You are on page 1of 11

Forest Policy and Economics 21 (2012) 81–91

Contents lists available at SciVerse ScienceDirect

Forest Policy and Economics


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

Evaluation of harvesting methods for Sustainable Forest Management (SFM) using


the Analytical Network Process (ANP)
Ismael Ghajar a, Akbar Najafi b,⁎
a
Dep. of Forestry, Faculty of Natural Resources, Tarbiat Modares University, Noor, PO Box: 64414-356, Iran
b
Dep. of Forestry, Faculty of Natural Resources, Tarbiat Modares University, Noor, PO Box : 64414-356, Iran

a r t i c l e i n f o a b s t r a c t

Article history: Forest harvesting is a basic component of Sustainable Forest Management (SFM). Consideration of all aspects
Received 21 June 2011 of harvesting operations in an effort to improve forest practices requires an approach that first addresses
Received in revised form 17 December 2011 multiple criteria and incorporates a wide range of data. Second, the approach should provide a framework
Accepted 10 January 2012
to evaluate both quantitative and qualitative criteria. In this study, the Analytical Network Process (ANP) is
Available online 21 February 2012
utilized to evaluate three existing harvesting methods by integrating the main components of the FAO
Keywords:
Model Code of Harvesting Practice as an international mechanism for local management. The ANP framework
Analytical Network Process helps forest managers to prioritize all alternatives and criteria with respect to each other and to develop their
Decision making corresponding preferences. This study is an analysis of the environmental, economic, and social context of
Forest harvesting harvesting operations with the goal of making forest practices more sustainable and leading forest utilization
Sustainable Forest Management to SFM. Although the criteria are calibrated for Caspian forests (northern Iran), the key principles applied
here can be used or developed in other countries or regions.
© 2012 Elsevier B.V. All rights reserved.

1. Introduction As a result of the critical focus provided by the Earth Summit, a va-
riety of forest certification programs were developed around the
The choice of timber harvesting method is crucial for the world including: the Canadian Standard Association (CSA), Forest
profitability of forest operations, but it is also one of the components Stewardship Council (FSC), Lambaga Ekolabel Indonesia (LEI),
of forestry practices that must be chosen carefully in order to realize Malaysian Certification Council, National Timber Certification Council
Sustainable Forest Management (SFM). The economic, ecological, (NTCC) and Pan European Forest Certification Council (PEFC).
and social differences between harvesting methods require forest Certification programs have been designed to develop interim har-
managers to evaluate each method in all existing conditions. Over vesting standards for countries with poor track records regarding the
the past two decades, a number of instruments have been developed quality of harvesting practices, and with recognition that in many
to improve forest management, increase wood recovery from areas contractors/concessionaires have limited financial resources
harvesting operations, and minimize costs and forest damage as a (Pulkki et al., 2002). A number of developments spurred countries to
result of logging operations (Durst et al., 2003). The United Nations provide forest certification programs in an effort to promote SFM,
Earth Summit, held in Rio de Janeiro in 1992, spawned the develop- including: 1) economic focus at the highest levels of forest manage-
ment of variety of new tools and mechanisms to address these ment; 2) a general consensus about the necessity to sustainable forest
problems, including: management; 3) recognition that wood markets are demanding
responsible logging practices; and 4) governmental imposition of
• criteria and indicators for SFM; stricter regulations on harvesting.
• certification standards for forest management; In relation to these programs, a number of approaches, collectively
• principles and requirements for compulsory forest management termed reduced-impact logging (RIL), were developed for decreasing
plans; the effect of harvesting on tropical forests (Spong, 2007). Although
• reduced impact logging techniques; and forest harvesting generates more revenue than all other forest uses,
• codes of practice for forest management and forest harvesting it also has the most significant short- and long-term impacts on the
(Durst et al., 2003). site (Applegate et al., 2004). Likewise, soil disturbances, erosion, com-
paction, and the amount of land surface needed for skid trails, roads,
and landings (all concerns for SFM) often receive little planning to
⁎ Corresponding author. Tel.: + 98 122 625 3101; fax: + 98 122 625 3499.
minimize their impacts (Mengistu, 2002). In relation to this problem,
E-mail addresses: ismael.ghajar@modares.ac.ir (I. Ghajar), a.najafi@modares.ac.ir the FAO Model Code of Forest Harvesting Practices (Dykstra and
(A. Najafi). Heinrich, 1996) was established as an ‘instrument designed to

1389-9341/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.forpol.2012.01.003
82 I. Ghajar, A. Najafi / Forest Policy and Economics 21 (2012) 81–91

