Journal of Environmental Management 302 (2022) 114099

Contents lists available at ScienceDirect

Journal of Environmental Management

journal homepage: www.elsevier.com/locate/jenvman

Improving forest management by implementing best suitable timber

harvesting methods
Leo Gallus Bont a, *, Marielle Fraefel b, Fritz Frutig a, Stefan Holm a, Christian Ginzler c,
Christoph Fischer d
a
Swiss Federal Institute for Forest, Snow and Landscape Research (WSL), Sustainable Forestry Group, Zuercherstrasse 111, CH 8903, Birmensdorf, Switzerland
b
Swiss Federal Institute for Forest, Snow and Landscape Research (WSL), GIS Group, Zuercherstrasse 111, CH 8903, Birmensdorf, Switzerland
c
Swiss Federal Institute for Forest, Snow and Landscape Research (WSL), Department of Land Change Science, CH 8903, Birmensdorf, Switzerland
d
Swiss Federal Institute for Forest, Snow and Landscape Research (WSL), Scientific Service NFI, Zuercherstrasse 111, CH 8903, Birmensdorf, Switzerland

A R T I C L E I N F O A B S T R A C T

Keywords: Efficient forest operations are required for the provision of biodiversity and numerous ecosystem services, such
Ecosystem services as wood production, carbon sequestration, protection against natural hazards and recreation. In numerous
Forest operations countries, under difficult terrain conditions, the costs of forest management and harvesting are not covered by
National forest inventory
timber revenue. One possible option to increase the cost-effectiveness of the forestry sector is the application of
Natural resource
Management
state-of-the-art harvesting and extraction techniques, so-called best suitable harvesting methods. We present a
Spatial decision support system case study from Switzerland, where a lack of competitiveness in the forestry sector is of particular interest, with
the aim of quantifying the efficiency gains if estimated best suitable harvesting methods were to be rigorously
applied instead of the currently applied harvesting methods. For this purpose, we developed a spatial decision
support system to allocate estimated best suitable harvesting methods to plots, while concurrently considering
hauling route limitations, extraction route properties and stand characteristics. Our approach was based on
productivity models and supported with expert-defined decision trees. The evaluation of the estimated best
suitable harvesting methods and the comparison with the currently applied harvesting methods were completed
for all 6500 National Forest Inventory (NFI) plots in Switzerland. We draw the following three major conclusions
from our study: First, our modeling approach is an effective method to allocate estimated best suitable harvesting
methods to NFI plots. Second, applying estimated best suitable harvesting methods would lead to cost reductions,
in particular in the regions that include steep terrain and where harvesting mainly relies on cable- and air based
extraction methods. Third, assuming an average timber price of 75 CHF m − 3, 64 % instead of 52 % of the forest
area could be harvested economically over the whole country if estimated best suitable methods were applied.
This advantage would mainly be caused by a shift towards more mechanized harvesting methods. Improving the
cost-effectiveness of the forestry sector is of high global relevance, as the increased use of domestic timber re­
sources is a cost-efficient way to reduce atmospheric carbon emissions. The methodological framework described
here was developed for Switzerland in particular, but it could be applied to Central Europe and other parts of
Europe with a large amount of mountain forests.

1. Introduction forestry sector is under discussion (Korhonen, 2016; Köhl and Linser,
2020), which means that revenue gained from timber sales is less than
Efficient forest operations are a requirement for the provision of the total harvesting costs. However, even if the logging operation itself is
biodiversity and numerous ecosystem services, such as wood produc­ not economical, forest management remains the most efficient way to
tion, carbon sequestration, protection against natural hazards and rec­ administer these services. This applies in particular to protection forests
reation (Blattert et al., 2018). Managed forests also lower the risk of in the mountains, which must fulfill a certain silvicultural requirement
disturbances such as wind storms (Temperli et al., 2020; Maurer and in order to have their optimal protective effect (Heinimann and
Heinimann, 2020). In several countries, the competitiveness of the Stampfer, 2003; Frehner et al., 2005). Therefore, even forests that have

* Corresponding author.
E-mail address: [email protected] (L.G. Bont).

https://doi.org/10.1016/j.jenvman.2021.114099
Received 12 April 2021; Received in revised form 2 November 2021; Accepted 10 November 2021
Available online 18 November 2021
0301-4797/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
L.G. Bont et al. Journal of Environmental Management 302 (2022) 114099

high harvesting costs are managed (Bont and Church, 2018). However, to fully mechanized systems would increase efficiency and lower the
forest management that does not cover costs jeopardizes the provision of number of accidents and stand damage. The lowest productivity in
ecosystem services in the long term (Bürgi et al., 2018). timber felling and processing in this study was reported in Spain, with
This applies in particular to Switzerland. The Swiss Forest Policy 1.3 m3 h− 1, while the highest was in Sweden, with 16.6 m3 h− 1. The
2020 indicator 3.6 (FOEN, 2013) states: “The economic efficiency and difference in productivity can be partly explained by a different grade of
performance of the forestry sector shall improve”. One way to increase mechanization (harvester vs. chainsaw). Cable-yarders were considered
economic efficiency would be to lower the expenses of timber harvest­ the most appropriate extraction technology in steep terrain, but it was
ing, as they account for about 60 % of the total cost of forest manage­ acknowledged that they require a well-developed road network for
ment of a forest enterprise (BAFU, 2017; Bürgi et al., 2018). We bringing the equipment into the forest.
hypothesize that the cost-effectiveness could be increased through the Similar studies have been conducted in other parts of the world, such
application of so-called estimated best suitable harvesting methods. By as in the Caspian Forest in Northern Iran, with the result that the ‘rubber-
‘estimated best suitable’ we mean a technically feasible method that is tired skidder’ is the best alternative (Jaafari et al., 2015). Kaakkurivaara
compliant with environmental and occupational health and safety de­ and Stampfer (2018) conducted a study in small-scale forests in steep
mands (e.g. soil protection) but also involves the most economical terrain in Thailand. They concluded that a shift from motor-manual
timber harvesting method (including tree-felling, processing, and off- cut-to-length methods towards partly-mechanized systems would
and on-road transportation). This is especially relevant for Switzerland, significantly improve productivity (11 % cost reduction), as well as
where about 70 % of the consumed timber is imported (Leyder et al., improve worker safety.
2021), because domestic products are comparably expensive due to high Some methods have already been developed to assign optimal or best
salaries. Leyder et al. (2021) also concluded that the use of local instead suitable harvesting methods to forest areas. Heinimann (1998) pre­
of imported timber for construction purposes leads to a 45 % reduction sented a model to determine whether ground- or cable-based extraction
in greenhouse gas emissions, a 30 % reduction in non-renewable energy systems should be applied under different terrain conditions. This
consumption, and an increase in local employment and gross value theoretical model assumed homogeneous terrain and did not provide
added. Thus, the application of best suitable harvesting methods would spatially explicit results. To deliver spatially explicit recommendations,
increase the economic efficiency of the domestic forestry sector and, at methods based on operations research (OR) were developed. The first
the same time, lower associated impacts, e.g. through the reduction of application of OR methods for designing harvesting layouts was pro­
fuel consumption in transportation, and shall therefore be encouraged. posed by Dykstra and Riggs (1977). With this approach it was possible to
The problem has high global ecological relevance. For example, 29–39 identify minimum-cost cable-yarding units and assign optimal
% of deforestation-related emissions were driven by international trade cable-based harvesting methods to forest areas. Further developments in
in 2010–2014, mainly of agricultural but also of forest products cable-based terrain were made by Chung and Sessions (2002), Chung
(Henders et al., 2015; Pendrill et al., 2019). et al. (2004), Bont and Heinimann (2012), Bont et al. (2014, 2019) and
Seidl et al. (2007) emphasized the importance of timber harvesting Bont and Church (2018). For trafficable terrain where ground-based
and sustainable forest management strategies that serve multiple de­ methods can be applied, OR methods are based on graph theory, as
mands, stating that substantial carbon offsets could be generated from presented by Suvinen (2006), Dupire et al. (2015) and Bont et al. (2018).
the potential substitution of fossil fuels and that carbon sequestration However, these approaches only check the soil trafficability of forest
through forest management can be a cost-efficient way to reduce at­ areas and ensure that these areas are connected with a forest road. They
mospheric CO2. The role of sustainable forest management becomes do not assign harvesting methods to the forest areas. One reason for this
even more important when time spans of more than 100 years are is that stand properties such as diameter class or conifer/broadleaved
considered. Werner et al. (2010), Mehr et al. (2018) and Leyder et al. distribution must be known in order to assign harvesting methods.
(2021) underlined that domestic timber harvesting, in combination with In addition to OR-based tools, expert-based approaches are also used,
carbon storage in wood products and substitution effects, outperforms especially in situations where a wide range of systems in both trafficable
no- and low-management strategies in forests, and further decreases and non-trafficable terrain need to be covered. Meyer et al. (2001)
systemic environmental impacts. No- and low-management strategies developed an optimized timber harvesting concept for four forest en­
lead to a large short- or mid-term accumulation of carbon in the forest terprises in Switzerland, in which the harvesting system assignment is
stand, which becomes increasingly unstable in the long term and rep­ based on expert assessment. Kühmaier and Stampfer (2010) conducted a
resents a major carbon source when those stands collapse (Seidl et al., comparable study but combined expert assessment with multi-attribute
2017). utility theory to fulfill several competing objectives.
Best suitable timber harvesting methods have been evaluated in All of the above-mentioned approaches lead to reasonable results.
several studies. Kühmaier and Stampfer (2010) developed a The disadvantage of OR methods, however, is that they do not cover the
multi-attribute decision-support system for selecting timber harvesting entire range of options, as the models would then become computa­
methods. This system suggests that the most suitable method consid­ tionally too complex. Further, OR methods require area-wide data,
ering stakeholder interests, technical feasibility and environmental which are usually not available for entire countries in homogeneous
conditions. The system suggests that a combination of increasing forest quality. This also limits their application. On the other hand, expert-
road density and changing to best suitable harvesting systems could lead based approaches are easier to implement but require a decision
to enhanced economic efficiency, less stand damage and fewer worker layout that is tailored for the particular study area. For example,
injuries. In a project area, through a combination of increasing forest currently used expert-based approaches, such as the one proposed by
road density and technology improvement (use of cable-forwarder), the Meyer et al. (2001), do not incorporate hauling costs, which are
contribution margin (revenue minus harvesting costs, without road costs particularly relevant when the forest road network consists of roads with
or overhead) was increased from 40 to 56 € m− 3 (+40 %). different standards.
Enache et al. (2016) assessed the current logging practices in Euro­ In summary, it can be said that: (1) existing studies indicate that,
pean mountain forests. They further highlighted existing efficiency gaps through the consistent use of best suitable, optimal timber harvesting
and identified opportunities for increasing efficiency in timber produc­ methods or technology improvement, timber harvesting could be car­
tion. Their recommendations to increase efficiency were based on a ried out more cost-effectively compared with current practice and (2)
comparison of multiple case studies. An automatic assignment of best approaches with tailored components for the specific regions, countries
suitable or optimal harvesting methods was not implemented. Enache or conditions are needed to determine best practices.
et al. (2016) reported that non-mechanized harvesting systems have the We therefore state the following overall study objectives: (1) to
lowest efficiency and the highest environmental footprint, while a shift develop a method to assign estimated best suitable timber harvesting

2
L.G. Bont et al. Journal of Environmental Management 302 (2022) 114099

methods to sample plots and (2) to quantify the efficiency gains if esti­ 4) Expert knowledge, which was used in particular for the definition of
mated best suitable timber harvesting methods were rigorously applied. decision trees for deriving the technical framework in which a
This leads to the following specific research questions: certain harvesting method can be applied.

