Material and energy flow indicators of wood raw material production for LCA of forest products

Transcription

1 Appendix 1 Material and energy flow indicators of wood raw material production for LCA of forest products Simo Kaila, Metsäteho With acknowledgments to: Jari Hynynen, Finnish Forest Research Institute Seppo Kellomäki, University of Joensuu Sari Pitkänen, University of Joensuu Risto Ranta, Forestry Development Centre Tapio Markus Strandström, Metsäteho

2 ABSTRACT 2 This report is a subproject report in the Nordic joint project LCA: Mark och Geografi. The goal was to develop serviceable indicators into LCA of forest products for energy and material flows brought about by wood raw material extraction in the forest. Modelling in the inventory phase was found to be an indispensable tool in handling forestry in LCA. Different options in handling time in modelling wood production were considered. The scheme of things, where forest regeneration is considered resulting from logging, was adopted as conforming with the purpose of LCA. The prospect for sophisticated ecosystem oriented modelling, with possibility to calculate versatile material flow indicators, was discussed. However, a simplified modelling approach focusing on future wood production at the logging sites of the wood raw material in question was still found more practicable. Indicators of the actual wood production potential, with the decrease due to possible departures from the optimum in logging and the consequent silvicultural operations, were proposed and demonstrated through a case study at Finnish conditions. The prospect of developing similar indicators based on Swedish and Norwegian data was discussed. In the context of LCA, the indicators can be used in LCIA as indicators of future biomass production potential of the product system in question, and used further in calculation of the accordant carbon flux. Keywords LCA, inventory, impact assessment, impact category, indicator, land use, modelling, forestry, forest management, wood production, productivity

3 3 CONTENTS 1 Optional approaches to wood raw material production in LCA of forest products Introduction Basic structure of LCA: possible problems Development of LCA Functionality of the present notion of land use in LCA and development needs Options for LCI modelling Forestry as modelling subject Basic options 9 2 Differences in material flows between different forest management regimes: effect of thinning regimes on forestry s carbon and nutrient flows 10 3 Practical assessment of forestry material flow indicators in LCA of forest products Feasibility of ecosystem-oriented modelling in LCI Simplified approach: wood production potential on logging sites succeeding logging Case study: forest regeneration and first thinning in Forestry Centre of Central Finland Data Results Review of the method and development aspects 22 4 References 25

4 1 Optional approaches to wood raw material production in LCA of forest products Introduction Basic structure of LCA: possible problems Life cycle assessment (LCA) claims to be a systematic technique for studying a product s entire life cycle i.e. compiling and evaluating all the relevant material and energy flows and their potential environmental impacts from raw material acquisition through production, use and disposal, as expressed in the introduction to the ISO standard. The framework for LCA comprises three successive phases: goal and scope definition; inventory analysis and impact assessment, and a fourth interpretation process that is carried out in interaction with each of these three phases. It is stressed that LCA is an iterative process, since each phase is conducted with close regard to the other phases. The quote above suggests possible limitations to LCA studies in that the definition does not cover the production of raw materials. This raises the question as to whether LCA be properly applied to products based on actively managed renewable natural resources. In forestry, silvicultural systems may have various environmental impacts according to the varying natural and technological conditions. If active forest management is practised as part of the production chain, with different sets of forest goods and services produced at different intensities, there are clearly grounds to consider wood production as part of the product system. In LCA terms, this would require the inclusion of forest management in the goal and scope definition phase as one of the unit processes to be examined in life cycle inventories (LCI). LCA is basically a time-unspecific method, intended for analysing unit processes only in terms of the quantities needed for a unit of the product, without focusing the analysis on events explicitly expressed in site and time. The totally different time-scale of the process of wood production compared to the logging and transportation processes is problematic here. The other parts of the life cycle of a product may be a fraction of the total production time of the raw materials in the product (Karjalainen & Welling, 2001). Many of the environmental impacts of raw material extraction only appear gradually, and this process must be investigated on a time-scale extending for decades into the future or retrospectively, depending on the line of thinking used Development of LCA Attempts to develop LCA, through the forums of SETAC 1 and COST E9 2, have not yet satisfactorily clarified in LCA terms the role of various aspects of land use, such as raw material production. LCI is the phase that should compile the information concerning all the different unit processes, allocating it to the functional unit of the product. In LCA mainstream development striving for a general analysis, temporal problems have been dealt with by dividing unit processes related to land use in LCI into land occupation and land use change 1 The Society of Environmental Toxicology and Chemistry: an independent, non-profit professional society that provides a forum for individuals and institutions engaged in study of environmental issues, management and conservation of natural resources, environmental education, and environmental research and development. 2 Life Cycle Assessment on Forestry and Forest Products (European Concerted Research Action)

