Address correspondence to: Ottar Michelsen Department of Industrial Economics and Technology Management Norwegian University of Science and Technology (NTNU) NO-7491 Trondheim, Norway firstname.lastname@example.org http://www.iot.ntnu.no/users/ottarm/
The forestry sector is experiencing an increasing demand for documentation about its environmental performance. Previous studies have revealed large differences in environmental impact caused by forestry operations, mainly due to differences in location and forestry practice. Reliable information on environmental performance for forestry operations in different regions is thus important. This article presents a case study of forestry operations in Norway. Environmental impact and value added of selected operations were assessed. This was done with a hybrid life cycle assessment (LCA) approach. Main results, including a sensitivity analysis, are presented for a set of four impact categories. The production chain assessed included all processes from seedling production to the delivery of logs to a downstream user. The environmental impact was mainly caused by logging, transport by forwarders, and transport by truck. These three operations were responsible for approximately 85% of the total environmental impact. The contribution to value added and total costs were more evenly distributed among the processes in the value chain. The sensitivity analysis revealed that the difference in environmental impact between the worst case scenario and the best case scenario was more than a factor of 4. The single most important process was the transport distance from the timber pile in the forest to the downstream user. The results show that the environmental impact from forestry operations in boreal forests was probably underreported in earlier studies.
The forestry sector is important in Norway. In 2004, wood and wood-based products to the value of more than 14 billion Norwegian kroner (NOK1) were exported. This constitutes 2.5% of the total export (Statistics Norway 2005). The forestry sector is also important for employment, especially in rural areas, and the woodworking industry is present in more than 70% of all municipalities in Norway (The Ministry of Agriculture 1998).
More than one third of the country is covered with forest, of which slightly more than 80% (74,000 km2) is productive. The timber volume in the productive forests is estimated at 705 million cubic meters (m3)2 (Hobbelstad et al. 2004). The forests are also important habitats for a range of species; almost half of the species on the Norwegian Red List are forest species (Kålås et al. 2006). Forestry is thus important for biodiversity.
Wood is a renewable material, and the regrowth in Norway is estimated at 24 million m3/year (Hobbelstad et al. 2004). Annual logging in the last decade has been approximately 8 million m3/year (Statistics Norway 2006), giving a net regrowth of 16 million m3/year for the time period. Still, the use of wood is debated, in particular due to the impact on the forest ecosystems and habitat loss (Petersen and Solberg 2005; Puettmann and Wilson 2005; Sanness 2003). The public's confidence in nonproven marketing statements, such as “environmentally friendly,” for example, applied on the basis of the renewable nature of wood products, is rapidly declining (Kuckartz 2000). Reliable information and data, combined with a consistent methodological framework, are thus required to compare forest products to other products, which primarily are nonrenewable. Examples could be wood-based biofuels and wood in the construction industry.
Several studies have compiled life cycle inventory (LCI) data on forestry operations. These reveal large differences in emissions depending on location, forestry management practices, and logging techniques (Berg and Lindholm 2005; Johnson and colleagues 2005; Schweinle 2000). As an example, Berg and Lindholm (2005) have shown that emissions of greenhouse gases during forestry operations are almost 40% higher in northern Sweden than in the south. Some of the differences between the studies might also have been caused by methodological issues, such as system boundary selection. Studies on products of which wood is a major component must thus take the large variations due to geography and management practice into consideration.
The purpose of this study is to assess the environmental impact from forestry operations in a given region in Norway. This is done with a hybrid life cycle assessment (LCA) approach, and the impact is related to a functional unit of one cubic meter of round wood logs under bark3 delivered to a downstream user. The focus is on spruce logging, primarily Picea abies, which is the dominant species in Norwegian forestry, constituting about 75% of the logged volume. Forestry activities are included, as well as transport to a downstream user. Average environmental impact is calculated, as well as a best and worst case scenario. The results are also compared to other LCA studies on forestry operations to reveal differences in environmental impact due to differences in location, forest practice, and methodology.
The forestry sector in Norway has experienced a significant increase in the pressure to provide information on environmental performance during the last decade. All sawmills4 in the region regularly receive questions about the environmental performance of their products. The paper-producing company Norske Skog has in particular been driven by export markets, in which important customers demand paper originating from “sustainable forestry.” There is thus a growing concern for environmental issues within the entire woodworking sector (Sverdrup-Thygeson et al. 2004), and it is assumed that environmental performance will become a competitive factor in the future.5
Norske Skog has passed this pressure on to its suppliers and encouraged them to introduce environmental management systems to document their performance (Sanness 2003). In 1999, 2000, and 2001, Norske Skog paid an additional NOK 7 per cubic meter if the timber came from environmentally certified forestry (ISO 14001, with the Norwegian PEFC6 standard “Living Forest”7 as the basis for the forestry performance; Sverdrup-Thygeson et al. 2004).