• Which harvesting methods would be used, and how frequently, if The Swiss NFI consists of a systematic 1.41 km × 1.41 km sampling
estimated best suitable methods were systematically applied grid covering the whole country, resulting in about 6500 plots in the
(substitution)? forest (Lanz et al., 2016, 2019; Brändli and Hägeli, 2019, Fig. 1). The NFI
• How large are the cost savings and how much additional forest area further categorizes the forest into production regions, as visualized in
can be harvested economically through the systematic application of Fig. 1: Jura, Plateau, Pre-Alps, Alps and Southern Alps. Originally, the
estimated best suitable methods (economic benefits and efficiency NFI was developed to measure the forest conditions, through measure­
gain)? Are there regional differences? ments such as standing volume and tree species distribution. More
recently, a survey of timber harvesting and forest access was performed
As an innovative feature of our methodology, we combined pro­ within the NFI framework (Keller, 2013). It reports the currently applied
ductivity models with expert knowledge to answer these questions. method if a cut was done in the period since the last NFI survey or, if this
For this study, we assumed that regular conditions prevail when no was not the case, according to the forester’s assessment. One example of
damage after extreme events, such as storms or bark beetle infestation, this survey is the attribute ‘extraction means’, which reports the means
needs to be cleaned up and timber harvesting can be carried out during used for timber extraction for each plot. This information is used
good conditions, e.g. when soils are not water saturated. together with other information to derive the trafficability of a plot, as
For this approach we used the country of Switzerland, situated in shown in the section ‘Harvesting Options’. Another survey attribute that
Central Europe along the Alpine arc, as a study area. Most forests in is relevant to derive the trafficability of a plot is the attribute ‘site type’,
Switzerland are managed to achieve multiple biodiversity and which contains information about soil moisture conditions. The NFI
ecosystem services, with a combination of economic, protective, recre­ interview survey took place in the office, but the foresters know the plot
ational and ecological services. About 32 % of the country is covered by location and properties, such as species and DBH distribution.
forest (Abegg, 2014). An important reason for choosing Switzerland as
the study area is the data basis, especially that provided through the 2.2. Estimated best suitable timber harvesting approach
Swiss National Forest Inventory (NFI). It delivers a wide range of in­
formation on forests using a combination of field assessments and an 2.2.1. Defining estimated best suitable timber harvesting methods
interview survey. Through the interview survey, which includes all local ‘Best suitable’ refers to harvesting methods that are most effective in
foresters in Switzerland (about 830), the NFI documents the currently terms of technical efficacy, economic efficiency, ‘environmental con­
applied harvesting method and is able to model timber harvesting costs sumption’ and socio-economic factors (work safety, expertise, available
for each field plot. Using these NFI data and derived information, we work force). Defining an estimated best suitable harvesting system for a
could thus contribute to the current discussion by modeling estimated particular site requires close attention to all of these factors. In partic­
best suitable harvesting methods, and by comparing these with the ular, system productivity is highly dependent on specific site conditions,
currently applied methods. Further, Switzerland is a well-suited test area silviculture treatment, work organization, and available workforce. Our
because it has a diverse topography, from flat areas, to moderately steep main focus in the selection of an estimated best suitable harvesting
areas, to high alpine and very steep conditions that rely on cable- or system was the economic efficiency, but we used other factors as con­
air-based harvesting means. Results from this study could serve as a straints in the selection process.
basis for informed forest management decisions. We aim to develop a Table 1 lists almost all harvesting systems currently applied in
framework that can be adapted for other regions by making minor ad­ Switzerland, which forms the basis for the selection of the estimated best
justments for local conditions or particularities. suitable methods. However, some methods are excluded in the imple­
The first step in this endeavor was to develop a general framework mentation of the estimated best suitable method. It is important to note
with tailored components to assign estimated best suitable harvesting that the methods used today are often applied because enterprises or
methods to sample plots. We used NFI field plots as sample plots in this contractors have the corresponding machinery and want to use their
study and refer them as ‘plots’ throughout this paper. We then compared own equipment and staff to full capacity (see also the Discussion sec­
the estimated best suitable harvesting methods with the currently tion). There are also a number of methods that used to be regarded as
applied harvesting methods. Along with reporting and discussing our promising but are no longer considered best suitable. The method PM_SK
findings, in this paper we discuss the wider applications of our approach. (motor-manual felling and full tree extraction with a skidder or tractor)
is not considered a best suitable method, as it is more cost-efficient to
2. Methods process (delimb and cut-to-length) in the stand (harvester) than to
extract full trees and process them on the forest road with an excavator
2.1. Data sources or a harvester. To underline this point, we took all 208 NFI plots where
PM_SK was reported in the NFI survey (Keller, 2013) and we used
Various data sources were used to derive the estimated best suitable HeProMo (Holm et al., 2020) to calculate the costs for PM_SK and the
harvesting methods. The data can be divided into several groups: fully mechanized MM_FW system. The latter system emerged as having
the lower costs in 90 % of the plots. In the other plots, however, the
1) GIS data, such as a digital terrain model with a resolution of 2 m, the differences are very small (see Appendix: Figure Appendix 01).
forest road network collected through the NFI (Müller et al., 2016), a Furthermore, motor-manual felling and processing is known as the most
forest boundary map (Waser et al., 2015) and the NFI field plot dangerous work phase (Axelsson, 1998; Kühmaier and Stampfer, 2010;
locations. Laschi et al., 2016; Kaakkurivaara and Stampfer, 2018), which also in­
2) NFI field measurements and observations on plots, such as site dicates that a more highly mechanized harvesting system is more suit­
characteristics, diameter at breast height (DBH) of trees, coniferous able. The application of PM_SK is a typical case for enterprises that own a
or broadleaved species, and obstacles for harvesting. Each plot tractor and want to use it to full capacity.
covers an area of 500 m2 (Düggelin et al., 2020). In the case of PM_LY (motor-manual felling and extraction with a
3) NFI interview survey data (Fischer and Fraefel, 2019) on currently long-distance yarder), the method is not considered the most suitable
applied harvesting methods and systems and forest access. system. Downhill yarding of full trees with LYs is technically not
feasible, because LYs are having a two-rope system. Downhill yarding of

3
L.G. Bont et al. Journal of Environmental Management 302 (2022) 114099

Fig. 1. Overview of the Swiss National Forest Inventory (NFI) plots and the NFI production regions considered in this study. (Coordinate System: WGS 84,
EPSG 3857).

full trees requires a three- or four-rope system, as pulling the carriage 2.2.2. Conceptual model
downhill might be necessary when full trees are touching the ground. The conceptual model applied in this study is visualized in Fig. 2. The
Uphill yarding is technically feasible, but it results in higher set-up and wood-harvesting process is captured, from trees standing in the plots to
dismantling costs compared with tower-yarders (Holm et al., 2020). For their delivery at a connection point, which is defined as the points where
this reason, (processor) tower-yarders (PTY/TY) are the preferred sys­ the route reaches the superordinate road network. This superordinate
tem when yarding full trees uphill. However, the cable road length of a road can always be used by the largest truck type considered in the
TY is limited to about 800–1000 m. Above this distance, a LY has to be study, as described below. The connection points were set, indepen­
used, but it usually has a low logging productivity, i.e. < 4–6 m3 h− 1, for dently of the plots, at each point where there is a forest road to super­
large yarding distances, as shown in the appendix (Table Appendix 01), ordinate road connection. At the beginning (Fig. 2), the [a] initial
which causes high processing costs per m3 timber. Those processing situation is given with the characteristics of the plots, the terrain and the
costs are higher than the additional costs of manual processing (cut-­ road network. Then, the workflow comprises the following subsystems:
to-length and delimbing), i.e. 10 to 25 CHF m− 3, in the stand, according [b] ‘harvesting options’, which identifies, for each plot, all landings (the
to Holm et al. (2020) (see Appendix: Figure Appendix 02). The impor­ term ‘landing’ is used to describe the transition point from off-road to
tance of matching the various system components to each other was also on-road transport) from which the plot can be accessed and the corre­
emphasized by Kizha et al. (2020). sponding possible harvesting method for timber extraction (plot to
The method MM_H (motor-manual felling and processing, logging of landing). We assumed that a landing can be installed anywhere along a
assortments with helicopter) involves flying assortments by helicopter. forest road, as done in Fraefel et al. (2021); [c] ‘harvesting cost’, which
This method was previously used with older types of helicopters, which estimates the cost of each harvesting and extraction technique; [d]
could only be flown with a low payload. Since modern helicopters have ‘hauling routes’, which identifies the best route from each landing to one
higher weight capacities and the stands are far apart and mostly in very of the available connection points; [e] ‘hauling costs’, which estimates
steep terrain (low productivity during processing, high risk of acci­ the cost of the timber on-road transportation. The computational results
dents), today almost all flown timber in Switzerland is full trees that are from these subsystems are used to identify the estimated best suitable
processed in a fully mechanized manner on the forest road. Nowadays, harvesting method [f], which is characterized through an extraction
assortments are only occasionally flown by helicopter for scattered use (off-road) and hauling (on-road) route, a harvesting system and the
(forced use after windthrow), where the woodcutters can process the corresponding cost estimation.
timber during waiting periods in the stand. Other outdated methods
include the walking harvester in combination with a TY or hand- 2.2.3. Harvesting options
skidding (by gravity), as found under the category ‘other’ in Table 1. The ‘harvesting option’ subsystem evaluates possible best suitable
The harvesting methods considered best suitable represent the state of harvesting and wood extraction options for each plot (Fig. 3). The
the art in Central Europe, covering flat areas as well as mountainous analysis was done using the open-source database PostgreSQL with the
terrain. The use of winch-assisted harvesters and forwarders was not GIS extension PostGIS.
considered as a separate category, as winches should only be used to Therefore, the following steps were performed:
increase safety and reduce soil compaction and not to extend the
maximum gradeability (Bont et al., 2018). 1. Define landings: As countrywide data about existing landings were
not available, we assumed that landings can be installed on each road

