**GIS-Based Models as Tools for Environmental Issues: Applications in the South of Portugal**

Jorge M. G. P. Isidoro, Helena M. N. P. V. Fernandez, Fernando M. G. Martins and João L. M. P. de Lima

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/48218

## **1. Introduction**

250 Cartography – A Tool for Spatial Analysis

Ecology 1: 153-162.

1088.

243–262

raton, USA.

structure, Oikos 68:571-3

Sciences. Ås, Norway.

Management 59(2):228-237.

[52] Roldán Martín MJ, De Pablo CL, Martín de Agar P, (2006) Landscape changes over time: comparation of lands uses, boundaries and mosaics, Landscape Ecology 21: 1075-

[54] O'Neill R, Krummel J, Gardner R, Sugihara G, Jackson D, Milne B, Turner M, Zygmunt B, Christensen S, Dale V, Graham R, (1988) Indices of landscape pattern, Lanscape

[55] Marull J, Mallarach JM, (2005) GIS methodology for assessing ecological connectivity: application to the Barcelona Metropolitan Area, Landscape and Urban Planning 71(2-4):

[56] Bennett G (2004) Integrating Biodiversity Conservation and Sustainable Use: Lessons learned from Ecological Networks. Gland, Switzerland and Cambridge UK. UICN: 55 p. [57] Bodin Ö, Norberg J (2007) A network approach for analyzing spatially structured

[58] IUCN (1994) Guidelines for protected areas management categories CNPPA and

[59] Gómez-Sal A, Belmontes JA, Nicolau JM, (2003) Assessing landscape values: a proposal for a multidimensional conceptual model, Ecological Modelling,168: 319-341. [60] Gómez-Sal A, Álvarez J, Muñoz-Yanguas MA, Rebollo S, (1993) Patterns of change in the agrarian landscape in the area of the Cantabrian Mountains (Spain). Assessment by

[61] Paoletti MG, Editors: Landscape ecology and agrosystems. Lewis Publishers, Boca

[62] Taylor DR, Fahrig L, Henein K, (1993) Connectivity is a vital element of landscape

[63] Opdam P, Foppen R, Vos C, (2002) Bridging the gap between empirical knowledge and

[64] Fahrig L (2003) Effects of Habitat Fragmentation on Biodiversity, Annual Review of

[65] Fry G (2005) Lectures on Landscape Ecology at the Norwegian University of Life

[66] Beier P (1995) Dispersal of juvenile cougars in fragmented habitat, Journal of Wildlife

[67] Berggren A, Birath B, Kindvall O (2002) Effect of corridors and habitat edges on dispersal behaviour, movement rates and movement angles in Roesel's Bush-Cricket

spatial planning in landscape ecology, Landscape Ecology 16: 767–779.

populations in fragmented landscape, Landscape Ecology 22: 31–44.

WCMC, IUCN, GLAND, Switzerland and Camdbrige, UK.

transition probabilities. In: Bunce RGH, Ryszkoewski L.

Ecology, Evolution and Systematics 34: 487-515.

(*Metrioptera roeseli*), Conservation Biology 16: 1562–1569

[53] Pielou EC (1977) Mathematical Ecology. Wiley, London, UK.

Geographical Information Systems (GIS) simulate a given geographic space in a computational environment, allowing to store, map and analyse large amounts of georeferenced data (*e.g.*, Umbelino *et al.*, 2009). GIS were converted into a powerful tool in regional natural resources assessment, as it permits a speedy integration and representation of several biophysical attributes (*e.g.*, Bastian, 2000; Bocco *et al.*, 2001) from diverse origins such as, *e.g.*, topographic, cartographic, photogrametric, GPS and remote sensing.