This study was performed in cooperation with ALLSKOG BA, a co-operative society with 9,250 part owners. These are relatively small forest owners operating in five counties in the northern parts of Norway (see figure 1). Until 2006, ALLSKOG was a forest owner organization known as Skogeierforeninga Nord (SN).
The forest owners are responsible for all forestry operations, but most forest owners subcontract some or all of the operations to ALLSKOG, such as planning, planting, silviculture, logging, and sales. In 2005, SN had the responsibility for logging 557,000 m3. Of this, 94% was logged with harvesters8 and transported to forest roads by forwarders9 (SN 2005). Other logging techniques, such as the use of chainsaws and cable yarding, are not taken into account in this study, due to their small contribution.
In total, SN sold 759,000 m3 locally logged timber in 2005 (SN 2005), which was 70% of the timber logged in this region (Statistics Norway 2006). Fifty-two percent of the timber is sold as sawn timber, while the rest is pulp wood, primarily for paper production, chips, and firewood. In the end, more than half of the timber ends up as pulp wood, as the residuals from sawmills are to a large extent used in paper production (figure 2).
ALLSKOG has ownership interests in two sawmills in the region as well as some subsidiary companies: a wholly owned subsidiary producing and delivering chips to smelting plants, 50% of a company for marketing and delivery of wood pellets for heating, and 50% of a transport company responsible for almost all timber transport in the region. Some of the sawmills have ALLSKOG as their only supplier. ALLSKOG is therefore an important actor in the forestry and woodworking sector in the region.
In 2005, SN had, in total, 515 customers. Most of these are small, and the 10 largest purchase almost 95% of all logged timber. This percentage has increased during the last years, as some of the largest customers have experienced a significant growth in through-put. As shown in figure 2, Norske Skog is by far ALLSKOG's most important customer, purchasing almost one third of all timber. Partly as a consequence of the pressure from Norske Skog, SN became ISO 14001 and ISO 9001 certified in December 2000 (Rametsteiner and Simula 2003).
The pressure to provide environmental information is to a large degree related to the impact on biodiversity from forestry operations (Hanski and Walsh 2004; Seppälä et al. 1998). At present, there is no agreed-on methodology for including the impact on biodiversity from land use activities in LCA (Milà i Canals et al. 2007). This is thus not discussed in this article but is addressed in a separate publication (Michelsen 2008), and the focus here is on emissions from forestry operations.
The functional unit in this study is the production of one cubic meter of round wood logs under bark delivered at the gate of a downstream user. The user might be a sawmill, a pulp producer, or a chip producer (figure 2). The system boundaries include planning of forestry operations, seedling production, soil scarification and planting, silviculture (mechanical cleaning of undesirable vegetation, fertilization, chemical cleaning, and weed combating and drainage), harvesting (felling, pruning, and cutting into logs), transport to pile at forest road, construction and maintenance of forest roads, and transport from pile to a factory (figure 3).
Hybrid LCA Systems
Previous LCAs on forestry operations have focused on traditional LCIs (Berg and Lindholm 2005; Johnson et al. 2005; Schweinle 2000). However, the combination of physical LCIs with input–output (I-O) data has received significant interest over the last few years. Suh and colleagues (2004) provided a rationale for the application of hybrid LCA approaches based on system boundaries issues in traditional LCA.
The hybrid LCA system used in this article is formulated by an adaptation of the notation of I-O analysis (IOA). An LCA can then be expressed in a single equation yielding a vector d of different types of environmental impacts:
This is a linear model of process interdependence. Central to its understanding is the requirements matrix, or coefficients matrix, matrix A. The columns of the A matrix describe the intermediate inputs a production process requires from itself and other processes to produce one unit of output. These inputs can be expressed in any units, be they monetary, mass, or energy. In common LCA terminology, the A matrix contains the inventory data of interprocess relations. The remaining inventory data, the emission intensities for each of the processes, are given in the F matrix. The C matrix contains characterization factors for the various emissions, describing how much each of the emissions contributes to the different environmental impact categories. The functional unit of the system is defined by the y vector, and I is the identity matrix.