improve forest management, to ensure better wood recovery from (6) ANP utilizes the idea of a control hierarchy or a control net-
harvesting operations, and to minimize forest damage arising from work to deal with different criteria, eventually leading to the
logging operations’ (Durst et al., 2003). The goal of this model code analysis of benefits, opportunities, costs, and risks (Chang et
was to decrease and manage any negative social and environmental al., 2009).
impacts of harvesting operations while also improving efficiency
and increasing economic benefits to a wide range of constituents Setting priorities and trade-offs among goals and criteria, measur-
(Spong, 2007). ing all tangible and intangible criteria (i.e. qualitative variables such
Poor and ambiguous economic evaluation of forest operations as slash distribution) in the model (Momoh and Zhu, 1998) and
could lead to misleading assessments of links and subsequently using the ratio scale of human judgment instead of arbitrary scales
poor policy; harvesting methods play a primary role in those assess- (Kim et al., 1997; Saaty, 1999), makes ANP easy to use by managers
ments. Wheeled skidders and ground skidding appeared in the early and other decision makers.
1970s and are now the most commonly used harvesting methods
worldwide (Naghdi et al., 2005). Ground skidding systems are widely
1.2. Application
used in the forests of developing countries because of high flexibility,
low investment, and low skidding costs, but this system can be highly
A great deal of research has been carried out on effects of harvest
destructive (Pinard et al., 2000), and severe impacts can destroy
machinery operations in forest stands of mountainous regions (Han
current and future commercial production (Spong, 2007). However,
et al., 2004; Adebayo Adebola et al., 2007; Jamshidi et al., 2008).
economic returns to the landowners are important considerations
Most of these studies concentrated solely on one aspect of harvesting
and costs, productivity (volume of logs produced per hour during a
operation effects, calculating the productivity, costs or environmental
harvesting operation), and overall profitability can be greatly affected
effects of harvesting methods, despite the fact that forest managers
by different harvesting systems and the different methods used for
need approaches to solve multi-criteria decision problems involving
felling, processing, loading, and transporting logs to the landing area
multiple actors. In fact, for decisions regarding forest harvesting
and eventually to the mill (Han et al., 2004).
methods, the socio-economic and ecological factors should be
Accordingly, we need a flexible and comprehensive framework to
simultaneously taken into consideration. There are a lot of tangible
simultaneously model and rank existing harvesting methods. To
and intangible (i.e. quantitative and qualitative, respectively) criteria
improve policy design and implementation, socio-economic and eco-
related to forest management that should be measured or estimated
logical evaluation of forest harvesting methods should facilitate the
and then integrated. In this study, all criteria are compared pairwise
comparison of the respective social, economic, and ecological bene-
based on the results of prior studies and expert knowledge. Many
fits, costs, opportunities and risks and allow assessment of tradeoffs
decision-support tools that have been developed over the years that
and synergies.
use only numerical variables as input data for evaluation, whereas
SFM decisions are based on a complex mix of multiple criteria,
different groups, economic data, and other inputs that combine both
1.1. Analytical Network Process
quantitative and qualitative data (Spong, 2007). The criteria that are
proposed by the FAO Model Code of Forest Harvesting Practices (i.e.
The concepts and methods of Multi-Criteria Decision Making
economic, environmental, and social factors) can be used as the
(MCDM) present a framework that incorporates multiple, conflicting
main criteria for selection of the alternatives in forest operations. Se-
criteria into planning (Komarov et al., 2002; Miettinen, 2006).
lection of the appropriate harvesting method using decision-making
Problems with feedback and intangibles in MCDM can be addressed
tools based on SFM policy should improve the sustainability of forest
by decision modeling methods such as the Analytic Network Process
operations. The main objective of the paper is to demonstrate that the
(ANP). In the mathematical approaches such as Linear Programming
contribution of the key principles of international criteria and the
(LP), mathematical programming, and multi-objective programming,
ANP structure (including factors identified in prior studies) can be
there are one or more objective functions including known quantita-
used to select the best harvesting method for more efficient forest
tive variables. The function in these systems should be optimized in
management in the given conditions.
the given data space by taking in account the system constraints.
The Analytic Network Process (ANP) (described by Saaty, 1999)
However, the qualitative (intangible) criteria could not be included
was applied to evaluate the overall preference of common harvesting
in these kinds of models. ANP has the ability to capture complex
methods in mountainous forests of Iran. By performing pairwise com-
systems and allows for complex interdependent relationships
parisons, we were able to prioritize the form in which wood is deliv-
among elements. This evaluation model can show the best alternative
ered to the logging access road, which is an especially important
by integrating interconnections between criteria, alternatives, and
factor for forest economics. Incorporating all effective factors in the
feedback into the decision system. ANP has a number of strengths
model is imperative in order to make a comprehensive decision and
as a decision modeling method (Saaty, 1999). They are listed as
to achieve the strategic goals of forest harvesting.
follows:

(1) ANP is built on the widely used AHP by allowing for 1.3. Forest operations in Iran
dependence.
(2) ANP deals with dependence within a set of elements (innerdepen- The Caspian forest is located in northern Iran, covering the north
dence), and among different sets of elements (outerdependence). facing slopes of the Alborz mountain ranges and is classified as
(3) Unlike a hierarchy, the looser network structure of ANP makes mountain forest. Approximately 60% of this forest is used for timber
it possible to represent any decision or problem without con- production. Limited forest area and rare plant species diversity in
cern for what comes first and what comes next. this area mean that it is necessary to harvest forests in a methodical
(4) ANP is a non-linear structure that deals with sources, cycles, and responsible way. Forest managers are trying to find the best
and sinks having a hierarchy of linear form with goals in the way to optimize productivity with respect to the potential and exist-
top level and the alternatives in the bottom level. ing conditions of the forest. Ground skidding systems, cut-to-length
(5) ANP portrays a real-world representation of the problem under and tree-length methods have been used on the Caspian forests
consideration by prioritizing not only just the elements but since machinery was first introduced to forest management in Iran.
also groups or clusters of elements, as is often necessary. In addition, animal skidding is commonly applied on steep slopes
I. Ghajar, A. Najafi / Forest Policy and Economics 21 (2012) 81–91 83