The integration of Digital Terrain Models (DTM) in GIS leads to the emergence of methodologies to represent and simulate the real-word, complementing the use of thematic environmental information (*e.g.*, Felicísimo, 1999). A DTM is a numerical representation of a variable obtained from a discrete set of points, with well-known cartographic coordinates, which distribution allows calculating, by interpolation, that variable for any arbitrary point (Fernandez, 2004). If the mesh of points is altitude-related, the DTM is designated as a Digital Elevation Model (DEM). From a DEM, it becomes easy to attain topographic-derivate models (*e.g.*, slopes, orientations, curvature and visibility, shadowed and exposed areas).

Many authors have used DEM processing techniques to automatically extract geographic features (*e.g.*, Herrington & Pellegrini, 2000; MacMillan *et al.*, 2000; Burrough *et al.*, 2001; Jordán *et al.*, 2007b; Zavala *et al.*, 2005a, 2007), hydrologic structures (*e.g.*, Flanagan *et al.*, 2000; Maidment, 2000), erosive processes (*e.g.*, Zavala *et al.*, 2005b), vegetation habitats (*e.g.*, Anaya-Romero, 2004; Anaya-Romero *et al.*, 2005; Jordán *et al.*, 2007a; Pino *et al.*, 2010), among other uses.

© 2012 Isidoro et al., licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 Isidoro et al., licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

In this chapter three GIS-based models were used, relying on different methodologies and techniques, to illustrate visualization and quantification of the geomorphic processes. This valuable input for decision-makers is attained through the versatility of GIS. Geo-form mapping in a coastal lagoon catchment, rainfall erosivity on a mountain ridge and urban flood delimitation show the potential usefulness of DEM/DTM based cartographic models for helping to solve environmental issues. All case studies presented are from the south of mainland Portugal.

GIS-Based Models as Tools for Environmental Issues: Applications in the South of Portugal 253

**2. Location and classification of geo-forms in the Ria Formosa estuary** 

has happened in other Mediterranean areas (Kosmas *et al.*, 2000).

hierarchy is made in terms of size, order and geometric complexity.

The study of the risk of soil degradation is the starting point for development and sustainable land management. The global warming and the land use changes expected in the XXI Century predict loss of quality and reduction of soil's productivity (Cerdan *et al.*, 2010). In the Mediterranean area the natural and semi-natural vegetation is sclerophyllous, which is well suited for the local climatic conditions, however, extreme weather and human activity can cause imbalances in the ecosystem (Kosmas *et al.*, 2000). The southern Portugal is a region where the balance between the natural environment and human activity is very sensitive to erosion and desertification (Gonçalves *et al.*, 2010). Thus, it is necessary to employ control, prevention and correction measures to preserve the soil and prevent the emergence and intensification of desertification processes, which can become irreversible, as

This study emphasizes geomorphologic processes, because they describe the natural space, the dynamics of occupations and the anthropogenic changes. According to Hammond (1954, 1964), geomorphologic studies of the earth's surface can be carried out over large areas (small scale) and based on the analysis of the land features, topographic maps or directly through field measurements. The variables considered should be quantitative, been used for

Later, with the appearance of the GIS, Dikau (1989) and Dikau *et al.* (1991), the Hammond procedures were rectified and automated in some regions in the United States. The DEM becomes the most important tool in the process of identifying landforms. Using the "moving-window" process and through algorithms based on local operators (spatial filters) it is possible to create models (such as slope, local relief and relative position). This allows a more detailed study, with a large number of variables in a large scale. The landforms

Many researchers have used this methodology, although with some modifications as explained by Martínez-Zavala *et al.* (2004) and Jordan *et al.* (2005), in studies conducted in Spain and Mexico in a detailed scale. Drãgut and Blaschke (2006), and Gerçek *et al.* (2011) classified the landforms in regions of Transylvania (Romania), Germany and Turkey, respectively based in decision rules of fuzzy logic. Another technique widely used was the non-supervised classification based on levels of information. For example, Sallun *et al.* (2007) in the catchment of Alto Rio Paraná in Brazil extracted information by applying the Principal Component Analysis (PCA) in multispectral satellite images. Oliveira and Santos (2009), applied spatial filters of high pass and low pass in DEM in Feira de Santana (state of Bahia, Brazil). Teng *et al.* (2009) extracted the landforms in Shaanxi Province (China) with

In the present study we intend to carry out the mapping of landforms in the catchment of the Ria Formosa located in the south of Portugal (Algarve), using the methodology followed by Jordán *et al.* (2005), modified in order to take in consideration the specific features of this region.

