• Open Access

Effects of forest management on the carbon dioxide emissions of wood energy in integrated production of timber and energy biomass

Authors


Johanna Routa, tel. +358 10 211 3297, fax +358 10 211 3251, e-mail: johanna.routa@metla.fi

Abstract

The aim of this work was to study the sensitivity of carbon dioxide (CO2) emissions from wood energy to different forest management regimes when aiming at an integrated production of timber and energy biomass. For this purpose, the production of timber and energy biomass in Norway spruce [Picea abies (L.) Karst] and Scots pine (Pinus sylvestris L.) stands was simulated using an ecosystem model (SIMA) on sites of varying fertility under different management regimes, including various thinning and fertilization treatments over a fixed simulation period of 80 years. The simulations included timber (sawlogs, pulp), energy biomass (small-sized stem wood) and/or logging residues (top part of stem, branches and needles) from first thinning, and logging residues and stumps from final felling for energy production. In this context, a life cycle analysis/emission calculation tool was used to assess the CO2 emissions per unit of energy (kg CO2 MWh−1) which was produced based on the use of wood energy. The energy balance (GJ ha−1) of the supply chain was also calculated. The evaluation of CO2 emissions and energy balance of the supply chain considered the whole forest bioenergy production chain, representing all operations needed to grow and harvest biomass and transport it to a power plant for energy production. Fertilization and high precommercial stand density clearly increased stem wood production (i.e. sawlogs, pulp and small-sized stem wood), but also the amount of logging residues, stump wood and roots for energy use. Similarly, the lowest CO2 emissions per unit of energy were obtained, regardless of tree species and site fertility, when applying extremely or very dense precommercial stand density, as well as fertilization three times during the rotation. For Norway spruce such management also provided a high energy balance (GJ ha−1). On the other hand, the highest energy balance for Scots pine was obtained concurrently with extremely dense precommercial stands without fertilization on the medium-fertility site, while on the low-fertility site fertilization three times during the rotation was needed to attain this balance. Thus, clear differences existed between species and sites. In general, the forest bioenergy supply chain seemed to be effective; i.e. the fossil fuel energy consumption varied between 2.2% and 2.8% of the energy produced based on the forest biomass. To conclude, the primary energy use and CO2 emissions related to the forest operations, including the production and application of fertilizer, were small in relation to the increased potential of energy biomass.

Introduction

Forests act as both sinks and sources of carbon dioxide (CO2) (IPCC, 2000) and therefore they affect the atmospheric CO2 concentration and the global climate. This emphasizes the importance of understanding how the dynamics of the forest ecosystem and management interact in controlling the carbon balance in forests. This is especially the case when assessing the role of forest biomass in substituting fossil fuels in energy production.

Energy generation based on forest biomass is a part of the global carbon cycle. This energy is considered to be carbon neutral, because the combustion of biomass releases the same amount of CO2 as was captured in its growth (IPCC, 2000). This is true in the long term, however, in the short term CO2 and other greenhouse gases (GHG) are emitted in different phases of the supply of energy biomass. Fossil fuels are required in the production and harvesting of the biomass, its processing and handling, as well as in operating the bioenergy plants and transporting feedstock and biofuels. On the other hand, the management and harvest operations lead to changes in carbon stored in the forest ecosystems. Searchinger et al. (2008) and Melillo et al. (2009), for example, have questioned the carbon neutrality of renewable biomass, like forest biomass, in energy production owing to high indirect GHG emissions in different phases of the biomass supply for energy production.

Carbon in forests is bound to the soil and vegetation. Globally, forest soil is a remarkable carbon stock (Jobbagy & Jackson, 2000; Smith et al., 2006). In the boreal zone, its share of the total forest carbon stock is particularly significant (Liski et al., 2002). The growth and removal of biomass determine the carbon stock aboveground, while accumulating and decomposing litter determine the carbon stock in the soil (Johnson & Curtis, 2001; Yanai et al., 2001). In harvesting biomass, the total carbon balance (trees, soil) is disturbed, because the litter accumulation on the soil will substantially change due to the harvesting of branches and foliage, stumps and roots and top parts of the stems. Consequently, the amount of organic matter in the soil decreases in the long term.

In the Nordic countries, like Finland, the harvesting of energy biomass (logging residues) is integrated with the harvesting of timber (pulpwood, sawlogs). This integration, with the production of traditional industrial wood, makes the production of energy biomass cost-efficient. When producing industrial timber such as sawlogs, management usually aims at fast diameter growth. This is achieved if spacing in the tree stands is wide throughout the rotation, with stocking lower than needed to maximize the biomass production. It is still under discussion regarding how to optimize the spacing when aiming at integrating the production of timber and wood energy based on forest biomass. In this respect, the spacing in planting and thinning used in timber production is probably too wide for the efficient production of energy biomass. On the other hand, the optimal rotation for timber production extends over several decades, e.g. in Finland a rotation of 80 years is widely used in timber production for Norway spruce and Scots pine in southern and central boreal conditions on fertile and medium-fertile sites (Recommendations for Forest Management in Finland, 2006). In the integrated production of timber and energy biomass, the rotation length is determined by the timber production, because the value of timber still greatly exceeds that of energy biomass.

Generally speaking, the addition of nitrogen (N) to the boreal forest ecosystem will enhance the growth of trees and thus, increase the litter fall onto the soil (Aber et al., 1989, 1998; Hyvönen et al., 2007). This is especially the case for the boreal forests in northern Europe, where the limited supply of N greatly restricts forest growth (e.g. Tamm, 1991; Vitousek & Howarth, 1991; Magill et al., 1997). However, the growth response to the N fertilization is clearly lower in stands growing on soils with a high N supply than on soils with a low supply (Ingerslev et al., 2001). This implies that the effect of N fertilization on carbon sequestration in trees depends on both the dose of the N addition and the site fertility.

