=Paper=
{{Paper
|id=Vol-2030/HAICTA_2017_paper81
|storemode=property
|title=Assessing Forest Species Biogeophysical Suitability by Spatial Multicriteria Analysis: A Study Case for the Centro Region of Portugal
|pdfUrl=https://ceur-ws.org/Vol-2030/HAICTA_2017_paper81.pdf
|volume=Vol-2030
|authors=Luís Quinta-Nova,Cristina Alegria,Teresa Albuquerque,Natália Roque
|dblpUrl=https://dblp.org/rec/conf/haicta/Quinta-NovaAAR17
}}
==Assessing Forest Species Biogeophysical Suitability by Spatial Multicriteria Analysis: A Study Case for the Centro Region of Portugal==
https://ceur-ws.org/Vol-2030/HAICTA_2017_paper81.pdf
Assessing forest species biogeophysical suitability by
spatial multicriteria analysis: a study case for the Centro
region of Portugal
Natália Roque1, Isabel Navalho1, Cristina Alegria1, 2, Teresa Albuquerque 3, Luís
Quinta-Nova1, 4
1
Polytechnic Institute of Castelo Branco, School of Agriculture, Quinta da Senhora de
Mércules, Apartado 119, 6001-909 Castelo Branco, Portugal.
2
CERNAS - Centro de Estudos de Recursos Naturais, Ambiente e Sociedade, Polytechnic
Institute of Castelo Branco, Agriculture School, Castelo Branco, Portugal.
3
Polytechnic Institute of Castelo Branco, Technology School, Castelo Branco, Portugal.
4
GeoBioTec - Geobiociências, Geoengenharias e Geotecnologias, University of Aveiro,
Aveiro, Portugal.
Abstract. It is generally agreed that the choice of the most suitable uses based
in soil and climatic factors, complemented with socio-economic criteria,
promotes sustainable use of rural land. There are, however, different
methodologies for defining soil suitability to agroforestal systems or natural
and seminatural ecosystems, including agricultural uses, forest plantations,
agroforestry areas and priority areas for conservation. Many of these methods
rely on decision support systems based on multicriteria spatial analysis. In this
study it was intended to determine the different levels of suitability for the
most representative forest species in the central region of Portugal. For that
purpose a set of climatic, soil and topographic variables based in a geographic
information system, soil and biogeographic mapping were used. A stochastic
approach was undertaken in order to estimate several bioclimatic indices and
the associated spatial uncertainty. Results were duly discussed and addressed
in this framework. In the end, forest species suitability evaluation was
performed using the Analytic Hierarchy Process (AHP), methodology that
allows the exploitation of natural fitness of the territory, contributing to a
reflection on the adequacy of current and future occupations face the carrying
capacity of the environment. The use of this methodology will be important as
a supporting tool to public administration agents that work in forestry planning
and management.
Keywords: suitability, forest management, GIS, AHP.
681
1 Introduction
Understanding the spatial relationships between the different territorial functions
through the establishment of relations of continuity and connectivity between the
elements under consideration, together with geographical and alphanumeric
information allows the realization of spatial analysis to determine the degree of
functionality of each element in analysis (Guiomar et al. 2007).
Multicriteria analysis is an alternative assessment tool, particularly interesting
when you explore the various possible combination of criteria, allowing consider
different scenarios assessment. To Roy (1996) it is a mathematical tool that allows
you to compare different alternatives or scenarios, based on various criteria, in order
to help decision-makers choosing an alternative.
The Analytic Hierarchy Process (AHP) is one of the methods of multi-criteria
spatial analysis more widely used, developed in the 1970s by Thomas L. Saaty
(Steiguer et al., 2003). This process is based on mathematics and psychology and
provides a comprehensive and rational framework for structuring a decision problem,
allowing the representation and quantification of its elements, in order to relate these
elements with general objectives and evaluate alternative solutions (Saaty, 1980).
