Published online 6 July 2006
Published in J Environ Qual 35:1439-1450 (2006)
DOI: 10.2134/jeq2005.0159
© 2006 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
Greenhouse Gas Emissions from Forestry Operations
A Life Cycle Assessment
Edie Sonne*
College of Forest Resources, University of Washington, P.O. Box 352100, Seattle, WA 98122
* Corresponding author (edie.sonnehall{at}weyerhaeuser.com)
Received for publication April 29, 2005.
 |
ABSTRACT
|
|---|
Most forest carbon assessments focus only on biomass carbon and assume that greenhouse gas (GHG) emissions from forestry activities are minimal. This study took an in-depth look at the direct and indirect emissions from Pacific Northwest (PNW) Douglas-fir [Pseudotsuga menziesii (Mirbel) Franco] forestry activities to support or deny this claim. Greenhouse gas budgets for 408 "management regimes" were calculated using Life Cycle Assessment (LCA) methodology. These management regimes were comprised of different combinations of three types of seedlings (P + 1, 1 + 1, and large plug), two types of site preparation (pile and burn, and chemical), 17 combinations of management intensity including fertilization, herbicide treatment, pre-commercial thinning (PCT), commercial thinning (CT), and nothing, and four different rotation ages (30, 40, 50, and 60 yr). Normalized to 50 yr, average direct GHG emissions were 8.6 megagrams (Mg) carbon dioxide equivalents (CO2e) ha–1, which accounted for 84% of total GHG emissions from the average of 408 management regimes. Harvesting (PCT, CT, and clear cutting) contributed the most to total GHG emissions (5.9 Mg CO2e per 700 m3 harvested timber), followed by pile and burn site preparation (4.0 Mg CO2e ha–1 or 32% of total GHG emissions) and then fertilization (1.9 Mg CO2e ha–1 or 15% of total GHG emissions). Seedling production, seedling transportation, chemical site preparation, and herbicide treatment each contributed less than 1% of total GHG emissions when assessed per hectare of planted timberland. Total emissions per 100 m3 averaged 1.6 Mg CO2e ha–1 over all 408 management regimes. An uncertainty analysis using Monte Carlo simulations revealed that there are significant differences between most alternative management regimes.
Abbreviations: CO2e, carbon dioxide equivalents • CT, commercial thinning • DBH, diameter at breast height • FVS, Forest Vegetation Simulator • GHG, greenhouse gas • GPG, Good Practice Guidance • GWP, global warming potential • LCA, Life Cycle Assessment • LMS, Landscape Management System • LULUCF, land-use, land-use change, and forestry • PCT, pre-commercial thinning • PNW, Pacific Northwest • TPH, trees per hectare
|
INTRODUCTION
|
|---|
THE KYOTO PROTOCOL and its subsequent accords recognize seven land-use, land-use change, and forestry (LULUCF) activities (United Nations Framework Convention on Climate Change, 2002)1, one of which is forest management. Globally there are 187 million ha of plantations that are included in the definition of a managed forest (Food and Agriculture Organization of the United Nations, 2000). These lands vary in their productivity and in their management practices. Past forest carbon assessments have focused only on changes in biomass carbon and assume that greenhouse gas (GHG) emissions from forestry activities are minimal. This assumption not only omits a potentially significant source of emissions from forest management, but also it precludes evaluation of differences in emissions from alternative forest management intensity choices by forest landowners. Whether a party to Kyoto and its subsequent accords or not, it is important to be able to quantify both GHG emissions and carbon sequestration from forest management.
The goal of this paper is to provide an in-depth assessment of GHG emissions from forestry operations in the PNW. This is done by (i) developing a framework for a detailed GHG inventory from forest management activities; (ii) examining the relative contribution of direct emissions, which occur within the forest management system boundary, to the total life cycle upstream impact of forestry activities; and (iii) examining the relative global warming impact of different forest management decisions.
The global warming impact for 408 "management regimes," comprised of the production and transportation of three types of seedlings (P + 1, 1 + 1, large plug), two types of site preparation (pile and burn, chemical), 17 different growth management intensities that include three types of initial planting density (865, 1235, and 1729 trees ha–1) (350, 500, and 700 trees acre–1) and various combinations of fertilization, herbicide treatment, pre-commercial thinning (PCT), commercial thinning (CT), and no treatments, and four different rotation ages (30, 40, 50, and 60 yr) were calculated using Life Cycle Assessment methodology. Uncertainty analysis using Monte Carlo simulation was used to discern actual differences in total GHG emissions from different forest management regimes.
|
BACKGROUND
|
|---|
Although there have been a large number of global, national, regional, and local forest carbon assessments, very few have taken into account GHG emissions from forest management activities. The majority of carbon assessment studies in managed forests have excluded non-biomass emissions altogether (Dewar, 1990; Evrendilek, 2004; Harmon and Marks, 2002; Hoen and Solberg, 1994; Kurz et al., 2002; Masera et al., 2003; Murray, 2003; Nilsson, 1997; Thompson and Matthews, 1989). Some have assumed a general ratio intended to account for some or all of greenhouse gas emissions from forest management activities (Schlamadinger and Marland, 1996; Sikkema and Nabuurs, 1995). Thornley and Cannell (2000) include nitrous oxide emissions from fertilization and nitrogen deposition, but did not include any other emissions. Schwaiger and Zimmer (2001) compare fuel consumption and GHG emissions for timber harvesting, hauling, and transport in different European countries. They found that harvesting and hauling GHG emissions represent a fraction (0.5%) of carbon sequestration during forest growth, but log transportation accounts for an additional 0.5 to 3.2%. Seppala et al. (1998) include harvesting, ditching, site preparation, and forest fertilization in their LCA of the Finnish forest industry, but because the functional unit is based on the whole annual production of the Finnish forest industry, emissions from all activities except harvesting are so small they are excluded from the analysis. In one of the few detailed studies on GHG emissions from a part of forestry, Athanassiadis et al. (2000) examine spare part material consumption in different types of harvesting machinery.