In this study, an approach for establishing hybrid inventories developed by Strømman and Solli (2006, 2008) was applied. This approach enables the estimation of missing inventories from I-O data, utilizing knowledge of product prices. The method ensures 0% cutoff with respect to costs. Principally, I-O-based data are combined with original key data and adapted to represent the input structure of the processes in question. The application of Leontief's (1949) price model is essential in doing this.
The method of Strømman and Solli (2006, 2008) requires an identification of which sectors of the economy the various foreground processes belong to. The input structure of these sectors is used as the model for the missing inputs. Further, the structures are scaled so that they, together with the original key data, satisfy the Leontief (1949) price model. The resulting hybrid LCA structure is then a model that is valid in both the primal and the dual form. That is, it has a consistent representation of both the flow structure and the cost structure.
A purchase from a system such as this instigates purchases in the next production layer, and then in the next layer, and so on, producing an infinite number of production paths that add up to the total production.10 A structural path analysis (SPA) is applied to identify the most significant contributions to environmental impact. The SPA identifies the contribution to impacts caused by individual chains according to the functional unit. Because the number of process chains is infinite, pruning techniques and ranking of paths are necessary to identify the most important paths. The results from the SPA provide an efficient basis for quality control of the inventory. (See Strømman and Solli [2006, 2008] for more details on the applied methodology and Treolar[1997, 2002] and Peters and Hertwich  for more details on application of SPA.)
For this study, the Norwegian I-O matrix for 2000 was applied (Statistics Norway 2003). The matrix includes capital and imports and is compiled in basic prices plus trade, transport, and financial intermediation services indirectly measured (FISIM) margins. The vector of value added is supplied additionally by Statistics Norway (see Appendix S1 in the Supplementary Material on the Web). The environmental stressor intensities of each sector include emissions of 20 components contributing to global warming, photochemical oxidation, acidification, eutrophication, and human toxicity (table 1). Characterization factors are taken from the CML 2 baseline method11 for all impact categories except human toxicity, which is from research by Hertwich and colleagues (2001).
Table 1. Absolute and relative impacts from the processes included in the product system, related to the functional unit of 1 m3 round wood logs under bark delivered to a downstream user
Transport by forwarder
Forest road construction
Note: HTP = human toxicity potential; NOK = Norwegian kroner.
Global warming potential
Photochemical oxidation potential
HTP air, cancer
kg benzene eq.
HTP air, noncancer
kg toluene eq.
Assumptions and Data Sources
Time is a critical element in LCIs of timber production. In a boreal forest, the rotation period might be 100 years or more. Planting and silviculture are thus carried out long before logging and in most cases under different management principles and methodologies than are common today. Similarly, the areas planted or left for natural regeneration will not be logged for about a century, and it is not possible to know for sure which principles will apply at that time.
Despite this time lag, the present level of planting and silviculture was allocated to today's level of logging, because no better options are available with current knowledge. The amount of planting, silviculture, logging, and forest road construction was based on annual average data from the period 2000–2004. These are the best available data for the given region. Other assumptions and data sources are given in Appendix S1 in the Supplementary Material on the Web.
Average values for environmental impact and value added were calculated in accordance with a functional unit of one cubic meter round wood under bark delivered to a downstream user. In addition, a worst case and a best case scenario were assessed on the basis of three assumptions (table 2). First, the size of the log has a major impact on the diesel consumption during logging (Kjøstelsen and Lileng 2006), and average size in a logging area was here assumed to range from 0.1 to 0.5 m3/log. Second, the distance from the logging spot to the forest road is important. Here, it was assumed that this ranges from 50 m to 3,000 m. Finally, the transport distances from pile to factory registered in the region range from 12 to 301 km. It was also assumed that the loading factor would be somewhat higher on long distances, while there would be no return cargo at the shortest distances (giving a loading factor of 50%). Estimated diesel consumption in the processes resulting from these assumptions is also shown in table 2. Potential changes in the carbon content in the soil were not included in the analysis (Reijnders and Huijbregts 2003).
Table 2. Estimated diesel consumption and assumptions made in the calculation of worst case and best case scenarios
Note: One cubic meter (m3, SI) ≈ 1.31 yd3. One liter (L) = 0.001 m3≈ 0.264 gal. One kilometer (km, SI) = 1,000 m ≈ 0.621 mi.
