia are diverse and include peat versus mineral soils as main ecological contrast, large‐scale plantations versus smallholders as socio‐economic contrast and plantations developed from forest (after logging) versus plantations developed on lands with low current C‐stocks due to previous degradation. The sector (or at least the more progressive parts of it) is now willing to engage in debate on environmental impacts and GHG emissions, but needs support in understanding carbon (and other GHG) accounting procedures and in capacity development for data collection and analysis. Data for currently feasible ‘best practice’ will play a key role in the self‐ adjustment of the industry (closing the gap between average and best practice) and in efforts to maintain market share. Early signs of a shift in consumer concerns, especially in Europe, from a focus on the quality of palm oil to sustainable production started some 5 years ago. Current demand is to secure good environmental practices, at the same time giving more attention to social and economic viability and maintaining high product quality and safety. In response to these concerns, a new type of public‐private‐NGO partnership has emerged in the form of the Roundtable on Sustainable Palm Oil (RSPO). By developing voluntary standards, the RSPO has contributed to improvement of the industry as well as its public image. From the perspective of the industry, however, implementation of RSPO Sustainable Certification System means compliance with more than 120 indicators (major and minor) and burdensome. No price premium has so far emerged for RSPO compliant palm oil. In this very confused stage the Indonesian Palm Oil Board (IPOB), in line with its duty to assist the Indonesian producers, decided to focus on understanding the system and to implement it according to the EU buyer’s demand. During the preparation for RSPO Certification another regulation emerged in the form of the EU Directive of the European Parliament and Council on the promotion of the use of energy from renewable resources. The objectives of this Directive differs from RSPO, and is focused on mitigation of climate change through the reduction of greenhouse gases emissions, sustainable development, secure supplies of energy, and the development of knowledge based industry creating jobs, economic growth, competitiveness and regional and rural development. The new EU directive means that planting in high biodiversity land which are untouched forest, protected forest and high biodiversity grass land are not allowed and planting is prohibited in high C‐ stock areas such as peatland, wetland and continuously forested area of more than one ha; forest is defined on the basis of a tree height of at least 5 meters and a canopy cover of more than 30%. Another important article in the directives is that GHG emission saving from the use of biofuels and bioliquids shall be at least 35%, restricting life‐cycle emisisons attributed to each unit of biofuel to less than 65% of the fossil fuel saved in its use. This rule requires data on life‐cycle emissions for each type of product. The origin of the data presented in this proposed directive for palm oil are not very clear and the scientific base is not convincing. Overall, palm oil is considered to be (just) short of the target. The new EU guidelines on biofuel are in reach for current good practice on mineral soils, not for oil palm on peat soils. Differentiation of product by geographic or ecological origin requires detailed data obtained with standardized methods and open to public scrutiny. Key decisions on reference dates in the international climate policies (1990 for reforestation, possibly 2000‐2005 for REDD, 2008 for EU footprint accounting of biofuel) are still under discussion, as is the scope of the methods (deforestation/forest degradation or full C (+ equivalent GHG’s) accounting; inclusion of peat drainage issues); a comprehensive approach to data collection is needed, including timelines of C‐stock loss ahead of oil palm plantations. Life‐cycle analysis involves three main phases: plot history (main issue: loss of C stocks), oil palm production stage (main issue: N2O emissions on mineral soils and CO2 on drained peat soils) and milling/processing (main issue: CH4 emissions) + transport (fossil fuel use). Previous research and ongoing agronomic efforts of the industry focus on production stage, while the engineering aspects of mills for CH4 capture have received attention for technical CDM projects; the plot history part is the biggest unknown and requires focus of new efforts. The “avoided Deforestation with Sustainable Benefits” study of ICRAF + ASB Partnership for the Tropical Forest Margins in 2007 has provided analysis for 3 provinces of the 1990 – 2005 change in land cover, loss of C stock and economic gains; a more detailed analysis of the main oil palm producing areas can follow along the same lines of analysis. While the industry can and will do the ‘lobby’ work, solid data collection and scientific independence is needed in the interest of all. After many communications and discussions with the EU partners, the Indonesian industry would like to give positive input to the EU by giving real data as result from a study. Preparations have started for “Carbon emissions of biofuel production from oil palm in Indonesia: A pilot study”. A full scope study will be based on methodology and lessons learnt from this particular pilot study. The industry is keen to cooperate on measurements in ‘current best practice’, recognizing that in reality much production is not yet reaching this management level; data to estimate the performance gap against such standards will be a second step, and may best be done by self‐appraisal through capacity building. Our approach allows for such stepwise improvement of practice, with technical standards that are considered to be technically realistic. The overall outline of the study (Figure 1) is designed to reach the following objectives are: 1. To estimate historical and ongoing carbon emission from land conversion to oilpalm plantation; 2. To estimate current carbon emission from oilpalm plantation establishment, management and processing to biofuel; 2. Integrated Assessment Scheme for Carbon and GHG Emissions Carbondioxide (CO2) and other greenhouse gas emissions due to the production of palm oil can be attributed to three phases of the production process (see Box I): A. the initial conversion of preceding vegetation into a palm oil plantation, usually based on ‘land clearing’, leading to a ‘C debt’ B. the balance of emission and absorption during the growth cycle of the oil palms, depending on growth rate, green manure and organic waste management and fertilizer practices, leading to a time‐averaged C‐stock that