Because of the recognition of the terrestrial biosphere as a dynamic carbon sink, there is interest in understanding the potential for carbon mitigation projects in managed and unmanaged forests. The complexity of terrestrial carbon processes has rendered a host of assumptions for forest carbon measurement (which carbon pools to include) and carbon accounting (how to keep track of carbon changes over time). For this study accounting principles outlined in Good Practice Guidance for Land-use, Land-use Change, and Forestry (Intergovernmental Panel on Climate Change, 2003) were followed, referred to hereafter as the GPG. Methods for reporting biomass carbon dioxide, methane, and nitrous oxide emissions from site preparation, fertilization, and harvesting are included in these guidelines. The most critical assumption in these guidelines is that all biomass immediately decomposes on harvest. The GPG recognizes that this assumption is not true for solid wood products and even for on-site slash decomposition but it was decided that it is a "legitimate, conservative assumption for initial calculations" (Intergovernmental Panel on Climate Change, 2003)2. The impact of accounting for wood product carbon is assessed in Sonne Hall (2005). For all non-biomass based emissions, standards outlined in the Greenhouse Gas Protocol (World Resources Institute, 2004) were followed.
|
MATERIALS AND METHODS
|
|---|
Life cycle assessment (LCA) was used because it presents a logical, transparent, and compact way to examine the global warming impact of all forestry activities. Developed from system-level studies on energy use, which gained prominence in the 1970s, LCA has become a standardized protocol to examine the environmental performance of a product, process, or service over its entire life cycle or any parts within its life cycle (cradle-to-grave or cradle-to-gate) (Ross, 2003). Furthermore, different alternatives can be examined separately to compare the relative impact of each management regime. In this study, a forest manager has choices in seedling type, site preparation method, planting spacing, and growth intensity, and the combinations of each of these alternatives will be examined separately.
Functional Unit
The function of the system under study is the production of timber and carbon storage on a sustainable basis. The functional unit for this study is 1 ha of forestland managed for 50 yr. Because forestry, like agriculture, both uses and produces natural resources, it does not fall into a traditional "cradle-to-grave" process that is acceptable for industrial processes (Haas et al., 2000). The functional unit can be identified on the farm level, area (hectare) level, or a product level, and the choice of a functional unit is dependent on the goal of the impact assessment and the scope of the study (Haas et al., 2000). For example, an LCA based on a product unit assumes there is ample land available to produce the product and the acreage can change according to product input intensity. An LCA based on the farm or area level assumes the acreage is fixed and product output can change. Because this study is concerned with the impact of a particular landowner or land-base, the functional unit is based on area and production will vary.
No matter whether the functional unit is chosen to be based on a fixed area or a fixed level of production, the LCA data are the same and the results are essentially presented from a different point of view. For example, Haas et al. (2000) found that emissions were much lower on organic farms when compared on a farm or area functional unit, yet intensive farming yielded lower emissions per product unit due to lower production in organic farms. In this paper, results will also be displayed per unit of timber production to compare production efficiency.
System Boundary
The scope of this study is limited to managed plantations of Douglas-fir owned by the private forest industry on the west side of the Cascade Mountains in Washington and Oregon. Although the results of the study can be applied to all ownership types (industry, non-industrial private, state, tribal, federal), industrial forestlands supply the majority of data used in the study. The primary reason for relying on industrial forestland data is because the management goals of this ownership type are consistent with maximizing economic value. There are approximately 3.2 million ha of timberlands owned by the forest industry in western Oregon and Washington (Briggs and Trobaugh, 2001). Approximately 71% of these timberlands are Douglas-fir (Azuma et al., 2002). The majority of forest industry timberland acreage (40%) is classified as Site Index II, which is 35 to 41 m in height after 50 yr (Briggs and Trobaugh, 2001).
Unit Processes
The forest management regimes defined in this paper are suites of decisions that affect the growth rate and production volume of timber. The system boundary extends from cradle-to-harvest and includes fuel, electricity, and fertilizer production but does not include production at sawmills, transportation to mills, or the construction of facilities and equipment (Fig. 1). The life cycle sub-stages included within the system boundary are seedling production and transportation, site preparation, growth enhancements, and harvesting (Table 1). Within each sub-stage certain process alternatives have been identified as common practices.
The benchmark management regime, which will be used compare changes from current practice, was defined as P + 1 seedling type with chemical site preparation grown at 1235 trees per hectare (TPH) for 50 yr with one fertilization, herbicide treatment, and CT performed during the rotation (Briggs and Trobaugh, 2001). The benchmark regime was the most prevalent response for each category taken from the Stand Management Survey.
For each life cycle stage there are on-site GHG emissions that occur via combustion or volatilization, as well as emissions that occurred before entering the system boundary. For example, during forest fertilization, nitrous oxide is emitted on nitrogen fertilization and carbon dioxide is emitted during combustion of jet fuel during helicopter application. Additionally, nitrous oxide, methane, and carbon dioxide are all emitted during the production and transportation of the nitrogen fertilizer and during production and transportation of the jet fuel. The GHG Protocol separates these emissions into direct ("emissions from sources that are owned or controlled by the company") and indirect ("emissions that are a consequence of the activities of the company but occur at sources owned or controlled by another company") (World Resources Institute, 2004) (Table 2). Although a landowner most likely will only be responsible for, and therefore get credit for, changes in on-site, or direct emissions, from a global warming impact perspective it is important to understand the complete ramifications of the landowner decision.