4
L.G. Bont et al. Journal of Environmental Management 302 (2022) 114099

Table 1 beyond, and is 1500 m long (maximum cable road length). This is
Overview of currently applied harvesting methods and estimated best suitable necessary because the properties of an entire slope must be consid­
harvesting methods in Switzerland. The abbreviations of the harvesting methods ered for cable-based methods in order to determine the estimated
are defined as follows: [1] felling and processing mode (MM, PM, FM) and [2] best suitable harvesting method. The characteristics of the EER are
extraction means (SK, FW, TY, LY, H), with MM = motor-manual felling and only relevant for cable-based harvesting systems and not for ground-
processing, PM = motor-manual felling in the stand and fully-mechanized pro­
or air-based harvesting systems. Table 2 gives an overview of the
cessing at the landing (partially-mechanized), FM = fully-mechanized felling
characteristics that were taken from the (extended) extraction routes
and processing, SK = skidder, FW = forwarder, TY = tower-yarder, LY = long-
distance yarder, and H = helicopter. The IDs are only used for estimated best and used as input for step 4.
suitable harvesting methods. 4. Apply decision trees: The calculated properties of the potential
extraction routes, plot and interview data from NFI (Table 2) were
Abbreviation Description of harvesting Currently Selected as
method applied estimated best
used as input for expert-defined decision trees. Applying the decision
suitable trees, we assessed the suitability of each potential extraction route
for different timber harvesting methods. One potential estimated
MM_SK Motor-manual felling and x x
processing, skidding with best suitable harvesting method was assigned to each potential
skidder (assortments) extraction route (Fig. 3.3 and 3.4).
MM_SK_FW Motor-manual felling and x
processing, pre-skidding with
skidder, forwarding
(assortments) 2.2.3.1. Derivation of the slope length. Some attributes listed in Table 2
MM_FW Motor-manual felling and x are not trivial to derive and need some explanation, such as the length of
processing, forwarding the entire cable road on a slope (referred to as ‘slope length’). This
(assortments)
parameter is used to specify different cable-based methods. To deter­
PM_SK Motor-manual felling, skidding x
with skidder (full tree),
mine the slope length, we first evaluated the maximum feasible length of
processing at forest road a cable road with a given number of intermediate supports (five being
FM_FW Fully mechanized felling and x x the default) (Fig. 4a). This maximum feasible length further depends
processing with harvester (or upon the properties of both the terrain and the yarding system (e.g.
tracked/crawler harvester),
breaking strength of the skyline, minimum clearance between the
forwarding (assortments)
MM_TY Motor-manual felling and x x skyline and the ground, maximum skyline length). To compute the
processing, cable-based skyline properties, we implemented the design approaches of Pestal
harvesting (tower-yarder, (1961) and Zweifel (1960), running the former by default because of its
assortment yarding)
better calculation efficiency (Bont and Heinimann, 2012). In a second
MM_LY Motor-manual felling and x x
processing, yarding with long-
step, we reduced the maximum possible slope length to a plausible slope
distance yarder (assortments) length (= effective cable road length), as visualized in Fig. 4. In this
PM_TY Motor-manual felling, yarding x x calculation, if the maximum feasible cable road crossed another forest
with tower-yarder, processing road or ended outside the forest, the effective cable road end was
(full tree), or yarding and
shortened to the forest road (Fig. 4c) or to the forest edge (Fig. 4d).
processing with tower-yarder
with mounted processor
PM_LY Motor-manual felling, yarding x 2.2.3.2. Assigning harvesting methods using expert-defined decision trees.
with long-distance yarder, Although optimization models are available as an aid for selecting the
processing (full tree)
harvesting method, we decided to conduct the selection based on expert-
MM_H Motor-manual felling and x
processing, logging with defined decision trees combined with productivity models. The
helicopter (assortments) following reasons were decisive. First, most models that optimize the
PM_H Motor-manual felling, logging x x harvesting layout are based on spatial data (e.g. Bont et al., 2014, 2019;
with helicopter, processing (full Bont and Church, 2018). Since we conducted the study over an entire
tree)
Other Other, walking harvester and x
country, it would have been difficult to combine data of varying quality
tower-yarder, mobile chipper, from different sources. Second, the choice of a timber harvesting system
‘hand-skidding’ has to be made on the basis of several criteria. An ideal solution is
technically feasible, economically efficient, environmentally friendly
and ergonomic. For this purpose, both the terrain and silvicultural
segment of the forest road network. To keep the modeling effort
treatment must be taken into account. All these factors have yet to be
within reasonable bounds, the forest road network was split into 100-
integrated into an optimization model. A simplified approach for the
m-long road segments. The locations of the possible landing were
automatic delineation of harvesting methods was develop by Bont et al.
defined at the start and end of each road segment, resulting in one
(2019). However, incorporating all relevant factors, as well as decision
landing every 100 m along a forest road (Fig. 3.1).
making, remains difficult, even if there are productivity models avail­
2. Evaluate extraction routes: Potential extraction routes were
able that can be used to evaluate the productivity and cost of different
calculated for each plot by connecting it with the potential landings
systems (Fulvio et al., 2017).
(Fig. 3.2). Each potential extraction route had a maximum slope
The scheme we used to assign a harvesting system to each extraction
length of 1500 m (maximum cable road length of a LY; the term
route consisted of three steps and followed the order of preference
‘cable road’ is used to refer to the transport line on which the timber
recommended by Heinimann (2003): (1) ground-based, (2) cable-based
is yarded). The extraction routes always ended at a landing. Potential
and (3) air-based. First, we evaluated whether a ground-based harvest­
extraction routes from different plots could end at the same potential
ing system could be used, or whether a cable- or air-based system was
landing. For simplicity, the extraction routes are drawn as straight
required (Fig. 5). Then, we determined the specific method, either under
lines in Fig. 3.2.
the ground-based category (Fig. 6) or under the cable- or air-based
3. Determine extraction route characteristics: Although the extrac­
category (Fig. 7). Every extraction route was assigned to only one har­
tion route connects the landing with the plot, characteristics over a
vesting method. However, several harvesting methods could be assigned
longer distance were analyzed, i.e. over an ‘extended extraction
per plot.
route’ (EER) that starts at the landing, goes through the plot and

5
L.G. Bont et al. Journal of Environmental Management 302 (2022) 114099

Fig. 2. Conceptual model for the assignment of the estimated best suitable harvesting method and hauling route. Landings are not explicitly drawn, as in our case
each forest road segment also serves as a landing.

Trafficable terrain is a basic requirement for ground-based harvest­ harvesting in a broadleaved stand results in slightly lower productivity
ing. Factors affecting trafficability are the frequency and size of obsta­ compared with fully mechanized harvesting (Krč et al., 2015), the
cles, such as rocks, the slope and the soil bearing capacity (Ahlvin and former is nevertheless more cost-effective (Liepiņš et al., 2015). As
Haley, 1992; Soman et al., 2019). Further, the soil bearing capacity detailed studies are not available, the thresholds in Fig. 6 are based on
changes periodically because of meteorological events such as rain, frost the one hand on the research experience of the authors, reflecting their
and snowfall (Eichrodt, 2003). For our trafficability evaluation, we assessment, and on the other on discussions with various enterprises in
assumed that conditions were basically favorable. This means that the Switzerland. There is a certain degree of uncertainty in this threshold
soil water content was low, i.e. no precipitation had fallen in the pre­ setting, in that the values could easily be offset by ±10 %. In one of the
vious few days. Modeling trafficability over large areas is challenging; author’s experience, a skidder rather than a forwarder is used for stands
methods do exist, such as those proposed by Niemi et al. (2017), but they dominated by broadleaves and with a mean DBH >50 cm, for load
are inadequate in most situations due to a lack of basic data, such as soil weight reasons.
load-bearing capacity, plasticity, shear strength, skeletal fraction (min­ Fig. 7 shows the expert-defined decision tree used to assign the
eral soil components that are larger than 2 mm), permeability and specific harvesting method from the various cable- and air-based
water-holding capacity. Therefore, we derived the trafficability based on methods. The decision tree was based on a review of best suitable ap­
the NFI interviews (Fischer and Fraefel, 2019) and a nationally available plications of timber harvesting methods in steep and non-trafficable
soil suitability map (BfS, 2000), as outlined in Fig. 5. In a first step terrain (Frutig and Thees, 2011). The values indicated in the decision
(Fig. 5a and b), the site type and the recorded extraction mean was tree also include a certain degree of uncertainty and may vary, as they
checked to evaluate the trafficability of the plot. In a second step are based on the research experiences of the authors, thus reflecting
(Fig. 5c), the spatial course of the extraction route was defined and its their assessment, as well as on discussions with various forest contrac­
feasibility checked, according to the method described in Fraefel et al. tors in Switzerland.
(2021). In this way, the maximum slope of the extraction route was We specified that the following conditions must be met for the use of
compared with a soil-specific threshold for trafficability (slope percent) a cable-based harvesting method (Fig. 7a). The extraction distance
derived from the soil suitability map, which lies between 20 % and 35 % should not exceed 1500 m, which is an economic criterion based on
depending on the soil classification. Features like streams, if indicated in productivity (see). The maximum inclination of the cable road is a
the soil suitability map, were also included in the analysis. feasibility criterion (Aggeler, 2002) and is limited to 100 %. In addition,
If classified as a trafficable extraction route, the assignment of a we required the forested slope length (FSL) to be at least 60 %, because
specific ground-based harvesting method did not depend on the topog­ of feasibility (availability of trees as intermediate supports) and cost
raphy, but on species distribution (share of conifers, Fig. 6) and mean effectiveness. Otherwise the timber accumulation might be too small for
DBH. Harvester (head) development was mainly conducted for softwood an economic operation. According to Heinimann (2003), the economic
harvesting, for which it reflects the best suitable method in terms of costs threshold to switch the best suitable harvesting method from
and worker safety (Mederski et al., 2018). In contrast, fully mechanized cable-yarder to helicopter is at about 0.4 m3 of wood accumulation per
processing of deciduous trees is more complex, because of large meter cable road. Then a check is performed (Fig. 7b) if a cable-based
branches, forks, severe crooks and sweeps, and a diverse crown archi­ extraction method was reported in the NFI survey (Keller, 2013): if
tecture, resulting in a lower productivity compared with conifer pro­ ‘yes’ cable-based methods remain possible. The selection of the best
cessing (Labelle et al., 2018). Although the use of motor-manual suitable cable-yarding systems is carried out according to the criteria

6
L.G. Bont et al. Journal of Environmental Management 302 (2022) 114099

Fig. 3. Intermediate results of the steps defined in the workflow for the derivation of the estimated best suitable harvesting method. The abbreviations for the
harvesting methods are defined as follows: [1] felling and processing mode (MM, PM, FM) and [2] extraction means (SK, FW, TY, LY, H), with MM = motor-manual
felling and processing, PM = motor-manual felling in the stand and fully-mechanized processing at the landing (partially-mechanized), FM = fully-mechanized felling
and processing, SK = skidder, FW = forwarder, TY = tower-yarder, LY = long-distance yarder, and H = helicopter.