influences ‘C debt’ and repay time, C. transport to the refinery followed by processing and further transport to the end users; CPO and kernel production, transesterification into biofuel, and transportation. A comprehensive accounting system on carbon and other GHG emissions of biofuel production of oil palm has to include the whole life cycle assessment (LCA) through a life cycle inventory (LCI) ( ISO, 1997). In the pilot phase for a broader sampling of Indonesian palm oil sector, two plantations on mineral soils were analyzed, in order to test the ‘lifecycle’ approach and provide initial estimates of what is feasible under ‘good agricultural practice’ in well‐managed plantation conditions. Carbondioxide (CO2) and other greenhouse gas emissions due to the production of biofuel can be attributed to three phases of the production process: A. the initial conversion of preceding vegetation into a biofuel feedstock plantation, usually based on ‘land clearing’, B. the balance of emission and absorption during the growth cycle of the plants, depending on growth rate, green manure and organic waste management and fertilizer practices, and C. transport to the refinery followed by processing and further transport to the end users. Emission estimates require data on all three steps: E = total emissions from the use of the fuel = el + eec + ep + etd + eu – esca– eccs– eccr– eee el =annualised emissions from carbon stock changes caused by land-use change; eec =emissions from the extraction or cultivation of raw materials; esca =emission saving from soil carbon accumulation via improved agricultural management; ep =emissions from processing; etd =emissions from transport and distribution; eu =emissions from the fuel in use; eccs =emission saving from carbon capture and geological storage; eccr =emission saving from carbon capture and replacement; and eee =emission saving from excess electricity from cogeneration. Box I. Three phases of production process and their emission components The total emission of palm oil (A + B + C) per unit palm oil produced has become a focus of debate, especially where palm oil is used as ‘biodiesel’. Substituting for fossil fuel diesel saves CO2 emissions, but the total emissions per unit production need to be subtracted from these savings before a net emission reduction (or enhancement) can be calculated. The net emissions due to steps A and B depend on the land use history of the plantations as well as its soil and soil management. Step C depends on technical specification of the mill and the transport from the mill to the nearest port. Five nested scales of assessment can be distinguished as follows: 1) Life‐cycle approach to production of fresh fruit bunches (FFB) at plot level, with existing return flows of organic waste (or ‘by‐products’) in the from of empty fruit bunches (EFB) and/or palm oil mill effluent (POME); policy‐wise a distinction needs to be made between the first (I) and subsequent (II) production cycle, as they differ in preceding C stock of the land. Major differentiation occurs by i) preceding vegetation and its C stock (with specification of ‘attribution’ of the resultant C flows between resource use for logging and land‐clearing), ii) soil types and associated CO2 emissions, with primary distinction between mineral soil and peat, and further distinctions by soil texture, pH, elevation and C/Cref on mineral soils, and peat depth and drainage regime on the peat soils, iii) management regime, including level of fertilization, yield levels and organic recycling iv) producer: nucleus plantation under company management, plasma plantation (planted by company, managed by smallholders) and independent smallholder producers (with wider variation in management intensity) 2) Mill and its sources of fresh fruit bunches, generally consisting of three fractions: nucleus (FFFBNu), plasma (FFFBPl)and independent smallholders (FFFBInd), with FFFBNu + FFFBPl + FFFBInd = 1 3) All mills in a production area that contribute to a recognizable trade‐flow (e.g. holding company, export trademark), with its characteristic pre‐oil palm vegetation, ratio of mineral‐to‐peat soils, first or subsequent production cycles and producers (FFFBNu, FFFBPl, FFFBInd). 4) All mills in Indonesia as CPO source area, with its characteristic pre‐oil palm vegetation, ratio of mineral‐to‐peat soils, first or subsequent production cycles and producers (FFFBNu, FFFBPl, FFFBInd). 5) All CPO exporting countries, with its characteristic pre‐oil palm vegetation, ratio of mineral‐to‐peat soils, first or subsequent production cycles and producers (FFFBNu, FFFBPl, FFFBInd). We started at scale 1, a life‐cycle analysis of a plantation, with a focus on the net annual emissions per unit area or per unit product. Existing EU‐policies relate to scale 5 and its emission profile per unit biodiesel derived from CPO, with optional differentiation at scales 4 and 3. For scale 5, a weighted average is needed of all current production conditions in Indonesia. It may represent the global impact of oil palm production in Indonesia, but due to its aggregegation provides little incentive to improve production practices. A focus on individual mills (scale 3) will provide incentives for improvement of exporters who want to meet the standards, but may have little overall impact, given the large current market share of exports to countries where no questions are asked. Overall calculations show the importance of the various parameters and the relative sensitivity of the end result to uncertainty in each of the contributing parameters. Current ‘default’ value refers to knowledge at inception stage and will be modified by the full‐scale assessment (Table 1). Calculations of the area needed to make a dent into current fossil fuel use quickly showed that it cannot be a substantial contribution to energy issues without requiring large areas and interfering with markets for food crops. If biofuel production extends beyond current agriculture, it will often increase emissions of carbondioxide. The net effect will be often a lower estimate of emission reduction than expected, but if high C‐stock land is cleared, biofuel use can also increase net emissions. The debate on such emission enhancement has focussed on oil palm in the humid tropics of SE Asia, where forest and peatland conversion currently lead to large emissions – with or without a specific role for oil palm expansion. The public debate, however, has linked the two issues. The EU provided guidance to countries on minimum standards that should be used when biofuels are included in national renewable energy plans. Until 2017, a minimum emission reduction of 35% has to be achieved for any fuel included in the scheme, shifting to 50% by 2017 and 60% beyond. Default estimates are