where skid road construction is not practical and in conditions where on findings of prior studies in mountainous forest. In previous studies,
machinery is not economically viable. statistical analysis and simulation models were used to separately
estimate the total time and cost of skidding, loading, and transporta-
2. Material and methods tion. One of the principles in this method was to divide the work into
smaller elements that allowed for more precise measurements and
2.1. Study area also separated productive work (productive time) from unproductive
work (unproductive time) (Bjorheden, 1991). To estimate production
The research was performed using the data of previous conducted and cost per unit volume of wood in each method and three
researches and theses in the 2nd district of Neka Zalemroud mentioned stages, the work–study technique was applied.
watershed in the Iranian Caspian forest where situated between In this study, the calculated productive time of each harvesting
longitudes 53°37′56″E–59°29′55″E and latitudes 36°23′21″N–36°26′ method estimated from prior studies was used as the foundation for
15″N. economic comparison of methods. The physiographic conditions
were included in these productivity models, so variables such as
2.2. Building the ANP harvesting method model steepness of slope were not included as a separate criterion in the
ANP model.
The entire ANP model consists of a two level decision-making Basic factors such as the initial investment cost of machinery and
network. The top-level structure has four Benefit–Opportunity– repair and maintenance costs have not been incorporated directly in
Cost–Risk (BOCR) merits (Fig. 1). The subnets under each of the the model but are collected as the cost of each cubic meter wood in
four BOCR merits are composed of their respective clusters and each stage (i.e. skidding, loading and transportation to the mill).
elements (Figs. 2 to 5). The goal of the model is selection of the best Other variables such as tree size, removal density, and extraction
harvesting method. The strategic criteria are created along with the distance were the same throughout the study. Variables remained
goal. There are economic, ecological, and social criteria. In addition, constant so that there was no effect on cost and productivity of differ-
seven strategic subcriteria, along with the strategic criteria, are devel- ent harvesting methods. This condition was needed to compare the
oped to evaluate the priorities of the BOCR merits. The “economic” harvesting methods properly and to obtain accurate final results. It
subcriteria are maximizing production, maximizing revenue and should be mentioned that it would be useful to include a conceptual
sustainable production. The “ecological” subcriteria are minimizing range of ‘all possible’ factors for harvesting operations in this model,
soil distribution, minimizing vegetation damage, and sustainable eco- even if some rarely occur and might not show up as statistically
system. The “social” subcriteria include improving employment significant in related models (Niemiraa and Saaty, 2004).
(number of jobs) and rural development. Any decision has several favorable and unfavorable factors to con-
In general, there are both quantitative and qualitative parameters sider. Some of these are definite; others are less certain but have a
in forest management. The first can be calculated using scientific certain likelihood of materializing. The favorable definite factors are
methods (tangible criteria), but making decisions about the second labeled Benefits and include items such as production per hour in
group (intangible criteria) is more variable in different conditions. skidding, slash distribution and so on, while the unfavorable ones
In this case, local experts can compare alternatives in the model are labeled Costs and include items such as loading costs per cubic
according to local realities and judge which alternative is best. meter of wood. The positive uncertain factors of a decision are the
In our model, the pairwise comparisons, with respect to quantita- Opportunities that the selection of each alternative can create; for
tive criteria, were performed by a group experienced experts based example, we have employment or the number of workers in this

Goal

The best harvesting method

Strategic criteria

Economic Ecological Social

Economic subcriteria Ecologicalsubcriteria Social subcriteria


Maximizing production Minimizing soil disturbance Improving employment
Maximizing revenue Minimizing vegetation damages Rural development
Sustainable production Sustainable ecosystem

Model

Benefits Costs Opportunities Risks


subnet subnet subnet subnet

Fig. 1. Top-level network with BOCR.


84 I. Ghajar, A. Najafi / Forest Policy and Economics 21 (2012) 81–91

Alternatives Advanta ges


Cut-to-length Skidding productivity
Tree-length Loading productivity
Animal skidding Productivity of transportation to the
mill

Others
Maximum skidding distance
Slash distribution
The value of yield timber

Fig. 2. Micro view of the Benefits network.

study. The fourth classification of general factors considered in all Step 2: The decision subnet for each BOCR is built. After linking the
decisions is the Risks entailed in the decision. Each of these four nodes and clusters appropriately, the pairwise comparisons with
concerns (BOCR) utilizes a separate structure for the decision, respect to the parent clusters and nodes inside each of the BOCR
beginning with a benefits control structure and the network of are performed separately. To obtain priorities of elements, we
interdependencies that belong to it, and ending with a risks control
first prioritize the clusters constituting each of the BOCR. For
structure. Also, each of these concerns contributes to the merit of a
example, under the Benefits subnetwork there are three clusters,
decision and must be evaluated (rated) individually on a set of (prior-
itized) criteria (Saaty et al., 2006). The model for finding the optimal ‘alternatives,’ ‘advantages,’ and ‘others’ (Fig. 2). The influence of
harvesting method was designed based on BOCR models. Super a cluster on the other clusters is represented in matrix form in
Decisions software v.1.6.0 was used for the analysis due to the differ- Table 1. Thus the ‘alternatives’ cluster is influenced by ‘advantages’
ent variables that needed to be considered. and ‘others’ clusters (column 1 in Table 1), ‘advantages’ is
The ANP model for evaluating the best harvesting method com- influenced by ‘alternatives’ and ‘others’ (column 2) and ‘others’
prises several steps that will be discussed below. is influenced by ‘alternatives’ (column 3). All of the comparisons
are based on a scale of relative importance with the option to
Step 1: Comparisons should be performed in the top-level model express preference between two elements on a ratio scale from
(with the BOCR). Here, we have a goal cluster with ‘the best equally important (i.e. equivalent to a numeric value of 1) to
harvesting method’ node, a strategic criteria cluster with the absolute preference (i.e. equivalent to a numeric value of 9) of
main criteria, and a cluster for each of the main criteria containing one element over another (Saaty, 1977).
their strategic subcriteria (Fig. 1). The main criteria are compared These influences are prioritized. Given a cluster (for example, ‘al-
with respect to the goal and the strategic subcriteria are pairwised ternatives’) the question is: which cluster influences it more, ‘advan-
with respect to their strategic criterion (parent node) (The rows tages’ or ‘others,’ and how much more? Table 2 shows the matrix of
‘weight’ and ‘global weight’ in Table 10). this paired comparison.

Alternatives Outlays
Cut-to-length
Tree-length Loading cost
Animal skidding Skidding cost
Cost of transportation to the mill

Othe r costs
Access road requirement
Need to supervision
Road side landing requirement
Skid trail planning
Training the personnel

Fig. 3. Micro view of Costs network.


I. Ghajar, A. Najafi / Forest Policy and Economics 21 (2012) 81–91 85

Opportunities
Number of employment
Coordination to selection thinning
Alternatives
Coordination to group selection
Cut-to-length Sorting and storage of various wood
Tree-length assortments
Animal skidding Number of work days per year
Organization
Easy establishment and camping
Accessibility of labors

Fig. 4. Micro view of the Opportunities network.