The following section summarizes data collections and analyses methods; for a more complete description see Sonne Hall (2005).
Seedling Production
Ninety-five percent of all Douglas-fir seedling types planted in western Washington and Oregon timberlands are P + 1 (60%), 1 + 1 (25%), and large plug (10%) (Briggs and Trobaugh, 2001). A P + 1 seedling is grown in a 2.2-cm plug container for about 4 mo and is then planted in nursery field at a spacing of about 25 seedlings m–2 (6 ft–2) for roughly 15 mo. The mortality rate is about 10% each year (personal communication with N. Khadduri, forest nursery technician, Webster Forest Nursery, Washington State Department of Natural Resources, Olympia). A 1 + 1 stock is grown for 1 yr in a seedbed at a density of 1 790 750 seedlings ha–1, and it is then harvested, root pruned to 13 cm, and transplanted back into a bed at a larger spacing (402 610 seedlings ha–1). The average height reached is 46 cm (N. Khadduri, personal communication). Large plug seedlings are grown in larger plugs [164 cm3 (10 in3)] for 1 yr. The mortality rate is 12% (N. Khadduri, personal communication). Table A1 (in the appendix) lists the inputs into these seedling types, which were converted into forms compatible with the LCA model. Pesticides were categorized into their respective chemical families, which is the level where production inputs and emissions data are available. Emissions from fertilizer production were estimated by calculating the nitrogen, phosphorus, and potassium contents and forms using the building block method outlined in Kongshaug (1998). Data for pesticides, fertilizers, and transportation to regional storage data were estimated from Ecoinvent Data Version 1.1 (Frischknecht and Jungbluth, 2004). Fuel and electricity production and truck transportation were estimated from the Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation Model Version 1.6 (Wang, 2001). Nitrous oxide, nitrogen oxides, and ammonia emissions were obtained using emissions factors outlined in Ecoinvent (Frischknecht and Jungbluth, 2004), which are based on emissions factors specified in Intergovernmental Panel on Climate Change (2001) and Asman (1992), respectively. Although there is evidence that N fertilization significantly reduces the natural methane sink into soils (Castro et al., 1994), this loss is not included in the analysis because, as of yet, there is not a standard method with which to quantify (Intergovernmental Panel on Climate Change, 2003).
In addition, although the medium for the P + 1 and large plug seedling stock is peat, this study does not include emissions from peat harvesting. Draining peatlands can result in both decreases in methane emissions and increases in carbon dioxide and nitrous oxide emissions (Intergovernmental Panel on Climate Change, 2003). Roulet (2000) found that for the 100-yr global warming potential of carbon dioxide and methane, Canadian peatlands are neither a source nor a sink of GHGs. (Annual flux of CO2 is –0.111 Pg CO2–C yr–1 and annual flux of CH4 is 0.005 Pg CH4–C yr–1.) Because of the uncertainty in the direction of the flux and lack of information on the origin of the peat used in the nursery, peat production was omitted from this analysis. However, this uncertainty presents an opportunity for further analysis.
Site Preparation
In western Washington and Oregon the two principal methods of site preparation are chemical site preparation and pile and burn, accounting for 56 and 20% of all practices, respectively (Briggs and Trobaugh, 2001). In chemical site preparation, herbicide application is used to kill undesirable vegetation that will compete with seedlings; the slash left over from the previous harvest remains on the ground for slow decomposition. In pile and burn site preparation the prior harvests slash is piled mechanically and then burned using a propane–diesel mix. The amount of slash found on an acre of timberland is highly variable and depends on the volume of the prior stands and the harvest removal efficiency. Many naturally seeded sites can have between 15 and 220 Mg slash ha–1 (Intergovernmental Panel on Climate Change, 2003); the survey for this study found an average slash pool of 54 Mg ha–1 (22 Mg a–1) from a mix of naturally seeded and planted sites. Planted sites are assumed to have smaller amounts of slash because the live branches and foliage are the only principal contributors to the amount.
The principal source of emissions from pile and burn site preparation is carbon dioxide emissions from biomass combustion. However, because standards assume immediate decay on harvest these emissions do not need to be counted again. Methane and nitrous oxide emissions, which are not accounted for on harvest, do need to be considered. Methane, nitrous oxide, nitrogen oxides, and carbon monoxide emissions from biomass burning are estimated using Equation 3.2.20 of the GPG, and emissions factors found in Table 3A.1.16 (Intergovernmental Panel on Climate Change, 2003). A combustion efficiency of 62% was used (according to Table 3A.1.12). Diesel and propane production and combustion information was estimated from GREET Version 1.6 (Wang, 2001). Information on herbicide production was estimated from Ecoinvent Version 1.1 (Frischknecht and Jungbluth, 2004). Table A2 lists the associated reference flows for pile and burn and chemical site preparation.
Transportation of Seedlings
Emissions data for seedling transportation are expressed in tkm (ton-kilometers). Distance, truck-loads, and truck capacity were obtained from a survey sent to western Washington and Oregon forest managers, and emissions information was obtained from the GREET model, using data for a medium–heavy duty truck with a 6-Mg capacity (Wang, 2001). Table A3 lists these reference flows.
Tree Growth
Seventeen different growth trajectories were considered, accounting for various combinations of planting density (865, 1235, 1729 TPH), fertilization, herbicide treatment, PCT, CT, or plant with no further treatments. The Landscape Management System (LMS) was used to run simulations that projected both stand volume and biomass carbon for the 17 management intensities and four rotation ages (McCarter et al., 1998, McCarter, 2001). Stand volume used in LMS was calculated using the Forest Vegetation Simulator (FVS) growth and yield model (Dixon, 2002) and biomass carbon was calculated using diameter at breast height (DBH)–based allometric equations based on Jenkins et al. (2004).