N fertilization may affect the pool of soil organic carbon (SOC) through increased litterfall as a consequence of increasing foliage mass and the higher production of other organs. The role of fine roots is more complicated, because it is unclear whether the increased N availability influences their growth (Hyvönen et al., 2008). On the other hand, the N fertilization and deposition may affect SOC storage by enhanced or reduced heterotrophic respiration as a response to litter C/N ratio. Fertilization experiments show that the addition of N may decrease microbial biomass and activity in the forest soil, resulting in slower decomposition of the soil organic matter (Södeström et al., 1983; Martikainen et al., 1989). Thus, the addition of N may increase the amount of carbon both in the humus layer and mineral soil (Mäkipää, 1995; Hyvönen et al., 2008).

In decision making in forestry, growth and yield models are essential tools, e.g. they allow the analyses of the sensitivity of stem wood production to different silvicultural treatments (e.g. spacing, thinning, fertilization) and varying environmental conditions (e.g. Kellomäki et al., 1992; Hynynen et al., 2005). Such models could also be used to support the identification of optimal management in forest planning (Hynynen et al., 2005; Hyytiäinen et al., 2006). This holds also for the problems regarding how the harvesting of forest biomass for energy production affects the long-term dynamics of the forest ecosystem and how the use of forest biomass can reduce the GHG emissions in energy production. In this regard, environmental life cycle assessment (LCA) can be used to assess how the use of energy biomass effects the environment, for example, regarding the CO2 emissions and leaching of nutrients over the whole biomass supply chain (e.g. Consoli et al., 1993; Lindfors et al., 1995; UNEP, 2003).

LCA is a tool for quantifying burdens for the environment over the whole life cycle of a product, material, service or facility in relation to a functional unit. Its strength is that a consistent tool is used, which quantifies all possible environmental burdens in relation to a functional unit. Its weakness is that the results have a low spatial and temporal resolution, and that social and economic aspects are not taken into account (Owens, 1997; Udo de Haes et al., 2004). Despite these limitations, LCA facilitates the comparative analysis of how different management strategies in the production of energy biomass may affect the forest environment and what are the benefits and drawbacks compared with the use of fossil fuels (e.g. coal) (Cherubini et al., 2009).

The aim of this work was to study the sensitivity of CO2 emissions of wood energy to different forest management regimes when aiming at an integrated production of timber and energy biomass. For this purpose, the production of timber and energy biomass in Norway spruce [Picea abies (L.) Karst] and Scots pine (Pinus sylvestris L.) stands was simulated using an ecosystem model (SIMA) on sites of varying site fertility under different management regimes, including various thinning and fertilization treatments over a fixed simulation period of 80 years. The simulations included timber (sawlogs, pulp), energy biomass (small-sized stem wood) and/or logging residues (top part of stem, branches and needles) from first thinning, and harvesting residues and stumps and roots from final felling for energy production. In this context, a life cycle analysis/emission calculation tool was used to assess the CO2 emissions per unit of energy (kg CO2 MWh−1) which was produced based on the use of wood energy. Wood energy refers to energy based on forest biomass (energy biomass). The energy balance (GJ ha−1) of the whole supply chain was also calculated. The evaluation of CO2 emissions and energy balance of the supply chain considered the whole forest bioenergy production chain, representing all operations needed to grow and harvest biomass and transport it to a power plant for energy production.

Material and methods

Outlines of the ecosystem model and its simulations

Outlines of the ecosystem model. In this work, the ecosystem model SIMA (Kellomäki et al., 1992; Kolström, 1998) was used. In the simulations, the growth of a tree is based on diameter growth, which is the product of the potential diameter growth and environmental factors. The model incorporates four subroutines describing the site conditions; i.e. temperature sum (degree days), within-stand light conditions, soil moisture and soil N. These factors affect the demographic subroutines (birth, growth, death); i.e. G=Go×M1×⋯×Mn, where G is the growth and/or regeneration, Go the growth and/or regeneration in the optimal temperature, light, N and water conditions and M1, …, Mn the multipliers for different environmental factors. The death of the trees is determined by the crowding with the consequent reduction in growth, which determines the risk for a tree to die at a given moment. Furthermore, the random mortality was included, with a small fraction of the trees dying each year.

Litter and dead trees end up on the soil to be decomposed, with the resulting release of N in the long run. In decomposition, soil organic matter is divided into litter (divided into foliage, stem, branches, roots) and humus. In both cases, the CO2 emitted in relation to the loss of weight in such a way that the carbon content of 50% of the weight was assumed. Decomposition of litter is affected by soil moisture and temperature conditions in terms of annual evapotranspiration. This, along with the quality of litter (carbon, N and lignin contents), drives the decomposition rate. When the N concentration of any litter cohort reaches a critical concentration, the litter cohort is converted into humus. Decomposition of humus is also a function of annual evapotranspiration. Furthermore, N deposited from the atmosphere and added as fertilizer affects the amount of available N (Kellomäki et al., 1992). The simulation is based on the Monte Carlo simulation technique; i.e. certain events, such as the birth and death of trees, are stochastic events. Each time when such an event is possible (e.g. it is possible for a tree to die every year) the algorithm selects whether or not the event will take place by comparing a random number with the probability of the occurrence of the event. The probability of an event is a function of the state of the forest ecosystem at the time when it is possible. Each run of a Monte Carlo code is one realization of all possible time courses of the forest ecosystem. Therefore, the simulation of succession in the forest ecosystem must be repeated several times in order to determine the general tendency of variations over time for the forest ecosystem.