Table 2
Input variables needed for the derivation of the estimated best suitable method.
Abbreviations of sources: ER = extraction route, EER = extended extraction
route, NFI = National Forest Inventory records, Cat = categorical variable.
Attribute Shortcut Unit Source

Transport direction TD [] ER Code: 1 uphill, 2

downhill
Slope SI [ %] EER/
ER
Extraction distance (slope ED [m] ER
distance)
Slope length SL [m] EER
Forested slope length (length FSL [ %] EER
of the slope with a forest
cover)
Crossing of a forest road CFR [] Code: 1 = true, 0 Fig. 4. Derivation of the slope (cable road) length. The maximum possible
= false slope length (a) is reduced to a plausible slope length by considering obstacles
Site type ST Cat NFI (b), other forest roads (c) or forest cover (d).
Extraction method EM Cat NFI ground-based,
cable-based, air-
based listed in boxes (c)–(e) of Fig. 7. According to Heinimann (2003), systems
Volume share of broadleaves VB [ %] NFI with a small share of personnel costs are preferable to systems with a
Volume share of conifers VC [ %] NFI large share; this applies in particular to Switzerland, where the pro­
Mean diameter at breast mDBH [cm] NFI
height
duction factor “labor” is very expensive. Furthermore, more highly
mechanized systems are also preferable in terms of work safety (Laschi
et al., 2016). The choice of harvesting systems is therefore made with the
following priority: (1) PM_TY (full tree), (2) MM_TY (cut-to-length) and

7
L.G. Bont et al. Journal of Environmental Management 302 (2022) 114099

Fig. 7. Expert-defined decision tree to assign the specific harvesting method

from the various cable- and air-based methods (see Table 1 and Table 2 for
explanations of the abbreviations). (*) If there are several criteria, then these
are logical AND operations, i.e. all conditions must be fulfilled for a Yes [1]. If
only one condition is not fulfilled, the result is No [0]. The abbreviations of the
timber harvesting system are defined as follows: [1] felling and processing
mode (MM, PM) and [2] extraction mean (SK, TY, LY, H), with MM = motor-
manual felling and processing, PM = motor-manual felling in the stand and
fully-mechanized processing at the landing (partially-mechanized), SK =
skidder, TY: tower-yarder, LY: long-distance yarder, and H: helicopter.
Fig. 5. Expert-defined decision tree for differentiating between ground-based
and cable- or air-based harvesting methods. yarding of full trees represents a risk of greater stand damage. As illus­
trated below, we have added a constraint regarding the slope for this
issue. For most processor-tower-yarders, an 800 m cable road length is
the technical operational limit and, based on the authors’ experience,
this technical limit is not exhausted in downhill yarding; therefore, a
length of 600 m was set. The mean diameter at breast height (mDBH) is
limited to 70 cm and the volume share of broadleaved trees to 50 %,
both due to the size and hardwood limitations of current processor heads
(Labelle et al., 2018). When yarding full trees downhill, the slope of the
cable road has been limited to 50 %. On the one hand, the full trees must
be attached at the thick end, which requires felling in the uphill direc­
tion, but this is both difficult and dangerous with increasing slope. On
the other hand, more damage to the remaining stand is also risked with
increasing slope with downhill yarding (Heinimann, 2003). If a
processor-tower-yarder cannot be applied, then a tower-yarder or a
long-distance yarder are used, which are selected based on cable road
length. The value of 800 m reflects the feasibility limit of most
tower-yarders (Frutig and Thees, 2011). If no cable-based system is
selected, either the helicopter (PM_H) or the winch (MM_SK) is used
(Fig. 7f). The latter is used if the plot is located next to the road: 50 m for
downhill or 80 m for uphill terrain, according to Meyer et al. (2001).

2.2.4. Harvesting costs

The model subsystem ‘harvesting cost’ involved the estimation of the
Fig. 6. Expert-defined decision tree to assign the specific harvesting method harvesting and extraction cost [CHF m− 3] for each extraction route
from the various ground-based methods (see Table 1 and Table 2 for further identified in the component ‘harvesting options.’ To model productivity,
explanations of the abbreviations). The abbreviations of the timber harvesting we used the HeProMo model collection (Frutig et al., 2016; Holm et al.,
system are defined as follows: [1] felling and processing mode (MM, FM) and 2020) for the following reasons. First, these models had already been
[2] extraction mean (SK, FW), with MM: motor-manual felling and processing,
used in the NFI to calculate costs for the currently applied harvesting
FM: fully-mechanized felling and processing, SK: skidder, FW: forwarder.
methods. Using the same models for both current and estimated best
suitable methods ensured comparability. Second, these models are
(3) MM_LY (cut-to-length). At this point, it would also be quite reason­ particularly tailored to Central European and Alpine conditions, since
able to generally prefer MM_TY with uphill yarding to PM_TY with the databases for these statistical models originate in Switzerland and
downhill yarding, as proposed by Heinimann (2003), as downhill

8
L.G. Bont et al. Journal of Environmental Management 302 (2022) 114099

neighboring Germany (Bavaria, Baden Wuerttemberg and such as bridges, or by the road geometry, such as small radii. Both
Rhineland-Palatinate). And third, these models were derived using connection points and hauling routes were selected as described in Bont
up-to-date data. We calculated the harvesting costs for each potential et al. (2018) by means of a network analysis. The hauling route was
harvesting system for each plot. The estimated best suitable harvesting chosen in such a way that, as a first priority, road limitations caused by
methods were modeled using the productivity models listed in Fig. 8. geometrical dimensions (truck type, bearing capacity) were as low as
The models require the input variables listed in Table 3. The timber possible and, as a second priority, the distance from the road segments to
volume, the stand properties and the terrain properties were taken from the connection point was as short as possible. A similar methodology for
the recorded NFI plot data. To derive the timber volume, we assumed – modeling logistics was used by Kizha. et al. (2015) and Montgomery
as traditionally done in NFI evaluations – that all trees on a plot will be et al. (2016).
harvested, to ensure comparability with other NFI studies. However, this
assumption does not necessarily reflect reality, as it occurs that not all 2.2.6. Hauling costs
trees on a plot are harvested in a cut. In contrast, the timber volume per The cost of hauling timber covers transport costs from the landings
cable road, which is used as input for cable-based methods, corresponds on the forest road to a storage facility, plant or sawmill. However, since
to the forested slope length (share of the cable road within the forest) the ultimate destination of the timber is not known, we defined the
multiplied by a timber volume of 1 m3 per meter of cable road and connection point (connection to the superordinate road network) as the
multiplied by the cable road length. This value is independent of the interface. The transport costs were calculated as follows and included
volume of the plot, because the whole slope along a cable road is always additional costs with regard to the limitations caused by geometrical
harvested. Along with a stand with a low volume density, there can also dimensions and bearing capacity. Input parameters are listed below and
be a stand with a high-volume density along the same cable road, and their values can be found in Table 4. The limitation of each hauling route
vice versa. The extraction route properties are based on the length­ was based on the maximum number of axles allowed per truck (ax).
–elevation profile, and the cable road length amounts to the slope length ( )
ctot (ax, d) = ctrfix,ax * ttot + 2 * ctrvar,ax * d *a−ax1
(Table 2).
Fixed costs for transporting the machines to and from the site
with:
(relocation costs) are not included in the calculation. We thus assume
( ) /
not only that a single plot is harvested in isolation, but that the same ttot = td e + tload * aax + td l + tunload * aax *1 60
method is subsequently applied on other areas in the neighborhood. This
corresponds to reality in almost all cases. The assumptions for the ma­ td e = 60* d*vavg − 1

chine and labor costs can be found in the Appendix (Table Appendix 02).
1
td l = 60*d*vavg −
2.2.5. Hauling options
The model component ‘hauling route’ identifies the ‘best’ hauling where:
routes from the landings to one of the available connection points. The ctot (ax,d): total cost for one turn to the connection point [CHF m− 3],
choice of hauling routes was based on the NFI dataset Forest Access written as a function with the number of axles per truck (ax) and the one-
Roads 2013 (Müller et al., 2016; Fraefel and Fischer, 2019). This dataset way distance (d) as input.
contains the area-wide forest roads that are relevant for timber trans­ ttot : total time for one turn to the connection point [h]
portation and can be used by trucks. The data were collected by means of td e : time driving empty [min]
a survey of approximately 800 forest officers. Among other things, td l : time driving loaded [min]
characteristics regarding drivability were recorded on the basis of the tunload : time for unloading at the connection point [min m− 3] (≈1.0,
following truck categories: 3-axle truck with 26 t total weight, 4-axle according to Holzleitner et al., 2011)
truck with 28 or 32 t total weight, and 5- or 6-axle truck with 40 or tload : time for loading at the landing [min m− 3] (≈1.9, according to
44 t total weight. If the use of 5- or 6- axle trucks is not possible, Holzleitner et al., 2011)
transport can be considerably more expensive, especially if the transport d: one-way distance [km]
distance is long. Hauling limitations could be caused by constructions, ctrfix,ax : fixed truck cost [CHF h− 1] for trucks with ax axles (Table 4)

Fig. 8. Overview of the processes, machines and productivity models used to calculate harvesting costs for each potential harvesting system for each plot.

9
L.G. Bont et al. Journal of Environmental Management 302 (2022) 114099

Table 3
Input parameters for the different productivity models. CR = cable road, DBH = diameter at breast height.
Name of Timber Stand properties Extraction route properties Terrain properties CR properties
productivity volume
model
Plot CR Species Mean stem Mean Hor. Vert. Slope up/ Slope Terrain CR No. of
type volume DBH dist. dist. dist. down- difficulties/ length supports
hill obstacles

Motor-manual x x x x
felling
Motor-manual x x x x
Harvester x x
Skidder x x x
Forwarder x x x x x
Processor-tower- x x x x x x
yarder
Tower yarder x x x x x
Long-distance x x x x x x x
yarder
Helicopter x x x x x

2.2.7. Selecting the estimated best suitable harvesting method

Table 4
As far as possible, harvesting systems and extraction routes that link
Net payload and cost parameters of different truck types for Switzerland, according to
a plot with a forest road segment (landing) were determined in the
Hirt (1997) and Kissling (2020) (1 EUR = 1.09 CHF, 1 USD = 0.90 CHF, Date: June
16, 2021).
subsystem ‘harvesting options’, and possible paths for transportation
from the landings to the connection points were identified in the sub­
Number Total Weight of Net Variable truck Fixed
system ‘hauling route’. To select the estimated best suitable harvesting
of axles weight truck [t] payload costs [CHF truck costs
[t] [t] km− 1] [CHF h− 1] method, the whole transportation chain from the plot to the landing to
the connection point was configured as described below. For each plot,
5 40 17 23 1,67 60
4 32 15 17 1,41 58
the cost of each possible extraction route (connection from plot to forest
3 26 13 13 1,17 58 road) was added to the cost of hauling from the forest road to the
2 18 10 8 1,01 57 connection point. Thus, the costs of all possible connections from the
plot to the connection point were known. The estimated best suitable
method for each plot was then defined as the option with the lowest total
ctrvar,ax : variable truck cost [CHF km− 1] for trucks with ax axles
cost, based on the modeling framework.
(Table 4)
vavg : average speed [km * h− 1]
aax : amount of timber per truck load [m3] for truck with ax axles, aax 2.3. Currently applied harvesting method
= net payload [t]/density fresh timber [t m− 3] (≈0.9) (Table 4)
The parameter vavg refers to the model in Holzleitner et al. (2011), The Swiss NFI has assessed and implemented a wider variety of
where average speed is reported for different functional road classes. We timber harvesting methods than we considered for the estimated best
assumed an average speed of 20 km h− 1, which is a mixed value for local suitable methods, because the Swiss NFI has to cover the full range of
roads (26 km h− 1) and forest roads (15 km h− 1) as they occur in techniques actually used in Switzerland. A complete overview is pro­
Switzerland, irrespective of terrain conditions. Unfortunately, there are vided in Fischer and Stadelmann (2019) and in Table 1 (left side). The
no data available to refine the speed classification further by road type, currently applied timber harvesting method of each plot was determined
as done by Svenson et al. (2016). through the interviews done by the NFI. The software HeProMo (Frutig
If the transport distance is long and there are bottlenecks in the road et al., 2016; Holm et al., 2020) was used for the calculation of the cur­
network, it may also be cheaper to reload onto a larger truck as soon as rent timber harvesting costs for each plot, as done for the estimated best
the bottleneck is passed. suitable harvesting methods.