The matrix of priorities of all cluster comparisons under the Bene- where W is the weighted supermatrix, N indicates the sequence,
fits subnetwork is given in Table 3. and k is the exponent determined by iteration (Wolfslehner et al.,
Next the elements of a cluster are prioritized with respect to the 2005).
elements of the other clusters that have influence on it. For example, After calculating the limit matrix, the priorities of alternatives are
for the Benefits subnetwork, given advantage 1, ‘loading productivi- idealized by dividing by the largest priorities (Table 8). Doing this for
ty,’ and two alternatives, ‘cut-to-length’ and ‘tree-length,’ which all merits (BOCR) yields the first four columns of Table 11. The overall
alternative (element) has more hour productivity in the loading synthesized priorities for alternatives under BOCR are also given in
step of wood extraction and how much more? The answer to this this Table.
question is given in the (1, 2) position in the matrix of paired compar- Step 3: After identifying the ideal alternative under each merit, the
isons (Table 4). ratings of BOCR are performed. Generally, for rating BOCR (or any al-
The obtained relative priorities of alternatives are inserted in the ternative) correctly, they must be independent of one another. Other-
unweighted supermatrix given in Table 5. These priorities are then wise, the presence or absence of an alternative must have no effect on
multiplied by the weight of the cluster ‘alternatives’ in the cell (alter- how one rates any of the others; this kind of ranking with respect to
natives, advantages) from Table 3. The results of these processes are an ideal is called absolute measurement or rating (Saaty et al., 2006). In
highlighted in Tables 5 and 6. The values of Table 6 are used to obtain order to rate BOCR with respect to an ideal, intensity levels are creat-
the limiting priorities of elements in the clusters under the Benefits ed (for example, very high, high, medium, low, and very low in our
subnetwork (Table 7). Mathematically the limit of the supermatrix case). Then, to establish priorities, they are compared pairwise. The
is processed by raising the entire supermatrix to powers until conver- resulting priorities are normalized by dividing by the largest value
gence in terms of a limes (i.e., a Cesaro sum): among them, so that very high would have a value of 1.000 and
others would be proportionately less (Table 9). The prioritized strate-
 X gic subcriteria are used to rate the BOCR by first taking the ideal alter-
1 N K
Lim w native for each merit obtained in Step 2 and then selecting the
N k−1
appropriate intensity from categories that we had created before,

Negative effects on the soil


Alternatives
Cut-to-length Ground disturbance (dry conditions)
Tree-length Ground disturbance (wet conditions)
Animal skidding Road side landing impact
Soil compaction
Runoff

Damage to vegetation
Other risks
Accidents Damage to regeneration by browsing
Possibility of breakage Damage to residual regeneration
Damage to machinery (skidding routes)
Dirt contamination Damage to residual regeneration
(winching stripes)
Damage to residual trees (skidding
routes)
Damage to residual trees (winching
stripes)
Environmental pollution

Fig. 5. Micro view of the Risk network.


86 I. Ghajar, A. Najafi / Forest Policy and Economics 21 (2012) 81–91

Table 1 Table 3
Cluster influences under the Benefits subnetwork “x” represent the relationship be- Cluster priorities of the Benefits subnetwork.
tween the clusters.
Clusters Advantages Alternatives Others
Benefits Alternatives Advantages Others
Advantages 0 0.875 0.5
Alternatives 0 x x Alternatives 1.0 0 0.5
Advantages x 0 0 Others 0 0.125 0
Others x x 0

also which of the children nodes (elements) was stronger than anoth-
i.e. very high, high, medium, low, very low, (Table 9) for that ideal er. In detail, preference values are calculated within three matrices:
alternative, on each strategic subcriterion (Table 10). The selected (i) the unweighted supermatrix is derived directly from pairwise
intensity for each merit should describe it best on each subcriterion. comparison ratios; (ii) within the weighted supermatrix, the values
For example, to select an intensity for Benefits in relation with the are multiplied by cluster weights and normalized by column; and
subcriterion ‘maximum production,’ the question is how much of (iii) the limited supermatrix calculates the priorities of the alterna-
the ideal alternative under Benefits (‘tree-length’) can realize this tives by converging the supermatrix.
goal? The answer is very high. Doing this for all subcriteria yields
the selected intensities of Benefits (the highlighted cells in 3.1. BOCR results
Table 10). Likewise, the intensities of Opportunities, Costs, and Risks
are selected. The priorities of ratings of BOCR under each subcriteria 3.1.1. Benefits model
are given as the numbers in parentheses in Table 10. A score is com- Fig. 2 illustrates the clusters and their respective elements in the
puted for each merit by multiplying the priority of selected intensity benefits network. The ‘advantages’ cluster refers to productivity per
times by the priority of the criterion (global weight in Table 10) and hour in each of the three parts of the harvesting operation. This clus-
summing for all the subcriteria (shown in the total column in ter affects the ‘alternatives’ cluster because it decides which method
Table 10). The priorities of BOCR are obtained by normalizing the is the best one in terms of productivity of each stage. The ‘others’ clus-
total score column by dividing by the sum of value in it (Table 10). ter contains some other benefits that are favorable for forest manage-
The obtained normalized values are used for the priorities of the ment. This cluster affects both ‘advantages’ and ‘alternatives’ clusters
BOCR to do synthesis in the top-level network (Fig. 1). and is influenced by ‘alternatives.’
Step 4: Relative importance values for the alternatives are deter- The results of the overall synthesis of judgment in the benefits
mined based on two formulas, the additive and the multiplicative. network (Table 11) indicated that the alternative ‘tree-length’ gives
The additive and multiplicative formulas can be given as bB + oO − the highest benefit and has a considerably larger value than other al-
cC − rR and {B b O o[(1/C)Normalized] c[(1/R)Normalized] r}, respectively, ternatives from a benefits point of view. Following ‘tree-length,’ ‘cut-
where B, O, C and R represent the overall synthesized priorities for al- to-length’ was identified as the second best alternative and only has a
ternatives under BOCR; whereas b, o, c, and r are BOCR rates (22, 27, 0.288 priority.
28). The structure of BOCR subnetworks is given in Figs. 2 to 5. These
subnets (subsystems) are made up of components (clusters) and each 3.1.2. Costs model
component consists of elements (nodes). Fig. 3 illustrates the clusters in the costs network and their respec-
To derive priorities for elements, pairwise comparisons are done tive elements. The ‘outlays’ cluster refers to costs of each cubic meter
for weighting the criteria and alternatives and estimating the direc- of wood in each of the three stages of the harvesting operation. The
tion and importance of influences of one element on other both in ‘outlays’ cluster was affected by the ‘alternatives’ and ‘other costs’
the top-level network and BOCR subnetworks. clusters and simultaneously was impacted by the ‘alternatives’ clus-
Generally there are three types of components (Saaty et al., 2006): ter. ‘Other costs’ was impacted by the ‘alternatives’ cluster. Choosing
a harvesting method also affected these costs.
1– components which no arrow enters (source components) The results of synthesized judgment in the costs network
2– components from which no arrow leaves (sink components) (Table 11) showed that, with respect to costs of the network, the ‘an-
3– components which arrows both enter and exit (transient imal skidding’ alternative had the highest costs in forest manage-
components). ment. The ‘cut-to-length’ alternative had the smallest impact on
harvesting costs.
In this study, we just had the third component and most of com-
ponents form a cycle of two components that feed back and forth 3.1.3. Opportunities model
into each other. Types of connection are different. Some components The structure of clusters and their respective elements in the op-
have loops that connect them to themselves (innerdependent such as portunities model are shown in Fig. 4. There were two clusters with
‘other’ in the risks subnetwork). All other connections represent de- two-way links in this model. The opportunities cluster referred to un-
pendence between components (outerdependent). certain favorable opportunities that could be created depending on
the harvesting method that was chosen. The opportunities cluster af-
3. Synthesis and results fected the alternative cluster and would also be impacted by each
alternative.
Comparison of each couple of nodes with respect to a parent node Synthesis of judgments in the opportunities model can be seen in
in the model showed these elements influenced the parent node and Table 11. These results indicated that from an opportunities point of