given for major current or potential sources of biofuel. A procedure was established to calculate emission reduction factors, using a lifecycle approach. Specific market flows of biofuels can apply for exception from the 'default' for the commodity. These procedures create the need for exporting countries and entities to understand the steps in calculation and to do the research needed to get reliable data. The net result is very sensitive to the preceding vegetation. For the oil palm example, a minimum emission reduction efficiency of 35% can only be reached in a 2nd production cycle, or when oil palm replaced vegetation of less than 40 t C/ha. Investment in CH capture at the mill can improve the situation. Where peat soils are used, the effects of drainage on emissions usually means the target efficiency cannot be met. A third factor with considerable influence is the use of N fertilizer in relation to yield. Increase in N use efficiency can lower costs as well as help reaching the fossil fuel substitution efficiency. Table 1. Defaut value of parameters of integrated assessment scheme for carbon and GHG emissions Phase Symbol Parameter Default Based on: A Tcycle Accounting period for plantation (yr) 25 Policy decision, ideally linked to typical production cycle A Cbefore Attributable time‐averaged C stock before the 60 Pre‐condition (can range from 250 to 0 t C/ha plantation crop was planted [t C/ha] B Coilpalm Time‐averaged C stock of the plantation crop [tC/ha] 40 Pilot‐phase field assessments; value depending on management style B FN2O N‐fraction of fertilizer‐N lost as N2O 0.04 Literature value to be updated by actual emission studies; 0.01 is current IPCC default, 0.04 is based on Crutzen et al. (2008) B EPeatPerDrainDepth Peatland CO2 loss per cm drain depth, Mg CO2/(ha.yr) 0.8 Literature value to be updated by new findings B ESoil Mineral soil CO2 loss (depending on EFB and POME 0 Assumption linked to soil C studies recycling to maintain Corg levels) B SPeat? Peatland 0 Pre‐condition B SPeatDrainDepth Peatland drain depth, cm 50 Management choice C Nfert N fertilizer use (kg N/ha, averaged over lifecycle) 150 Management choice C YFFB FFB yield Mg per ha/yr (averaged over life cycle) 21.1 Depending on management style C YOER CPO extraction rate (OER), % CPO per FFB 20.5 Technical coefficient C YpKER Kernel extraction rate (pKER), % Ker per FFB 5.2 Technical coefficient C YPKO PKO kernel oil per kernel extracted 0.5 Technical coefficient C CCPO C concentration of CPO 0.6 Technical coefficient C CKER C concentration of Kernel oil 0.6 technical coefficient C FCH4Mill Mill emissions of CH4 expressed as CO2eq/t C extracted 0.6 Mill dependent C FTransportEmissions CO2eq emissions processing and transport , t CO2eq/t C 0.2 Depends on distance to port C FBiodieselConversion Biodiesel production per t CPO 0.88 Technical coefficient C FFossilFuelEquivalence Biodiesel / fossil fuel diesel equivalence ratio 1 Technical coefficient Constants CO2/C CO2/C = 44/12 3.67 N2O/N N2O/N = 44/28 1.57 GWPN2O GWP of N2O relative to CO2 296 Table 2. Equations and results for the default Symbol Explanation Equation Default YOilC Annual oil harvest, t C/(ha.year) : YOilC = YFFB * (YOER * CCPO + YpKER * YPKO * CKER ) / 100 2.93 ETotPerHa Annual CO2e emissions due to ETotPerHa = Nfert * FN2O * N2O/N * GWPN2O /1000+ (1 ‐ SPeat?) * ESoil + SPeat? * 2.79 production [t CO2eq/(ha.year)] EPeatPerDrainDepth * SPeatDrainDepth [t CO2e ha‐1 yr‐1] Annualized Cdebt Annualized Cdebt = (CO2/C) * (Cbefore ‐ Coilpalm)/ Tcycle Fossil fuel emission substitution per YFFB * (YOER * CCPO + YpKER * YPKO * CKER ) * (1 ‐ FCH4Mill ‐ FTransportEmissions) * ha through Phase C: FBiodieselConversion * FFossilFuelEquivalence TPayback Payback time [years] TPayback = max(0, (Cbefore ‐ Coilpalm) * (CO2/C) / (YOilC * (CO2/C ‐ FCH4Mill ‐ 13.1 FTransportEmissions) ‐ ETotPerHa )) CNetSeq_I Net C sequestration during first CNetSeq_I = (CO2/C) * (Cbefore ‐ Coilpalm)/ Tcycle + YOilC *( CO2/C ‐ FCH4Mill ‐ 2.67 production cycle (I): FTransportEmissions ) ‐ ETotPerHa tCO2eq/(ha.year) CNetSeq_II Net C sequestration during second CNetSeq_II = YOilC *( CO2/C ‐ FCH4Mill ‐ FTransportEmissions ) ‐ ETotPerHa 5.60 production cycle (II): tCO2eq/(ha.year) ESubstFossiFuel_I Fossil fuel emissions substituted by ESubstFossiFuel_I = CNetSeq_I * CCPO / (FBiodieselConversion * FFossilFuelEquivalence * YOilC ) 0.62 biodiesel, 1st production cycle (I): tCO2eq /t biodiesel ESubstFossiFuel_II Fossil fuel emissions substituted by ESubstFossiFuel_II = CNetSeq_II * CCPO / (FBiodieselConversion * FFossilFuelEquivalence * YOilC ) 1.31 biodiesel, 2nd (or subsequent) production cycle (II): tCO2eq /t ESubstEff_I Fossil fuel substitution efficiency ESubstEff_I = ESubstFossiFuel_I / (CO2/C) 0.17 CO2eq/CO2eq, 1st cycle (I) ESubstEff_II Fossil fuel substitution efficiency ESubstEff_I = ESubstFossiFuel_II / (CO2/C) 0.36 CO2eq/CO2eq, 2nd cycle (II) 2.1. Carbon debt from land use conversion Figure 2 capture the typical trajectories of oil palm plantation development in the tropics and the associated C‐stock dynamics with each human activities and disturbances that might happen. The furthest back in the trajectory is started with undisturbed natural forest, which has its own C‐stock dynamics, despite of the sequestration/growth due to natural causes such as drought and natural fire. Selective wood harvesting while land remains to be forested can happen recurrently over the period of decades, until forest reaches a very degraded state and C‐stock is significantly lower than the previous undisturbed state. Paths usually diverge from here; the degraded forest can be cleared due to further biomass harvesting and then left either as imperata grassland, planted with some tree species either for timber, fibre or fruits, more intensified croplands, settlement or other uses. Eventually the imperata grassland areas might be converted to either of the uses as well. 1st Drought logging year 2nd logging Clear felling, End of 1st oil Carbon stock, Mg C/ha fire palm rotation Annual grass Land fire clearing Natural forest logging cycles Imperata Oil palm Responsibility of entity issuing permits and/or not Full responsibility oil monitoring implementation palm producers Figure 2. Trajectories of land uses and the dynamics of C-stock Two issues associated with the trajectories with regards to environmental impact of oil palm industry are to quantify the carbon emissions from the land conversion and to attribute which part of emissions to whose responsibilities. Whilst the first one only needs technical steps, the second issue will take political negotiations among the entire stakeholders as to how far back in the trajectories and how large outside the plantation nucleus should the responsibility stops. This study will mainly cover the first issue but targeting on a longer timeframe than immediate land conversion to oil palm and larger areas than the nucleus boundary to accommodate the follow‐up discussions and negotiations on attribution to take place. 