cttt (ax1 , d1 , ax2 , d2 ) = ctot (ax1 , d1 ) + ctot (ax2 , d2 ) + ctransship 3. Results

with: 3.1. Substitution of harvesting methods on the NFI plots

( ) /
ctransship = ttransship * ctrfix,ax1 + ctrvar,ax2 *1 60
Fig. 9 shows the substitution of the NFI survey timber harvesting
where: methods (left side) with the modeled estimated best suitable timber
ctr : truck-to-truck reload cost [CHF m− 3] harvesting methods (right side), and the frequencies in each case.
ttransship : time for truck-to-truck reload [min m− 3] (≈1.2, according to The percentage of plots with air-based harvesting methods (H)
Holzleitner et al., 2011) decreased slightly (from 18 % to 16 %), whereas the percentage of plots
ax1 : truck size allowed on the first part of the hauling route (with with cable-based harvesting methods (LY/TY) increased (from 26 to 32
limitations), indicated as the number of axles, ax ≤ 4. %). Plots with ground-based harvesting methods decreased slightly (54
ax2 : truck size allowed on the second part of the hauling route % to 52 %). Within the cable-based plots there was a shift towards use of
(without limitations), indicated as the number of axles, ax = 5. the tower-yarder (TY; from 15 % to 21 %), whereas the percentage of
d1 , d2 : one-way distance on the first/second part of the hauling route plots in which the long-distance yarder (LY) is used remained constant at
[km] 11 %.
Both transportation options were computed, one with truck-to-truck Further, in plots that are harvested using ground-based methods, a
reload, cttt (ax1 , d1 , ax2 , d2 ), and one without, ctot (ax,d), and the one with shift towards more mechanized harvesting methods was observed. The
the lower cost was used for the cost estimation. use of a harvester/forwarder (FM_FW) increased from 8 % to 26 % of the
plots, whereas logging with skidders decreased from 35 % to 16 %

10
L.G. Bont et al. Journal of Environmental Management 302 (2022) 114099

Fig. 10. Comparison of the distribution of timber harvesting costs between the
NFI survey (currently applied methods) and modeled estimated best suit­
able methods.

Fig. 9. Diagram showing substitutions in the timber harvesting system between

the NFI survey (left side) and the modeled estimated best suitable timber har­
vesting system (right side), with their percentages. The abbreviations of the
timber harvesting system are defined as follows: [1] felling and processing
mode (MM, PM, FM) and [2] extraction means (SK, FW, TY, LY, H), with MM =
motor-manual felling and processing, PM = motor-manual felling in the stand
and fully-mechanized processing at the landing (partially-mechanized), FM =
fully-mechanized felling and processing, SK = skidder, FW = forwarder, TY =
tower-yarder, LY = long-distance yarder, and H = helicopter.

(MM_SK).
In Fig. 9 the corresponding substitution for each plot is visualized. It
can be observed that even if the proportion of a harvesting method did
not change much, the method applied could still change in a consider­ Fig. 11. Comparison of timber harvesting costs for Swiss production regions
able number of plots. This was the case with helicopters, for example. A and percent of forest area, including sampling errors. The estimated best suit­
able harvesting method is the upper bar and the currently applied harvesting
considerable proportion of the plots declared as helicopter-harvested in
technique is the lower bar.
the survey were assigned to a cable-based method in the modeling, and
vice versa. Some plots also switched towards a less mechanized har­
vesting method, for example from a fully mechanized (FM_FW) to a increased from 30 % to 39 % with estimated best suitable methods and
motor-manual (MM_SK) method. for costs of less than 75 CHF m− 3 from 52 % to 64 % (1 EUR = 1.09 CHF,
1 USD = 0.90 CHF, Date: June 16, 2021). Assuming an average timber
price of 75 CHF m− 3 (sold from forest road), 12 percentage points more
3.2. Economic benefits and efficiency gain

Fig. 10 depicts the distribution of harvesting costs according to the Table 5

3
NFI interview survey and the modeled estimated best suitable method. Percentage of forest area that can be harvested for less than 75 CHF m− in
different Swiss regions.
The figure shows that the cost curve shifted to the left, i.e. that the costs
decreased, when the estimated best suitable harvesting methods were Currently Estimated Increment for estimated best
applied [ best suitable suitable method
applied. The average cost has decreased from 91.5 to 70.3 CHF m− 3. The
%] method [ %]
difference was tested with a t-test and found to be significant (with a p- Absolute Relative
value of 2.2 10− 16). From our experience, the absolute values must be increment increment [
[percentage %]
taken with caution, as productivity models are unprecise. However,
points]
since both comparative values (estimated best suitable, currently
Switzerland 52 64 12 23
applied) are based on the same models, we think that the relative dif­
Jura 84 90 6 7
ferences are more reliable and the difference is significant. Plateau 85 91 6 7
In Fig. 11 the timber harvesting and extraction costs for currently Pre-Alps 49 65 16 33
applied and estimated best suitable methods are compared for all of Alps 22 44 22 100
Switzerland and for individual regions. For all of Switzerland the per­ Southern 15 26 11 73
Alps
centage of forest area that can be harvested for less than 50 CHF m− 3

11
L.G. Bont et al. Journal of Environmental Management 302 (2022) 114099

forest area can be harvested economically if estimated best suitable allow the use of trucks with at least two axles (18 t total weight) for the
methods are applied (Table 5). On the Swiss Plateau, where ground- entire transportation route. Since this was no precondition for the
based harvesting methods are mainly applied, the percentage of forest assignment of the currently applied method, this requirement may have
area that can be harvested for less than 75 CHF m− 3 increases from 85 % led to an increase in the distance from a plot to the forest road, and thus
to 91 % with estimated best suitable methods. The regions Pre-Alps, Alps to the elimination of possible methods (e.g. choice of long-distance
and Southern Alps, which are comprised of steep terrain and mainly rely yarder instead of tower-yarder) when modeling the estimated best
on cable- and air-based methods, could benefit most from rigorously suitable method.
applied estimated best suitable harvesting methods. In these areas, the
percentage of forest area that can be harvested for less than 75 CHF m− 3 4.1.2. Slope length
rises from 49 % to 65 % (Pre-Alps), from 22 % to 44 % (Alps) and from The maximum slope length in our decision tree for a tower-yarder to
15 % to 26 % (Southern Alps) (Table 5). Fig. 11 also indicates the be considered was 600 m for downhill yarding (MM_TY) and 800 m for
sampling errors, which are around 1–2 % (Lanz et al., 2019). uphill yarding (PM_TY), which applies to most systems. However, there
are yarder types on the market (e.g. Valentini V 1000) that allow longer
4. Discussion cable roads. This further explains the shift of methods from tower-
yarders to long-distance yarders.
The objective of our study was twofold: we explored the shift in
harvesting systems of individual plots, and we investigated the influence 4.1.3. Maximum extraction distance
(efficiency gain) of rigorously applying estimated best suitable har­ The maximum extraction distance considered in our analysis was
vesting methods. 1500 m. However, although in theory not economical, in reality longer
distances are feasible and sometimes necessary, for example with long-
4.1. Shift in harvesting methods distance yarders (cable road lengths of up to 2.5 km have been reported
in Switzerland), which could explain the shift from long-distance yarder
We observed the following patterns in the shift in harvesting to helicopter.
methods of the individual plots. First, there would be a general trend
towards more mechanized harvesting methods if estimated best suitable 4.1.4. Stand properties (plots with regeneration)
methods were applied. Second, while the shares of air-based and The timber volume or the mean DBH for samples harvested with
ground-based harvested plots would decrease slightly, the share of ground-based methods or helicopters corresponded to the values of the
cable-based harvested plots would increase. Third, there would also be sample, whereas a length-dependent value was assumed for an extrac­
individual cases of shifts towards less mechanized harvesting methods. tion route with cable-based harvesting. If samples in the regeneration
A general trend towards more mechanized methods was expected. It stage (all small trees) were harvested with ground-based means, very
is expressed, for example, by fully mechanized methods with a harvester low productivity would follow, due to the small tree dimensions and the
and forwarder replacing partially mechanized or motorized methods as low volume. Cable-based harvest extraction routes, on the other hand,
the most common methods in trafficable terrain. are less sensitive and can therefore show higher productivity in cases
One of the reasons why the estimated best suitable harvesting where young stands are harvested. However, this occurs only in situa­
method is currently not applied in many situations is that enterprises tions in which the stock has not actually been considered for thinning or
want to use their own machines and personnel. However, this is also harvesting in reality. This could explain the shift from ground-based to
valid for forest contractors who own expensive machines for fully cable-based methods. In the end, this aspect should also be negligible for
mechanized harvesting, such as harvesters, forwarders and processor- answering our research question (gain of efficiency), as it only concerns
tower-yarders. They must acquire production equipment to cover all samples with a very small standing volume.
or most of the conditions in their area. Depending on their scale of
operation, they therefore select a system that can operate under the full 4.1.5. Model boundaries
set of conditions. Flexibility is therefore important. For example, motor In the decision trees, strict boundaries were drawn between the
manual felling and extraction with a skidder is not the most efficient on methods. For example, in trafficable terrain, the proportion of conif­
all sites, but it is probably the most versatile. Further, scheduling is erous timber was used for delineation. In reality, however, these
easier with universally useable machines. There are also other reasons boundaries are not as strict. If, for example, a stand with 40–50 %
preventing contractors from investing in new equipment. Investments coniferous wood is harvested with a harvester (FM_FW, instead of a
already made must be amortized, and towards the end of the payback motor-manual method, such as the likely proposed estimated best suit­
period these machines may no longer be part of an estimated best able method MM_FW), it can still be expected that the harvesting costs
suitable harvesting method. Every investment is also associated with a will not differ substantially. We therefore believe that these shifts in the
business risk; the machines must be able to be operated at full or high methods only have a very small influence on the outcome of our research
capacity. Further, especially in Switzerland, there are many small forest question (gain of efficiency).
owners and logging is not coordinated between them. As a result, the
machines have a lot of unproductive time, during which they are 4.1.6. Survey reliability
transported on a low loader to a new location. In addition, limitations on The survey has its strengths and weaknesses. It takes place in the
the forest road network, which affects about a quarter of the forested office and the foresters use the available planning documents and their
area in Switzerland (Bont et al., 2018; Fraefel et al., 2021), could restrict memory, as well as the specifications of the NFI, so the statements are
the use of a low loader for transport of equipment, and a considerable usually trustworthy. However, there are also features that are not so
amount of driving on own axles would be required to move harvester reliable. For example, the exact stand composition of the plot is uncer­
und forwarders. tain, as the NFI refers to the main stand and the forester may have in
Interpreting the other two patterns, the increase in the share of cable- mind another stand in the same area.
based harvested plots and the shift towards less mechanized harvesting
systems, is more difficult because they depend on several factors, as 4.1.7. Available machines
listed below: Once equipment for fully mechanized harvesting methods is avail­
able, the forest contractors want to use it to the full extent. This can lead
4.1.1. Road network to the use of fully mechanized methods in situations where other, less
In order to derive the best practices, we only considered roads that mechanized methods would actually be more efficient.

