2.1 Outline for forest ecosystem model SIMA
In this study a gap-type forest ecosystem model, called SIMA (Kellomäki et al. 2008) was applied for simulation of forest stand dynamics. The model was developed for the main forest tree species (Scots pine, Norway spruce, birch spp) growing on upland forest sites in Finland between N 60° and N 70° latitudes and E 20° and E 32° longitudes. In SIMA model, environmental conditions of the site are described by temperature sum, soil moisture, nitrogen and light availability. Optimal conditions for growth and regeneration assume no shade and no limitations in soil moisture and nitrogen availability. Environmental factors are linked to the demographic processes (birth,
growth, death) by multipliers as G = G0 x M1, …, Mn, where G is growth, G0 is growth under optimal conditions, and M1, …, Mn are multipliers of environmental conditions, i.e.
temperature sum, light conditions and soil water and nitrogen supply. At the same time such factors as atmospheric CO2 concentration, nitrogen deposition and diameter of tree at breast height (1.3 m above the ground) also affect tree growth. Height and mass of tree are calculated based on diameter.
Fig. 1 Outlines for forest ecosystem model SIMA used in the simulations (adopted from Kellomäki et al. 2008).
Growth responses of different tree species to the temperature sum are modeled according to tree species-specific minimum, optimum and maximum temperature sum values for growth. Temperature sum, site fertility type and input of litter and dead wood define amount of soil organic matter and available nitrogen. Soil moisture available for tree growth on different soil and site types is affected by the precipitation and evaporation.
Simulation of forest stand dynamics require initial stand density and average tree diameter and tree species as input data. Different management operations such as intensity and type of thinnings and timing of final cut need to be specified. Model calculations are based on Monte Carlo technique with one year time step and performed on an area of 100m2. In this study, average simulation results were obtained using 50 iterations for simulations.
2.2 Model calibration for Russian sites
The use of SIMA model for growth and yield predictions of forests in Moscow area required some changes in model parameters depending on geographical location. This was due to the fact, that Moscow area has lower latitude than Finland, so the sun radiation activity also differs. That affects the growth of trees.
The study area was - located in central Russia federal region in the North-East part of Moscow area. Forest inventory and climate data used in model calibration were taken from Shelkovsky training forest compartment managed by Moscow State Forest University.
Climate data represented by mean monthly temperatures and precipitations measured by the local meteorological station over the period 1962-1984. Climate of Shelkovsky area was characterized by warm summer and cold winter. Mean annual maximum temperature was in July (17,4°) and mean annual minimum in January (-10,3°).
Fig. 2 Average temperature and precipitation curves. Moscow area climate = solid line, Joensuu climate = dot line.
Precipitation is highest during the growing season from May to October (325mm on average). While mean annual amount of precipitation is around 549mm. The highest amount of precipitation refers to July. In general, climate of Moscow area is very similar to that of Joensuu, in Eastern Finland (Fig. 2). Thus, finally Joensuu climate was used in all calculations.
The growth parameters used in SIMA model were originally estimated based on old growth and yield tables in Finland (Kellomäki et all. 1992). To make model calibration for Moscow area, the growth and yield tables representing main forest tree species of Moscow area (Scots pine, Norway spruce and Silver birch) were available from Shelkovsky training forest (see Appendx A). These growth and yield tables were based on data from sample plots established in 1936-1940 and 1946 and contain the information on stand characteristics such as average tree height and diameter, stand density, basal area and stand volume for every tenth year during rotation, starting from 20 years age for conifers and 10 years age for birch. The data was organized by tree species and yield classes (bonitet). To relate Russian forest stand classification based on yield classes to Finnish classification based on site fertility types it was used “Forestry organization and development project of Shelkovsky training forest compartment 1984-1985”, which includes characteristics of forest undergrowth. As a help, it was used detailed description of Finnish site fertility types (Cajander 1926).
The model parameters were calibrated by comparison between SIMA simulation results and growth and yield table values. Simulations were initiated using initial stand data from growth and yield tables. As a result, some changes were done for potential growth, height and mortality parameters.
For the potential growth calculation, following equation was used:
PotGr = EXP(-1,307-1,643/(CO2/100))*DBH*EXP(DGRO*DBH) (1)
where DBH is the stem diameter at the breast height [cm] and DGRO is a species specific parameter, which has been modified (Table 1) and CO2 is atmospheric CO2 content (ppm).