5 5 (Lindeijer et al., 1998). Land occupation is defined as continuous land use of a specific type for a certain period of time, implying ecological stability, with spatial and temporal dimensions included in LCI (m 2 * year). Land use change denotes any change in land use from one type to another, such as the conversion of an area of forest to agricultural use, and only the spatial dimension is included in LCI (m 2 ). Land use issues are made operational in life cycle impact assessment (LCIA) through this distinction in the structuring of LCI. This involves classification, characterisation, and optional elements (normalisation, weighing) that may be left out of studies, as presented in ISO Classification is the phase that sorts interventions originating from unit processes into impact categories. In characterisation, the contribution per unit of intervention to impact categories is defined. In the case of land use, the unit of intervention might refer to a square meter of a certain transformation, for instance, or m 2 * year of a certain occupation (Voet, 2001). Category indicators are used as tools for this characterisation. LCIA phase modelling can be applied to proceed from the results of LCI to category indicators and on to the appropriate endpoint. According to the standard, in the environmental mechanism of the impact category, category indicators can be defined anywhere between the intervention and the endpoint inclusively. Choosing an indicator close to the intervention diminishes the level of uncertainty, but this uncertainty will then shift to the environmental relevancy of the indicator (Udo de Haes et al., 1999a). SETAC-Europe s proposed list of impact categories is as follows: - Input-related: - extraction of abiotic resources - extraction of biotic resources - land use - Output-related: - climate change - ozone depletion - human toxicity - eco-toxicity - photochemical ozone creation - acidification - nutrification It is also proposed that land use should include the following sub-categories: - Increase of land competition - Degradation of life support functions - Biodiversity degradation Impact categories that could be associated with wood production include extraction of biotic resources, various sub-categories of land use, and climate change. SETAC-Europe has suggested the following category indicators or outlines for their development (Udo de Haes et al., 1999b): A possible category indicator for the extraction of biotic resources could be developed to express the scarcity and regeneration rate of the resources. Suggested indicators for the sub-category increase of land competition include area of exclusive land use over a period of time, and change in area of exclusive land use. The sub-category degradation of life support functions could be indicated by vegetation cover, standing crop of natural vegetation, productivity of natural vegetation, and soil permeability, again with a possible distinction between a more permanent situation and a one-off change. For possible biodiversity indicators a similar distinction is suggested to

6 6 be drawn between more permanent situations and changes. The loss of vascular plant species is also suggested for study, as well as any opportunities to characterise impacts in this sub-category in a comparable way to the emission impacts on biodiversity. Climate change is a newer concept that will replace the formerly used global warming, with radiative forcing proposed as a direct category indicator Functionality of the present notion of land use in LCA and development needs LCI and LCIA, as presented above, are not readily applicable to forestry in reality. Defining forestry in LCI in terms of land occupation and land use change is already problematic. Forestry practices are continually being developed at a much faster rate than the process of wood production itself. Practices and silvicultural systems have undergone sweeping changes over recent decades, thanks to technological and economic developments, and more recently also due to the increasing importance given to natural values. The forest landscape is the product of former forestry practices, as are the environmental impacts of forestry. Major changes in forestry practices are likely to lead to new kinds of impacts, which may be immediate or delayed decades. The distinction between the concepts of land occupation and change clearly oversimplifies this reality, implying that environmental impacts would be static and compiled through direct observation, which is not possible in practice (cf. Voet, 2001). In practice, a forest s natural energy and material flows can feasibly be depicted as processes through modelling (Karjalainen & Welling, 2001). Modelling is an established technique in forestry, since the sustainability of forest management must be defined with regard to natural processes. Forestry planning has routinely involved inventories of the growth and yield of forests and models of the development of stands, as well as the consequences of logging and silvicultural operations under different stand conditions. Biodiversity, another key issue in sustainability, is not subject to analysis on the process level, due to problems with measurement and scale regarding the complex nature of the whole concept (Alvarado et al., 2002). The role of modelling in LCA deserves closer examination. It can be used in the calculation of results for LCI, and in LCIA for assessments of category indicators, for characterisation, and to aggregate impact categories into the endpoints. The task and the phase where the modelling is to be used determine the modelling environment. As discussed earlier, a category indicator can be defined at any stage of the environmental mechanism of the impact category from intervention to endpoint, with consideration given to the certainty and relevancy of the indicator. Moreover, it should be noted that the process of calculating results in LCI through modelling is in part interchangeable with the modelling of category indicators in LCIA. The former should be extended as far along the environmental mechanism as is favourable in terms of the information related to the product system improving results; whereas modelling that disregards the product system is used for LCIA. Impact category assessment cannot handle any positive effects of land use such as the regeneration of low-yield forests and carbon sequestration, so LCI modelling is needed to bring out the net effects, otherwise subsequent phases will be based on information from inaccurate generalisations, unrelated to the specific features of the product system. Returning to the proposed LCIA impact categories, the modelling results in LCI are obviously related to climate change, as the flows of carbon dioxide into and out of the product system play a role here. It is not yet clear to what extent the results could be used

