Objectives of the model
- To estimate the carbon stocks of stands and forests (in living and dead biomass and soil), and any associated harvested wood products
- To estimate the greenhouse gas emissions avoided through the use of wood products that displace fossil fuels and fossil-fuel intensive materials.
The model is applicable at the stand, forest and national level. It uses as input data, estimates of stand structure and growth obtained from yield tables that are applied at the stand level (Edwards and Christie, 1981). When stand-level carbon estimates are combined with area/age-class information, forest and national carbon stocks can be estimated. CARBINE can be used to estimate historical forest carbon stocks (if information on area is available), as well as current and future carbon stocks under different forest area and management scenarios. Using the same set of yield tables for all estimates assumes the same growth rates/patterns would be observed at any time: historic, current or future. This means that changes that might affect growth rate or form are excluded, such as the improvement of planting material or better site quality. Carbon stock changes are inferred from differences in carbon stock estimates at different times.
The model consists of four sub-models or ‘compartments’ which estimate carbon stocks in the forest, soil, and wood products and, additionally, the impact on the greenhouse gas balance of direct and indirect fossil fuel substitution attributable to the forestry system.
The model is able to represent all of the introduced and native plantation and naturally-occurring species relevant to the UK. The forest carbon sub-model is further compartmentalised to represent fractions due to tree stems, branches, foliage, and roots. The impact of different forest management regimes can only be assessed for the range of tree species, yield classes and management regimes represented in published yield tables (Edwards and Christie, 1981). However, at present not all of these are implemented in CARBINE.
Wood products are represented as long-lived and short-lived sawn timber, particleboard and paper. Carbon in harvested stemwood is allocated to these wood product categories using an assortment forecasting model that accounts for variation in product out-turn due to tree species and tree size class distribution at time of harvest (Rollinson and Gay, 1983). Wood products in primary use are assumed to decay over time with no account taken of carbon stocks in landfill or greenhouse gas emissions (due to wood products) from landfill. ‘Inherited’ emissions from wood products replaced are not considered.
The soil carbon sub-model runs independently of the forest sub-model. Initial soil carbon is estimated based on land use/cover and soil texture (sand, loam, clay and peat). The timecourse of any soil carbon stock change is assumed to follow an exponential form with the magnitude of the stock change and rate constant dependent on the soil type and on the particular land-use transformation imposed (e.g. arable agriculture to forest or grassland to forest). This information is based on published literature. There is no explicit representation of a litter compartment or sub-model.
CARBINE relies on UK Forestry Commission yield tables to provide basic input data on stand growth and structure. Currently these models are limited to the representation of pure-species, even-aged stands. As a result, neither model can be used to evaluate mixed-species stands, and no account is taken of understorey species. Both models can be ‘forced’ to approximate uneven-aged or mixed tree stands by assuming a large number of even-aged patches of different species and varying age class.
The forest carbon pools included in the sub-model are stem, branches, foliage, and roots. The model utilises pre-existing yield models developed for each species, yield class and management regime to estimate the development of merchantable stem volume at an annual time-step. Potentially, there are over 1000 different yield models available for different combinations of species, yield class and management regime. However, in general only one or two examples for each tree species are represented in CARBINE, although a comprehensive range of species relevant to the UK is covered.
The choice of yield class represented for a given species takes into account typical site productivity and any genetic improvements due to breeding (notably for poplar). The yield models are based on an extensive permanent sample plot network maintained across the UK and are considered to be robust. However, a recent study has highlighted possible deficiencies in the predictions of volume development made by the models, particularly for the latter stages of a rotation typical in the UK at present (Matthews, 2003). The yield tables take stem mortality throughout the rotation into account, and predictions are given for live stem volumes, i.e. net of mortality. The fate of dead standing trees or litter from any mortality (trees, branches or foliage) is allocated to the ‘waste’ compartment in the wood products sub-model.
Volume estimates obtained from yield tables are multiplied by relevant forest areas to give merchantable stem volume for the entire area. It is assumed that the area provided is the net planted area. Volume is converted to total tree volume (including stem, branches, foliage and roots) using total/merchantable (T/M) ratios (commonly referred to as expansion factors) for each of the major species groups. These expansion factors are assumed to vary with stand top height. The variation of T/M ratio with top height is assumed to handle site- and age-dependent effects. The T/M ratios are derived from published biomass studies, but the studies were not designed explicitly to address carbon sequestration issues. Consequently there is some uncertainty as to the robustness of the conversion from stem to total tree volume.