Table 2 Table 4
Cluster comparison with respect to the alternative cluster under the Benefits Relative priorities of alternatives with respect to loading productivity.
subnetwork.
Alternatives Cut-to-length Tree-length Animal skidding Priorities
Clusters Advantages Others Priorities
Cut-to-length 0 7 5 0.1734
Advantages 1 7 0.875 Tree-length 1/7 0 9 0.7719
Others 1/7 1 0.125 Animal skidding 1/5 1/9 0 0.0545
I. Ghajar, A. Najafi / Forest Policy and Economics 21 (2012) 81–91 87

Table 5
Unweighted supermatrix of the Benefit subnetwork.

Advantages Alternatives Others

Loading Skidding Productivity of Cut- to- length Tree-length Animal Maximum Slash The value of
productivity productivity transportation skidding skidding distribution yield timber
to the mill distance

Advantages Loading productivity 0 0 0 0.2555 0.6266 0.2255 0.4705 0 0


Skidding productivity 0 0 0 0.6738 0.2796 0.6738 0.0588 0 0
Productivity of 0 0 0 0.1006 0.0936 0.1006 0.4705 0 0
transportation to
the mill
Alternatives Cut- to-length 0.1734 0.1811 0.3196 0 0 0 0.2872 0.2176 0.3568
Tree-length 0.7719 0.7535 0.5584 0 0 0 0.6348 0.0914 0.5891
Animal skidding 0.0545 0.0653 0.1219 0 0 0 0.0779 0.6909 0.0540
Others Maximum skidding distance 0 0 0 0.1772 0.2049 0.1811 0 0 0
Slash distribution 0 0 0 0.0852 0.0727 0.7535 0 0 0
The value of yield timber 0 0 0 0.7374 0.7222 0.0653 0 0 0

view, the alternative ‘cut-to-length’ would create more positive op- subcriteria in Table 10. It can be seen that 7 strategic subcriteria
portunities than both ‘tree-length’ and ‘animal skidding.’ There are were used to rate the BOCR. A five-step scale design for rating the
no great differences between the three alternatives, as priorities strategic subcriteria was: very high, high, medium, low and very
show. low. Table 9 shows the priorities of intensities in ideal form that
was normalized by dividing each by the largest.
3.1.4. Risks model The BOCR ratings were obtained according to the ideal alternative
Since the risks model contains more interactions and risk factors under each of the benefits, opportunities, costs and risks (the first
occur more often in the forest operation, the structure of this model four columns of Table 11). After selecting the appropriate intensity
is more complicated. The inner- and outerdependencies of clusters for the ideal alternative from Table 9 for each strategic subcriterion
and their respective elements in the risks model are shown in Fig. 5. and making comparisons, the BOCR were ranked. Resulting
The ‘soil’ cluster returns negative effects of harvesting operations on priorities as unnormalized and normalized are shown on the left in
the ground. The ‘vegetation’ cluster refers to risks of vegetation dam- Table 10.
age that may occur as a result of applying a harvesting method. The Relative importance values and the final ranking of the alterna-
‘other’ cluster contains personnel, machinery, and wood risks that tives are shown in Table 11. These priorities were computed by
can affect each other (innerdependency). These three clusters affect- using four normalized values of B, O, C and R based on two different
ed alternatives and were also impacted by harvesting methods. Ele- formulas: the multiplicative (ratio) and the additive (total).
ments of ‘soil’ and ‘vegetation’ clusters affected each other.
Synthesis of judgment in the risks model is shown in Table 11. Re- 3.3. Sensitivity analysis
sults showed that the alternative ‘cut-to-length’ had the highest risks.
The least risky alternative was the ‘animal skidding’ method. The results of sensitivity analysis of alternatives for each of the
main control criteria identified as the benefits, opportunities, costs,
3.2. Rating and synthesis of the entire model and risks associated with the goal of the best harvesting method are
shown in Figs. 6–9. Ranking of alternatives varies when priority rat-
The importance of benefits, opportunities, costs, and risks is deter- ings are modified. Effects of this modification could be studied by per-
mined by a rating process. Before rating, preference of subcriteria or forming sensitivity analysis.
subgoals should be calculated in the top-level network, because will-
ingness to realize one subgoal may be less than another subgoal 3.3.1. Benefits
(Erdogmus et al., 2006). Super Decision software computed the The benefit sensitivity analysis, illustrated in Fig. 6, showed that
weights that express the willingness level of strategic subcriteria. the greatest benefits would always be achieved through the alterna-
These weights were given in parenthesis under strategic criteria and tive ‘tree-length.’ With less than 35% emphasis on benefits, this

Table 6
Weighted supermatrix of the Benefit subnetwork.