7 7 for other impact categories, such as the extraction of biotic resources and land use. Wood from managed forests should not be considered as a resource in the former category, because it originates from within boundaries of the product system (Klöpffer, 1999), though biodiversity could be incorporated here (Baitz & Kreissig, 2001). As to the life support function, biomass production is generally considered a valid indicator (Voet, 2001). However, the use of productivity as well as biodiversity has also been criticised, given no quality measure is applied to determine the natural-artificial scale; the essential content of the land use impact category has rather been seen as the loss of habitats and species (Klöpffer, 1999). In summary, some relevant features of what is generally defined as land use can evidently be broken down into material flows related to life support functions in land use, and to climate change in other impact categories. It seems that the proposed definitions for impact categories and indicators should still be thought over, taking into consideration the interface between LCI and LCIA. Important issues include determining both the factors that can be calculated in LCI, and what should be required of the results to assure the semantic integrity and modularity of the whole structure, as well as how best to utilise the available information related to renewable natural resources as raw materials. LCI seems to have been shown little interest in developments in LCA (cf. Voet, 2001). One problem with LCA development work seems to be that the LCA template as a whole has not been discussed. Such discussion should include the roles of goal and scope definition, the boundaries of the product system, and LCI. 1.2 Options for LCI modelling Forestry as modelling subject Due to the problems discussed above, it seemed to be advisable with forestry and forest products to concentrate on developing LCI, and trying to find workable applications for this phase of LCA. Research-based models with the capacity to handle forestry s unit processes over time seemed to be a valid approach. In modelling forestry, there are different options for handling the time-scale. Results can thereafter be averaged out over the time period selected. The choice of the final application will also depend on the capacity of models and the availability of data. In modelling forestry, a distinction should be made between the two main types of forestry. Rotation management or even-aged forestry deals basically with even-aged stands characterised by periodic thinnings, regeneration cuttings after specified rotations, and regeneration through natural reproduction or replanting. In compliance with the notion of normal forest, the stands form an age-class distribution with roughly equal areas of each developmental stage to ensure that yields remain roughly stable over time. The control of such systems is relatively easy to manage. Continuous cover management or unevenaged forestry operates in forests of indefinite age, where single trees are harvested selectively. This system is more difficult to control in management, as age-based measures of productivity and valuation are lacking (Gadow, 2000). The first type of silviculture is widely used in northern forests, and the second type is more typical in tropical regions, although in recent years it has also gained in popularity in Central Europe. In practice, forestry may exhibit features of both types, and it may be difficult to draw a line between the continuous cover management system and a system of extensive forest use with selection felling and little silvicultural input.

8 8 The following discussion will exclusively concern the rotation management type, used as part of the forest industry s product system. This type of forest management consists of the distinct processes of wood production, logging, and transportation. Logging and transportation involve the extraction of wood raw material, and can be covered straightforwardly in LCA studies through simple calculations based on inventories. These processes have a short time-scale in Nordic wood procurement systems from a few weeks to a few months. The processes are purely technological, and are covered by readily available and accurate information. The primary data sources are the wood procurement organisations planning and control systems. In Sweden and Finland, LCI data on the amounts of fuels, oils, etc. needed for logging and transportation have been collected into databases, together with their respective emission coefficients. These figures are used to calculate energy consumption and emissions of. CO 2 or NO x, for example, for specific forested areas, for organisations, or for product systems (Berg & Lindholm, 2001). Data is available from many European countries, enabling comparisons of different forestry systems and technologies (Schwiger & Zimmer, 2001). It is not obvious how silvicultural operations (i.e. operations related to wood production, forest improvement, and road construction) should be dealt with in integrating forestry as unit processes in LCA. Karjalainen & Welling (2001) make a distinction between biological and technical production systems, with technical systems comprising the chain from timber harvesting to product disposal, but no reference is made to silvicultural operations. In practice, if this basic distinction is made, silvicultural operations can with good reason be included in databases on forestry operations, as in the works mentioned above. The contribution of silvicultural operations within LCI is fairly low when compared to logging and transportation, and can be allocated to the functional unit on the basis of mass, for instance. Wood production, if juxtaposed with logging and transportation, can on the other hand be considered an explicit unit process, consisting of growing operations such as site preparation, planting or seeding, young stand cleaning and thinning, and biological processes such as tree growth and decay which will show up in physical form as flows and stocks. Unlike in logging and transportation, more or less natural ecosystems are the actual production units, with their natural processes regulated, not driven, by human action. Many aspects of these processes are complex, and information may be scarce and uncertain. The data for LCI is not absolutely process-associated, since it is available from nonoperative sources such as forestry planning or the monitoring of forestry operations. This all means that wood production cannot directly be covered by inventories alone. Wood production operates on a time scale of several decades. In modelling, decisions must be made on how the time-scale is handled, and what the subject of modelling is to be. The first choice is whether the process preceding harvesting, or the process succeeding it should be regarded. The key to this choice is how causality is seen between the processes of wood production and logging: the wood production process can be seen either as the former phase of wood raw material production, or the latter phase that retains the productivity of the forests after logging. The second choice to be made is whether analysis should concern the logging sites that yield the functional unit s raw material alone, or on a larger forest area that will produce a more sizeable set of raw materials, including those in the functional unit. If a larger forest area is chosen, a third time-scale choice in which wood production is defined as a current process becomes feasible. Forest improvement can be handled as part of the wood production process, defined as above. Operations like fertilisation can be easily integrated, while other measures such as drainage and afforestation modify or start new phases in the process. Forest road construction, sometimes seen as belonging to forest improvement, is most logically included