Total tree volume is converted to carbon using published values of specific wood density (Lavers and Moore, 1983) and an assumed carbon content of 50% (Matthews, 1993). The same density is assumed for all woody tree components. The same carbon content is assumed for all woody components, based on the review of Matthews (1993). Default carbon content values generally can be used, but they should be supported by a validated sample (IPCC 2000).
Forest areas and age-class distributions can be used by CARBINE to estimate total (estate-level) forest carbon. Results are presented as total forest carbon stocks and are not broken down into the individual pools of stem, branches, foliage and roots.
CARBINE may be biased in terms of the forest management regimes represented. The standard thinning regime assumed for most species is based on recommended practice (Edwards and Christie, 1981). However, in reality actual forest management departs significantly from these recommendations due to economic constraints and/or requirements to meet varying objectives in different localities within the UK. Variations in thinning regime will result in variations in forest carbon stocks for which predictions based on the assumption of standard management will be biased. This may not be a large source of bias in national carbon estimates, if the alternative management regimes vary both positively and negatively around the ‘average’. On the other hand, CARBINE may not predict carbon stocks well for individual stands or even districts when forest management is different from the management options available in the model. More information may be required on current and potential future management practices to justify the use of the ‘average’ regime in all circumstances, or to inform modification of assumptions. Unmanaged or ‘semi-natural’ forest is poorly modelled as it is assumed to follow the same growth patterns as unthinned productive forest up to the maximum potential carbon stock.
CARBINE contains a very basic sub-model to estimate carbon stocks and stock changes in this pool. This sub-model runs completely independently of other sub-models. Initial soil carbon is estimated based on land use/cover and soil texture (sand, loam, clay and peat). Changes in soil carbon are assumed to take place in response to land-use change and the magnitude and timecourse are estimated according to soil type (texture) and major land use category. This information is based on RothC, a UK soil carbon model, and published literature (Coleman et al., 1997).
The model assumes that any harvested wood products make a contribution that is additional to current consumption. In reality, it is likely that some products manufactured will merely replace other wood products, hence there may be less change in carbon stocks than is predicted by the models. How long the model continues to overestimate stocks will be affected by product service lives and the period over which the model is run.
At thinning and harvest, the CARBINE model allocates merchantable stem volume to various wood products, while the remainder is transferred to the waste pool. The ‘end-use’ wood products represented are:
- Long-lived sawn timber
- Short-lived sawn timber
During wood processing, conversion losses are assumed and enter the waste stream and decay within a year. The amount of carbon allocated to the raw stemwood product, categories of each of the wood products is estimated by first inputting the merchantable stem carbon derived from the forest yield model to a stand volume assortment forecasting model which estimates the volume allocated to sawn timber, roundwood and waste. This is implemented in CARBINE as a set of functions derived from the output of a more general and flexible assortment forecasting program known as ASORT (Rollinson and Gay, 1983). There is no reason why the full ASORT program could not be integrated into CARBINE as a subroutine, providing much greater flexibility. Having allocated some of the stem carbon to sawn timber, roundwood and waste, fractions of the first two categories are further allocated, in different proportions, to the four ‘end-use’ wood product categories specified above. The proportions differ depending on the species harvested. This information is based on expert opinion rather than data or scientific research. A carbon retention curve is used to estimate product decay and return of carbon to the atmosphere. Each wood product category has its own carbon retention curve based on estimated service lives, taking into account not just the decay rate of wood products but the service life as influenced by socio-economic factors. The functions are used to calculate the amount of carbon retained in wood products in successive years after harvest.CARBINE does not include a compartment which represents the carbon dynamics of wood products disposed of to landfill.
Wood products can contribute to greenhouse-gas emissions reductions in two ways, through:
- Direct substitution, in which wood is used as a direct source of energy (i.e. bioenergy) in place of fossil fuels
- Indirect substitution, in which wood is used in place of more energy-intensive materials, with implied reductions in fossil fuel consumption.
Both types of substitution can be taken into account using CARBINE. Each wood product is assumed to have a characteristic potential (emissions savings factor) to displace alternative materials or fossil fuels, thus determining the magnitude of avoided emissions. This is calculated by assuming an end use for wood products. These assumptions are based on expert opinion rather than data or scientific research. Greenhouse gas emissions over the life cycle of the wood product are compared to the greenhouse gas emissions for the most likely alternative product (for example steel or concrete), allowing for the possibility that the service lives of wood and non-wood products may be different .
Currently the emissions savings factor for wood waste to bioenergy is assumed to be zero. This sub-model of CARBINE was developed in the early 1990s. At this time there was very limited research globally on greenhouse gas emissions over the life cycle of a bioenergy system. The substitution sub-model of CARBINE has not been updated since it was first developed.