2.2. Timeaveraged Cstock of, and emissions from plantation Time‐averaged C‐stock of plantation estimation is conducted comprehensively, taking into account all components of biomass of oil palm, soil, preceding necromass, recycling and other additional organic inputs if applicable, from plot measurement upscaled to plantation. Estimation of N2O emission and CH4 oxidation from management regime is to be conducted by modelling. This pilot study covers the estimation of time‐averaged C‐stock of plantation from the field measurement and the data analysis, however for the N2O emission and CH4 oxidation from management regime, this pilot study only identifies the data input required and prepares the modelling platform to be used. Figure 3 illustrates the components of C‐stock in oil palm plantation and its time‐averaged over a planting cycle. 90 80 Oil palm Necromass 70 previous necromass vegetation Carbon stock, Mg C/ha 60 50 Oil palm canopy 40 30 Stem: annual 20 increment ~ 10 40 cm Plant/replant cycle 0 Root biomass 10 Soil organic matter: decomposition, 20 root turnover, surface inputs Figure 3. Time-averaged C-stock in oil palm plantation from each component 3. Methodology and activities conducted in the pilot study The two pilot areas are located in Sumatra (Site 1) and Kalimantan (Site 2). A field trip was conducted in each of the pilot area in order to understand the systems and assess the variability in bio‐physical characteristics, land use/cover types and agricultural practices. In determining the sampling scheme, we planned to use stratified random sampling. The strata are decided based on factors determining C‐stock and carbon emissions. The original plan was to stratify based on: vegetation cover, soil type, actors and means of land clearing, as presented below. Because of anticipated variation, we recommended 5 replicates for each, i.e, 450 sample plots. Using the land cover map of most recent year, map of soil type, key informant and a focus group discussion, these sample plots can be identified and located randomly. However, this original plan is not feasible to be carried out because of the current actual land use and land cover systems in the field. The two landscapes of the pilot estates and their surrounding areas are hugely dominated by oil palm, except for minor patches of areas allocated for conservation. The two estates are of mineral soil only. Site 1 has plasma and is surrounded by independent smallholder plantation whilst Site 2’s mills is basically only dependent on their nucleus plantation for raw materials. The distribution and areas of independent smallholders’ oil palm production that go to the mills of Site 1 are difficult to identify without some more in‐depth survey. For this pilot study, with the recommendation of the estate management, only nucleus and plasma area are measured for plot level data collection. Therefore, the number of plot samples is reduced most significantly due to this situation since we have almost no freedom to select plot samples. Table 3. Sampling design Factors Descriptions Number of classes (tentative) Land cover • Age of oil palm (5 classes) 10 • Land cover types converted to establish the oil palm plantation (e.g., undisturbed forest, logged over forest, rubber garden) ( 5 classes) Soil type • Mineral soil 3 • Peatland (by depth if it varies a lot in the study area (2 classes) Actors • Company 3 • Plasma • Independent smallholders Means of land clearing • Fire 3 • Heavy equipment • Others 3.1. Carbon debt from land use conversion Carbon debt from land use conversion is estimated by: 1. looking at the actual trajectories of land uses in the pilot estates and surrounding area through remote sensing imagery interpretation in order to quantify changes in areas between one land cover types to others within the study period. Groundtruthing (recording geo‐referenced information) of current land use/cover types that includes all existing and historical variation in the areas was conducted as part of data input of image interpretation. Key informant interviews were conducted to understand the history and local contexts; 2. estimating time‐averaged C‐stock of each of existing land use/cover types; 3. up‐scaling changes of C‐stock to the whole pilot areas and sub‐regions of interests (e.g., nucleus, plasma) based on area changes and time‐averaged C‐stock of each land use/cover types. Despite of the ideal study period that should be selected, the choices are largely driven by the availability of relatively cloud‐free imageries. On the contrary to the plan, which was written prior to any preliminary fieldwork, the existing variation of land use/cover types are extremely small. For the two pilot areas, the landscape and vast surrounding areas are largely dominated by oil palm at present, and by logged‐over forest prior to the plantation establishment. The selection of plot samples is therefore extremely limited and basically we are left with only the areas allocated for conservation areas within the estate. We have then to take an assumption that the C‐stock of current existing conservation area of logged‐over forests and shrubs resemble levels of C‐stock in the similar land cover types in the past. If the conserved area is high in C‐stock compared to those prior to plantation establishment then the estimation of C‐debt from land conversion his overestimated and vice versa. While conceiving the similarities of C‐stock levels between similar land cover presently and in the past is the best assumption we can take under the unavailability of C‐stock data in the past, we have to keep in mind that this assumption may cause uncertainty in the estimation and we need to address this during the next study phase. In addressing in‐house capacity building of estate in estimating carbon emissions, the field teams in two pilot areas were trained in the class and in the field in conducting plot measurement of C‐stock for the purpose of estimating time‐averaged C‐stock of each land use systems. When this interim report is written, trainings on GIS/Remote Sensing to study land use trajectories and up‐scaling from plot and pixels to landscape are still yet to be conducted. 