9 9 in the process of logging and wood transportation, since the standards of roads used in wood production are much lower Basic options The modelling options for forestry in LCA differ in their handling of the process of wood production, whereas in logging and transportation there are no reasons to depart from the current process. The following three options have been outlined (Kaila, 2000): Option Modelling subject Processes to be modelled Logging and transportation Wood production Handling of time scale in modelling 1 Logging sites Current process Process preceding logging 2 Logging sites Current process Process succeeding logging 3 Forest area Current process Current process The first option, looking at wood production as the process preceding logging, is basically correct and logical. The starting point is the current logging process for the raw material of the functional unit, together with real events in the wood production process since the beginning of the rotation. Although logical and concrete as modelling basis, this option lacks practicability and purpose. The process must be covered back to the beginning of the rotation, and the uncertainty grows the farther back the task proceeds. Little is usually known about the earlier silvicultural state of any stand, and it is impossible to say how much timber a stand has yielded in previous cuttings. This gives rise to additional difficulties in allocating previous inputs between earlier raw material uses and the present functional unit. Moreover, a retrospective assessment going back for decades does not address the purpose of an LCA study, with regard to product and process development and improvement. These problems would seem to make this approach unsuitable for LCA. The second option considers wood production as a consequence of the raw material extraction for the functional unit. As in the first option, the starting point is the current logging process in logging sites, but this is assessed together with the predicted wood production process starting from logging to the end of the next rotation after it. The logic that the wood production process is a consequence of harvesting is inherent in normal forest planning, as the allowable cut is set first on the basis of forest growth potential, and silvicultural operations are planned with fellings as the starting point. The weakness of this option is the uncertainty of predictions made decades in advance. To be feasible, this requires stable social and economic conditions with regard to forestry, as well as the application of consistent established practices, and appropriate modelling tools with sufficiently detailed data collection. This option s major advantage is its conformity with the purpose of LCA: it recognises the need for sustainability in raw material use. In conclusion, this approach is suitable for LCA, but a proper application will require adequate data collection and a competent analysis system for LCI. The third option analyses current processes in logging and transportation, and in wood production, but operates on a larger scale, covering a greater area of forest. This option is again concrete and logical. Over a larger area, processes like forestry operations and growth can be monitored and inventoried over a suitable time period, such as one year. They can also be predicted through forest planning tools, where management options are concerned. Forestry operations and timber flows can be analysed on a scale relevant for practical wood procurement, also since such data may be available in the form of regional forest management planning and administration statistics. However, current events

10 10 in wood production are not in any direct cause-effect relationship with the functional unit at hand; rather they affect the current structure of the forests. This causes difficult allocation problems in LCI, so it must be concluded that this approach is suitable for ecobalance assessment, given the availability of a proper data collection system. 2 Differences in material flows between different forest management regimes: effect of thinning regimes on forestry s carbon and nutrient flows The impacts of different forest management systems on carbon and nitrogen balances were demonstrated, in order to show the relationships between material flows in forestry operations and forest growth. The simulations were carried out using the Impact Model, a process-based tree-soil system model developed by Russian scientists (Chertov et al., 1999) that comprises soil dynamics as described in Chertov et al. (2000), further adapted at the University of Joensuu for Finnish conditions. The quantities of carbon absorbed in tree growth and reallocated into different parts of standing crop during the rotation and carbon and nitrogen in soil were calculated with the simulator under different cutting regimes. Emissions caused by forest regeneration, logging, and transportation operations were also included in the calculations related to carbon binding. Emission coefficients used by Metsäteho were applied here. The demonstration cases comprised two tree species on a typical site: Norway spruce on a good site (OMT) and Scots pine on a medium quality site (VT). The cutting regimes for both tree species and sites were as follows: - a normal thinning programme with three thinnings, in accordance with forest management recommendations by the Forestry Development Centre Tapio - one heavy thinning in the middle of a rotation - selection thinning, removing 15 % of the largest trees The starting point of the simulations was set as the same, i.e. 2,080 seedlings with a height of 2 metres at the age of 9 years, and a rotation time of 80 years. The results can be seen in Figures 1 and 2. The differences in the total amounts of carbon absorbed into the wood under different cutting regimes for the same tree species were rather small: for the whole rotation period the figures were about 11 tonnes per hectare for Norway spruce and 2 tonnes per hectare for Scots pine. It can be assumed that comparing different forest regeneration regimes would have resulted in much larger differences, but such comparisons by modeling were considered to involve too many uncertainties, because the model had not been fully calibrated to early development phases of a stand. The results show that the amounts of carbon absorbed during wood growth and the differences between these amounts due to the use of different thinning regime are orders of magnitude greater than the carbon emissions caused by forestry operations during the thinning regimes, which in the cases studied made up 1,4 % - 1,7 % of the total quantity of carbon involved. It is therefore clear that calculations of the eco-efficiency of forestry systems should include the consequences of the management methods applied to future production. The model enables the quantities of carbon and nitrogen in the soil to be calculated, as well as the amount of carbon in the trees. In the cases studied, implicating normal mineral soil conditions, the calculations revealed no drastic changes over time (Figures 3 and 4).