12
L.G. Bont et al. Journal of Environmental Management 302 (2022) 114099

4.2. Efficiency gain of rigorously applying estimated best suitable ‘Harvesting options’: First it must be ensured that the set of po­
harvesting methods tential estimated best suitable harvesting methods is complete, which
might be the case under many Central European conditions. Incorpo­
The second objective of this study was to quantify the economic rating only winch-assisted harvesters and forwarders (Holzfeind et al.,
benefits and efficiency gains if best practices were to be rigorously 2020) might be considered. However, for countries with a completely
applied. An efficiency gain can result from two factors: (1) a shift in the different silviculture philosophy, a newly defined set of estimated best
harvesting operations (with different harvesting method or extraction suitable harvesting methods would be necessary. Further, to determine
distance) or (2) a shift in the hauling route (with different hauling dis­ trafficability we relied partially on survey data, which might not be
tance or weight limit). However, the hauling route was not recorded in available in most countries. In these cases soil trafficability maps, such
the NFI survey, and the comparison was therefore only possible based on as those proposed by Schönauer et al. (2019) and Salmivaara et al.
the first factor. (2020) could be used as an alternative. Further, the expert-defined de­
The results show that, through the consistent use of estimated best cision trees must be carefully checked and adapted to specific
suitable harvesting methods, (1) harvesting costs can be reduced and (2) conditions.
a larger proportion of the forest can be managed to cover costs or ‘Harvesting and hauling costs’: The productivity models used in
generate profits. This applies to the country as a whole, but especially to this study require specific input in terms of salaries and machine costs,
regions with forests in steep terrain. The results are reasonable and can which must be adapted to country-specific values. However, the pro­
be explained by the following factors. First, we observed a shift towards ductivity models represent the European timber harvesting technique; if
more mechanized methods, especially in trafficable areas; the FM_FW applied outside Europe other productivity models would have to be
method, which is based on the use of harvesters and forwarders, would utilized. Analogous to the adjustment of the wood harvesting costs,
become the most frequently used method in trafficable areas if estimated country-specific cost values must also be used as a basis for the hauling
best suitable methods were widely applied. A shift towards higher costs.
mechanization was observed in steep terrain in our estimated best
suitable harvesting method simulation. The percentage of plots in which 4.4. Uncertainty assessment
processor-tower-yarders are used increased from 15 to 21 %, making it
the most commonly used method in steep terrain. Second, the percent­ The entire modeling of the best practices and the associated cost
age of plots harvested with cable-based methods would increase if estimation contain several sources of uncertainty and error, respec­
estimated best suitable harvesting methods were applied. On the one tively. These are errors in the data recordings, vague threshold values in
hand, the area with helicopter use would decrease, and on the other the decision trees, uncertainties in the productivity models and sampling
hand, former ground-based areas would be assigned to cable-based errors.
methods. In many steep areas (with a dense forest road network), The sampling errors can be accurately specified, according to Lanz
winches on skidders are traditionally used because the enterprises often et al. (2019), and are between 1 % and 2 %, as shown in the results
want to use their own personnel and machines, but the tower-yarder section. For the NFI inventories, a quality target was formulated for the
would be the more suitable method. DBH specifying that 98 % of the measured values have less than 1 cm
Our findings are consistent with the findings of several studies. deviation, which is met (Traub et al., 2016). The species richness
Enache et al. (2016) aimed to assess the current practices in logging identification (which includes all species) does not meet the re­
operations and to identify the efficiency gaps in timber production, and quirements of the NFI (Traub and Wüest, 2020), but we think that a
they found that inappropriate technology was often used, in particular in broadleaved/conifer differentiation is less complex and can be per­
steep terrain. Kühmaier and Stampfer (2010) also concluded that a shift formed in adequate quality.
towards more highly mechanized technology leads to a higher contri­ Uncertainties in the productivity models are larger and depend on
bution margin (revenue minus harvesting costs, without road costs or the model component used. A compilation of the underlying databases
overhead), lowers the damage rate in the remaining stand, and lowers can be found in Holm et al. (2020). All models are based on a large
the laborer injury rate (accidents). Meyer et al. (2001) also saw an amount of data, for example, the motor manual felling and processing
improvement in the contribution margin by applying the resulting model is based on 8 million m3 of timber. The harvester, skidder and
estimated best suitable harvesting methods. forwarder models rely on between 0.1 and 0.5 Mio. m3 of harvested
timber, whereas the models for the cable-based systems are based on
4.3. Wider application amounts between 1800 and 66,000 m3 timber. The adjusted R-square
serves as a measure of the proportion of variance explained by the
We see three main areas of application for the model. First, it could model, which is 0.62 (harvester), 0.4 (processor-tower-yarder), 0.37
be used to estimate the potential of wood or biomass, as well as their (forwarder), 0.67 (motor manual felling and processing model) or 0.72
supply costs. Especially in the context of the bioeconomy, this issue is (skidder). Values for the long-distance yarder and the tower-yarder are
very relevant, for example, when it comes to establishing new sawmills, not reported, but similar productivity models such as those used by
biorefineries or combined heat power plants (Erni et al., 2020). Second, Schweier et al. (2020) and Spinelli et al. (2020) result in adjusted
the methodology could be used to better target subsidies for forest R-squared values between 0.6 and 0.75.
management. For example, in the Alps, protective forest management is The uncertainty in the decision trees and the associated threshold
not cost-covering and needs to be supported with public subsidies. Until values must be considered separately for each decision tree. In reality,
now, flat-rate contributions have been granted, regardless of the diffi­ the distinction between the different ground-based methods (Fig. 5) is
culty of harvesting. With the application of this method, the subsidies not as sharp as shown in our decision tree, but rather involves smooth
could be adjusted to the difficulty of the timber harvest and thus be transitions, in particular the threshold related to the share of broad­
allocated more efficiently and in a more focused manner. Third, this leaves. However, we think that although another mix of estimated best
method could also be a valuable tool for forest managers, as it could be suitable methods is outputted when different threshold values are used,
used as a tool to select the optimal harvesting method for logging op­ this has only a marginal influence on the final costs. The borders be­
erations or machine evaluation. tween the two methods are blurred, and the two methods probably
The method presented here has been tailored to conditions in perform similarly under those conditions. The same principles are valid
Switzerland. The method could be applied in other regions with modi­ for the decision tree for steep terrain (Fig. 6). In this decision tree, we
fications. Doing so would require an adaptation of the following think that the threshold values are more robust, because, for example,
subsystems: the cable road lengths or the DBH limits are based on the feasibility

13
L.G. Bont et al. Journal of Environmental Management 302 (2022) 114099

limits of the respective systems. Similarly, as seen in Fig. 5, the ‘share of However, effects (efficiency gains) of investments in the forest road
broadleaves’ is quite an indistinct boundary. infrastructure were not considered in this study. The approach presented
The combination of the productivity models with the expert-defined here can be a good basis for studies quantifying such effects. In a further
decision trees leads to individual productivity models that are restricted step, scenarios for the upgrading of forest road infrastructures would
in their application, namely in the value range on which the empirical have to be defined. These scenarios could be included in future analyses
data basis is recorded. If the delimitation of the estimated best suitable to determine to what extent it would be possible to further reduce log­
methods would only be carried out based on the productivity models, ging and transport costs.
the productivity models would also have to be applied over a value The method presented here could be further developed, such as
range for which an insufficient empirical data basis exists, and thus the refining the underlying expert rules or using remote sensing data to
outputs would be subject to a higher degree of uncertainty. In our obtain wall-to-wall forest properties (such as volume estimates; Bont
opinion, the combination of productivity models and expert-based de­ et al., 2020) or tree type classifications (Waser et al., 2017)) around the
cision trees leads to more robust results than if they were based on plots, e.g. to better predict properties over the entire cable road length.
productivity models alone. Our method could also be used outside the NFI framework. In this case,
where comparability with previous methods is not necessary, some as­
5. Conclusions sumptions need to be reconsidered: it should not be assumed that all
plots will be harvested or that the entire volume is always removed from
We draw the following major conclusions from our study. First, the a plot. For example, a minimum stock level could be defined, below
model we developed represents effective method to allocate estimated which the sample is not considered for harvesting.
best suitable harvesting methods to plots. Second, applying estimated Uncertainty should also be better determined so that a quantitative
best suitable harvesting methods would lead to a larger percentage of confidence interval can be given. It is also conceivable that at the
forest area that can be harvested below a certain cost threshold, in indistinct borders between different best suitable methods, not only one
particular in the Swiss regions Pre-Alps, Alps and Southern Alps, which method is indicated as output, but several potential methods of equal or
include steep terrain and where harvesting is mainly based on cable- and similar value are displayed.
air-based methods. Assuming an average timber price of 75 CHF m− 3, an Our approach was developed for Switzerland in particular, but the
additional 12 percentage points, or 64 % instead of 52 % of the forest methodological framework could be used in other regions. In Europe,
area, could be harvested economically over the whole country. More where similar silvicultural philosophies exist and management princi­
specifically, 16 percent points more could be harvested in the Pre-Alps ples are mainly based on single-tree or selective cutting, minor modifi­
(65 % instead of 49 %) and 22 in the Alps (44 % instead of 22 %). cations would be necessary to apply the approach. Applying the
This is caused by a shift towards more mechanized harvesting methods. resulting estimated best suitable harvesting methods and therefore
Within the scope of this study, the uncertainty of the results could not be strengthening the domestic and sustainable forest industry also have
precisely quantified. However, the qualitative analysis of the individual global relevance. Forests provide not only numerous traditional prod­
uncertainty factors shows that robust and plausible results can be ex­ ucts, such as wood for material and energetic purposes, but also the raw
pected by combining expert-based decision trees with productivity material for innovative bio-based products (such as foams and resins),
models. contributing to a circular bio-economy. Demand for these products has
To our knowledge, the approach presented here is the first to allocate increased continuously for years, and greater quantities of the limited
best suitable harvesting methods to plots, while concurrently consid­ resources will likely be in demand in the future. Forests also fulfill
ering hauling route limitations, extraction route properties and stand increasingly important biodiversity and forest ecosystem functions,
characteristics, all based on productivity models and supported with while at the same time being affected by climate change and a growing
expert-defined decision trees. human population. Forest management has to adapt to these changes in
Our findings have several implications for the public sector and for order to provide much-needed products and services in a sustainable and
administrators and practitioners. They provide indications that more competitive way, e.g. by changing the intensities and frequencies of
wood could potentially be harvested and that more forest area could be thinning and final felling operations. Applying estimated best suitable
accessed at a lower cost than with current practices. By increasing the harvesting methods helps to achieve three positive outcomes in partic­
use of highly mechanized methods, timber harvesting could be made ular: (1) cost-effective biomass production, freeing resources for sus­
more cost-efficient. However, the switch towards more mechanized tainable forest management with improved processes (e.g. without
harvesting methods is not trivial. Highly mechanized methods, such as damage to soils and remaining stands) and efficient machines; (2)
FM_FW using a harvester and forwarder, are associated with large in­ lowering of potential negative impacts, e.g. through lower fuel con­
vestments. For these investments to be profitable, the machine must be sumption due to more efficient processes (Labelle and Lemmer, 2019),
operated at high capacity. This is made more difficult by the fact that or by avoiding soil compaction (Bustos and Egan, 2011); and (3)
such specialized machines have a relatively narrow focus of application reduction of deforestation-related emissions driven by international
and are unsuitable for other purposes. For example, it is not advisable to timber trade. Therefore, identifying and applying estimated best suit­
use them during rainy periods. On the other hand, MM_SK, using able harvesting methods is not only economically reasonable, but also
chainsaws and skidders, is one of the most versatile production systems benefits the environment and society.
in use today. It is easy to operate (labor qualification), requires only
small investments, can sustain periods of low activity (the workforce or Credit author statement
the skidder can be used for other purposes) and can be operated under a
large range of site conditions. The estimated best suitable harvesting Leo Gallus Bont: Conceptualization, Methodology, Software, Vali­
system could even prove detrimental to contractors or enterprises if dation, Supervision, Visualization, Writing – original draft, Writing –
conditions are not exactly as “estimated” by the models. Future research review & editing, Resources, Formal analysis, Investigation. Marielle
should therefore aim to establish an optimal portfolio of machines for Fraefel: Writing – original draft. Fritz Frutig: Conceptualization, Meth­
entrepreneurs, which allows the full range of operations to be carried odology, Writing – review & editing, Writing – original draft. Stefan
out as efficiently as possible (with estimated best suitable harvesting Holm: Software, Writing – review & editing, Writing – original draft.
methods), but also allows flexibility if conditions are not as predicted Christian Ginzler: Resources. Christoph Fischer: Conceptualization,
and also considers the financial risks of contractors. Software, Validation, Supervision, Visualization, Writing – original
Harvesting a larger percentage of forest area below a certain cost draft, Resources, Formal analysis, Investigation.
threshold can be achieved by changing the applied technology.