3.1.1. Land use and cover trajectories Time coverage, spatial resolution, and amount of cloud cover are three main criteria used in selecting the best satellite images for this study. As mentioned above, the time coverage had to cover the period before and after plantation establishment. Spatial resolution is the smallest size of object in earth surface that can be recognized in satellite image. Since the size of study area is quite small, high resolution satellite image should be a good option. However, historical data availability is the main constraint for this type of data. Therefore middle resolution satellite image such as Landsat (30m resolution) and SPOT (20m resolution) was chosen for this study. Analysis of land use and cover trajectory (ALUCT) was conducted on the basis of time series land cover maps produced from satellite images. In the context of understanding carbon debt of Site 1 and Site 2, the data is required to cover a sufficient time period of before and after plantation establishment. To get a complete picture of the area, it is also necessary to quantify the changes in the plantation’s surrounding area. Therefore, three main outputs from the analysis are: 1. Time series land cover maps from covering time period before and after oil palm establishment 2. Land cover change quantification of the estate area and its surrounding 3. Land cover trajectories for the period of analysis Field reference data collected during groundtruthing as part of the field work in each of the study sites is used as geo‐referenced information of various land cover type in the field, recorded using global positioning system (GPS) receiveres. This data serve two purposes, as guidance in image interpretation process and as a reference to calculate accuracy of the land cover map produced. ALUCT workflow can be classified into three stages: (1) Image pre‐processing, (2) Image classification, and (3) Post interpretation analysis. The first stage, Image pre‐processing, aims to rectify geometric distortion in satellite images using ground control point (GCP) collected from reference datasets. In this case, orthorectified Landsat ETM from United States Geological Survey (USGS) in each study sites is used as reference data. Minimum of 20 GCP were used in geometric correction, ensuring geometric precision of 0,5 pixel (<15m) for all images. Figure 4. Overall workflow of ALUCT The second stage of ALUCT is image classification. The objective is to produce time series land cover maps through satellite image interpretation. Object‐based hierarchical classification approach is used in this stage. In this approach, image classification steps begin with a series of image segmentation process. The purpose is to produce image objects, a group of pixel with a certain level of homogeneity in term of spectral and spatial. Image objects had to be able to represent actual feature on satellite image, therefore several phase of segmentation was conducted to get the required levels of detail. The result of these phases is called multiresolution image segments which serve as a basis for hierarchical classification system. Illustration of segmentation process is showed in Figure 5. Following segmentation process, image classification is conducted using hierarchical structure showed on Figure 6. The hierarchy is divided into three levels, where in each level land cover types is interpreted using spectral and spatial rule. Details and complexity of land cover types is increase in each level, therefore each of them has different set of rules applied. Level 1 consist of general classes such as: Forest, Tree based system, Non tree based system and Non vegetation. These classes can be easily distinguished using visual inspections and simple vegetation index. Vegetation index is a ratio of spectral value between vegetation‐sensitive channel (near infra red spectrum) and non vegetation‐sensitive channel (visible spectrum) in satellite image. Result of Level 1 is further classified in Level 2, this time field reference data is required and Nearest Neighborhood algorithm is used to distinguished total of 9 land cover types: forest, swamp forest, oil palm, shrub, grass, agriculture, cleared land, and settlement. Figure 5. Segmentation process Some of the classes in Level 2 are classified into more details in Level 3. In this level, spectral value is not the only parameters used, spatial characteristic such as distance to settlement, proximity to logging road, forest concession, and plantation map was used as a rule in classification. Forest is classified into undisturbed forest, logged‐over high density, and logged‐over low density based on proximity to observed logging road, forest concession map, and estimated vegetation density derived from vegetation index value. Using the same approach, Swamp forest is classified into undisturbed swamp forest and logged‐over swamp forest. Oil palm area is classified into young oil palm, mature oil palm and old oil palm. These classes can be considered as a proxy to plantation age, in which the level of canopy cover is differentiated. Mature oil palm approximately relates to the beginning of productive stages, but before the full canopy is reached. The old oil palm relates with the stage where full canopy is reached. Classification is conducted based on vegetation canopy density value and detail planting maps acquired from Site 1 and Site 2. Post classification analysis process is the last stage of ALUCT. It consists of two processes, accuracy assessment and land cover change analysis. The objective of accuracy assessment is to test the quality of information derived from image classification process. It is conducted by comparing field reference data with the most recent land cover map produced in each site. Minimum accuracy level should be above 80%. The last step in ALUCT is the land cover change analysis itself. Two form of land cover change analysis is conducted for each study site: are‐ based changes analysis and trajectories analysis. An area‐based change is a simple analysis conducted by comparing total area of land cover types in each time period. This analysis will conducted in 3 analysis windows: (1) plantation area, (2) plasma area (if any) and (3) all area outside plantation and plasma. The result will show a clear indication of overall trend of land cover change in the area and its surrounding. However there are no information on the location and trajectories of changes provided. Trajectories analysis is conducted to solve this particular problem. Trajectories of changes are the summary of changes sequence over all time period observed at pixel level. In the context of understanding carbon budget from oil palm plantation, types of trajectories is designed to be able to capture changes in C‐stock caused by land cover changes. Trajectories types are classified into 10 classes: 1. Undisturbed forest to logged‐over forest to oil palm 2. Logged‐over forest‐high density to oil palm 3. Logged‐over forest‐low density to oil palm 4. Undisturbed swamp forest to oil palm 5. Logged‐over swamp forest to oil palm 6. Non forest to oil palm 7. Non oil palm‐related trajectories 