11 11 Total Case 1 Commercial timber Natural removal Logging residue Carbon, tonnes/hectare Total Case 2 Commercial timber Natural removal Logging residue Carbon, tonnes/hectare Total Case 3 Commercial timber Natural removal Logging residue Carbon, tonnes/hectare Figure 1. Absorbed carbon. Simulation results with Norway spruce on a good site (OMT): Case 1: normal thinning; Case 2: one heavy thinning; Case 3: selection thinning (see text). Total amounts of carbon are shown, along with their allocation during rotation. The negative side of the total carbon column shows the emissions of all forestry operations, including forest regeneration, logging and transportation.

12 12 Total Case 1 Commercial timber Natural removal Logging residue Carbon, tonnes/hectare Total Case 2 Commercial timber Natural removal Logging residue Carbon, tonnes/hectare Total Case 3 Commercial timber Natural removal Logging residue Carbon, tonnes/hectare Figure 2. Absorbed carbon. Simulation results with Scots pine on a medium quality site (VT). Explanations as in Figure 1

13 Carbon, tonnes/hectare Year Figure 3. Carbon in the soil. Norway spruce on good soil (OMT). Case 1: normal thinning (see text) Nitrogen, tonnes/hectare Year Figure 4. Nitrogen in the soil. Norway spruce on good soil (OMT). Case 1: normal thinning (see text)

14 3 Practical assessment of forestry material flow indicators in LCA of forest products Feasibility of ecosystem-oriented modelling in LCI The ideal LCI framework for wood production covers the biomass production of forest ecosystems together with their energy, carbon, and nutrient flows and balances. Sophisticated models supported by research are required, together with suitable data systems to provide the necessary set of parameters. Empirical growth models based on regression between sites and changes of tree or stand are widely used in forest research and management, often within simulators. These models focus on wood production in terms of stem dimensions and volumes. Over the past decade, ecophysiological process-based modelling has begun to be used as a tool for research and development work, since it can handle biomass production in components and is more detailed in its response to conditions than empirical models. By combining empirical and process-based models, new sets of hybrid models have also been devised. Process-based and hybrid models are suitable tools for an ecosystem oriented approach as they allow the direct simulation of biomass production with all its components, as well as carbon, energy, and nutrient flows in trees and stands, and their dependence on forestry operations. These models can produce results of the relevant order of magnitude with the necessary sensitivity to distinguish between optional management systems. The Impact Model used in calculating the cases in Section II of this paper is one example of a process-based forest stand simulator that covers tree growth, cutting regimes, and carbon and nutrients in trees and soils. The state of the art in modelling of the dynamics of forests has been assessed by Mäkelä et al. (2000), who conclude that process-based causal models with a carbon balance approach can cover the essential features of the development of stands and trees, even though the models may focus on different parts of the tree-stand system. These models can be improved by adding on empirical elements. By using empirical components such as the total productivity of the site, carbon allocation and competition, and tree mortality and regeneration, it is possible to quantify the carbon balance frameworks. Calibrations by whole-system data seem adequate. Such models, devised for development purposes, have been built for specific tasks, but their operational use in forestry may still be a few years away. Proper documentation and coding, calibration and evaluation, and well considered scope with respect to the phenomena included and site factors are mentioned as vital prerequisites (Mäkelä et al., 2000). The current situation with the Impact Model used in the simulations presented in Section II can be rated similarly. In addition to improvements in calibration and evaluation, there is work to be done concerning certain issues such as mixed forest structures and the spacing of trees, although the model has shown an ability to cope with them. The dynamics of other vegetation than trees should also be included to ensure all components of biomass production are covered. Peaty soils, which are common in Finland, and other nutrients besides nitrogen should also be included to make the simulator more universally applicable in forest ecosystems. The data needed as the starting point for calculations and parameters constitutes another problem where ecosystem-oriented models are concerned. The data collected for operational purposes in forestry largely lacks information on site conditions and the state of stands. Data is also collected for updating forest planning after operations, but is far too