Advantages Alternatives Others

Loading Skidding Productivity of Cut-to-length Tree-length Animal Maximum Slash The value of y
productivity productivity transportation skidding skidding distribution ield timber
to the mill distance

Advantages Loading productivity 0 0 0 0.1973 0.5483 0.1973 0.2352 0 0


Skidding productivity 0 0 0 0.5895 0.2447 0.5895 0.0294 0 0
Productivity of 0 0 0 0.0880 0.0819 0.0880 0.2352 0 0
transportation to
the mill
Alternatives Cut-to-length 0.1734 0.1811 0.3196 0 0 0 0.1436 0.2176 0.3568
Tree-length 0.7719 0.7535 0.5584 0 0 0 0.3174 0.0914 0.5891
Animal skidding 0.0545 0.0653 0.1219 0 0 0 0.0389 0.6909 0.0540
Others Maximum skidding 0 0 0 0.0221 0.0256 0.0226 0 0 0
distance
Slash distribution 0 0 0 0.0106 0.0090 0.0941 0 0 0
The value of yield timber 0 0 0 0.0921 0.0902 0.0081 0 0 0
88 I. Ghajar, A. Najafi / Forest Policy and Economics 21 (2012) 81–91

Table 7
Limit matrix of the Benefit subnetwork.

Advantages Alternatives Others

Loading Skidding Productivity of Cut-to- Tree- Animal Maximum Slash The value of
productivity productivity transportation to the length length skidding skidding distribution yield timber
mill distance

Advantages Loading productivity 0.2263 0.2263 0.2263 0.2263 0.2263 0.2263 0.2263 0.2263 0.2263
Skidding productivity 0.1701 0.1701 0.1701 0.1701 0.1701 0.1701 0.1701 0.1701 0.1701
Productivity of 0.0444 0.0444 0.0444 0.0444 0.0444 0.0444 0.0444 0.0444 0.0444
transportation to the
mill
Alternatives Cut-to-length 0.1027 0.1027 0.1027 0.1027 0.1027 0.1027 0.1027 0.1027 0.1027
Tree-length 0.3571 0.3571 0.3571 0.3571 0.3571 0.3571 0.3571 0.3571 0.3571
Animal skidding 0.0370 0.0370 0.0370 0.0370 0.0370 0.0370 0.0370 0.0370 0.0370
Others Maximum skidding 0.0122 0.0122 0.0122 0.0122 0.0122 0.0122 0.0122 0.0122 0.0122
distance
Slash distribution 0.0078 0.0078 0.0078 0.0078 0.0078 0.0078 0.0078 0.0078 0.0078
The value of yield 0.0420 0.0420 0.0420 0.0420 0.0420 0.0420 0.0420 0.0420 0.0420
timber

alternative is followed by ‘animal skidding’ and after that by ‘cut-to- data are amassed through participatory collection methods before
length.’ It should be observed that by increasing the emphasis on they are evaluated and passed on to farmers, foresters, government
the benefits of selecting the best harvesting method, the priority of policy makers, NGOs, etc. (Grant and Sear, 1999). The main strategic
‘tree-length’ method increases while simultaneously decreasing for criteria that were targeted in this study for selecting the appropriate
both alternatives. harvesting method (i.e. economic, social, and environmental/ecologi-
cal) were based on SFM guidelines established by the Food and Agricul-
3.3.2. Costs ture Organization of the United Nations (FAO) and the International
The cost sensitivity analysis illustrated in Fig. 7 indicates that the Tropical Timber Organization (ITTO) (ITTO, 2002). Separate evaluation
highest costs will be presented through ‘animal skidding’ except in of alternatives with respect to each merit in their respective subnet-
primary emphasis on costs. The next costly alternative is ‘cut-to- work, and rating the BOCR based on strategic subcriteria (subgoals)
length’ which is followed by ‘tree-length’ as the least costly choice. that are derived from three main strategic criteria (main goals), helps
After approximately 70% emphasis, the costs alternative ‘tree-length’ the manager or any decision maker arrive at the suitable conclusion,
becomes the best harvesting method. supported by the collected data.
Although the tree-length method was identified as the best har-
3.3.3. Opportunities vesting method within the conditions of this study, it is inevitable
The opportunity sensitivity analysis (Fig. 8) indicates that from that this method would be applied in combination with other
primary emphasis to the end, alternatives move toward convergence. methods, especially animal skidding. Since forest production units
The alternative ‘tree-length’ yields the highest opportunity until an are different from industries in Iran, preference of the forest indus-
85% emphasis and after that ‘cut-to-length’ begins to provide the tries was considered as ‘the value of yield timber’ delivered to the
highest positive opportunities. mill.
The amount of production is dependent on the condition of har-
vesting operations (i.e. the factors and variables that affect efficiency
3.3.4. Risks
and cost of production). These variables are mean volume of logs,
The risk sensitivity analysis (see Fig. 9) indicated that, similar to
stand density per hectare, extraction distance, operator skills, and,
benefits, the model is very sensitive to changes in weight of the
of course, topography and micro topography (Kellogg and Spong,
risks, as ‘animal skidding’ was the riskiest alternative when the
2004). Factors such as percent slope on skid trails, number of logs
importance of risks was low. With an approximate 36% emphasis on
per cycle, and volume per cycle affect skidding time (Sobhani and
risks, it was the least risky alternative in comparison with both
Ghasemzadeh, 1990), and consequently change the cost of producing
other methods. ‘Cut-to-length’ creates the highest risks by perform-
a cubic meter of wood in harvesting operations. According to these
ing the harvesting operation.
findings, calculating each parameter under different conditions
leads to different quantities. Using ratio-scale based techniques
4. Discussion
helps the managers to select the best option in a given condition.
The final synthesized priorities of alternatives (i.e. measures)
To continue the progress of forest management toward SFM an
obtained in this research may vary in other physiographic situations,
‘intensively planned and carefully controlled implementation of
but creating a complete decision framework that takes into account
harvesting operation for minimizing the impact on forest stands and
all possible criteria in the model and applies all four control criteria
soil, especially in individual tree selection cutting’ (Bull et al., 2001) is
required. In the context of SFM, social, economic and environmental

Table 9
Deriving priorities for intensity levels.
Table 8
Priorities of alternatives for the Benefit subnetwork in Ideal form. Very high High Medium Low Very low Priorities Idealized

Very high 1 3 5 7 9 0.5128 1.000


Alternatives Priorities (ideal form)
High 1/3 1 3 5 7 0.2614 0.5099
Cut-to-length 0.2877 Medium 1/5 1/3 1 3 5 0.1289 0.2515
Tree-length 1.000 Low 1/7 1/5 1/3 1 3 0.0633 0.1235
Animal skidding 0.1037 Very low 1/9 1/7 1/5 1/3 1 0.0333 0.0650
I. Ghajar, A. Najafi / Forest Policy and Economics 21 (2012) 81–91 89

Table 10
Rating BOCR with respect to strategic subcriteria.