**2.1. Introduction** 

a hierarchical classification.

hillslope units from DEM.

The first case study is used to illustrate the main geo-form classes of the catchment including the Ria Formosa. The Ria Formosa is a shallow coastal lagoon covering an area of about 16,000 ha in the south of Portugal. It is protected by EU and national Laws, and is classified as a Wetland of International Importance under the RAMSAR convention (PORTUGAL Ramsar Site 212). This study aims to establish a method based on the Hammond hierarchical criteria and geographical information related to soft-slopes, local topography and terrain profiles, to locate and classify the geo-forms present in the Ria Formosa catchment.

The second case study focuses on the use of DEM/DTM based on climate models to obtain and analyze isohyetal maps, and to identify how rainfall distribution influences water erosion. Rainfall distribution, which is highly variable in space and time, is difficult to study, due to the lack of good quality data (*e.g.*, insufficient or poorly-distributed gauges in the study areas; non-homogeneous rainfall data series; dubious readings from nonautomated gauges; lack of radar coverage). This issue may be partially addressed by geographic models (*e.g.*, DEM/DTM) and climate data related to rainfall. Parameters such as curvature, slope and orientation of hillslopes, which influence local climate, can also be obtained from these geographic models. Isohyetal maps, through multilinear regression analysis, can then be created using the DEM/DTM and their derivative models, the hillshading potential and some specific data (*e.g.* distance to coastline). From these elements and using the Modified Fournier Index (MFI), rainfall erosivity can finally be quantified. This technique was used to assess the soil erosion risks in Serra de Grândola, which is a north-south oriented mountain ridge with an altitude of 383 m, located in southwest mainland Portugal.

The third case study demonstrates the use of cartographic information to produce flood delimitation maps. The city of Tavira (10,600 inhabitants) in the south of Portugal embraces the outfall of the Séqua/Gilão River into the Atlantic Ocean. A GIS-based hydrologic model of the 221 km2 Séqua/Gilão river catchment was first created to obtain soil use and type regional parameters. Afterwards, and to identify the maximum water heights for those return periods, a hydraulic model of the rivers' last 9.5 km was produced. These maximum water heights were compared with the observed values (flood level marks, photographs and video records) of the 3rd December 1989 flood and used to validate the model. Mean sea level changes due to climate change were also considered. With this procedure, it was finally possible to produce flood delimitation maps for the Tavira urban area. This type of modelling may provide a useful tool for urban planners and city authorities.

#### **2. Location and classification of geo-forms in the Ria Formosa estuary**

#### **2.1. Introduction**

252 Cartography – A Tool for Spatial Analysis

mainland Portugal.

Formosa catchment.

mainland Portugal.

In this chapter three GIS-based models were used, relying on different methodologies and techniques, to illustrate visualization and quantification of the geomorphic processes. This valuable input for decision-makers is attained through the versatility of GIS. Geo-form mapping in a coastal lagoon catchment, rainfall erosivity on a mountain ridge and urban flood delimitation show the potential usefulness of DEM/DTM based cartographic models for helping to solve environmental issues. All case studies presented are from the south of

The first case study is used to illustrate the main geo-form classes of the catchment including the Ria Formosa. The Ria Formosa is a shallow coastal lagoon covering an area of about 16,000 ha in the south of Portugal. It is protected by EU and national Laws, and is classified as a Wetland of International Importance under the RAMSAR convention (PORTUGAL Ramsar Site 212). This study aims to establish a method based on the Hammond hierarchical criteria and geographical information related to soft-slopes, local topography and terrain profiles, to locate and classify the geo-forms present in the Ria