The model has been parameterized for Scots pine (P. sylvestris L.), Norway spruce [P. abies L. Karst], birch (Betula pendula Roth. and Betula pubescens Ehrh), aspen (Populus tremula L.) and grey alder (Alnus incana (L) Moench.) growing between the latitudes 60 and 70°N and longitudes 20 and 32°E within Finland (Kellomäki et al., 1992; Kellomäki & Kolström, 1993; Kolström, 1998). The model is run on an annual basis and its computations are applied to an area of 100 m2. The procedure for treatments includes thinning, fertilization, rotation length and harvesting of timber and energy biomass (foliage, branches, stumps and top part of the stem not suitable for timber). In the case of fertilization, the effect of the total amount of fertilizer (NT, kg ha−1) added in a single fertilizing event was allocated over several years in terms of addition to the available N. The N addition in the fertilizing event will gradually reduce and the effect of fertilizing will finally disappear. Annually, the fraction of fertilizer [F(t)] from NT for the year t is

image(1)

where k is the time in years since fertilization, n is the length of time in years with any addition to available N [F(k)<0.01] and the factor B is a function of the total amount of fertilizer

image(2)

The values of the parameters were estimated on the basis of Jonsson (1978) and Kukkola & Saramäki (1983). In the year k, the amount of fertilizer [NA(k), kg ha−1] made available for growth is NA(k)=F(k) × NT. The addition of N to the available N through fertilization is demonstrated in Fig. 1.

Figure 1.

 Change in the available nitrogen as a function of the amount of fertilizer and the time since fertilizer application.

Performance of the model. The validation of the model has been previously discussed in detail by Kolström (1998) and Kellomäki et al. (2008). Furthermore, Mäkipääet al. (1998) have earlier demonstrated that the long-term growth response of trees to the addition of N in fertilization is in good agreement with the measured responses to the N additions in field conditions. Nevertheless, we further studied the performance of the model by comparing the simulations for the growth of Norway spruce and Scots pine to the parallel simulations with the statistical growth and yield by the Motti model (Hynynen et al., 2005).

Growth dynamics in the Motti model are based on tree growth data representing a large number of sample plots (forest inventory plots) and this model comparison provides a good benchmark to assess the performance of the SIMA model throughout Finland. In this model validation work, we calculated the growth (m3 ha−1 a−1) for 13 different sites throughout Finland on medium-fertile sites occupied by Norway spruce and Scots pine (Table 1). The temperature sum varied between 360 and 650 degree days (d.d.) in the north (Ivalo) and 1300–1370 d.d. in the south (Helsinki). Planting density was 1800 trees ha−1 in Norway spruce and 2000 trees ha−1 in Scots pine stands. The simulation period was 80 years and the thinning rules followed those currently recommended for different tree species, site fertility types and regions of Finland (Recommendations for forest management in Finland, 2006). Fertilization was performed twice during the rotation by adding 150 kg N ha−1 at the time of first thinning and the second time 10 years later. Figure 2 shows a fairly good correlation between the simulated growth values for the Motti and SIMA models regardless of tree species. However, the SIMA model seems to give slightly (10–20%) lower growth values compared with the Motti model.

Table 1.   Location, temperature sum and precipitation of simulation sites throughout Finland for model comparison work
 LocationTemperature sum (d.d.)Precipitation (mm a−1)
Helsinki60°19′N,24°55′E1300–1370600–700
Tampere61°26′N,23°40′E1200–1250500–600
Lappeenranta61°04′N, 28°18′E1250–1300600–700
Jyväskylä62°23′N, 25°49′E1150–1200500–700
Joensuu62°39′N, 29°37′E1150–1200500–700
Kruunupyy63°44′N, 23°31′E1050–1100400–500
Kajaani64°16′N, 27°45′E1000–1050500–600
Oulu64°58′N, 25°36′E1050–1100400–500
Suomussalmi64°52′N, 28°56′E850–900500–600
Kuusamo65°59′N, 28°56′E800–850500–600
Kemi65°49′N, 28°59′E950–1000500–600
Rovaniemi66°35′N, 26°05′E850–900400–500
Ivalo68°37′N, 27°24′E360–650400–500
Figure 2.

 Relationship between simulated mean annual growth of Scots pine and Norway spruce on different sites throughout Finland by the SIMA and Motti models for unfertilized and fertilized (2 × 150 kg N ha−1) stands.

The performance of the SIMA model was further studied against the measured data for the annual mean forest growth over the period 1996–2003 for the ten southernmost Forest Centres in Finland by the National Forest Inventory (Finnish Forest Research Institute, 2005). The corresponding growth for Norway spruce and Scots pine was calculated with the SIMA model based on a total of 1855 permanent sample plots. However, only plots on the upland mineral soils were included in analyses. The number of plots for each Centre used in the analyses followed their share of the total forest area in Finland. Figure 3 shows that there is a close correlation between the simulated and measured values.

Figure 3.

 Simulated growth values of Scots pine and Norway spruce for different Forestry Centres against the growth values provided by the National Forest Inventory between the years 1996 and 2003 (Finnish Forest Research Institute, 2005).

Simulations of management regimes. The simulations were made for the Joensuu region in eastern Finland (62′39° N, 29 ′37° E). The mean temperature sum in this area is 1150–1200 d.d. Simulations for Norway spruce were performed for the sites of high [Oxalis-Myrtillus type (OMT)] and medium fertility [Myrtillus type (MT)], and for Scots pine for the sites of medium (MT) and low fertility [Vaccimum type (VT)] (Cajander, 1926). The initial soil organic matter was 69 Mg ha−1 on the OMT site, 67 Mg ha−1 on the MT site and 59 Mg ha−1 on the VT site regardless of tree species (Kellomäki et al., 2008). In the simulations, an annual N deposition of 6.0 kg ha−1 was used, which is the long-term mean for the simulation site (Järvinen & Vänni, 1994). The average diameter of the seedlings at the beginning of the simulation was 2 cm at the height of 1.3 m aboveground, with the stand density varying between 1800 and 4500 seedlings ha−1 depending on the management regimes used in a particular simulation.