Forest Vegetation Simulator (FVS) is an individual-tree distance-independent model that grows stands using a few inputs, including location, species, site index, elevation, slope, aspect, and stand density (Dixon, 2002). The growth variant used was Pacific Northwest coast (PN) and the stand was grown from bare ground using the ESTAB key-word. The model was adjusted to (i) calibrate growth during the first stage of stand development to be consistent with yield curves published in Anderson et al. (1994), (ii) account for growth due to fertilization, and (iii) account for growth due to herbicide treatment. For more information on how and why these adjustments were made see Sonne Hall (2005). Emissions for PCT, CT, and harvesting are based on the volume of timber harvested. For every cubic meter that is harvested 2.57 L of diesel are consumed (Johnson et al., 2002). This value accounts for hand-felling and cable yarding to a landing site but does not include transportation of logs to a mill.
Simulations were run separately for rotation ages of 30, 40, 50, and 60 yr, and each management regime was harvested at the end of its respective rotation. Comparisons were made both between and within rotation ages (within rotation ages allowed for comparison of management intensity alone). The average amount of on-site carbon storage was calculated over each respective rotation age for each management intensity. On-site carbon included stem, bark, limbs, crown, foliage, and roots. Biomass carbon was calculated using allometric equations based on DBH found in Jenkins et al. (2004).
Soil carbon and litter–dead wood carbon were not included in this analysis. Although some studies have found management practices, such as harvesting and site preparation, to be a negative influence on soil carbon (Harmon and Marks, 2002), others have indicated that the effect is minimal (Johnson and Curtis, 2001). The Tier 1 estimate for the GPG assumes that the soil carbon pool remains stable for a given area and management regime over time (Intergovernmental Panel on Climate Change, 2003). Dead wood and litter carbon also were given a stable default assumption by the GPG, although there is recognition that these pools contain highly variable yet significant sources of carbon.
Process Matrix
The LCA was conducted according to the methodology outlined in Heijungs and Suh (2002). Specifically, the basic inventory model for LCA is comprised of unit processes, represented as column vectors, each of which describes the activity of a single operation or group of operations (e.g., urea production, helicopter take-off and landing) needed to fulfill the demand vector f, which is dictated by the reference flows as specified by the functional unit. These unit process vectors together form the process matrix P, where each column represents a unit process and each row represents a specific type of exchange. To account for the large number of matrices needed to examine the 408 scenarios of the forest management "regimes," the large process matrix was first divided into sub-matrices. The sub-matrices were: 1 + 1 seedling production, P + 1 seedling production, large plug seedling production, pile and burn site preparation, chemical site preparation, transportation to field, herbicide treatment, fertilization, and harvesting (thinning). The scaling vector (s), which determines the amount of each economic flow required in the system, and the inventory vector (g), which determines the total amount of environmental emissions in the system, were calculated for each sub-matrix following the methods outlined in Heijungs and Suh (2002) [see Sonne Hall (2005) for more details]. These calculations were performed in Microsoft Excel (Microsoft, 2003).
The results from the sub-matrices were pooled together to form one large process matrix. The demand vector was varied to reflect different combinations of seedling production, site preparation, growth enhancements, and rotation ages. To find the inventory vector for the 102 regime combinations (not including rotation ages), a loop code was written in Matlab 7.0 (The MathWorks, 2004). This loop was run eight different times to vary rotation age (four rotation ages) and direct or total emissions.
|
RESULTS
|
|---|
The greenhouse gases considered in this LCA include carbon dioxide, methane, and nitrous oxide. For a 100-yr time horizon the global warming potential (GWP) for CO2, CH4, and N2O are 1, 23, and 296, respectively, reflecting revised numbers from the Third Assessment Report of the IPCC (Intergovernmental Panel on Climate Change, 2001). These numbers weight emissions from each specific greenhouse gas appropriately so all emissions can be reported as carbon dioxide equivalents (CO2e), calculated by multiplying each GHG emission by its GWP (
CO2 + CH4 x 23 + N2O x 296). Emissions for the unit processes, which reflect production and application emissions from the inputs described in Tables A1
through A3, are found in Tables 3 
through 5. These emissions factors can be adjusted according to study specific parameters. For example, for seedling production, dividing the emissions factors by the number of seedlings produced yields emissions per seedling. The pile and burn site preparation numbers, and associated emissions factors for methane and nitrous oxide, can be adjusted up or down according to the specific study parameters. If a rotation has more than one application of fertilization, emissions should be multiplied accordingly. Finally, emissions factors for harvesting are based on 1 m3 of harvested wood.
The total global warming impact (direct and indirect emissions included) of forest management activities for 102 different management regimes at rotation age 50 yr varied between 5.93 Mg CO2e ha–1 (for seedling type 1 + 1, chemical site preparation, 1235 TPH, PCT) and 13.59 Mg CO2e ha–1 (for large plug seedling type, pile and burn site preparation, 1729 TPH, CT, herbicide treatment, and fertilization). The average global warming impact for the 50-yr rotation was 9.8 Mg CO2e ha–1. For direct emissions only, the average global warming impact for the 50-yr rotation was 8.2 Mg CO2e ha–1. For more information see Sonne Hall (2005).
Average emissions varied by rotation age, with the 30-yr rotation having the lowest emissions and the 60-yr rotation having the highest. However, when normalized to the functional unit (50 yr) the order is reversed (Table 6). (Normalization is done by multiplying the 30-yr rotation by 5/3, multiplying the 40-yr rotation by 5/4, and multiplying the 60-yr rotation by 5/6.) In all rotations, 84% of emissions are categorized as direct emissions (see Discussion).