8. Stable forest 9. Stable swamp forest 10. Stable oil palm Figure 6. Hierarchical classification structure 3.1.2. Timeaveraged Cstock of existing land use types Based on the land cover identification, there are three type land cover systems inside Site 1 (oilpalm, shrub or young secondary forest 18 years old and logged‐over forest). Another logged‐over forest found in plasma area. In Site 2 we found logged‐over forest in the conservation belt along the river and imperata grassland in the surrounding area. C‐stock measurement was done in 14 plots altogether in Site 1 and Site 2. In Site 1, sample plots were set up in young secondary forest (2 plots), logged‐over forest inside plantation (2 plots) and logged‐over forest in plasma area (3 plots) (Figure 7A). In Site 2, sample plot were set up in logged‐over forest (3 plots) and imperata grassland (4 plots) (Figure 7B). Figure 7. Plot sample position in Site 1 (A) and Site 2 (B) (SF = secondary forest; LOF = Logged-over forest; IMP = Imperata grassland) Four types of carbon pool (tree biomass, necromass, understorey and litter) were measured during the observation. 20 m x 100 m plot was set up in each land cover with nested plot 5 m x 40 m and 0. 5 m x 0.5 m (Figure 8). 5 m * 40 m (0.5 m x 0.5 m) 20 m * 100 m Figure 8. Plot sample C‐stock can be derived from total biomass of live tree, dead tree (necromass), understorey and litter. Default value for carbon content from biomass vegetation is 46%. Tree biomass was estimated using allometric equation developed by Kettering et al (2001) on the basis of stem diameter at 1.3 m above the ground (dbh): W = 0.11ρD 2+c where ρ is the wood density and the coefficient c is based on the allometrict relation between tree height (H) and tree diameter (D); H = aDc (default value for c = 0.62) Necromass (dead tree) was estimated through calculate dry weight material based on diameter, length and wood density: DW = (π / 40) ρHD 2 where ρ is the wood density; H is height/length of dead tree; D is diameter of dead tree. The default value of ρ (wood density) used is 0.7 for forest and 0.62 for shrub based on previous study conducted in East Kalimantan (Lusiana et al., 2005) and the real value taken from the wood density database (http://www.worldagroforestry.org/sea/Products/AFDbases/WD), if the species can be identified. Biomass understorey and litter were calculated based on the destructive sample from 0.5 m x 0.5 m. All of understorey and litter inside this plot were removed, then separated between stem and leaves before weighted it of fresh and dry oven 100°C 48 hours. Live and dead trees with more than 30 cm diameter were measured in plot 20 m x 100 m, but for 5‐30 cm diameter were measured in plot 5 m x 40 m. Trees less than 5 cm diameter is considered as understorey. A manual that covers the field protocol in detailed has been written and distributed widely (Hairiah et al., 2001; Hairiah and Rahayu, 2007). 3.1.3. Upscaling The up‐scaling step requires the output of ALUCT and time‐averaged C‐stock of each land use/cover types as the data input. The primary step for calculating carbon density is substituting land use/cover classes with their associated time‐averaged C‐stock per unit area. The results can be presented as maps. As for carbon emission calculation we use the following formula: ∆C = (∑i Ai x ∆Ci ) / T where ∆C is the total annual carbon emission at the landscape, is are land use and cover types, Aj is the area of land use/cover type i in the beginning of study period that was converted to oil palm during the study period, ∆Ci is the difference of time‐averaged C‐stock of land use/cover type i and time‐averaged C‐stock of oil palm plantation per unit area, T is the length of study period. A simple spread sheet macro is suitable for this calculation. 3.2. Timeaveraged Cstock of oil palm plantation All components of biomass of oil palm, soil, preceding necromass, recycling and other additional organic inputs if applicable are included in accounting for C‐stock of oil palm. Therefore, sampling for measurement is designed to cover variation in factors that determine each of the components. In Site 1 planting took place between 1989 to 1991; in this area variation in topography is quite marked. In Site 2, planting takes place since 1997 up to now; the topography is flat and two main soil types are found: inceptisol and ultisol. In both estates, management practice of empty fruit bunch application was found. Table 4 below presents the sampling design for the two estates. Table 4. Design of sampling measurement of the two sites Site Block Year of Type of soil Topography Management planting (Age of palm) L21 1997 (11) Ultisol Flat M21 1998 (10) Ultisol Flat K27 2001 (7) Ultisol Flat K34 2005 (3) Ultisol Flat J16 1997 (11) Inceptisol Flat Site 2 M28 1998 (10) Inceptisol Flat M38 2001 (7) Inceptisol Flat N38 2005 (3) Inceptisol Flat L22 1998 (10) Flat With EFB (substitution) N34 2000 (8) Flat With EFB (supplement) I6 1989 (19) Steep (25%) Without EFB Site 1 I27 1990 (18) Flat With EFB Plasma 1991 (17) Sampling measurement of above and belowground of C‐stock besides was made represent the variation of the sites, was also made represent four spatial zones of WaNuLCAS model set up: zone 1. circle or fertilizer application zone/weeded zone, zone 2. empty fruit bunch (EFB) application or grass zone, zone 3. frond stack zone and zone 4. harvesting path zone, with two samples per strata per tree (Figure 9). In Site 2, remnants of forest necromas (whether in windrow , tree stumps or log remnants, are sampled in a transect). Where land‐clearing before planting made use of windrows into which tree and other necromass was piled, the necromass transect length must be adjusted to the distance between two windrows (Figure 10). The measurement covers above and belowground C‐stock. Aboveground measurement includes litter production, biomass of understorey, palm biomass and necromass. In Site 1, one plot was represented by 48 palms, while in Site 2 was represented by 20 palms. Necromas, tree stumps or log remnants, are sampled in a transect. Belowground measurement includes soil bulk density at 0 – 15 cm soil depth of each zone, soil carbon and palm root biomass. Frond stack Green replicon (or EFB) zone Weeded circle Palm Path Figure 9. Sampling measurement of above and belowground of C-stock represent four spatial zones: zone 1. circle or fertilizer application zone/weeded zone, zone 2. empty fruit bunch (EFB) application or grass zone, zone 3. frond stack zone and zone 4. harvesting path zone. s m as 0 c rom > 1 ec ood tn res ct (w Fo nse r) e tra met d ia Figure 10. Sampling measurement of necromass In calculating total biomass of palm, the total biomass was partitioned into three components: trunk biomass, rachis biomass (including petiole) and frond bases biomass. The palm biomass was estimated through allometric equation. The allometric equation was developed by measuring, palm height, palm diameter, total number of leaf, frond base biomass and leaves biomass as shown in Table 6. Cylindrical shape of trunk, allow us simply use volume of cylinder to estimate trunk biomass based on measured data of trunk diameter and height and data of wood density (Porankiewicz et al., 2005). In Site 1, 48 palms was measured for each block (144 palms in total), while in Site 2 10 palms was measured for each block (100 palms in total). There is difference of trunk diameter measurement between Site 2 and Site 1. In Site 1, the measurement was done by removing the frondbase, thus actual trunk diameter was got. While in Site 2, the measurement was done without removing the frondbase. Some calibration is then made in order to make comparable measurement from the two pilot areas. For estimating total palm above ground standing biomass, allometric equation as a function of palm height is developed. Empty fruit bunches which are returned to the sites from mills are taken into account. The below ground carbon pool from soil and root and necromass are added to calculate total time‐averaged C‐stock, which is then upscaled to the whole estate area to get time‐averaged C‐stock of plantation. Table 5. Component of palm measured to developed allometric equation Parameter Site 2 Site 1 Trunk height √ √ Trunk diameter √ √ Number of leaf * √ Rachis length √ √ Petiole: - Diameter √ √ - Depth - √ - Length √ - Cumulative frond base production DW ** √ Leaf DW √ √ √ measured * assumed number of leaf: 48 leaves (< 5 years old), 40 leaves (5 – 9 years old) and 32 leaves (> 10 years old) (pers. Comm. SMARTRI researcher, 2008) ** estimated based on Site 1 data 3.3 Activities Field trip to Site 1 was conducted from 6‐12 November 2008 and Site 2 from 10‐14 November 2008, by ICRAF team, IPOB staff, estate senior researchers and estate field team. Groundtruthing and plot measurement for land use and cover types other than oil palm were conducted during the field trips as a join effort between the three teams. Trainings for field teams were also conducted during the trip. Measurement protocol and sampling design were established and communicated during the trip. Following up to the trip, data collection within oil palm plantation was conducted by estate field team under the supervision and guidance of the estate senior researchers for the period of November 2008 to March 2009. Data was sent for further analysis to ICRAF for time‐averaged C‐stock of oil palm plantation. Literature survey was conducted as the analysis was conducted. For the C‐stock estimation of other land use and cover types, the 14 plot data measurement was further analyzed in ICRAF. Remote sensing imagery interpretation and ALUCT were conducted by ICRAF team during the period of October 2008 and March 2009. ALUCT training for estate research teams are yet to be conducted in the future. Upscaling and synthesis were conducted in March to April 2009. 4. Results from two pilot studies Figure 11 shows the indicative location of the two sites to give some illustrations of the historical differences of the two sites rather than referring to any management practices or particular companies. Figure 11. The two sites of the pilot study 4.1. Land cover trajectories Land cover trajectories analysis (Figure 12) of Site 1 (established in the early 1990’s) estate clearly showed that more than 40% of conversions within the plantation area were from logged‐over forest (Figure 13). Nearly half of it was high‐density logged‐over forest area. In plantation‐plasma area, almost 50% of oil palm was converted from forest, with 27% of it was from high‐density logged‐over forest and 5% from undisturbed swamp forest. In the surrounding area, 67% of oil palm was converted from forest. From that amount, 12% was undisturbed swamp forest and 34% was high density logged‐over forest. Figure 12. Time series land cover map of site 1 estate 100% Non oilpalm‐related 90% trajectories Stable swamp forest 80% Stable oilpalm 70% Stable forest 60% 50% Non forest to oilpalm 40% Log over forest‐low density to oilpalm 30% Log over forest‐high density to oilpalm 20% Log over swamp forest to oilpalm 10% Undisturbed swamp forest to oilpalm 0% Plantation area Plasma area Image area Figure 13. Summary of land cover trajectories in Site 1 estate and surrounding area In Site 2 (established in the early 2000’s), the surrounding area was still undergoing some logging activity (Figure 14). Conversions from undisturbed forest to logged‐over forest is a strong indication of this on‐ going process. Conversion to oil palm was only located in less than 35% of the observed area. Inside plantation area, more than 90% of oil palm area were converted from forest, 30% of it was high density logged‐over forest (Figure 15). Figure 14. Time series land cover map of site 2 estate 100% Non oilpalm‐related 90% trajectories Stable swamp forest 80% Stable forest 70% Non forest to oilpalm 60% Log over swamp forest to 50% oilpalm 40% Undisturbed swamp forest to oilpalm 30% Log over forest‐low density to oilpalm 20% Log over forest‐high density to oilpalm 10% Undisturbed forest to log over forest to 0% Image area Plantation area Figure 15. Summary of land cover trajectories in Site 2 estate and surrounding area 4.2. Cstock estimation in land covered by vegetation other than oil palm at plot level Above ground C‐stock in logged‐over forests in Site 1 and Site 2 are markedly different. Logged‐over forests in Site 1 contain much higher number of large trees which leads to much higher C‐stock than those in Site 2, due to harvesting. It is interesting to note here that whilst the total aboveground C‐stock in logged‐over forest in Site 1 nucleus plantation is almost double than those in Site 2, those from living biomass is comparable. 300 Necromass Biomass 250 Total Carbon stock (ton/ha) 200 150 100 50 0 Shrub- Logged-over Logged-over Logged-over Imperata-S Buatan Forest (in) Forest (out) forest-S Rungau Buatan Buatan Rungau Figure 16. C-stock from necromass and standing biomass in non-oil palm landcover within nucleus and plasma oilpalm plantation in Site 1 and within nucleus plantation in Site 2 4.3. Timeaveraged Cstock of oil palm at plot level Total biomass of palm was partitioned into three components: trunk biomass, rachis biomass (including petiole) and frond bases biomass. The total palm biomass was estimated through allometric equation. The allometric equation was developed by measuring, palm height, palm diameter, total number of leaf, frond base biomass and frond biomass (Figure 17). Figure 17. Allometric equation as a function of palm height to estimates palm biomass Based on stem diameter, stem height and frond canopy biomass, aboveground C accumulation in oil palm biomass was estimated of about 5 t C ha per year. The aboveground time‐averaged C‐stock of oil palm plantation is similar between the two estates i.e., 38.8 ton ha and 39.2 ton ha respectively for Site 1 and Site 2, with 25 years planting cycle. This calculation takes into account tree biomass and empty fruit bunches that are returned from the mills to the plantation (Figure 18). 