In the simulations, the thinning rules followed those currently recommended for the different tree species, site types and regions of Finland (Recommendations for forest management in Finland, 2006). The basic idea of thinning recommendations is that whenever a given basal area threshold at a certain dominant height (i.e. the average of the heights of the 100 tallest trees) is reached, thinning is performed and the basal area is reduced to the recommended value. Thinning was performed from below, regardless of the management regime used. In addition, N fertilizer was applied either two or three times (or not at all) over the 80-year simulation period depending on the management regime. Altogether, 30 different management regimes were used for each tree species on two sites (Table 2).

Table 2.   Alternative management regimes used in simulations for Norway spruce (NS) and Scots pine (SP)
 Management regimeStand density
(trees ha−1) after
precommercial
thinning/first
commercial thinning
with hdom (m)
Possible N
fertilization treatments
are marked as follows:
150 kg N ha−1,
twice (f2) or three (f3) times during rotation
  1. All simulations were also performed without any N fertilizer treatment.

1Basic thinning, baseline precommercial stand densityNS: 1800/900 with hdom 13f2, f3
SP: 2000/1000 with hdom 13
2Basic thinning, with moderate precommercial stand densityNS: 2300/900 with hdom 13f2, f3
SP: 2500/1000 with hdom 13
3Early energy wood thinning with moderate dense precommercial standNS: 2500/1800 with hdom 8f2, f3
SP: 3000/1800 with hdom 8
4Late energy wood thinning with moderate dense precommercial standNS: 2500/1800 with hdom 10f2, f3
SP: 3000/1800 with hdom 10
5Early energy wood thinning with dense precommercial standNS: 3000/1800, with hdom 8f2, f3
SP: 3500/1800, with hdom 8
6Late energy wood thinning with dense precommercial standNS: 3000/1800, with hdom 10f2, f3
SP: 3500/1800, with hdom 10
7Early energy wood thinning with very dense precommercial standNS: 3500/1800, with hdom 8f2, f3
SP: 4000/1800, with hdom 8
8Late energy wood thinning with very dense precommercial standNS: 3500/1800, with hdom 10f2, f3
SP: 4000/1800, with hdom 10
9Early energy wood thinning with extremely dense precommercial standNS: 4000/1800, with hdom 8f2, f3
SP: 4500/1800, with hdom 8
10Late energy wood thinning with extremely dense precommercial standNS: 4000/1800, with hdom 10f2, f3
SP: 4500/1800, with hdom 10

The basic management regimes (case numbers 1–2), aiming at the production of timber, included a precommercial thinning to density of 1800–2500 trees ha−1, with the first commercial thinning at a dominant height of 13 m to a density of 900–1000 trees ha−1. The regimes aiming at the integrated production of timber and energy biomass (case numbers 3–10) included precommercial thinning to a density of 2500–4500 trees ha−1. In this case, the energy wood thinning was performed at a dominant height of 8 or 10 m to a density of 1800 trees ha−1. Second and third (if required) thinning were performed following the thinning recommendations based on the basal area and dominant height. In energy wood thinning, all the industrial-size wood and logging residues were used as energy biomass. Additionally in basic management regimes logging residues were collected in first thinning. In final felling, logging residues (top of stem, branches, needles) and stumps (and coarse roots) were collected in all management regimes.

In a single fertilizer treatment, the application of N was 150 kg N ha−1. The number of applications of fertilizer varied depending on the management regime (no fertilizing, fertilizing two or three times during the rotation). The first fertilization was performed at the time of first thinning; i.e. at the dominant height of 8, 10 or 13 m. Thereafter, fertilization was performed once or twice at intervals of 12 and 10 years depending on the management regime. Regardless of the management regime (i.e. different spacing and thinning treatments with and without fertilizing), the 80-year rotation length was used in all the simulations. This was performed to determine how large differences could exist between different management regimes for studied variables over a rotation length, which is typically used for Scots pine and Norway spruce in Finnish conditions.

CO2 emission calculation tool used in the study

The life cycle analysis/emission calculation tool (Kilpeläinen et al., 2011) was used to calculate the energy used and the CO2 emissions for different phases of forest biomass production and its use in energy production. The CO2 emitted during each phase was included through the energy consumption (diesel or electricity) in regard to the average productivity of the machinery used.

The calculations included (Fig. 4): (i) the establishment of seedling stand with production and transportation of seedlings, site preparation and planting, (ii) the management and harvest operations in thinning, final felling and forest haulage, and the production, transportation and application of fertilizers and (iii) the long-distance transportation of biomass and its chipping at the storage area of the power plant (Fig. 4). Emissions from manufacturing of fertilizers include N2O emissions converted to carbon dioxide equivalents (CO2e). Emissions in the application of fertilizers exclude the N2O emissions from fertilized soil. Additionally, the emissions from the manufacturing and maintenance of machinery were also excluded.

Figure 4.

 System boundaries for calculating carbon and energy balances from forest bioenergy supply chains.

The calculations were performed on an annual basis as indicated by the following equation:

image(3)

where Cnet is a sum of the carbon uptake in growth (Cseq) and the carbon emissions from management and harvesting (Cman), decomposition of soil organic matter (Cdecomp) and from the burning of energy biomass (Charv). In calculations, the uptake (Cseq) has negative values (carbon is flowing from atmosphere to forest) and the emissions Cman, Cdecomp and Charv have positive values (carbon is flowing from forests to atmosphere).

The calculations included biomass harvested from energy wood thinning [logging residues (top part of stem, branches, needles)] and small-sized stem wood and, if basic thinning was applied, logging residues from first thinning. Logging residues from final felling and stumps and roots were also included. The loss of needles in harvesting was assumed to be 30% [Eqn (3)]. Emissions were allocated according to the proportion of biomass components during the rotation. The unit for emissions and uptake of CO2 was g CO2 m2 a−1.