Carbon dioxide accounted for the majority of GHG emissions (67%), followed by N2O (23%), and CH4 (10%) (Table 7). For direct emissions only, nitrous oxide emissions increased to 27%. This is almost five times the total radiative forcing of N2O emissions globally, and this increase can be attributed to nitrogen fertilization, which is the main source of anthropogenic N2O emissions. The methane contribution in this study is lower than global methane radiative forcing, which is about 20%. (Including the loss of the soil methane sink from nitrogen fertilization may increase this study's methane contribution slightly.)
|
DISCUSSION
|
|---|
Direct versus Indirect (Upstream) Emissions
Overall, the upstream, or indirect, emissions in this study accounted for only 16% of the total contribution to global warming impact. This means that the emissions that occurred on site via combustion of fossil fuels, biomass burning, or nitrogen application were almost five times greater than the emissions that were released from the production and transportation of fuels and chemicals used on site. For pile and burn site preparation, 97% of emissions are direct. For harvesting, fuels used during the production and transportation of harvest material account for 17% of its total global warming impact. However, upstream emissions from fertilizer production and transportation account for 30% of the 1.9 Mg CO2e ha–1 attributed to the fertilization unit process. Fertilizer production is energy intensive and generates considerable GHG emissions, principally carbon dioxide from ammonia production and nitrous oxide from nitric acid production (Wood and Cowie, 2004).
Contribution of Each Unit Process
There are three key contributors to GHG emissions from forestry activities (Table 8). Harvesting emits 2.5 Mg CO2e per 300 m3 harvested wood. The harvest method considered was hand-felling and bucking at stump with cable yarding to the landing (Johnson et al., 2002). This accounts for about 51% emissions if pile and burn site preparation and fertilization are used, and more if they are not. Feller-bunchers and skidders require 15% more fuel than hand-felling and cable yarding, contributing to even higher GHG emissions per cubic foot of harvested timber (Johnson et al., 2002). If transportation of harvested wood to a mill is included, an additional 5.5 Mg CO2e per 300 m3 harvested wood are emitted, based on transportation consumption rates of 5.53 L diesel per cubic meter harvested wood (Johnson et al., 2002). [Note that fuel consumption rates for harvesting and transport from Johnson et al. (2002) are higher than those published in Schwaiger and Zimmer (2001). Johnson et al. (2002) use data from the Pacific Northwest and Schwaiger and Zimmer (2001) use European data.] Pile and burn site preparation contributed 4.0 Mg CO2e ha–1 or 32% and fertilization contributed 1.9 Mg CO2e ha–1 or 15% of total global warming impact. Herbicide treatment and chemical site preparation each contributed about 1% to total GHG emissions, and the rest of the unit processes, seedling production and seedling transportation, contributed less than 1%.
Stands that are fertilized typically emit 2.5 Mg CO2e ha–1 more over the rotation age (Fig. 2). Although the contribution of fertilizer production and application only accounts for 1.9 Mg CO2e ha–1, the increase in growth causes an increase in yield, which leads to more fossil fuel emissions to run the harvesting machinery, according to the assumptions of harvest emissions in this study (Johnson et al., 2002). Management regimes that included pile and burn site preparation also emitted, on average, 2.0 Mg CO2e ha–1 more than those whose site preparation method included chemical treatment (Fig. 3). Over 90% of these emissions are due to methane and nitrous oxide emissions from biomass burning. The emissions will increase if more slash is burned during site preparation. Differences in seedling type and initial stand density were minimal. Nitrous oxide emissions from fertilizer application accounted for 1.3 Mg CO2e ha–1.
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 2. Average global warming impact of fertilized stands by rotation, normalized to 50 yr (per ha–1).
|
|
The magnitude of nitrous oxide emissions from fertilizer application is highly variable and can be manipulated by changing fertilizer type or application method. Switching to a slow release fertilizer that eliminates nitrous oxide emissions from fertilizer application could reduce GHG emissions on industrial lands in western Washington and Oregon timberlands by 73 000 Mg CO2e yr–1 [55 870 ha are fertilized per year in coastal Oregon and Washington (Briggs and Trobaugh, 2001)]. Similarly, eliminating pile and burn site preparation altogether would reduce GHG emissions by 67 000 Mg CO2e yr–1 [17 000 ha are prepped using pile and burn in coastal Oregon and Washington (Briggs and Trobaugh, 2001)].
Greenhouse Gas Emissions per 100 m3 Harvested Timber
The functional unit in this LCA was defined such that GHG emissions were reported per hectare for each management regime. However, stand volume varied dramatically both between management intensities and between rotation ages. In this section GHG emissions are displayed as a function of volume by dividing the total GHG emissions for each management regime by the total yield associated with the management regime (see Table 9). Greenhouse gas emissions per 100 m3 were highest in the 30-yr rotation regimes and lowest in the 60-yr rotation regimes. The 40- and 50-yr rotation regimes were almost identical. The "no treatment" regimes (865_NA and 500_NA) had the lowest rates of GHG emissions per harvested timber and the 1729_PCT_CT_herb_fert had the highest rates. However, the differences in emissions among the different management intensities were smaller when assessed per unit harvested material than per hectare. This discrepancy is because the regimes with larger GHG emissions often yielded more volume.
Uncertainty Analysis
The uncertainty surrounding GHG emissions from each management regime include uncertainty in the parameter inputs and outputs due to natural variability, measurement uncertainty, process specific variations, temporal variations, uncertainty due to the appropriateness of the input or output flows to describe the specific unit process, appropriateness of the functional unit and choice of allocation procedure, and uncertainty due to the possible omission of important flows (Pedersen Weidema and Wesnaes, 1996; Sonnemann et al., 2003).