Average size of log (m3)
Diesel consumption (L/m3)
Average distance to forest road (m)
Diesel consumption (L/m3)
Distance to factory (km)
Loading factor (%)
Diesel consumption (L/m3)
Diesel consumption (L/m3)
For each process in the production chain, the basic prices are identified. The basic price is not equal to the price the purchaser has to pay because the value of subsidies (in particular related to planting, silviculture, and forest road construction) and taxes is not included (United Nations 1999) but represents a cost that has to be covered to run the system. The resource rent paid to the forest owner is also not included. Both subsidies and the resource rent fluctuate due to shifting policy and market possibilities, and it was decided here to relate the environmental impact to the actual and fixed costs. This assumption does not influence the environmental assessment. The method does, however, presuppose a cost breakdown in which the value added (VA) is identified. In the forestry operations, VA is primarily salaries and, to some degree, dividends and retained profit (cf. Sturm et al. 2003).
Average impacts for all environmental impact categories and costs are shown in table 1. The impact from soil scarification is included in the impact from planting, while silviculture is the sum of fertilization, mechanical and chemical cleaning, and drainage. For silviculture, more than 90% of the impact was caused by mechanical cleaning (data not shown). Absolute and relative values for the different processes are shown. As the table indicates, the emissions were primarily caused by logging operations, transport by forwarder, and transport to factory. For emissions of greenhouse gases, these three processes were responsible for almost 84% of all emissions, and, similarly, for acidification and eutrophication, they were responsible for 85% and 89%, respectively. When it comes to emissions of photo oxidants and human toxic compounds, forest road construction was also of high importance.
Costs, in particular value added, were more evenly distributed in the system. Logging, transport by forwarder, and transport on truck accounted for 66% of the total costs and only 52% of the value added. Forest road construction was also of importance here, and for value added, seedling production, planting, and silviculture also made significant contributions.
The relationships between environmental impact and total costs (basic prices) and value added are shown in figure 4 and figure 5, respectively. The steeper the line is, the higher was the environmental impact per unit cost. In figure 6, the environmental impacts for the best case and the worst case are shown relative to the average values. Four impact categories are included. In the best case, the total costs summed up to NOK 210, and the value added was NOK 99. In the worst case, the numbers were NOK 564 and NOK 221, respectively.
The most important processes when it comes to emissions were logging and transport operations. This is in accordance with previous findings (Berg and Lindholm 2005; Johnson et al. 2005). Our results show higher emissions than reported by Berg and Lindholm (2005), even though both forestry practice and climatic conditions for forestry are comparable in these studies. As an example, our results show average emissions slightly above 25 kg CO2-eq/m3 (table 1), which is more than 40% higher than reported from a similar system in northern parts of Sweden (Berg and Lindholm 2005).
The difference in comparable inputs was far less. In fact, according to Berg and Lindholm (2005), the diesel consumption for logging operations and transport to factory was 4.69 L/m3 in northern Sweden, while our results show an average on 4.59 L/m3 for the same operations (table 2). Berg and Lindholm (2005) stated that their results showed significantly higher emissions from forestry operations than were earlier reported in the Scandinavian countries, but their emissions were also probably underestimated. This is consistent with recent literature on hybrid LCA (Strømman et al. 2006; Suh et al. 2004). Given that hybrid LCA studies generally have more complete upstream system descriptions than standard LCA inventories, they capture a larger share of the total impacts generated (Strømman et al. 2006; Suh et al. 2004).
Issues related to system boundary selection are relevant for the comparison of our results with those of Berg and Lindholm (2005). These authors did not include the life cycle of capital goods or the transport of energy carriers, and a more narrow system boundary was thus applied. This at least partly explains the difference from our results.
We found significantly larger variations in transport to factory than in forest operations (logging and transport to forest road). The worst case scenario had 16 times as high fuel consumption as the best case scenario when it comes to transport by truck, while the differences in logging and transport in the forest were less than a factor of 5. The relative importance of these operations is comparable with findings from Sweden (Berg and Lindholm 2005). The diesel consumption in that study was 1.48–1.78 L/m3 for forest operations and 2.13–2.91 L/m3 for transport by truck. In addition, Berg and Lindholm included some transport by electric railway, which corresponded to a diesel consumption of up to 0.24 L/m3. Johnson and colleagues (2005) reported similar diesel consumption for logging operations in the Pacific Northwest of the United States (1.70 L/m3)12 but significantly higher consumption for road transportation (6.30 L/m3).
For the time being, the worst case scenario for forest operations is highly hypothetical. The estimated basic price (costs) of logging and transport by forwarders was NOK 280 per cubic meter, which would have made the logging unprofitable. The range for logging operations is thus even smaller than the scenarios indicate. This is not the case for transport by trucks, as it was based on empirical data. The differences from site to site are thus primarily a result of transport distances from the pile to the factory.