Finally, the net emissions for wood energy were converted into CO2 emissions per produced MWh (kg CO2 MWh−1) over the 80-year rotation period. In this calculation, the energy content of the fuel (diesel) was 38.6 MJ L−1 and the carbon content 0.857 kg L−1. Wood density of 400 kg m−3 was used in the calculations and C content in the dry biomass was assumed to be 50% of the dry mass. This was also assumed in the combustion of the biomass.

In this work, energy balance was defined as the difference between the energy produced per hectare of forest and the energy necessary to produce it. The latter one includes the energy used in the establishment of the seedling stand through the production and transportation of seedlings, site preparation and planting, the management operations with thinning, final felling, forest haulage and production, transportation of fertilizers and transportation of biomass as well as its chipping at the storage of the power plant. The energy produced is the energy obtained when the biomass is burnt for energy.

Results

Stem wood production between different management regimes

Norway spruce on most fertile and medium-fertile sites. On the most fertile site (OMT), the stem wood production for Norway spruce (including saw wood, pulp, m3 ha−1) varied in the range of 87–108% compared with the average for all the management regimes without fertilization. When applying 150 kg N ha−1 two or three times during the rotation the stem wood production (m3 ha−1) increased up to 6% and 9%, respectively (average 4% and 7%) compared with otherwise similar management regimes but without fertilization.

On the medium-fertile site (MT), the stem wood production of Norway spruce (m3 ha−1) varied in a similar range; 93–106% compared with the average for all the management regimes without fertilization. Again, fertilization clearly increased the stem wood production. When applying 150 kg N ha−1 two or three times during the rotation, the stem wood production (m3 ha−1) increased, i.e. up to 8% and 12%, respectively (average 6% and 10%) compared with otherwise similar management regimes but without fertilization.

Regardless of site fertility type, the stem wood production during the rotation was the highest for Norway spruce when the precommercial stand density (stand density after precommercial thinning at the height of 3–5 m) was very or extremely dense and late energy wood thinning was applied together with fertilization of 150 kg N ha−1 three times during the rotation (management regimes 10f3 and 8f3), i.e. being 14% and 11% higher than the averages for the OMT and MT sites, respectively (range 371–500 and 347–432 m3 ha−1). In addition, the amount of sawlogs was also the highest when the precommercial stand density was very or extremely dense. In general, fertilization clearly increased the stem wood production. Similarly the amount of energy biomass [including logging residues (top part of stem, branches, needles), small-sized stem wood and stump wood] was the highest when the precommercial stand density was very or extremely dense, in addition fertilization increased the amount of energy biomass (range 83–125 and 81–113 Mg ha−1, respectively) in the most and medium-fertile sites (Appendix S1).

Scots pine on medium- and low-fertility sites. On the medium-fertility site (MT), the stem wood production of Scots pine (including saw logs, pulp, energy biomass, m3 ha−1) varied in the range of 93–107% compared with the average for all the management regimes without fertilization. In general, fertilization clearly increased the stem wood production, as it did for Norway spruce. When applying fertilizer of 150 kg N ha−1 two or three times during the rotation, the stem wood production (m3 ha−1) increased the most, i.e. up to 5% and 6% (average 3% and 4%), respectively, compared with otherwise similar kinds of management regimes but without fertilization.

On the low-fertility site (VT), the stem wood production for Scots pine (m3 ha−1) varied in a similar range, 97–105%, compared with the average for all the management regimes without fertilization. Again, fertilization clearly increased the stem wood production as it did on the medium-fertile site. When applying 150 kg N ha−1 two or three times during the rotation, the stem wood production (m3 ha−1) increased the most, even more than on the medium-fertile site, i.e. up to 14% (low fertility) and 22% (medium-fertile site) (average 10% and 15%, respectively), compared with otherwise similar management regimes but without fertilization.

Regardless of site fertility type, the stem wood production during the rotation was the highest for Scots pine when the precommercial stand density was very or extremely dense and late energy wood thinning was applied together with 150 kg N ha−1 three times during the rotation (management regimes 10f3 and 6f3), i.e. being 9% and 13% higher than the averages for the medium- and low-fertility sites, respectively (range 317–432 and 239–323 m−3 ha−1). In Scots pine stands the amount of sawlogs was the highest when the basic management regime was applied with fertilization of 150 kg N ha−1 three times during the rotation. In general, fertilization also clearly increased the stem wood production in Scots pine. The amount of energy biomass [including logging residues (top part of stem, branches, needles), small-sized stem wood and stump wood] was the highest when the precommercial stand density was very or extremely dense, and fertilization increased the amount of energy biomass in Scots pine stands (range 53–68 and 46–68 Mg ha−1, respectively) on medium- and low-fertility sites (Appendix S2).

Net CO2 emissions of wood energy between different management regimes

Net CO2 emissions of Norway spruce. On the most fertile site (OMT) the net CO2 emissions for Norway spruce varied in the range of 84–130% compared with the average for all management regimes (on average 117.7±17.8 kg CO2 MWh−1), if fertilization was not applied (Fig. 5). When applying 150 kg N ha−1 two or three times during the rotation, the net CO2 emissions decreased up to 24% and 32%, respectively (average 16% and 22%) compared with similar management regimes without fertilization (average 97.5±19.8 kg CO2 MWh−1) (Fig. 5).

Figure 5.

 Relative differences (%) in net CO2 emissions in Norway spruce grown on fertile (OMT) and medium-fertile (MT) sites for different management regimes without fertilization compared with the average overall these management regimes (top) and corresponding differences between management regimes with fertilization of 150 kg N ha−1 two or three times during rotation compared with corresponding management regimes without fertilization (bottom). 1 and 2 are basic management regimes and 3–10 are wood energy management regimes with varying stand densities.