Aside from being able to quantify natural variability by repeated measurements, the other sources of uncertainty are inherently qualitative. To quantify these other sources of uncertainty, Pedersen Weidema and Wesnaes (1996) propose a matrix of data quality indicators and corresponding coefficients of variation (see Table 10). The basic uncertainty of a parameter can be adjusted to reflect these other sources of variation by calculating the square root of the sum of the squares of the individual coefficients.
The adjusted coefficients of variation, converted to the square of the geometric standard deviation, were reported in Ecoinvent following the same procedure outlined above. For the data not found in Ecoinvent the same procedures were followed. For example, the number of 1 + 1 seedlings produced had the following data quality score, based on the descriptions in Table 10. For reliability, a score of 3 was given because the number of seedlings was based on assumptions of spacing at the nursery. The data were representative of the seedling type but taken from a small number of sites (one) and for a short period of time (1 yr), and therefore was given a completeness rating of 4. Temporal correlation, geographical correlation, and technical correlation were all given the highest rating because the data were from a forest nursery on the west side of the Cascades and the information was collected in 2004. However, technical correlation was given a rating of 2 because seedling data were taken from a state forest nursery, which may have different practices from an industrial nursery.
To ascertain how the uncertainty of individual input and output parameters affect the overall global warming impact for each management regime, probabilistic means and associated standard deviations were calculated using Monte Carlo simulation. In this study, the uncertainty analysis was performed only on the direct emissions found in this life cycle inventory. This decision was justified for two reasons. First, landowners will only be responsible, and hence will only get credit for, changes in direct emissions within their system boundaries. Second, direct emissions accounted for, on average, 84% of the total contribution to global warming. Thus, the uncertainty contribution of the indirect emissions will be small in comparison to the contribution of the direct emissions.
The simulation was run for 1000 runs for a number of management regimes using a Monte Carlo application found in Simapro Version 6.0 (PRé Consultants, 2004). The probabilistic means and standard deviations for the reference flow (P + 1, chemical, 1235/CT/herb/fert, 50 yr), the lowest emission regime (P + 1, chemical, 1235/PCT/CT/, 50 yr), and the highest emission regime (large plug, pile and burn, 1729/CT/herb/fert, 50 yr) are summarized in Table 11.
The mean and standard deviations for the global warming impact for each of these regimes was found by summing the means and by taking the square root of the sums of the squared variances. Two-tailed t tests were used to detect statistical differences between the benchmark regime and each of the two alternative regimes (Zar, 1999; Drusano et al., 2000). This test could have been repeated 407 times to test the null hypothesis that the benchmark regime is the same as all other alternative regimes. However, in the interest of efficiency, an alternative way to estimate statistical differences was adopted. The method reversed calculations of the t test to find the minimal detectable difference given a sample size and a variance (Zar, 1999). The sample size was the number of runs performed during the Monte Carlo simulation (1000). The sample size is large, resulting in the ability to detect small differences. When the number of simulation runs was reduced to 100, the minimal detectable difference increased by 0.004 Mg (or 3.5%). The pooled variance was the average pooled variance between the benchmark regime and the highest and lowest alternative regimes, respectively. Although the minimal detectable difference is not statistically accurate for each individual run (because the pooled variance does not correspond to each individual run), it can serve as an adequate filter approach for understanding the accuracy of the differences between two management regimes. All total GHG emission differences between the 407 alternative management regimes and the reference regime that were greater than the minimal detectable difference were assumed to be statistically different from the reference regime.
At a 0.05 level of significance with 1000 simulation runs, there is a 95% chance of detecting a significant difference between means that are greater than or equal to 0.14 Mg CO2e ha–1. Out of the 102 management regimes for the 50-yr rotation, 90 (88%) differed from the reference regime by more than the minimal detectable difference. The 12 regimes that were statistically similar were: 1235_CT_fert/chemical/P + 1, 1 + 1, large plug, 1235_CT_herb_fert/chemical/P + 1, 1 + 1, large plug, 1729_CT_fert/chemical/P + 1, 1 + 1, 1729_PCT_CT_fert/chemical/large plug, and 1729_ PCT_CT_herb_fert/chemical/P + 1 and 1 + 1.
Emissions as Percentage of Carbon Storage
Average carbon storage is defined as the average amount of carbon stored per acre for each management regime, and is calculated by averaging carbon storage in 5-yr increments of the rotation age. Prior studies have assumed that GHG emissions from forestry operations are either minimal or negligible when compared with changes in on-site carbon sequestration. When compared with the average carbon storage of each management regime, GHG emissions represent approximately 4.5% of the value of on-site average carbon storage (Table 12). The percentage of emissions to sequestration varies between 2.5% (average of 60-yr rotation ages) and 6.8% (average of 30-yr rotation ages). If transportation of harvested materials is included in the analysis, GHG emissions increase to 9.4% of average carbon storage.
View this table:
[in this window]
[in a new window]
|
Table 12. Percent of greenhouse gas (GHG) emissions to average carbon storage by rotation age (normalized to a 50-yr rotation).
|
|
|
CONCLUSIONS
|
|---|
A Life Cycle Assessment of forestry operations in Pacific Northwest coastal Douglas-fir plantations was completed for 408 different management regimes, representing three types of seedling production and transportation, two types of site preparation, 17 different growth management intensities, and four rotation ages. Direct emissions, or emissions that occur "on-site," account for 84% of total global warming impact. Though GHG emissions from forestry operations are only a fraction of carbon sequestration from tree growth, there are opportunities for GHG reductions within forest management. The biggest contributors to GHG emissions are harvesting, followed by pile and burn site preparation, and fertilization. Switching to a slow release fertilizer that eliminates nitrous oxide emissions from fertilizer application could reduce GHG emissions on industrial lands in western Washington and Oregon timberlands by 73 000 Mg CO2e yr–1. Similarly, eliminating pile and burn site preparation altogether would reduce GHG emissions by 67 000 Mg CO2e yr–1 in western Washington and Oregon. Finally, reducing emissions from timber harvesting and timber transportation could yield significant reductions in GHG emissions.
|
APPENDIX
|
|---|
Reference Flows for Unit Processes
|
NOTES
|
|---|
Funded by the National Council on Air and Stream Improvement.