15 15 general for these purposes, and does not normally cover privately owned forests. One feasible source of this type of information could be the data collected by the authorities for supervision purposes, but the data specifications are general for the need, and more detailed measurements would have to be out of place considering the main purpose of monitoring. 3.2 Simplified approach: wood production potential on logging sites succeeding logging An advanced ecosystem approach was considered to involve too many uncertainties related to the new elements of forest modelling for LCI use at present. The uncertainties concern the ability of the models to produce relevant results for different components of biomass production, or for carbon and nutrient flows and balances within the relevant range of forest structures and conditions. Model calculations restricted to wood production are basically much safer in this respect because of their established research basis, the possibility to evaluate the models against long-term research results, and the greater availability of data. Therefore, a simplified approach concerning only the modelling of wood production was adopted. In this approach the task is to produce LCI results of the actual wood production potential on logged sites during the rotation concerned, and the reductions because of possible departures from optimum logging and directly consequent silvicultural operations. In the context of LCA, these results are meant for use in LCIA as indicators of future biomass production potential, and also to serve in the calculation of indicators for carbon flux from the atmosphere into the forestry system. With this approach the problems discussed above are considerably reduced. An established forest stand simulator used for forest planning, based on empirical growth models used for comparing management options and forestry data will meet the modelling task. All types of fellings were selected for analysis as follows: - regeneration felling: potential wood production during the rotation succeeding the operation and the decrease caused by possible shortcomings in regeneration - first thinning: potential wood production during the remaining part of the rotation succeeding the operation, also considering the decrease caused by possible shortcomings in standing crop density - other thinnings: potential wood production during the remaining part of the rotation Final fellings with subsequent regeneration and first commercial thinnings were given special attention as the most critical operations to the productivity of forests. The success of forest regeneration will largely decide the prospect of wood production of the next rotation after final felling, and the outcome can roughly be modelled by a forest stand simulator. As to thinnings, the first commercial thinning is most decisive for the productivity of the stand during the remainder of the rotation. It is essential here that the volume of the stand remains high in logging, since excessive thinning will affect growth. Later thinnings were left out of the calculation of the explicit decrease, because the timing of exploiting the standing crop at this later stage of stand development does not affect the total wood production of the rotation so much. However, because of larger tree size in felling and heavier loads in forwarding, lower quality work is more likely to affect future wood quality within the stand due to stem and root damage. In regeneration and first thinning the results are suggested to be calculated as two classes, one representing good practices and another unsatisfactory quality if identifiable in the data. Stand developments succeeded by the operations are to be modelled to the end of the rotation hypothesizing that the operations in the later phases are performed duly.

16 The approach is proposed to be realised by the following procedure: i) In regeneration felling - Specifying types of regeneration areas resulting from final fellings that serve as raw material sources for different forest industries - Selecting the regeneration areas accordingly from the data base - Calculating the wood production potential over the subsequent rotation time and the decrease due to the actual state of the forest regeneration areas ii) In first thinning - Specifying types of stands at first thinnings that serve as raw material sources for different forest industries - Selecting the logging sites accordingly from the data base - Calculating the wood production potential over the remaining rotation time and the decrease due to the actual state of the forests iii) In other thinnings - Specifying types of stands at other thinnings that serve as raw material sources for different forest industries - Selecting the logging sites accordingly from the data base - Calculating the wood production potential over the remaining rotation time iv) Calculating the results from different fellings per logged unit weighted by their proportions in wood procurement Case study: forest regeneration and first thinning in Forestry Centre of Central Finland Data The procedure was tested with data from the Forestry Centre of Central Finland. It was considered important that the handling of results should be possible on a regional level. Considering this, it is possible to select regions that are most comparable to the wood procurement area in the LCA study concerned. The variables needed in the data to calculate the above indicators by modelling are: i) Regeneration - Type of cutting removal in the preceding regeneration felling - Main tree species - Site class, climatic area - Number of viable, spatially properly distributed seedlings per hectare - Age of the stand from establishment ii) Thinnings - Type of cutting removal - Main tree species - Site class, climatic area - Basal area - Dominant height The data collected by Regional Forest Centres for monitoring of forest regeneration and cuttings fulfils basically the need as to variables. All of Finland s 13 regional forestry centres survey forest regeneration and thinning operations through specific field study systems for official supervision purposes, according to forest legislation. These surveys are based on the forest owners obligation to give written notice of the use of forests as well as completed forest regeneration (Nordiska Ministerrådet, 2001). Data is collected by the Forestry Centres field crews and compiled by the Forestry Development Centre Tapio into a database, which can be subjected to