The AHP decomposes a problem, question or decision, in all the variables that
constitute it, in a scheme of criteria and sub-criteria, then making pairwise
comparisons between them (Antunes, 2012). The comparison between criteria is
made using a scale of 1 to 9, wherein 1 is equaly preferred and 9 to highly preferred
(Saaty, 1980).
The AHP reverts comparisons on numerical values that can be processed and
compared to the full extent of the problem. The weight of each factor allows the
evaluation of each of the elements within the defined hierarchy. This conversion
capability of empirical data in mathematical models distinguish the AHP other
decision-making techniques (Saaty, 1980).
The Multicriteria Spatial Decision Analysis has been widely applied in various
studies in different fields, many of which are published and are cited by many
authors as processes of relevant decision-making. This is the case of Kangas et al.
(2000) referring to the use of GIS in the decision-making process through the
multicriteria analysis in the planning of forest resources conservation actions,
allowing actions are directed by the determination of the priority areas.
Quinta-Nova and Roque (2014) developed a model based on multicriteria spatial
analysis AHP in order to determine the suitability levels for agroforestry uses of the
sub-region of Beira Interior Sul. The criteria used were the soil potential, slope and
aspect. The authors note that this analysis identified the areas where the use of land
shall be subject to a conversion and/or a change of management. Thus having
endogenous variables such as soil, climate and terrain elevation was intended to
determine the suitability of various forest species based on an AHP model to evaluate
the different levels of agroforestry suitability in the Centro region of Portugal.
682
2 Materials and methods
2.1 Study area
The study area is the center region of Portugal (Fig.1a), which is divided in six
forest management regions (CL – Centro Litoral; DL – Douro e Lafões; PIN – Pinhal
Interior Norte; PIS – Pinhal Interior Sul; BIN – Beira Interior Norte; and BIS – Beira
Interior Sul) (ICNF, 2015). Forest area in the center region of Portugal (Fig. 1b,c) is
mainly composed by maritime pine (Pinus pinaster Aiton) (544.585ha; 51%),
eucalyptus (Eucalyptus globulus Labill.) (357.805 ha; 34%), oaks (e.g. Quercus
pyrenaica Willd., Quercus faginea Lam. e Quercus robur L.) (52.585 ha; 5%) and
cork oak (Quercus suber L.) (45.221ha; 4%). The remaining forest area is ocuppied
by holm oak (Quercus rotundifolia Lam.), stone pine (Pinus pinea L.), chestnut
(Castanea sativa Mill.), acacia (Acacia sp.), other broadleaved and other coniferous
(AFN, 2010).
For each one of the six management regions referred to above there is a
correspondent Forest Management Regional Plan (FMRP) (ICNF, 2015) that propose
forest composition goals for 2010, 2025 and 2045. Additionally, the National
Strategy for Forests (FNS) (DR, 2006) proposes also forest composition goals for
2030 by forest management region. In that view, for the study area these official
documents propose an important decrease of eucalyptus area, a slightly decrease of
maritime pine area and a strong increase of oaks areas (Fig. 1d). Therefore, having in
consideration both the current distribution of forest species over the study area and
forest composition goals for the future, six forest species were selected, namely:
maritime pine, eucalyptus, Pyrenean oak, cork oak, Portuguese oak and holm oak.
Afterwards, these species biogeophysical suitability was assessed in order to support
future landscape planning.