14
L.G. Bont et al. Journal of Environmental Management 302 (2022) 114099

Funding interests or personal relationships that could have appeared to influence

the work reported in this paper.
This research was conducted within the scientific project “Timber
harvesting costs and optimized forest accessibility” of the Swiss National Acknowledgements
Forest Inventory. The project was supported by the Swiss Federal
Institute of Forest, Snow and Landscape Research (WSL) and the Swiss We thank Melissa Dawes for English editing assistance and Janine
Federal Office for the Environment (FOEN). Schweier for helpful input on the manuscript. We also thank eight
anonymous reviewers for their constructive and helpful suggestions.
Declaration of competing interest

The authors declare that they have no known competing financial

Appendix
Table Appendix 01
Logging productivity, based on HeProMo (Holm et al., 2020) for long-distance yarders (LY) with different yarding distances and associated
processing costs with an excavator (excavator costs: 160 CHF h− 1, including operator). Processing costs [CHF m− 3] = 160 [CHF h− 1] * Logging
productivity− 1 [m3 h− 1].

Mean yarding distance Mean log volume Logging Productivity (LY) Processing costs (excavator)
3 3 − 1
[m] [m ] [m h ] [CHF m− 3]

500 0.4 6.51 24.6

1 8.80 18.2
1000 0.4 4.72 33.9
1 6.19 25.8
1500 0.4 3.72 43.0
1 4.79 33.4

Fig. Appendix 01. Cost difference between the PM_SK and the MM_FW harvesting method for all 208 NFI plots where PM_SK was reported in the NFI survey (Keller,
2013). For both harvesting systems, the cost calculation is based on HeProMo (Holm et al., 2020). PM_SK: Motor-manual felling, skidding with skidder (full tree),
processing at forest road, MM_FW: Motor-manual felling and processing, forwarding (assortments).

15
L.G. Bont et al. Journal of Environmental Management 302 (2022) 114099

Fig. Appendix 02. Boxplot of additional costs of motor-manual processing for all 285 NFI plots where PM_LY was reported in the NFI survey (Keller, 2013). The cost
calculation is based on HeProMo (Holm et al., 2020).PM_LY: Motor-manual felling, yarding with long-distance yarder, processing (full tree).

Table Appendix 02
Assumptions and values for operation costs (1 EUR = 1.09 CHF, 1 USD = 0.90 CHF, Date: June 16, 2021).

Model name Daily work Daily travel time and paid Cost Cost machine Cost Cost Cost Cost additional crane
time breaks labor operator machine helicopter chainsaw truck
1 1 1 1 1 1
min min CHF h− CHF h− CHF h− CHF min− CHF h− CHF h−

Motor-manual 540 60 65 14
felling
Skidder 540 60 70 130
Forwarder 540 60 70 120
Harvester 540 60 80 240
Tower-yarder 540 60 55 120 80
Long-distance 540 60 55 80 80
yarder
Helicopter 540 60 55 60 14 80

References Bont, L., Heinimann, H.R., 2012. Optimum geometric layout of a single cable road. Eur.
J. For. Res. 131, 1439–1448. https://doi.org/10.1007/s10342-012-0612-y.
Bont, L., Heinimann, H.R., Church, R.L., 2014. Optimizing cable harvesting layout when
Abegg, M., 2014. Viertes Schweizerisches Landesforstinventar - Ergebnistabellen und
using variable-length cable roads in central Europe. Can. J. Res. 44, 949–960.
Karten im Internet zum LFI 2009-2013 (LFI4b). http://www.lfi.ch/resultate/.
Bont, L.G., Church, R.L., 2018. Location set-covering inspired models for designing
(Accessed 2 March 2020). Accessed.
harvesting and cable road layouts. Eur. J. For. Res. 137, 771–792. https://doi.org/
Aggeler, R., 2002. Seilkrantechnik, vol. 3 (überarbeitete und erweiterte Ed).
10.1007/s10342-018-1139-7.
Ahlvin, R.B., Haley, P.W., 1992. Nato Reference Mobility Model: Edition II. US Army
Bont, L.G., Fraefel, M., Fischer, C., 2018. A spatially explicit method to assess the
Engineer Waterways Experiment Station, Vicksburg, MS, USA. NRMM User’s Guide.
economic suitability of a forest road network for timber harvest in steep terrain.
Axelsson, S.-Ă., 1998. The mechanization of logging operations in Sweden and its effect
Forests 9, 169. https://doi.org/10.3390/f9040169.
on occupational safety and health. J. Eng. 9, 25–31. https://doi.org/10.1080/
Bont, L.G., Hill, A., Waser, L.T., et al., 2020. Airborne-laser-scanning-derived auxiliary
08435243.1998.10702715.
information discriminating between broadleaf and conifer trees improves the
BAFU, 2017. Jahrbuch Wald und Holz 2017. Bundesamt für Umwelt, Bern.
accuracy of models for predicting timber volume in mixed and heterogeneously
BfS, 2000. Digitale Bodeneignungskarte der Schweiz. Bundesamt für Statistik,
structured forests. For. Ecol. Manag. 459, 117856 https://doi.org/10.1016/j.
Neuenburg.
foreco.2019.117856.
Blattert, C., Lemm, R., Thees, O., et al., 2018. Segregated versus integrated biodiversity
Bont, L.G., Maurer, S., Breschan, J.R., 2019. Automated cable road layout and harvesting
conservation: value-based ecosystem service assessment under varying forest
planning for multiple objectives in steep terrain. Forests 10, 687. https://doi.org/
management strategies in a Swiss case study. Ecol. Indicat. 95, 751–764. https://doi.
10.3390/f10080687.
org/10.1016/j.ecolind.2018.08.016.