17 17 various analyses. Results are normally reported in detail to forest owners and the organisations and forest companies concerned. The proportions of thinnings that do not meet the recommended norms are publicly reported by cutting type and Forestry Centre region, for instance. The data collected on regeneration areas includes: - general information - selection criteria (random; selected) and legal status - operation situation - site class - regeneration method - regeneration operations performed - presence of important habitats listed in the Forest Act - assessment of the appropriateness of regeneration measures performed - assessment of establishment success from sample plots: number of seedlings; proportions of tree species; mean height The data collected from thinnings includes: - general information - selection criteria (random; selected) and legal status - operation situation - assessment of forest use declaration - site class - development class of the stand - cutting method - main tree species - quality and volume of standing crop: logging damage; basal area; dominant height - presence of important habitats listed in the Forest Act The surveys account for 2-3 % of all regeneration areas, and 3-5 % of the year s thinning/cutting sites. The inspected sites include about 2000 regeneration areas, of which about 90 % have been completed as to regeneration operations, and about 1000 thinnings, of which about 40 % have been first commercial thinnings. Unlike in thinning, regeneration surveys do not cover the current year s operations, but those carried out over the previous 3-6 years. One reason for the timing of regeneration surveys is the legal deadline for the establishment of the new stand, which is five years in Southern Finland. Not all monitoring of regeneration has been based on random selections; a varying proportion has consisted of such cases, where there have been doubts about the outcome Results The simulations were performed at the Finnish Forest Research Institute by Mottisimulator, a stand simulation tool which is still under development but which incorporates the institute s findings on growth and yield research (Hynynen et al., 2002). There were doubts that sufficient quantities of data would be available, since only objects selected using random criteria could be used, so it was decided to use survey data from a period of four years ( ). The total amount of data still remained low though exceeding the average collected Forestry Centres in Southern Finland. A total of 64 first thinnings and 92 other thinnings had been surveyed for the Forestry Centre during the period, according to the random selection criterion. The first thinnings

18 18 were all Scots pine or Norway spruce stands and covered a total of 169 hectares. Among the other thinnings were four thinnings of other tree species, which have been omitted here. Scots pine and Norway spruce thinnings amounted to 213 hectares. The average situations in site classes comprising at least five logging sites were calculated to be included for simulation (table 1). Thinning Scots pine Norway spruce Mediumgood (MT) Medium (VT) Good (OMT) Mediumgood (MT) Logging sites First Other Table 1. Thinnings dominated by Scots pine or Norway spruce, surveyed by the Forestry Centre of Central Finland : number of stands in site classes represented by at least five stands. The first thinnings of the both tree species were analysed as two classes based on the quality of the stand after thinning: those rated as acceptable in field surveys and those rated as unsatisfactory or defective; the ratio of these was used as the reduction because of shortcomings in operation. The other thinnings were modelled as one class. The simulations were performed using the average dominant height and basal area of the classes (tables 2 and 3). Rotation times fulfilling the threshold values set in the recommendations of good forestry by Forestry Development Centre Tapio for the mean diameter of trees for regeneration were applied. Types Scots pine Norway spruce Medium-good (MT) Medium (VT) Medium-good (MT) H dom, m BA, Cases H dom, m BA, Cases H dom, m BA, Cases m 2 /ha m 2 /ha m 2 /ha Good 13,1 14, ,6 14, ,8 18,5 6 Unsatisfactory 12,5 11,9 6 12,8 12, ,5 14,8 2 Table 2. Simulation cases in first thinnings. Average dominant height, basal area, and number of surveyed cases according to types assessed Scots pine Norway spruce Medium-good (MT) Medium (VT) Good (OMT) Medium-good (MT) H dom, BA, Cases H dom, BA, Cases H dom, BA, Cases H dom, BA, Cases m m 2 /ha m m 2 /ha m m 2 /ha m m 2 /ha , , ,9 18, Table 3. Simulation cases in other thinnings. Average dominant height, basal area, and number of surveyed cases The average situations in first thinnings rated as unsatisfactory or defective compared with the thinning models of Forestry Development Centre Tapio standing for good forestry as seen in figure 5. First thinnings of Norway spruce stands had been cut slightly too intensively, on average. In Scots pine stands excessive thinning had been more common, affecting about one third of the thinnings, and the averages for basal areas were clearly under minimum levels in these cases.