683
Study area - NFI 2005 Other
300,000
Qx
250,000 Qr
200,000 Qs
Ec
Area (ha)
150,000 Pp
100,000
50,000
0
PIS
BIS
CL
DL
PIN
BIN
a) b)
Study area - NFI 2005 Study area - FMRP 2010 Other
Other
100% 100% Qx
Qx
90% 90%
Qr Qr
80% 80%
Qs
Forest composition
Qs
Forest composition
70% 70%
60% Ec 60% Ec
50% Pp 50% Pp
40% 40%
30% 30%
20% 20%
10% 10%
0% 0%
c) d)
PIS
BIS
CL
DL
PIN
BIN
PIS
BIS
CL
DL
PIN
BIN
Study area - FMRP 2025 Other Study area - FNS 2030 Other
100% Qx 100% Qx
90% 90%
Qr Qr
80% 80%
Qs Qs
Forest composition
Forest composition
70% 70%
60% Ec 60% Ec
50% Pp 50% Pp
40% 40%
30% 30%
20% 20%
10% 10%
0% 0%
PIS
BIS
PIS
BIS
CL
DL
CL
DL
PIN
BIN
PIN
BIN
d)
684
Study area - goals Other
Study area - FMRP 2045 Other 100%
Qx
100% 90%
Qx Qr
90% 80%
Qr Qs
Forest composition
80% 70%
Ec
Qs
Forest composition
70% 60%
Pp
60% Ec 50%
50% Pp 40%
40% 30%
30% 20%
20% 10%
0%
10%
FMRP 2010
FMRP 2025
FMRP 2045
FNS 2030
NFI 2005
0%
PIS
BIS
CL
DL
PIN
BIN
d)
Legend: Forest species - Pp – Pinus pinaster Ait., Ec – Eucalyptus sp., Qs – Quercus suber L., Qr –
Quercus rotundifolia Lam., Qx – Quercus pyrenaica Willd., Quercus faginea Lam. and Quercus
robur L Forest management regions - CL – Centro Litoral; DL – Douro e Lafões; PIN – Pinhal
Interior Norte; PIS – Pinhal Interior Sul; BIN – Beira Interior Norte; and BIS – Beira Interior Sul.
Fig. 1. Study area: a) Forest management regions (ICNF, 2015); b) Forest area (ha) in 2005 by
forest management region (NFI – National Forest Inventory) (AFN, 2010); c) Forest
composition (%) in 2005 (NFI) (AFN 2010); d) Forest composition goals (FMRP – Forest
Management Regional Plans and FNS –National Strategy for Forests) for 2010, 2025, 2030
and 2045 by forest management region and for the all study area (DR, 2006; ICNF, 2015).
2.2 Bioclimatic analysis
For the development of bioclimatic maps were used climatic data (Pp, Tp, Tmax,
Tmin, T, M e m) calculated from temporal series corresponding to climatological
normal of the period between 1981 and 2010 referring to 32 stations, located in
Portugal and Spain (Figure 2).
685
Fig. 2. Location of weather stations used in the preparation of the mapping of bioclimatic
indices.
The methodology now proposed results from the development of maps at the
national level, corresponding to each of the attributes to be used in the calculation of
bioclimatic indices (Figura 3).
Since the study variables can be assumed to be regionalized variables (Matheron,
1970), we began the study of the attributes by variogram analysis. The variogram is a
vector function, applicable to regionalized variables (Matheron, 1970), whose
argument is the distance vector h, which quantifies the variance of the increments of
the first order function (Soares, 2000). Experimental estimation function from the set
of experimental data carried out by applying the formula:
!(!)
1 !
(1)
𝛾 ℎ = 𝑍 𝑥! − 𝑍 𝑥! + ℎ
2𝑁 ℎ
!!!
Where Z(xi) and Z(xi + h) are the numerical values of the observed variable at the
points xi and xi + h, and N(h) is the number of pairs for a distance h. It is therefore
the average of the squared value of the differences between all pairs of points
existing in the geometrical field, spaced a distance h (Journel and Huijbregts, 1978).
686
The study of the variogram graphic behavior provides a description of the
structure of the spatial variation of the variable (Chica, 2005). The nugget effect
(Co), summarizes the behavior at the origin (Co), that is, the farther away from zero
the more random is the behavior of the variable. The other two parameters are the
platform (C1) and the amplitude (a) that define the area of influence and the
percentage of the total variance that will be used in the subsequent process of
interpolation or stochastic simulation.