16
L.G. Bont et al. Journal of Environmental Management 302 (2022) 114099

Brändli, U.-B., Hägeli, M., 2019. Swiss NFI at a glance. In: Fischer, C., Traub, B. (Eds.), Kizha, A., Han, H., Montgomery, T., Hohl, A., 2015. Biomass power plant feedstock
Swiss National Forest Inventory – Methods and Models of the Fourth Assessment. procurement: modeling transportation cost zones and the potential for competition.
Springer International Publishing, Cham, pp. 3–35. Calif. Agric. 69, 184–190.
Bürgi, P., Thomas, M., Pauli, B., Auer, N., 2018. Forstwirtschaftliches Testbetriebsnetz Kizha, A.R., Han, H.-S., Anderson, N., et al., 2020. Comparing hot and cold loading in an
der Schweiz: Ergebnisse der Jahre 2014–2016. Bundesamt für Statistik (BFS), integrated biomass recovery operation. Forests 11, 385. https://doi.org/10.3390/
Neuchâtel. f11040385.
Bustos, O., Egan, A., 2011. A comparison of soil compaction associated with four ground- Köhl, M., Linser, S., 2020. SoEF_2020.pdf. In: State of Europe’s Forests 2020. Ministerial
based harvesting systems. NJAF (North. J. Appl. For.) 28, 194–198. https://doi.org/ Conference on the Protection of Forests in Europe - FOREST EUROPE, 394p.
10.1093/njaf/28.4.194. Bratislava.
Chung, P.W., Sessions, J., Heinimann, H.R., 2004. An application of a heuristic network Korhonen, J., 2016. On the High Road to Future Forest Sector Competitiveness, PhD
algorithm to cable logging layout design. Int. J. For. Eng. 15, 11–24. https://doi.org/ Thesis. University of Helsinki.
10.1080/14942119.2004.10702485. Krč, J., Vranešič, U., Košir, B., 2015. Comparison OF mechanized and motor-manual
Chung, W., Sessions, J., 2002. Optimization of Cable Logging Layout Using a Heuristic cutting operation IN mixed stands OF southern Slovenia. Šumar List 139, 351–359.
Algorithm for Network Programming. PhD Thesis, Oregon State University Corvallis, Kühmaier, M., Stampfer, K., 2010. Development of a multi-attribute spatial decision
OR. support system in selecting timber harvesting systems. Croat J For Eng J Theory Appl
Düggelin, C., Abegg, M., Bischof, S., et al., 2020. Schweizerisches Landesforstinventar. For Eng 31, 75–88.
Anleitung für die Feldaufnahmen der fünften Erhebung 2018–2026. Labelle, E.R., Breinig, L., Sycheva, E., 2018. Exploring the use of harvesters in large-
Dupire, S., Bourrier, F., Monnet, J.M., Berger, F., 2015. Sylvaccess: a model for automatic diameter hardwood-dominated stands. Forests 9, 424. https://doi.org/10.3390/
mapping of the forest accessibility. Rev. Fr. 70, 111–126. https://doi.org/10.4267/ f9070424.
2042/57902. Labelle, E.R., Lemmer, K.J., 2019. Selected environmental impacts of forest harvesting
Dykstra, D.P., Riggs, J.L., 1977. An application of facilities location theory to the design operations with varying degree of mechanization. Croat J For Eng J Theory Appl For
of forest harvesting areas. IIE Trans. 9, 270–277. Eng 40, 239–257.
Eichrodt, A.W., 2003. Development of a spatial trafficability evaluation system. vdf Lanz, A., Abegg, M., Braendli, U.-B., et al., 2016. Switzerland. In: Vidal, C., Alberdi, I.A.,
Hochschulverlag AG. Hernández Mateo, L., Redmond, J.J. (Eds.), National Forest Inventories. Springer
Enache, A., Kühmaier, M., Visser, R., Stampfer, K., 2016. Forestry operations in the International Publishing, Cham, pp. 783–805.
European mountains: a study of current practices and efficiency gaps. Scand. J. For. Lanz, A., Fischer, C., Abegg, M., 2019. Sampling design and estimation procedures. In:
Res. 31, 412–427. https://doi.org/10.1080/02827581.2015.1130849. Fischer, C., Traub, B. (Eds.), Swiss National Forest Inventory – Methods and Models
Erni, M., Burg, V., Bont, L., et al., 2020. Current (2020) and long-term (2035 and 2050) of the Fourth Assessment. Springer International Publishing, Cham, pp. 39–92.
sustainable potentials of wood fuel in Switzerland. Sustainability 12, 9749. https:// Laschi, A., Marchi, E., Foderi, C., Neri, F., 2016. Identifying causes, dynamics and
doi.org/10.3390/su12229749. consequences of work accidents in forest operations in an alpine context. Saf. Sci. 89,
Fischer, C., Fraefel, M., 2019. Interview survey with the local forest service. In: 28–35. https://doi.org/10.1016/j.ssci.2016.05.017.
Fischer, C., Traub, B. (Eds.), Swiss National Forest Inventory – Methods and Models Leyder, C., Klippel, M., Bartlomé, O., et al., 2021. Investigations on the sustainable
of the Fourth Assessment. Springer International Publishing, Cham, pp. 187–195. resource use of Swiss timber. Sustainability 13, 1237. https://doi.org/10.3390/
Fischer, C., Stadelmann, G., 2019. Calculation of potential timber harvesting costs su13031237.
(HeProMo). In: Fischer, C., Traub, B. (Eds.), Swiss National Forest Inventory – Liepiņš, K., Lazdiņš, A., Liepiņš, J., Prindulis, U., 2015. Productivity and cost-
Methods and Models of the Fourth Assessment. Springer International Publishing, effectiveness of mechanized and motor-manual harvesting of grey Alder (Alnus
Cham, pp. 257–263. incana (L.) Moench): a case study in Latvia. Small-Scale For 14, 493–506. https://
FOEN, 2013. Forest Policy 2020. Visions, Objectives and Measures for the Sustainable doi.org/10.1007/s11842-015-9302-1.
Management of Forests in Switzerland. Federal Office for the Environment, Bern. Maurer, S., Heinimann, H.R., 2020. Framework for assessing the windthrow risk to
Fraefel, M., Bont, L.G., Fischer, C., 2021. Spatially explicit assessment of forest road Norway spruce forests in Switzerland. Eur. J. For. Res. https://doi.org/10.1007/
suitability for timber extraction and hauling in Switzerland. Eur. J. For. Res. https:// s10342-019-01251-w.
doi.org/10.1007/s10342-021-01393-w. Mederski, P.S., Bembenek, M., Karaszewski, Z., et al., 2018. Investigation of log length
Fraefel, M., Fischer, C., 2019. Compilation of a high-quality forest road data set. In: Accuracy and harvester efficiency in processing of Oak trees. Croat J For Eng J
Fischer, C., Traub, B. (Eds.), Swiss National Forest Inventory – Methods and Models Theory Appl For Eng 39, 173–181.
of the Fourth Assessment. Springer International Publishing, Cham, pp. 367–373. Mehr, J., Vadenbo, C., Steubing, B., Hellweg, S., 2018. Environmentally optimal wood
Frehner, M., Wasser, B., Schwitter, R., 2005. Nachhaltigkeit und Erfolgskontrolle im use in Switzerland—investigating the relevance of material cascades. Resour.
Schutzwald. Wegleitung Für Pflegemassnahmen Wäld Mit Schutzfunktion Conserv. Recycl. 131, 181–191. https://doi.org/10.1016/j.resconrec.2017.12.026.
Bundesamt Für Umw Wald Landsch Bern 564. Meyer, T., Stueckelberger, J., Hollenstein, K., et al., 2001. effor2: Holzerntekonzept für
Frutig, F., Holm, S., Pedolin, D., Thees, O., 2016. Holzernte Produktivitätsmodelle die Untersuchungsbetriebe im Kanton Schwyz. Professur für forstliches
HeProMo. Version 2.3. Eidg. Forschungsanstalt WSL, Birmensdorf, Switzerland. Ingenieurwesen. Eidgenössische Technische Hochschule, Zürich.
Frutig, F., Thees, O., 2011. Holzerntetechnik für das steile Gelände. Bündner Wald 7–15. Montgomery, T.D., Han, H.-S., Kizha, A.R., 2016. Modeling work plan logistics for
Fulvio, F.D., Abbas, D., Spinelli, R., et al., 2017. Benchmarking technical and cost factors centralized biomass recovery operations in mountainous terrain. Biomass Bioenergy
in forest felling and processing operations in different global regions during the 85, 262–270. https://doi.org/10.1016/j.biombioe.2015.11.023.
period 2013–2014. Int. J. For. Eng. 28, 94–105. https://doi.org/10.1080/ Müller, K., Fraefel, M., Cioldi, F., et al., 2016. Der Datensatz «Walderschliessungsstrassen
14942119.2017.1311559. 2013» des Schweizerischen Landesforstinventars. Schweiz. Z. Forstwes. 167,
Heinimann, H.R., 2003. Holzerntetechnik zur Sicherstellung einer minimalen 136–142. https://doi.org/10.3188/szf.2016.0136.
Schutzwaldpflege: Bericht im Auftrag des Bundesamtes für Umwelt, Wald und Niemi, M.T., Vastaranta, M., Vauhkonen, J., et al., 2017. Airborne LiDAR-derived
Landschaft (BUWAL), vol. 12. Interner BerichtETH For Eng. elevation data in terrain trafficability mapping. Scand. J. For. Res. 32, 762–773.
Heinimann, H.R., 1998. A computer model to differentiate skidder and cable-yarder https://doi.org/10.1080/02827581.2017.1296181.
based road network concepts on steep slopes. J For Res Jpn 3, 1–9. Pendrill, F., Persson, U.M., Godar, J., et al., 2019. Agricultural and forestry trade drives
Heinimann, H.R., Stampfer, K., 2003. In: Limbeck-Lilienau, B., Steinmüller, T., large share of tropical deforestation emissions. Global Environ. Change 56, 1–10.
Stampfer, K. (Eds.), Harvest Layout Planning for High-Altitude Protection Forests. https://doi.org/10.1016/j.gloenvcha.2019.03.002.
Schlaegl Austria, p. 13. Pestal, E., 1961. Seilbahnen und Seilkräne für Holz- und Materialtransporte. Georg
Henders, S., Persson, U.M., Kastner, T., 2015. Trading forests: land-use change and Fromme & Co, Wien und München.
carbon emissions embodied in production and exports of forest-risk commodities. Salmivaara, A., Launiainen, S., Perttunen, J., et al., 2020. Towards dynamic forest
Environ. Res. Lett. 10, 125012 https://doi.org/10.1088/1748-9326/10/12/125012. trafficability prediction using open spatial data, hydrological modelling and sensor
Hirt, R., 1997. Wer hat Angst vor 40 Toennern. Schweiz Ing Archit SIA 49, 4. technology. For Int J For Res 93, 662–674. https://doi.org/10.1093/forestry/
Holm, S., Frutig, F., Lemm, R., et al., 2020. HeProMo: a decision support tool to estimate cpaa010.
wood harvesting productivities. PLoS One. Schönauer, M., Talbot, B., Jaeger, D., 2019. TECH4EFFECT. Terrain Accessibility Maps
Holzfeind, T., Visser, R., Chung, W., et al., 2020. Development and benefits of winch- for 4 Case Study Areas.
assist harvesting. Curr For Rep 6, 201–209. https://doi.org/10.1007/s40725-020- Schweier, J., Klein, M.-L., Kirsten, H., et al., 2020. Productivity and cost analysis of tower
00121-8. yarder systems using the Koller 507 and the Valentini 400 in southwest Germany.
Holzleitner, F., Kanzian, C., Stampfer, K., 2011. Analyzing time and fuel consumption in Int. J. For. Eng. 31, 172–183. https://doi.org/10.1080/14942119.2020.1761746.
road transport of round wood with an onboard fleet manager. Eur. J. For. Res. 130, Seidl, R., Rammer, W., Jäger, D., et al., 2007. Assessing trade-offs between carbon
293–301. https://doi.org/10.1007/s10342-010-0431-y. sequestration and timber production within a framework of multi-purpose forestry in
Jaafari, A., Najafi, A., Melón, M.G., 2015. Decision-making for the selection of a best Austria. For. Ecol. Manag. 248, 64–79. https://doi.org/10.1016/j.
wood extraction method: an analytic network process approach. For Policy Econ 50, foreco.2007.02.035.
200–209. https://doi.org/10.1016/j.forpol.2014.09.010. Seidl, R., Thom, D., Kautz, M., et al., 2017. Forest disturbances under climate change.
Kaakkurivaara, N., Stampfer, K., 2018. Assessment for improvement: harvesting Nat. Clim. Change 7, 395–402. https://doi.org/10.1038/nclimate3303.
operations in small-scale forest on Thai steep terrain. Small-Scale For 17, 259–276. Soman, H., Kizha, A.R., Roth, B.E., 2019. Impacts of silvicultural prescriptions and
https://doi.org/10.1007/s11842-017-9386-x. implementation of best management practices on timber harvesting costs. Int. J. For.
Keller, M., 2013. Anleitung für die Umfrage- und Erschliessungserhebung 2013, Eng. 30, 14–25. https://doi.org/10.1080/14942119.2019.1562691.
Schweizerisches Landesforstinventar. WSL (Birmensdorf). Spinelli, R., Visser, R., Magagnotti, N., et al., 2020. The effect of partial automation on
Kissling, H., 2020. Fahrzeug-Selbstkostenrechnung. the productivity and cost of a mobile tower yarder. Ann. For. Res. 63, 3–14. https://
doi.org/10.15287/afr.2020.1883.

17
L.G. Bont et al. Journal of Environmental Management 302 (2022) 114099

Suvinen, A., 2006. A GIS-based simulation model for terrain tractability. Waser, L., Fischer, C., Wang, Z., Ginzler, C., 2015. Wall-to-Wall forest mapping based on
J. Terramechanics 43, 427–449. https://doi.org/10.1016/j.jterra.2005.05.002. digital surface models from image-based point clouds and a NFI forest definition.
Svenson, G., Flisberg, P., Rönnqvist, M., 2016. Using analytics in the implementation of Forests 6, 4510–4528. https://doi.org/10.3390/f6124386.
vertical and horizontal curvature in route calculation. IEEE Trans. Intell. Transport. Waser, L.T., Ginzler, C., Rehush, N., 2017. Wall-to-Wall tree type mapping from
Syst. 17, 1772–1785. https://doi.org/10.1109/TITS.2015.2503424. countrywide airborne remote sensing surveys. Rem. Sens. 9, 766 https://doi.org/
Temperli, C., Blattert, C., Stadelmann, G., et al., 2020. Trade-offs between ecosystem 10.3390/rs9080766.
service provision and the predisposition to disturbances: a NFI-based scenario Werner, F., Taverna, R., Hofer, P., et al., 2010. National and global greenhouse gas
analysis. For Ecosyst 7, 1–17. dynamics of different forest management and wood use scenarios: a model-based
Traub, B., Cioldi, F., Düggelin, C., 2016. Wiederholungsaufnahmen als Instrument zur assessment. Environ. Sci. Pol. 13, 72–85. https://doi.org/10.1016/j.
Qualitätssicherung im Schweizerischen Landesforstinventar. Schweiz. Z. Forstwes. envsci.2009.10.004.
118–127. https://doi.org/10.3188/szf.2016.0118. Zweifel, O., 1960. Seilbahnberechnung bei beidseitig verankerten Tragseilen. Schweiz
Traub, B., Wüest, R.O., 2020. Analysing the quality of Swiss National Forest Inventory Bauztg 78, 11.
measurements of woody species richness. For Ecosyst 37, 11. https://doi.org/
10.1186/s40663-020-00252-1.

18