On the medium-fertility site (MT), the net CO2 emissions for Norway spruce varied in a similar range, 86–132%, compared with the average for all the management regimes without fertilization (average 110.2±18.2 kg CO2 MWh−1) (Fig. 5). Again, fertilization clearly decreased the net emissions. When applying 150 kg N ha−1 two or three times during the rotation, the net CO2 emissions decreased the most, even more than on the most fertile site; i.e. up to 36% and 55%, respectively (average 22% and 39%) compared with similar management regimes without fertilization (average 85.3±17.4 kg CO2 MWh−1) (Fig. 5).

Overall the different management regimes, the net CO2 emissions varied in the range of 69–146% and 69–156% compared with the average on the most fertile and medium-fertile sites, respectively. The net CO2 emissions were the lowest on the most fertile site when precommercial stand density was extremely high and late energy wood thinning was applied together with fertilization of 150 kg N ha−1 three times during the rotation (management regime 10f3). In this case, the emissions were 31% lower (72 kg CO2 MWh−1) than the averages for all the management regimes (range 72–152 kg CO2 MWh−1).

On the medium-fertility site (MT), the net CO2 emissions were the lowest when precommercial stand density was very dense and late energy wood thinning was applied together with the fertilization of 150 kg N ha−1 three times during the rotation (management regime 8f3). In this case, the emissions were 31% lower (65 kg CO2 MWh−1) compared with the averages for all the management regimes (range 65–146 kg CO2 MWh−1). In general, fertilization clearly decreased the net CO2 emissions for Norway spruce.

Net CO2 emissions of Scots pine. On the medium-fertile site (MT), for Scots pine, the net CO2 emissions varied in the range of 92–110% compared with the average for all management regimes (average 99.3±6.7 kg CO2 MWh−1) if no fertilization was applied (Fig. 6). In general, the fertilization clearly decreased the net CO2 emissions, as it did in the case of Norway spruce. When applying 150 kg N ha−1 two or three times during the rotation, the net CO2 emissions decreased by 17% and 25% (average 11% and 16%, respectively) compared with similar management regimes without fertilization (average 87.6±7.6 kg CO2 MWh−1) (Fig. 6).

Figure 6.

 Relative differences (%) in CO2 emissions in Scots pine on medium-fertile (MT) and less fertile (VT) sites for different management regimes without fertilization as compared with the average overall these management regimes (top) and corresponding differences between management regimes with fertilization of 150 kg N ha−1 two or three times during rotation compared with corresponding management regimes without fertilization (bottom). 1 and 2 are basic management regimes and 3–10 are wood energy management regimes with varying stand densities.

On the low-fertility site (VT), the net CO2 emissions varied in the range 84–109% compared with the average for all the management regimes without fertilization (average 176.5±12.6 kg CO2 MWh−1) (Fig. 6). Again, fertilization clearly decreased the emissions. When applying 150 kg N ha−1 two or three times during the rotation, the net CO2 emissions decreased even more than on the medium-fertile site; i.e. 33% and 46% of the similar management regimes without fertilization (average 143.2±18.3 kg CO2 MWh−1) (Fig. 6).

Overall the management regimes (unfertilized and fertilized) the net CO2 emissions varied in the range of 85–119% and 78–125% compared with the average on the medium- and low-fertility sites, respectively. The net CO2 emissions were the lowest for Scots pine on the medium-fertile (MT) site, when precommercial stand density was extremely dense and late energy wood thinning was applied together with fertilizer of 150 kg N ha−1 three times during the rotation (management regime 10f3). In this case, the emissions were 15% lower (78 kg CO2 MWh−1) than the averages of all management regimes (range 78–109 kg CO2 MWh−1).

On the low-fertility site (VT), for Scots pine, the net CO2 emissions were the lowest when precommercial stand density was very dense and early energy wood thinning was applied together with fertilizer of 150 kg N ha−1, applied three times during the rotation (management regime 7f3), being 22% lower (120 kg CO2 MWh−1) than the averages for all management regimes (range 120–192 kg CO2 MWh−1). In general, fertilization clearly decreased the net CO2 emissions in Scots pine. Generally speaking, the net CO2 emissions were, on average, 10% lower on the medium-fertile sites compared with the most fertile sites for Norway spruce, on average 41% lower for medium-fertile Scots pine stands compared with pine stands of low fertility (Fig. 7). Fertilization and high precommercial stand density have positive effects on both stem wood production and forest biomass production for energy, and they are also indicated by the CO2 balance and the net CO2 emissions of the bioenergy chain with both tree species in all fertility types (Fig. 8).

Figure 7.

 CO2 emissions (kg CO2 MWh−1) of forest ecosystem over a 80-year rotation, when wood energy has been chipped and burnt at power plant with different management scenarios with and without fertilization.

Figure 8.

 Effect of management on stem wood production (including sawlogs, pulp and wood energy) and carbon dioxide emissions on Norway spruce and Scots pine stands. Numbers refers to different management regimes: 1 and 2 are basic management regimes and 3–10 are wood energy management regimes with varying densities. Symbols f2 and f3 with number means fertilization two or three times during the rotation.

Energy balance between different management regimes

Energy balance for Norway spruce. On the most fertile site (OMT) for Norway spruce, the energy balance varied in the range of 67–117% compared with the average for all management regimes without fertilization (average 1762±304 GJ ha−1). When applying fertilizer of 150 kg N ha−1 two or three times during the rotation, the energy balance increased up to 5% and 14% (average 2% and 6%), respectively, compared with the similar management regimes but without fertilization (average 1827±296 GJ ha−1).

On the medium-fertile site (MT), the energy balance for Norway spruce varied in a similar range; i.e. 72–114% compared with the average for all the management regimes without fertilization (average 1621±234 GJ ha−1). When applying fertilizer of 150 kg N ha−1 two or three times during the rotation, the energy balance increased up to 6% and 12%, respectively (average 1707±232 GJ ha−1).