The variogram study made to the study variables, did not allow to modeling
possible geometric anisotropy, and so were set omnidirectional models whose
parameters are summarized in Figure 5.
For the next step of interpolation was used ordinary kriging - KO (Journel e
Huijbregts, 1978; Soares, 2000).
The KO estimator of the found value is a linear weighting of experimental values
Z(xi) by unknown λi coefficients (Chica, 2005):
!
𝑍!" 𝑥 = λ! 𝑍 𝑥! (2)
!!!
687
Fig. 3. Representation of cartography at a national level, of the attributes used in the
calculation of the bioclimatic indices (Pp, Tp, Tmax, Tmin, T, M e m). It is also showed the
omnidirectional variogram, and the corresponding adjusted theoretical models in geostatistical
modeling by Ordinary Kriging.
688
Subsequently we calculated the following indices using map algebra:
continentality index (Ic = Tmax-Tmin); termicity index (It = 10*(T+M+m) and
ombrotermic index (Io = Pp/Tp). The SpaceStat 4.0.14 software (Biomedware) and
ArcMap 10 (ESRI) were used in the calculation process. The bioclimatic map was
obtained by the combination of three indices (Ic, It and Io) in a geographical analysis
function which groups the different ranges of each climatic domain in a new entry
that group in different combinations.
2.3 Diagnostic features of soil to the forest tree species
The species studied were the maritime pine (Pinus pinaster Ait.), eucalyptus
(Eucalyptus globulus Labill.), cork oak (Quercus suber L.), holm oak (Quercus
rotundifolia Lam.), Portuguese oak (Quercus faginea Lam. subsp. broteroi (P. Cout.)
A. Camus) and the Pyrenean oak (Quercus pyrenaica Willd.). Based on the
methodology developed by Ferreira et al. (2001), the soil units present in the study
area were classified in “diagnostic features” according to the soil conditions for the
development of forest species considered (Table 1). To each soil family was assigned
the corresponding diagnostic feature according to the limitation on forest
development.
Table 1. Species soil diagnosis characteristics (Correia and Oliveira, 2003; Dias et al., 2008).
Species Superior Reference Inferior
(3) (2) (1)
Textural Rocky outcrops
Pinus pinaster Ait. Expandable depth
discontinuity Unproductive areas
Eucalyptus globulus Labill. Expandable depth Rocky outcrops
_
Quercus pyrenaica Willd. Textural discontinuity Unproductive areas
Quercus suber L.
Textural Rocky outcrops
Quercus faginea Lam. subsp. Expandable depth
discontinuity Unproductive areas
broteroi (P. Cout.) A. Camus
Expandable depth Rocky outcrops
Quercus rotundifolia Lam. _
Textural discontinuity Unproductive areas
The development of soil interpretative maps and bioclimatic maps for each species
relied on the definition of three suitability classes, considering a reference class. In
the case of soil, the reference class is characterized by no constraints to the
development and growth of the tree species. Compared with the reference class, the
upper and lower than the reference classes have respectively less and more
restrictions for the survival, growth and development of forest species (Dias et al.,
2008).
The themes of diagnostic features and bioclimate were reclassified following the
methodology defined by Ferreira et al. (2001) and Dias et al. (2008) into three
classes: higher than the reference (3); reference (2); lower than the reference (1),
giving the most limiting classification.
The interpretative model of methodological procedures performed is presented in
Figure 4, where we present the working lines: i) Interpolation and geoprocessing for
determining slope, to ii) Spatial analysis to determine the diagnostic features
689
regarding forest use and iii) stochastic modeling using KO to calculate It, Ic and Io
indices.
Fig. 4. Geographic analysis and AHP methodology.