1 Afforestation, reforestation, and deforestation were designated as eligible activities under Article 3.3 of the Kyoto Protocol, and forest management, cropland management, grazing land management, and revegetation were accepted as additional activities under the Marrakesh Accords. 
2 Appendix 3a to the IPCC Good Practice Guidance for LULUCF proposes future methodological development for accounting for carbon in wood products.
|
REFERENCES
|
|---|
- Anderson, D., R. Alig, and D. Adams. 1994. Yield and inventory tables for western Washington. CINTRAFOR. AR-10. College of Forest Resources, Univ. of Washington, Seattle.
- Asman, W.A.H. 1992. Ammonia emissions in Europe: Update emission and emission variations. Rep. 227471009. RIVM, Natl. Inst. of Public Health and the Environ., Bilthoven, the Netherlands.
- Athanassiadis, D., G. Lidestav, and I. Wasterlund. 2000. Assessing material consumption due to spare part utilization by harvesters and forwarders. Int. J. For. Eng. 11(2):51–57.
- Azuma, D.L., L.F. Bednar, B.A. Hiserote, and C.F. Veneklase. 2002. Timber resource statistics for western Oregon, 1997. Resource Bull. PNW-RB-237. USDA For. Serv., Pacific Northwest Res. Stn., Portland, OR.
- Briggs, D., and J. Trobaugh. 2001. Management practices on Pacific Northwest west-side industrial forest lands, 1991-2000; With projections to 2005. Stand Management Cooperative Working Paper #2. College of Forest Resources, Univ. of Washington. Seattle.
- Castro, M.S., W.T. Peterjohn, J.M. Mellilo, P.A. Steudler, H.L. Gholz, and D. Lewis. 1994. Effects of nitrogen fertilization on the fluxes of N2O, CH4, and CO2 from soils in a Florida slash pine plantation. Can. J. For. Res. 24:9–13.
- Dewar, R.C. 1990. A model of carbon storage in forests and forest products. Tree Physiol. 6:417–428.[Abstract]
- Dixon, G.E. 2002. Essential FVS: A user's guide to the Forest Vegetation Simulator. Internal report. USDA For. Serv., Forest Manage. Serv. Center, Fort Collins, CO.
- Drusano, G.L., D.Z. D'Argenio, S.L. Preston, C. Baron, W. Symonds, S. LaFon, M. Rogers, W. Prince, A. Bye, and J.A. Bilello. 2000. Use of drug effect interaction modeling with Monte Carlo Simulation to examine the impact of dosing interval on the projected antiviral activity of the combination of Abacavir and Amprenavir. Antimicrob. Agents Chemother. 44(6):1655–1659.[Abstract/Free Full Text]
- Evrendilek, F. 2004. An inventory-based carbon budget for forest and woodland ecosystems of Turkey. J. Environ. Monit. 6:26–30.[CrossRef][Web of Science][Medline]
- Food and Agriculture Organization of the United Nations. 2000. FRA 2000 Assessing state and change in global forest cover: 2000 and beyond. Forest Resources Assessment Working Paper 31. FAO, Rome.
- Frischknecht, R., and N. Jungbluth (ed.) 2004. Overview and methodology. Rep. 1. Ecoinvent, Dubendorf, Switzerland.
- Haas, G., F. Wetterich, and U. Geier. 2000. Life cycle assessment framework in agriculture on the farm level. Int. J. Life Cycle Assess. 5(6):345–348.[CrossRef]
- Harmon, M.E., and B. Marks. 2002. Effects of silvicultural practices on carbon storage in Douglas-fir–western hemlock forests in the Pacific Northwest, U.S.A.: Results from a simulation model. Can. J. For. Res. 32:863–877.[CrossRef]
- Heijungs, R., and S. Suh. 2002. The computational structure of Life Cycle Assessment. Kluwer Academic Publ., Boston.
- Hoen, H.F., and B. Solberg. 1994. Potential and economic efficiency of carbon sequestration in forest biomass through silvicultural management. For. Sci. 40(3):429–451.
- Intergovernmental Panel on Climate Change. 2001. Climate Change 2001. The science of climate change. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge Univ. Press, Cambridge.
- Intergovernmental Panel on Climate Change. 2003. Good practice guidance for land-use, land-use change and forestry. Inst. for Global Environmental Strategies, Arlington, VA.
- Jenkins, J., D.C. Chojnacky, L.S. Heath, and R.A. Birdsey. 2004. Comprehensive database of diameter-based biomass regressions for North American tree species. GTR NE-319. USDA For. Serv. Northeastern Res. Stn., Portland, OR.
- Johnson, D.W., and P.S. Curtis. 2001. Effects of forest management on soil C and N storage: Meta analysis. For. Ecol. Manage. 140:227–238.[CrossRef]
- Johnson, L.R., B. Lippke, and J.D. Marshall. 2002. Forest resources—Pacific Northwest and Southeast. Appendix A. In J. Bowers, D. Briggs, B. Lippke, J. Perez-Garcia, and J. Wilson (ed.) Life cycle environmental performance of renewable building materials in the context of residential building construction. Phase I Interim Research Report. Consortium for Research on Renewable Industrial Materials (CORRIM), Seattle.