The data used in this study were a combination of process data and data based on I-O data. Six environmental impact categories were included (see table 1). The results for the human toxicity potential (HTP) categories are almost entirely based on I-O data. The HTP data are only included in table 1 and not in the figures. For estimating emissions from forest road construction and maintenance, data from the construction sector were used. Even though the data were adjusted through known purchases and refined with an SPA, the results might still be divergent from the real situation. In addition, the economic data for forest road construction are also uncertain (see Appendix S1 in the Supplementary Material on the Web). The data on impact from forest road construction should thus be used cautiously.
For improvements of the system, a further investigation of the trade-off between forest road construction and transport by forwarders should be performed. Long transport distances by forwarders make a significant contribution in the worst case scenario, but the impact from forest road construction is significant. The impact from forest road construction is, as pointed out above, uncertain. Also, impacts not included here, such as habitat loss due to fragmentation, should be considered in such analyses (Michelsen 2004, 2008). Other possibilities include assessing different logging techniques and also the trade-off between high planting frequency and natural regeneration. Planting is relatively costly (seedling production and planting summed up to almost 14% of the total costs) but shortens the rotation period and consequently makes logging in areas that are difficult to access less necessary.
The largest potential for improvements is in transport to a downstream user, as this caused almost half of the environmental impact. Here, several improvements could be foreseen. The loading factor, particularly on long distances, could be significantly increased, as there is almost no return cargo at present. Improved fuel efficiency and alternative fuels could also make significant improvements. Other transportation options, such as rail and boat, were not analyzed here. A transfer to these could make improvements but would require heavy investments in infrastructure. It must also be noted that the area in which ALLSKOG operates includes the most rural parts of Norway. Fuel consumption equal to what is found in Sweden and less than half of what is found in the Pacific Northwest of the United States indicates that the road transport is already comparatively efficient.
The possibilities to reveal potential environmental improvements also depend on the system boundaries. O'Rourke and colleagues (1996) stated that any system can be efficient as long as its definition is small enough. The converse may also apply; sometimes it is necessary to define the systems large enough to reveal the potential for improvement (Michelsen 2006). In the case presented in this article, it is obvious that the environmental performance for an average log delivered at factory gate will be improved if only the areas that are easiest to access are logged. Increased logging will therefore most likely increase the environmental impact from an average log, as less accessible areas also must be logged. This situation might, however, be totally different if timber is compared to other materials—for example, concrete in the construction industry. On the basis of this study, it is not possible to see whether timber logging is a significant part of the overall environmental problem of the construction industry (cf. Petersen and Solberg 2005; Puettmann and Wilson 2005; Sanness 2003) and should be improved, or whether logging is a part of the solution and should be increased to provide alternative materials (e.g., Deroubaix 2004; Lippke et al. 2004; Petersen and Solberg 2005).
Our results show that emission factors from forestry operations were probably underestimated in previous studies due to narrow system boundaries. In addition, there are large variations in emissions from forestry operations due to different log size, different transport distances in the forest, and, in particular, different transport distances from the pile to the factory. The sensitivity analysis used in this article clearly shows that average data on forestry operations should be used cautiously and avoided if possible.
One hundred Norwegian kronor ≈ Euro 11.95 ≈ US$14.70 in 2004.
One cubic meter (m3, SI) ≈ 1.31 cubic yards (yd3).
Round wood logs under bark refers to the volume of the logs under bark (true volume of the entire stem or a part of the stem excluding bark).
Based on questionnaires sent to all sawmills with an annual flow of more than 5,000 m3 in the region.
According to Sverre Thoresen, environmental corporate advisor in Norske Skog, in an interview in the magazine Norsk Skogbruk (Norwegian Forestry) in May 2006.
This project was partly funded by the Research Council of Norway through the project Productivity 2005—Industrial Ecology (NFR126567/230). We would like to thank the staff at ALLSKOG, in particular Magnus Mestvedt, for cooperating and providing data and also all others who have provided and estimated financial and environmental data for the different processes (see Appendix S1 in the Supplementary Material on the Web). We would also like thank the three anonymous referees, as well as LCA Editor Philippa Notten, for their valuable comments.
About the Authors
Ottar Michelsen is a postdoctoral researcher at the Department of Industrial Economics and Technology Management, Norwegian University of Science and Technology (NTNU), in Trondheim, Norway. Christian Solli is a researcher and Anders Hammer Strømman is an associate professor at the Industrial Ecology Programme at NTNU.