The best harvesting method

Criteria Economical Ecological Social Total Normalized


0.539 0.296 0.163 rate rate
Subcriteria Maximum Maximum Sustainable Minimum soil Minimum Sustainable Number of Rural
production revenue production disturbances vegetation ecosystem employment development
damages
Weight 0.090 0.661 0.244 0.225 0.100 0.673 0.750 0.250
Global weight 0.048 0.358 0.132 0.066 0.029 0.200 0.122 0.040
Benefits Very high Very high High (0.509) Very high (1.000) High (0.509) Very high Very high Very high 0.920 0.430
(1.000) (1.000) (1.000) (1.000) (1.000)
Opportunities High (0.509) Medium High (0.509) Medium (0.251) Low (0.123) High (0.509) High (0.509) Low (0.123) 0.375 0.175
(0.251)
Costs Medium Low (0.123) Medium Low (0.123) Very low (0.065) High (0.509) Very high High (0.509) 0.345 0.161
(0.251) (0.251) (1.000)
Risks Low (0.123) Very high Very low Medium (0.251) Very high (1.000) Low (0.123) Medium High (0.509) 0.499 0.233
(1.000) (0.065) (0.251)

as merits of decision can improve the design of policy and any deci- The sensitivity analysis of the costs model also indicated that by
sion for forest operations in other regions. increasing the importance of the costs, the tree-length method is
The sensitivity analysis showed that the priorities of alternatives the least costly option under approximately 70% emphasis on costs
change if the weight of each of the BOCR varies. In fact, emphasis on and after this point the cut-to-length method becomes the best
each merit (i.e. BOCR) reflects forest management considerations choice. Animal skidding was always the most costly alternative in
that are applied by the managers who make the model and decide this analysis. These results indicated that if forest managers perceive
about harvesting practices. As the results of sensitivity analysis for the costs as the most important criterion in making decision or have
benefits showed, the tree-length method was always the best alterna- limited financial resources, they would decide to apply the cut-to-
tive, except when very low importance was given to benefits. This length method. Meanwhile, if the weight of costs was decreasing,
means that if forest management was based on the most advanta- the final option would be the tree-length method.
geous harvesting operations, this method provides the maximum Sensitivity analysis of the third control criterion, namely opportu-
benefits for each utilization unit. nities, showed that creating positive opportunities for each of the
three methods are the same when the weight of this criterion is in-
creasing. There would not be any preference between alternatives ex-
cept in a primary emphasis on opportunities. The last criterion that
Table 11
Priorities for alternatives under BOCR and final synthesized results from the ANP
sensitivity analysis was performed on, based on the fact that it was
model. an independent variable, was risks. By looking at Fig. 9, one can see
that when risks are increased, animal skidding was the least risky op-
Alternatives Benefits Opportunities Costs Risks Final outcome Final outcome
tion, and tree-length and finally cut-to-length are the riskiest
using additive using
(norm) multiplicative methods for harvesting operations.
(norm) Unlike the initial thought that animal skidding was generally a
Cut-to- 0.288 1.000 0.536 1.000 0.273 0.136 risky option for forest harvesting, results of this study showed that
length damage to the soil and residual vegetation and also the number and
Tree-length 1.000 0.943 0.617 0.525 0.443 0.738 quality of accidents in cut-to-length and then the tree-length
Animal 0.104 0.915 1.000 0.190 0.283 0.126 methods were riskier by far. Finally when the weight of risks de-
skidding
creased (fewer than 35% emphasis), the tree-length method provides
Priority

Priority

Tree-length

Tree-length
Cut--length

Cut-to-length Animal skidding

Animal skidding

Emphasis on Benefits Emphasis on Costs

Fig. 6. Sensitivity analysis of Benefits. Fig. 7. Sensitivity analysis of Costs.


90 I. Ghajar, A. Najafi / Forest Policy and Economics 21 (2012) 81–91

• The cut-to-length method is suggested once the financial situation

Priority
of forest operation company is not good and stable.
• The cut-to-length method seems to create more job opportunities
when the emphasis on this criterion is high.
• The animal skidding method is recommended in sensitive situation
of topography and vegetation.

By making minor adjustments for local conditions, this process


can serve to identify appropriate harvesting improvements in many
Tree-length
other regions. In the case of a large number of tangible and intangible
criteria, making decision using Multi Criteria Decision Making
(MCDM) approaches could help to overcome problems of different
Animal skidding types of variables. However, more applied researches are required
Cut--length to approve the efficiency of MCDM methods in forest activities.