690
2.4 AHP pairwise comparison
Determination of criterion weights is crucial in multicriteria analysis. The AHP is
a mathematical method for this purpose when analyzing complex decision problems
(Saaty, 1980). It derives the weights through pairwise comparisons of the relative
importance between each two criteria. Through a pairwise comparison matrix, the
AHP calculates the weight value for each criterion (wi) by taking the eigenvector
corresponding to the largest eigenvalue of the matrix, and then normalizing the sum
of the components to a unity.It is necessary to verify the consistency of the matrix
after obtaining the weight values.
The consistency is judged on the basis of a consistency ratio CR. The
determination of CR value is critical. In our case study, we adopted a standard CR
threshold value of 0.10, which has been widely used as a measure of the consistency
in a set of judgments of AHP applications in literature. If CR <0.10, it deems that the
pairwise comparison matrix has acceptable consistency and the weight values
calculated are valid and can be utilized.
3 Results
3.1 Bioclimatic Indices
The aim of Bioclimatology is to determine the relationship between precipitation
and temperature values and the geographical distribution of species and plant
communities. The water availability in soil determines the distribution of plant
species within the limits defined by air temperature, and water is the factor that most
influences the physiology and plant morphology. The mapping of the calculated
bioclimatic indices - Ic Io It - are presented in Figure 5, showing the zoning
consistent with the Portuguese Centro region.
a) b) c)
Fig. 5. Modeling (Climatic normals 1981/2010): a) continentality index; b) ombrotermic
index; c) termicity index.
691
The contribution of the altitude in calculating the indices proved unpromising
since the correlation coefficients (r) were not very significant (Figure 6). The
continentality index shows a value of r of -58 %. Therefore in future works will be
used methods of stochastic simulation using altitude as auxiliary variable, in order to
characterize the associated spatial uncertainty.
Fig. 6. Correlation coefficients: a) continentality index; b) ombrotermic index; c) termicity
index.
3.2 Multicriteria Spatial Analysis using AHP
Figure 7 shows the input variables for the analysis of the diagnostic features and
bioclimate that were reclassified following the methodology defined by Ferreira et al.
(2001) and Dias et al. (2008) into three classes: high suitability (3); reference
suitability (2); Low or no suitability (1).
The result of AHP analysis in the studied species, together with the bioclimatic
and geomorphology characteristics indicate that the bioclimatic influence is
determinant in the development of these species, since the weighting of the AHP
analysis is 64.9 %, followed by the soil factor (diagnostic features) 27.9 %.
Fig. 7. Input data for multicriteria spatial analysis: a) bioclimatic map; b) diagnostic features
map; c) slope map.
The AHP process is completed by determining the relative importance of each
criteria/subcriteria and the validation of the consistency of these operations. If the
consistency ratio (CR) is less than 10 % (CR < 0.1) means that there is consistency in
the pairwise comparison matrix. In the following cartograms show the results of the
hierarchical analysis (Figure 8).
692
a) Pinus pinaster b) Eucalyptus globulus c) Quercus pyrenaica
d) Quercus suber e) Quercus faginea f) Quercus rotundifolia
High suitability
Reference suitability
Low or no suitability
Fig. 8. Representation of cartography, with the forest suitability levels for the tree species: a)
maritime pine; b) eucalyptus; c) Pyrenean oak; d) cork oak; e) Portuguese oak; f) holm oak.
4 Discussion
In this study it was developed a spatial model of spatial multicriteria evaluation in
GIS environment to determine the natural suitability for forestall tree species in
Centro region of Portugal (NUT II).
In this particular case it should be noted that despite having resorted to a limited
number of criteria in AHP analysis, this proved to be extremely important, since by
exploiting the land suitability and based on a set of biophysical factors, it is possible
assess the degree of importance of each criterion and identify the suitability of forest
species.
The AHP proved to be also suitable in assessing the suitability of the study area, to
allow the integration of the various criteria studied depending on the assigned
weights, being a very useful interactive tool in the analysis of the territory, which
enables decision-making and resolution problems.
This methodology allows the exploitation of natural territory suitability, based on
a set of biophysical factors contributing to a reflection on the adequacy of current and
future occupations due to the carrying capacity of the land.