Overall the different management regimes (including all unfertilized and fertilized ones), the energy balance varied in a range of 65–120% and 70–118% compared with the average on the most fertile and medium-fertile sites, respectively. The energy balance was the highest for Norway spruce on the most fertile site, when precommercial stand density was extremely dense and late energy wood thinning was applied together with fertilization of 150 kg N ha−1 three times during the rotation (management regime 10f3). In this case, the balance was 20% higher (2162 GJ ha−1) than the averages of all management regimes.

On the medium-fertility site (MT), the energy balance was the highest when precommercial stand density was very dense and late energy wood thinning was applied together with fertilization of 150 kg N ha−1 three times during the rotation (management regime 8f3); i.e. 18% higher (1972 GJ ha−1) than the averages of all management regimes. In general, the fertilization improved the energy balance in Norway spruce stands. The consumption of energy in the supply chain of forest bioenergy (all activities requiring fossil energy inputs, during stand growth as well as during harvesting and transportation of harvested products to the power station) was only 2.2–2.5% of the energy content of the harvested biomass.

Energy balance for Scots pine. On the medium-fertility site (MT), for Scots pine, the energy balance varied in the range of 90–115% compared with the average for all management regimes without fertilization (average 1008±88 GJ ha−1). When applying 150 kg N ha−1 two or three times during the rotation, the energy balance increased up to 3% and 4% (average 0.3% and 0.4%, respectively) compared with the similar management but without fertilization (average 1011±73 GJ ha−1). On the low-fertility site (VT), the energy balance of the Scots pine stands varied in the range 87–114% compared with all the management regimes without fertilization (average 812±78 GJ ha−1), when applying 150 kg N ha−1 two or three times during the rotation, the energy balance increased up to 16% and 21%, respectively (average 926±115 GJ ha−1).

Overall the different management regimes (unfertilized and fertilized), the energy balance varied in the range of 90–115% and 78–125% compared with the average on the medium- and low-fertility sites, respectively. The energy balance was the highest for Scots pine on the medium-fertility site, when precommercial stand density was extremely dense and late energy wood thinning was applied (management regime 10); i.e. 15% higher (1164 GJ ha−1) than the averages of all managements.

On the low-fertility site (VT), the energy balance was the highest for Scots pine when precommercial stand density was extremely dense and late energy wood thinning was applied together with fertilization of 150 kg N ha−1 three times during the rotation (management regime 10f3); i.e. 22% higher (1089 GJ ha−1) than the averages of all management regimes. In general, the fertilization clearly improved the energy balance in Scots pine stands in low-fertility sites, but in medium-fertile sites the energy balance was higher without fertilization. The consumption of energy in the supply chain of forest bioenergy was only 2.2–2.8% of that obtained when biomass was used in energy production.

Discussion and conclusions

The aim of this study was to analyze the effects of forest management on the carbon sink and source dynamics of the boreal forest ecosystem, with implications for CO2 emissions and energy balance of the supply chain, when aiming at integrated production of timber and energy biomass for Norway spruce and Scots pine on sites of varying site fertility. The calculation of net CO2 emissions for different management regimes provides information on the environmental impacts of the different regimes with regards to the use of forest biomass in energy production.

This study used a forest ecosystem model and an emission calculation tool (Kilpeläinen et al., 2011), which utilizes the results from the ecosystem model. The validity of the ecosystem model (SIMA) has earlier been discussed in detail (Kolström, 1998; Kellomäki et al., 2008). Furthermore, Routa et al. (2011b) also recently showed that the model was capable of simulating the dynamics and the consequent growth of Norway spruce and Scots pine in the boreal conditions. The emission calculation system boundaries included the whole chain of the forest bioenergy production, representing operations needed to produce and harvest biomass and transport them to a power plant for energy production.

Altogether 30 management regimes were considered in the simulations, representing varying precommercial and commercial thinning (timing and intensity) and N fertilization (number) treatments for each species on two sites with varying site fertility (on mineral soil). These management regimes were chosen on the basis of previous work by Routa et al. (2011b), which analyzed the impacts of thinning and fertilization treatments on timber and energy biomass production in Norway spruce and Scots pine. In that work, fertilization and high precommercial stand density were found to increase stem wood production in both Norway spruce and Scots pine, and the most in low-fertility sites (see Routa et al., 2011b).

In this work, the precommercial stand density ranged from 1800 to 2500 (representing basic management regimes) up to 4500 trees ha−1 (representing integrated management for timber and energy biomass). For the basic management regimes, the first commercial thinning was typically performed at a dominant height of 13 m regardless of tree species and site fertility, with the logging residuals (top part of stem, branches and needles) being harvested for wood energy. For the regimes aiming at the integrated production of timber and energy biomass, energy wood thinning was performed at a dominant height of either 8 or 10 m regardless of tree species and site fertility type (resulting in stand density of 1800 trees ha−1). In energy wood thinning, logging residues and also industrial-sized stem wood were used for wood energy. The other possible thinnings followed the currently applied thinning recommendations for Finnish conditions regardless of the management objectives (logging residues were not harvested). In the final felling, at age 80 years, logging residues and stumps and roots were extracted for wood energy.

A fixed rotation length of 80 years was used in this work, regardless of management regime. This was because this rotation length is widely used in Finnish forestry to maximize the production of sawlogs as the main objective in the management. On the other hand, this rotation length has been found to be quite optimal for reducing the emissions of energy production in Norway spruce on OMT and MT sites (Routa et al., 2011a). Nevertheless, a shorter rotation length would possibly be more optimal from a timber production point of view but not for the production of logging residues for energy biomass. There is work still to be performed for determining the optimal rotation for the combined production of timber and energy biomass.