- Kongshaug, G. 1998. Energy consumption and greenhouse gas emissions in fertilizer production. In IFA Tech. Conf., Marrakech, Morocco. 28 Sept.–1 Oct. 1998. Int. Fertilizer Industry Assoc., Paris.
- Kurz, W., M. Apps, E. Banfield, and G. Stinson. 2002. Forest carbon accounting at the operational scale. For. Chron. 78(5):672–679.
- Masera, O.R., J.F. Garza-Caligaris, M. Kanninen, T. Karjalainen, J. Liski, G.J. Nabuurs, A. Pussinen, B.H.J. de Jong, and G.M.J. Mohren. 2003. Modeling carbon sequestration in afforestation, agroforestry, and forest management projects: The CO2FIX V.2 approach. Ecol. Modell. 164:177–199.[CrossRef]
- McCarter, J.B. 2001. Landscape Management System (LMS): Background, methods, and computer tools for integrating forest inventory, GIS, growth and yield, visualization and analysis for sustaining multiple forest objectives. Ph.D. diss. Univ. of Washington, Seattle.
- McCarter, J.B., J.S. Wilson, P.J. Baker, J.L. Moffett, and C.D. Oliver. 1998. Landscape management through integration of existing tools and emerging technologies. J. For. 96:17–23.
- Microsoft. 2003. Microsoft Excel Version 2003. Microsoft, Redmond, WA.
- Murray, B.C. 2003. Economies of forest carbon sequestration. p. 221–241. In E.O. Sills and K.L. Abt (ed.) Forests in a market economy. Kluwer Academic Publ., Boston.
- Nilsson, L.-O. 1997. Manipulation of conventional forest management practices to increase forest growth—Results from the Skogaby project. For. Ecol. Manage. 91:53–60.[CrossRef]
- Pedersen Weidema, B., and M.S. Wesnaes. 1996. Data quality management for life cycle inventories—An example using data quality indicators. J. Cleaner Production 4(3–4):167–176.
- PRé Consultants. 2004. SimaPro Version 6.0. PRé Consultants, Amersfoort, the Netherlands.
- Ross, S. 2003. The use of Life Cycle Assessment in environmental impact assessment. Ph.D. diss. School of Anthropology, Geography, and Environmental Studies, Univ. of Melbourne.
- Roulet, N. 2000. Peatlands, carbon storage, greenhouse gases, and the Kyoto Protocol: Prospects and significance for Canada. Wetlands 20(4):605–615.[CrossRef][Web of Science]
- Schlamadinger, B., and G. Marland. 1996. The role of forest and bioenergy strategies in the global carbon cycle. Biomass Bioenergy 10(5/6):275–300.[CrossRef][Web of Science]
- Schwaiger, H., and B. Zimmer. 2001. A comparison of fuel consumption and greenhouse gas emissions from forest operations in Europe. Chapter 5. In T. Karjalainen, B. Zimmer, S. Berg, J. Welling, H. Schwaiger, L. Finer, and P. Cortijo (ed.) Energy, carbon and other material flows in the Life Cycle Assessment of forestry and forest products: Achievements of Working Group 1 of the COST Action E9. Discussion Paper no. 10. Eur. For. Inst., Joensuu, Finland.
- Seppala, J., M. Melanen, T. Jouttijarvi, L. Kauppi, and N. Leikola. 1998. Forest industry and the environment: A Life Cycle Assessment study from Finland. Resour. Conserv. Recycl. 23:87–105.
- Sikkema, R., and G.J. Nabuurs. 1995. Forest and forest productions: The challenge for a better carbon balance. In Life-cycle analysis—A challenge for forestry and forest industry. Proc. of the International Workshop, European For. Inst. and the Federal Res. Centre for Forestry and Forest Products, Hamburg, Germany. EFI, Joensuu, Finland.
- Sonne Hall, E. 2005. Greenhouse gas emissions from Pacific Northwest forestry operations: Implications for forest management. Ph.D. diss. Univ. of Washington, College of Forest Resources, Seattle.
- Sonnemann, G., M. Schuhmacher, and F. Castells. 2003. Uncertainty assessment by a Monte Carlo simulation in a life cycle inventory of electricity produced by a waste incinerator. J. Cleaner Production. 11:279–292.[CrossRef]
- The MathWorks. 2004. Matlab Version 7.0. The MathWorks, Natick, MA.
- Thompson, D., and R. Matthews. 1989. CO2 in trees and timber lowers greenhouse effect. Forestry and British Timber. 18–21.
- Thornley, J.H.M., and M.G.R. Cannell. 2000. Managing forests for wood yield and carbon storage: A theoretical study. Tree Physiol. 20:477–484.[Abstract]
- United Nations Framework Convention on Climate Change. 2002. A guide to the Climate Change Convention and its Kyoto Protocol. Climate Change Secretariat, Bonn, Germany.
- Wang, M.Q. 2001. Development and use of GREET 1.6 fuel-cycle model for transportation fuels and vehicle technologies. Center for Transportation Res., Energy Systems Div., Argonne Natl. Lab., Argonne, IL.
- Wood, S., and A. Cowie. 2004. A review of greenhouse gas emission factors for fertiliser production. IEA Bioenergy Task 38. IEA, Paris.
- World Resources Institute. 2004. Greenhouse Gas Protocol: A corporate accounting and reporting standard. Revised edition. WRI and World Business Council for Sustainable Development, Washington, DC.
- Zar, J.H. 1999. Biostatistical analysis. Prentice Hall, Upper Saddle River, NJ.
TOP
ABSTRACT
INTRODUCTION