References

Adebayo Adebola, B., Han, H.S., Johnson, L., 2007. Productivity and cost of cut-to-length
and whole-tree harvesting in a mixed-conifer stand. Forest Products Society. Forest
Emphasis on Opportunities Product Journal 57 (5), 59–69.
Applegate, G., Putz, F.E., Snook, L.K., 2004. Who pays for and who benefits from
improved timber harvesting practices in the tropics? Lessons learned and informa-
Fig. 8. Sensitivity analysis of Opportunities.
tion gaps: Center for International Forestry Research., 43.
Bjorheden, R., 1991. Basic time concepts for international comparisons of time study
reports. Journal of Forest Engineering 2 (2), 33–39.
the lowest risks. From the risks point of view, therefore, when the Bull, G.Q., Pulkki, R., Killmann, W., Schwab, O., 2001. Does it cost or does it pay? An
condition of topography and vegetation is very sensitive, animal skid- investigation of the costs and benefits of reduced impact logging. ITTO Tropical
ding is strongly recommended. Forest Update.11/2:3.
Chang, C.W., Wu, C.R., Chen, H.C., 2009. Analytic network process decision-making to
According to results of the sensitivity analysis for BOCR subnet-
assess slicing machine in terms of precision and control wafer quality. Robotics
works, the best choice under benefits and risks is always the tree- and Computer-Integrated Manufacturing 25, 641–650.
length method. Since obtained rates (emphasis) of costs and opportu- Durst, P.B., Brown, C., Enters, T., 2003. Development and implementation of national
codes of practice for forest harvesting in Asia and Pacific. International Expert
nities are fewer than 50%, the tree-length method once again be-
Meeting on the Development and Implementation of National Codes of Practice
comes the dominant alternative and the best harvesting method in for Forest Harvesting—Issue and Options. Kazua academia hall, Kisarazu city.
the mountain forests of northern Iran. Chiba Prefecture, Japan, 17–20 November.
Dykstra, D.P., Heinrich, R., 1996. FAO Model Code of Forest Harvesting Practice. FAO,
Rome, Italy.
5. Conclusions Erdogmus, S., Aras, H., Koc, E., 2006. Evaluation of alternative fuels for residential
heating in Turkey using analytic network process (ANP) with group decision
Design of the ANP model for comparison of three common har- making. Renewable and Sustainable Energy Reviews 10, 269–279.
Grant, I.F., Sear, C., 1999. Decision Tools for Sustainable Development. Natural
vesting methods in this study (i.e. selection of the criteria and Resources Institute, Kent, UK.
goals) was based on FAO and RIL guidelines. According to the results Han, H.S., Lee, H.W., Johnson, L.R., 2004. Economic feasibility of an integrated harvest-
of the present study it could be concluded that: ing system for small-diameter trees in southwest Idaho. Forest Product Journal 54
(2), 21–27.
International Tropical Timber Organization, 2002. ITTO guidelines for the restoration,
• ANP provided a suitable framework to incorporate all calculable and management, and rehabilitation of degraded and secondary tropical forests. ITTO
incalculable variables in the decision model and helped to achieve Policy Development Series No 13. 86.
Jamshidi, R., Jeager, D., Rafatniaa, N., Tabari, M., 2008. Influence of two ground-based
the goal of formulating forest policy. skidding systems on soil compaction under different slope and gradient
• Tree-length method provided the maximum benefits for forest uti- conditions. International Journal of Forest Engineering 19 (1), 9–16.
lization units. Kim, K., Park, K., Seo, S., 1997. A matrix approach for telecommunications technology
selection. Computers and Industrial Engineering 33 (3–4), 833–836.
Kellogg, L.D., Spong, B., 2004. Production and cost of cut-to-length thinning: experience
from the Willmette young stand project. Forest. Res. Lab., Oregon State Univ.
Priority

Corvallis, Oregon.
Komarov, A., Chertov, O., Andrienko, G., Andrienko, N., Mikhailov, A., Gatalsky, P., 2002.
DESCARTES & EFIMOD: an integrated system for simulation modeling and explora-
tion data analysis for decision support in sustainable forestry. In: Rizzoli, A.E.,
Jakeman, A.J. (Eds.), Integrated Assessment and Decision Support, Proceedings of
the First Biennial Meeting of the International Environmental Modeling and Soft-
ware Society, iEMSs. Manno, Switzerland.
Animal skidding Mengistu, K., 2002. Ethiopia country paper. Workshop on Tropical Secondary Forest
Management in Africa: Reality and Perspectives. ICRAF and CIFOR. Nairobi,
Kenya, pp. 9–13.
Miettinen, K., 2006. International society on multiple criteria decision making mission.
Webpage. http://project.hkkk.fi/MCDM/intro.html 27 September 2006.
Momoh, J.A., Zhu, J.Z., 1998. Application of AHP/ANP to unit commitment in the
deregulated power industry. IEEE, pp. 817–822.
Naghdi, R., Rafatniaa, N., Sobhani, H., Jalali, G.H., Hosseini, S.M., 2005. A survey of the
efficiency of timber jack 450C wheeled skidder in sheared forests. Journal of
Tree-length
Natural Resources 57 (4), 657–687 (In Persian).
Niemiraa, M.P., Saaty, T.L., 2004. An analytic network process model for financial-crisis
forecasting. International Journal of Forecasting 20, 573–587.
Cut--length Pinard, M.A., Barker, M.G., Tay, J., 2000. Soil disturbance and post-logging forest
recovery on bulldozer paths in Sabah, Malaysia. Forest Ecology and Management
130, 213–225.
Pulkki, R., Bull, G., Schwab, O., 2002. Connecting reduced impact logging and forest
Emphasis on Risks certification. Wood for Africa forest Engineering Conference.
Saaty, T.L., 1977. A scaling method for priorities in hierarchical structures. Journal of
Fig. 9. Sensitivity analysis of Risks. Mathematical Psychology 15, 234–281.
I. Ghajar, A. Najafi / Forest Policy and Economics 21 (2012) 81–91 91

Saaty, T.L., 1999. Fundamentals of the analytic network process. Paper Presented at the Spong, B. D., 2007. A decision framework for the implementation of appropriate logging
ISAHP. Kobe, Japan. practices in developing countries: case study—Ethiopia. Ph.D. Department of Forest
Saaty, T.L., Vargas, Luis, G., 2006. Decision making with the analytic network process. Engineering, Oregon State University. USA.
Economic, Political, Social, and Technological Applications with Benefits, Wolfslehner, B., Vacik, H., Lexer, M.J., 2005. Application of the analytic network process
Opportunities, Costs, and Risks. New York. in multi-criteria analysis of sustainable forest management. Forest Ecology and
Sobhani, H., Ghasemzadeh, R., 1990. The study of effects of skidding by wheeled Management 205, 157–170.
skidder. Journal of Natural Resources 41, 53–64 (In Persian).

You might also like