693
The instrumental point of view, the use of this methodology may take an interest
to stakeholders and other persons with roles in the planning and land management.
References
1. Guiomar, N., Fernandes, J.P., Neves, N., 2007. Modelo de Análise Espacial para
Avaliação do carácter Multifuncional do Espaço. In Atas do III Congresso de
Estudos Rurais (III CER). Faro, Universidade do Algarve, 1-3 Nov. 2007 - SPER.
Évora.
2. Roy, B. (1996), Multicriteria methodology for decision aiding. Dordrecht.
Kluwer Academic.
3. Steiguer, J. E., Liberti, L., Schuler, A., Hansen, B., (2003). Multi-Criteria
Decision Models for Forestry and Natural. USDA Forest Service, Northeastern
Research Station, pp. 8; 16-23.
4. Antunes, O. E. D., 2012. Análise Multicritério em SIG para Determinação de um
Índice Espacializado de Pressão Antrópica Litoral. Casos de Espinho, Caparica e
Faro. Dissertação de Mestrado em Gestão do Território. Área de Especialização
em Deteção Remota e Sistemas de Informação Geográfica. Universidade Nova de
Lisboa. Faculdae de Ciências Sociais e Humanas. Lisboa.
5. Saaty, T. L., (2008). Decision making whit the Analytic Hierarchy Process.
Internacional Journal of Services Scienes. Vol.1, Nº 1.pp.83-98.
6. Kangas, J., Store, R.L., Leskinen, P., Mehtatalo, L., 2000. Improving the Quality
of Landscape Ecological Forest Planning by Utilizing Advanced Decision-
Support Tools. Forest Ecology and Management, Amsterdam, v.132, p.157-171.
7. Quinta-Nova, L.C., Roque, N., (2014) Agroflorestal Suitability Evaluation of a
Subregional Area in Portugal Using Multicritéria Spacial Analysis. Internacional
Congress of Landscape Ecology – Understanding Mediterranean Landscapes
Human vs. Nature, 23-25 October. Antalaya. Turkey.
8. AFN, 2010. Inventário Florestal Nacional Portugal Continental. 5º Inventário
Florestal Nacional 2005-2006. Relatório Final. Autoridade Florestal Nacional.
http://www.icnf.pt/portal/florestas/ifn/ifn5/relatorio-final-ifn5-florestat-1.
Accessed on May 2015.
9. Matheron, G. (1970): “La théorie des variables régionalisées, et ses
applications”.Centre Géostatistique et Morphologie Mathématique. Ecole
Nationale Supérieure des Mines de Paris. Paris.
10. Soares, A. (2000): “Geoestatística para as ciencias da terra e do ambiente”.
Editorial Press. 206 pp.
11. Journel, A. G. y Huijbregts, C. J. (1978): “Mining Geostatistics”. Academic
Press. London.
12. Chica, M. (2005): “La Geoestadística como herramienta de análisis espacial de
datos de inventario forestal”. Actas de la I reunión de inventario y teledetección
forestal. Cuad. Soc. Esp. Cienc. For. 19: 47-55 (2005).
13. Correia, A. V.; Oliveira, A. C., 2003. Principais espécies florestais com interesse
para Portugal. Zonas de influência atlântica. Estudos e Informação n.º 322, DGF,
Lisboa, 187 p.
694
14. Dias, S., Ferreira, A., Gonçalves, A., 2008. Definição de Zonas de Aptidão para
Espécies Florestais com Base em Características Edafo-Climáticas. Silva
Lusitana, n.º especial 17-35.
15. Ferreira, A. G., Gonçalves, A. C., Pinheiro, A.C., Gomes, C.P., Ilhéu, M., Neves,
N., Ribeiro, N., Santos, P., 2001. Plano Específico de Ordenamento Florestal para
o Alentejo.2001. Universidade de Évora. Évora.
695