Fertilization and high precommercial stand density clearly increased the stem wood production (i.e. sawlogs, pulp and small-sized stem wood), but also the amount of logging residuals and stump wood and roots for energy use. However, there were clear differences between Norway spruce and Scots pine, as well as between sites for both species. On the low-fertile site (VT) for Scots pine, fertilization increased stem wood production and energy balance the most, while on the medium-fertile site (MT) for Norway spruce the CO2 net emissions decreased the most due to fertilization.

The lowest net CO2 emissions of wood energy use over the rotation were found for Norway spruce, regardless of site fertility, if the precommercial stand density was extremely or very dense and late energy wood thinning was performed with N fertilization of 150 kg N ha−1 three times during the rotation. The difference between the lowest and the highest CO2 net emissions without fertilization were 36% and 35% at the most fertile and medium-fertile sites for Norway spruce, respectively. In Scots pine, the corresponding values were 16% and 23% for medium- and low-fertile sites. The net CO2 emissions per unit of energy for Scots pine on the low-fertile site were 41% higher than on the medium-fertile site. This was because the growth in the latter case was higher than in the former.

The positive effects of fertilization and high precommercial stand density on both stem wood production and forest biomass production for energy are also indicated by the CO2 balance and the net CO2 emissions of the bioenergy chain. The emissions were the highest, on average, without fertilization as related to lower forest productivity. However, there were clear differences between Scots pine and Norway spruce regarding the relationship between stem wood productivity and emissions for the same site (MT), as well as between sites for both species.

The reduction of CO2 emissions per unit of energy could partly be explained by the increased litter production. The addition of N may also decrease microbial biomass and activity, resulting in slower decomposition of soil organic matter (Martikainen et al., 1989, Mäkipääet al., 1998). However, these impacts were not considered in this study. On the other hand, the harvesting of logging residues may reduce the availability of nutrients, which in turn may affect the long-term productivity of the forest ecosystem (Tamm, 1969; Mälkönen, 1976; Kuusinen & Ilvesniemi, 2008). According to Jacobson et al. (2000), whole tree harvesting reduces the volume growth in both Scots pine and Norway spruce stands (5% and 6%, respectively) during the first 10 years after thinning. Stumps may also play a significant role in retaining N after harvesting, and their removal for bioenergy may markedly affect the nutrient availability and nutrient cycling of boreal forests (Palviainen et al., 2009). Recent work by Sathre et al. (2010) also suggest that fertilization, in Norway spruce stands, can be used to compensate for the loss of soil carbon stock caused by biomass removal from the forest.

Over the life cycle, the net CO2 emissions per unit of energy are smaller for wood than those for fossil fuels; i.e. on average 99 kg CO2 MWh−1 for Norway spruce (range 65–152 kg CO2 MWh−1), 123 kg CO2 MWh−1 (range 78–192 kg CO2 MWh−1) for Scots pine and for coal 341 kg CO2 MWh−1, if the emissions for production and transportation of coal are excluded (Statistics Finland, 2005). Intensive management for timber and energy biomass clearly decreases net CO2 emissions in energy production. This is in line with the results of Sathre et al. (2010), who found that the forest fertilization can significantly reduce the net GHG emissions and increase the availability of primary energy. According to our results, the fertilization decreased net CO2 emissions for Norway spruce on OMT and MT sites by 17% and 23%, respectively, compared with unfertilized managements. For Scots pine, the reduction was 12% and 19%, on MT and VT sites related to the increased growth.

In general, forest bioenergy supply chains seem to be effective; i.e. the energy consumption was 2–3% of produced energy and the CO2 emissions are 4–7 kg CO2 eq MWhpa−1 (Wihersaari & Palosuo, 2000). This held also for this study, with the energy consumption varying in the range 2.2–2.8% of that produced in the energy supply chain. As a reference, the consumption of energy over the life time of coal varies between 3.5% and 14% of that produced in the energy supply chain (Kivistö, 1995; Uchiyama, 1996; Gagnon et al., 2002). The consumption of energy regarding harvesting, short and long-distance transportation, chipping and fertilization depended on the management regime. However, the differences between management regimes were small. Nevertheless, an intensive management for timber and energy biomass may decrease the net CO2 emissions per unit energy in energy production. In the calculations, we used the value 1 GJ ton−1 of energy consumption for the production of N fertilizers (based on personal communication with Timo Toivanen, 2010; Grönroos & Voutilainen, 2001). In the literature, other values are also shown, ranging from 7.6 to 12.9 GJ ton−1 (Mäkinen et al., 2006). Fortunately, its impact on final CO2 emissions and energy balance of the forest chain is small.

To conclude, this study indicates that increased forest growth, especially due to the fertilization and higher precommercial stand density in the early phase of stand development, induces most of the differences in the net CO2 emissions. Regardless of tree species and site fertility, the lowest CO2 emissions per unit of energy was obtained when the precommercial stand density was extremely or very dense and fertilization was performed three times during the rotation of 80 years. The highest stem wood yield and the smallest net CO2 emissions of wood energy were achieved with the same management regime. Intensive management for timber and energy biomass clearly decreases CO2 emissions in energy production. Thus, it seems possible to produce forest biomass for energy purposes with relatively low CO2 emissions by applying intensive management and in this way also substitute for fossil fuels (Sathre, 2007). However, additional research is still needed in greater detail to further understand how the forest productivity and CO2 emissions from the use of wood energy are affected by different tree species and site fertility types, as well as by the intensity of forest management (e.g. initial spacing, thinning, fertilization and, rotation length) and utilization of timber for different products and forest biomass for energy production.

Acknowledgements

This work is partly funded through the Finland Distinguished Professor Programme (FiDiPro) (2009–2012) of the Academy of Finland (Project No. 127299-A5060-06). Furthermore, the Graduate School in Forest Sciences, the School of Forest Sciences, University of Eastern Finland, and the Finnish Forest Research Institute, Joensuu Research Unit, are acknowledged for their support of this study. Furthermore, Dr David Gritten is thanked for revising the language of this paper.

Ancillary