120 y = 5.0141x + 15.947 y = 2.6679x + 29.676 2 2 R = 0.8752 R = 0.8752 100 DW palm biomass (Mg ha ) -1 y = 3.0876x + 24.3 2 80 R = 0.8752 60 40 20 0 0 5 10 15 20 Age of palm (years) This Study Corley et al., 1971 Khalid et al., 1999 Figure 18. Correlation between age of palm and palm biomass (Mg ha-1) 4.4. Upscaling and carbon debt from land use conversion In general Site 1 estate's emissions and sequestration per unit area are higher than those in Site 2 in each of the region under study (Table 6). The sequestration per unit area in Site 2 within the estate area is lower than that of Site 1 because of the differences in percentage of total areas which were planted by the end of this study period (91% in Site 1 estate and 84% in Site 2 estate). Emissions from plasma areas in Site 1 are 35 % lower than that of the estate due to more conversions from land cover of higher C‐stock initially. Table 6. Annual emissions and sequestration per unit area 5. Conclusions and next steps In order to address carbon debt, threes level of engagement from plantation companies could be taken, while international rules are still under discussion: • To avoid carbon debt, conversion should be conducted only from shrub and grassland with an aboveground C stock of less than 40 ton C ha . • To reduce/minimize carbon debt, companies should set aside conservation areas which are hot spots of C‐stock, to allow natural succession to happen and therefore to achieve co‐benefit of biodiversity conservation as well as reducing C‐stock emission. • To neutralize, rehabilitating larger areas in different places to achieve comparable sequestration or buying CER’s will be some options. The next phase of study will try to capturing generalities and specificities in Indonesia, reducing uncertainty of estimation in plot and estate level and being more comprehensive in including all components such as transportation and processings, especially in within plantation management through some modeling efforts. Socio‐economic impacts which are an important part of the equations of oil palm production in tropical countries will be studied comprehensively during the enxt phase. Acknowledgement This study and report was financially supported by IPOB and the Dutch Embassy with contributions from ICRAF. We are grateful for the excellent technical support, discussions, data collection and collaborations from the estate senior researchers and estate field team. References Chavalparit O, Rulkens W H, Mol A P J and Khaodhair S. 2006. Options for Environmental Sustainability of the Crude Palmoil Industry in Thailand through Enhancement of Industrial Ecosystems. Environment, Development and Sustainability, 8, 2, pp. 271 – 287. Corley R H V, Hardon J J and Tan G Y. 1971a. Analysis of Growth of the Oil Palm (elais guinensis jacq.) 1. Estimation of Growth Parameters and Application in Breeding. Euphytica 20: 307 – 315. Crutzen PJ, Mosier AR, Smith KA, and Winiwarter W. 2008. N O release from agro‐biofuel production negates global warming reduction by replacing fossil fuels. Atmos. Chem. Phys., 8, 389–395, 2008. Dewi S, Khasanah N, Rahayu S, Ekadinata A and van Noordwijk M. 2009. Carbon Footprint of Indonesian Palm Oil Production: a Pilot Study. Bogor, Indonesia. World Agroforestry Centre ‐ ICRAF SEA Regional Office. European Communities Comission, 2008. Proposal for a Directive of the European Parliament and of the Council on the promotion of the use of energy from renewable sources. Germer J and Sauerborn J. 2008. Estimation of the Impact of Palm Plantation Establishment on Greenhouse Gas Balance. Environ Dev Sustain 10: 697 – 716. Hairiah, K., Sitompul, S.M., Van Noordwijk, M. And Palm, C. 2001. Methods for sampling carbon stocks above and below ground. ASB Lecture Note 4B. ICRAF, Bogor, 23pp. Henson I E and Chai S H . 1997. Analysis of oil palm productivity. II. Biomass, distribution, productivity and turnover of the root system. JOPR 9(2). ISO, 1997. Environmental Standard ISO 14040, Environmental Management‐life Cycle Asssessment‐ principal and Framework, Reference Number: ISO 14040: 1997 (E). IPCC (2001b) Climate change 2001: the scientific basis. Technical summary of the working group I report , Geneva, 2001 Khalid H’, Zin Z Z and Anderson J M. 1999. Quantification of Oil Palm Biomass and Nutrient Value in a Mature Plantation. I Above‐ground Biomass. Journal of Oil Palm Research 1: 23‐32 Pleanjai S, Gheewala S H and Garivait S. 2004. Environment Evaluation of Biodiesel Production from Palm Oil in a Life Cycle Perspective. The Joint International Conference on “Sustainable Energi and Environment (SEE)” 1‐3 December 2004, Hua Hin, Thailand. Porankiewicz B, Iskra P, Sandak J, Tanaka C, Jo´ z´wiak K. 2006. High‐speed steel tool wear during wood cutting in the presence of high‐temperature corrosion and mineral contamination. Wood Sci Technol 40:673–682. Rahayu, S., Luasiana, B. and Van Noordwijk, M. 2004. Above ground carbon stock assessment for various lan use systems in Nunukan, East Kalimantan. In: Lusiana, B. et al (eds.). 2005. Carbon Stocks Monitoring in Nunukan, East Kalimantan: A Spatial and Modelling Approach. World Agroforestry Centre. p: 21‐34 Suhaimi M and Ong H K. 2001 Composting Empty Fruit Bunches of Oil Palm. Malaysian Agricultural Research and Development Institute (MARDI). Kuala Lumpur. Malaysia. http://www.agnet.org/library/eb/505a/ 7 April 2009. Van Noordwijk, M., Rahayu, S., Hairiah, K., Wulan, Y.C., Farida, A. and Verbist, B. 2001. Carbon stock assessment for a forest‐to‐coffee conversion landscape in Sumberjaya (Lampung, Indonesia: from allometric equation to land use change analysis, Science in China, 45: 75‐86. van Noordwijk M, Lusiana B. 1999. WaNuLCAS, a model of water, nutrient and light capture in agroforestry systems. Agroforestry Systems 43: 217‐242. van Noordwijk M, Lusiana B, Khasanah N .2004. WaNuLCAS 3.01: background on a model of Water, Nutrient and Light Capture in Agroforestry Systems. Bogor, Indonesia. World Agroforestry Centre ‐ ICRAF, SEA Regional Office. 246 p. Van Noordwijk, M. 2009.Biofuel Emission Reduction Estimator Scheme (BERES): Land use history, current production system and technical emission factors. Flyer.
Pages to are hidden for
"Estimating carbon footprint from biofuel production from oil palm "Please download to view full document