19 Norway spruce, MT (medium-good) site Basal area, m 2 Basal area, m Dominant height, m Recommended before thinning Upper limit after thinning Lower limit after thinning Criticisable and defective thinnings, on average Scots pine, MT (medium-good) site Upper limit before thinning Upper limit after thinning Lower limit after thinning Criticisable and defective thinnings, on average Dominant height, m Basal area, m Scots pine, VT (medium) site Upper limit before thinning Upper limit after thinning Lower limit after thinning Criticisable and defective thinnings, on average Dominant height, m Figure 5. Relation of criticisable or defective first thinnings to thinning models standing for good forestry practices, by Forestry Development Centre Tapio Altogether 326 completed regeneration areas, with a total area of 570 hectares, had been selected for surveying under random criteria over the period within the region supervised by the Forestry Centre of Central Finland. Of these, analysis focused on Scots pine and Norway spruce, with data also available on first thinnings. Areas dominated by Scots pine and Norway spruce were selected for modelling, since stands of both tree species on their typical sites can be assumed to be comparable with the previous final felling, in terms of their main tree species. This is something of a simplification, but for the time being the survey data system could not be linked with forest use and regeneration notice registers. In all, there were 80 and 150 stands dominated by Scots pine and Nor-

20 20 way spruce, respectively. Nearly all came under the same site classes as the thinnings, and those represented by at least five stands were selected for modelling (table 4). The time between the forest use notice and the survey was 4,6 years on average for Scots pine, and 3,4 years for Norway spruce. The average age of the new stands was thus 3-4 years. Stands had reached a mean height of 40 cm for pine and 37 cm for spruce with figures a little higher in planted sites than where sites were sown or left for natural regeneration. Direct sowing is not a standard regeneration method for Norway spruce, and had only been tried for experimental purposes. Regeneration method Assessment of regeneration Mediumgood (MT) Scots pine Medium (VT) Norway spruce Good (OMT) Mediumgood (MT) Natural Direct sowing Planting All Regeneration areas Acceptable Unsatisfactory 1 Acceptable Unsatisfactory 1 3 Acceptable Unsatisfactory 1 2 Acceptable Unsatisfactory Table 4. Surveyed regeneration areas dominated by Scots pine or Norway spruce in the Forestry Centre of Central Finland in All data measured and the site classes represented by at least five areas. Similarly as in thinning, the simulations used varied rotation times of years. The stands were considered too young for any evaluation of their height development rate. As the sole starting point in the modelling were chosen the amounts of seedlings, without splitting up the analysis by regeneration method. Correspondingly to the procedure with first thinnings, the simulations were performed in two classes according to the amount of seedlings, assessed either as acceptable or too low as in need of extra measures. The average numbers of seedlings in stands assessed as acceptable were per hectare for Scots pine in all regeneration methods and both site classes, and for Norway spruce This is well comparable with the Forestry Development Centre Tapio s recommendations of 2000 seedlings per hectare for Scots pine and 1800 for Norway spruce, and the numbers of the recommendation were used in simulation for the first class. The simulation for the second class was performed based on the average seedling numbers of the regeneration areas, but because the seedling stands were rather young, it was hypothesised that stands will still grow somewhat denser through natural restocking; the amount of this was assumed half of the current stem number (table 5).

21 Simulation cases Scots pine Norway spruce Medium-good (MT) Medium (VT) Good (OMT) Medium-good (MT) Seedlinglinglinglings Cases Seed- Cases Seed- Cases Seed- Cases /ha /ha /ha /ha Good 2000 *) *) *) *) 118 Unsatisfactory 564 **) **) **) ***) 640 ***) 1320 ***) Table 5. Simulation cases in regeneration by site class. Number of seedlings per hectare and number of surveyed regeneration areas *) With the seedling stands assessed as good the stem numbers in recommendation of good forestry by Forestry Development Centre Tapio were used. With the stands assessed as in need of extra measures the average numbers of seedlings **) were raised by half for simulation ***) (see text) Each hectare of the logging operations analysed will, according to the simulation results, have affected the wood production potential as presented in table 6. The results can be related to the functional unit of LCA by calculating the area logged for a unit of the wood raw material in question. That requires information of the distribution of different cuttings and the amounts of wood logged at them, which is available from the cutting statistics needed in the LCI for logging and transportation. 21 Operation Forest regeneration First thinnings Other thinnings Scots pine Norway spruce Medium-good (MT) Medium (VT) Good (OMT) Medium-good (MT) Good Def. Time Good Def. Time Good Def. Time Good Def. Time Production scale Production scale Production scale Production scale m 3 /ha yrs m 3 /ha yrs m 3 /ha yrs m 3 /ha yrs Table 6. Summary of the results of simulations. Indicator values of potential wood production with good and defective practices, based on operations of good / acceptable vs. defective / unsatisfactory quality and the duration time per one hectare of the operation in question. Calculated as case study from data Forestry Centre of Central Finland in Especially the quality of regeneration had a strong effect on the results. The reduction from acceptable to unsatisfactory regeneration was % for the years of the rotation. The effect of shortcomings in first thinning was lesser, 3-7 % for the years remaining of the rotation (table 7).