Table 1. Species specific values of DGRO parameter before and after modifications.
Parameter DGRO Pine Spruce Birch
Initial value -0.0719 -0.0502 -0.0956
Modified value -0.0590 -0.0500 -0.0950
In a height growth equation, parameter C was modified (Table 2).
where TS is local temperature sum (d.d), DBH is diameter at breast height (1,3 m form stem base), and A, B and C parameters.
Table 2. Species specific values of C parameter before and after modifications.
Parameter C Pine Spruce Birch
Initial value 0.4354000 0.6689000 0.7302000
Modified value 2.0856 2.94316 3.13986
For calculation of natural mortality, following equation was used
Mortality factor = (BA*density**a/37)/agemx(i) (3) where agemx(i) is a species specific parameter (for pine=350, spruce=180 and birch=110), BA is a basal area (m2ha-1), density is a stand density (stems ha-1) and a is a modified parameter, which was 0,7 before change and 1,05 after modification. According to the model design, tree dies if random number from 0 to 1 is smaller than factor.
2.3 Tree stands and climate data and management regimes used in simulations Climate and forest stand (initial) data used in simulations was the same as used for model calibration (see 2.2). To start simulations the initial stand data for pure even-aged Scots pine, Norway spruce and Silver birch stands growing on Oxalis - Myrtillus (OMT) and Myrtillus (MT) sites were used as input data. In total, the data represented 5 different forest stands.
Three alternative management scenarios: No management scenario, management according Finnish management guidelines and Russian forest management regulations were applied for each of 5 forest stands. In No management scenario (no thinning) only final cut was done. For Finnish recommendations for practical forestry (Äijälä et al.
2014) and Russian forest management regulations (Pravila uhoda za lesami 2017), timing, and intensity (and thus also frequency) of thinning operations were determined based on given thresholds for a basal area. Rotation lengths were taken according to
Height = 𝑇𝑆 1000
𝐶
𝑥 [1.3 + 𝐷𝐵𝐻2
𝐴 + 𝐵𝑥𝐷𝐵𝐻] (2)
Russian forest management recommendations in all cases, i.e. 100 years rotation was used for spruce and pine and 70 year rotation for birch (Ob ustanovlenii vozrastov rubok 2015).
2.4 Data analysis
First, reliability of model calibration results was analyzed based on average, minimum and maximum difference in percents between simulated values and growth and yield table’s values for each stand characteristic. Also standard deviation of the average difference was calculated using the following formula:
where xi s the average difference [%], µ s the mean of average difference [%], N is the total number of values.
Secondly, differences between management regimes were analyzed based on intensity and number of thinnings (Finnish and Russian ones) and the average volume of growing stock, volume growth and mortality, as well as volume of harvested saw and pulp wood in thinnings and final cut. Stand volume, average tree height and diameter at the end of rotation were also compared in all regimes.
The productivity of forest stands under different management regimes were compared based on total amount of harvested wood. Economic profitability assessment was based on Net Present Value (NPV) using following equation:
NPV = ∑ Ct/(1+r)t (5)
Where Ct is the net cash inflow during the period t; r = discount rate, and t = time of thinning or final cut. The net cash inflow was calculated as stumpage price of saw and pulp wood, for wood harvested during the thinning or final cut. For the calculations, an average prices for stumpage wood in Central Federal Region of Russia in spring 2018 were used (LesOnlajn 2018). Prices were found in rubbles and converted into euros at 73 rub to 1 euro conversion rate.
STD = 1
𝑁 (xi−μ)2
N
i=1
(4)
To consider uncertainties in Russian economy, two discount rates were applied in the profitability analysis. Current discount rate according to the Central Bank of Russian Federation is 7,25%, which is very high and not favorable for the long-term investments.
However, it is forecasted to decrease to the 4% rate in early 2019 (Trading economics 2018). It is still higher then typically used in Finnish forecasts, i.e. 2-3%.
Operational costs of commercial thinnings are higher than in final felling. Therefore, the price of thinning wood is 50% of that in final cut. Tending of seedlings was considered to be done manually by the forest owner, thus costs of these operations were excluded from calculations. Regeneration was also expected to happen naturally and its cost (incl.
possible soil preparation) were not considered. However, the average cost for tending of seedlings in Moscow area is about 200 euroha-1 (LesOnlajn 2014).