The second case study focuses on the use of DEM/DTM based on climate models to obtain and analyze isohyetal maps, and to identify how rainfall distribution influences water erosion. Rainfall distribution, which is highly variable in space and time, is difficult to study, due to the lack of good quality data (*e.g.*, insufficient or poorly-distributed gauges in the study areas; non-homogeneous rainfall data series; dubious readings from nonautomated gauges; lack of radar coverage). This issue may be partially addressed by geographic models (*e.g.*, DEM/DTM) and climate data related to rainfall. Parameters such as curvature, slope and orientation of hillslopes, which influence local climate, can also be obtained from these geographic models. Isohyetal maps, through multilinear regression analysis, can then be created using the DEM/DTM and their derivative models, the hillshading potential and some specific data (*e.g.* distance to coastline). From these elements and using the Modified Fournier Index (MFI), rainfall erosivity can finally be quantified. This technique was used to assess the soil erosion risks in Serra de Grândola, which is a north-south oriented mountain ridge with an altitude of 383 m, located in southwest

The third case study demonstrates the use of cartographic information to produce flood delimitation maps. The city of Tavira (10,600 inhabitants) in the south of Portugal embraces the outfall of the Séqua/Gilão River into the Atlantic Ocean. A GIS-based hydrologic model of the 221 km2 Séqua/Gilão river catchment was first created to obtain soil use and type regional parameters. Afterwards, and to identify the maximum water heights for those return periods, a hydraulic model of the rivers' last 9.5 km was produced. These maximum water heights were compared with the observed values (flood level marks, photographs and video records) of the 3rd December 1989 flood and used to validate the model. Mean sea level changes due to climate change were also considered. With this procedure, it was finally possible to produce flood delimitation maps for the Tavira urban area. This type of

modelling may provide a useful tool for urban planners and city authorities.

The study of the risk of soil degradation is the starting point for development and sustainable land management. The global warming and the land use changes expected in the XXI Century predict loss of quality and reduction of soil's productivity (Cerdan *et al.*, 2010). In the Mediterranean area the natural and semi-natural vegetation is sclerophyllous, which is well suited for the local climatic conditions, however, extreme weather and human activity can cause imbalances in the ecosystem (Kosmas *et al.*, 2000). The southern Portugal is a region where the balance between the natural environment and human activity is very sensitive to erosion and desertification (Gonçalves *et al.*, 2010). Thus, it is necessary to employ control, prevention and correction measures to preserve the soil and prevent the emergence and intensification of desertification processes, which can become irreversible, as has happened in other Mediterranean areas (Kosmas *et al.*, 2000).

This study emphasizes geomorphologic processes, because they describe the natural space, the dynamics of occupations and the anthropogenic changes. According to Hammond (1954, 1964), geomorphologic studies of the earth's surface can be carried out over large areas (small scale) and based on the analysis of the land features, topographic maps or directly through field measurements. The variables considered should be quantitative, been used for a hierarchical classification.

Later, with the appearance of the GIS, Dikau (1989) and Dikau *et al.* (1991), the Hammond procedures were rectified and automated in some regions in the United States. The DEM becomes the most important tool in the process of identifying landforms. Using the "moving-window" process and through algorithms based on local operators (spatial filters) it is possible to create models (such as slope, local relief and relative position). This allows a more detailed study, with a large number of variables in a large scale. The landforms hierarchy is made in terms of size, order and geometric complexity.

Many researchers have used this methodology, although with some modifications as explained by Martínez-Zavala *et al.* (2004) and Jordan *et al.* (2005), in studies conducted in Spain and Mexico in a detailed scale. Drãgut and Blaschke (2006), and Gerçek *et al.* (2011) classified the landforms in regions of Transylvania (Romania), Germany and Turkey, respectively based in decision rules of fuzzy logic. Another technique widely used was the non-supervised classification based on levels of information. For example, Sallun *et al.* (2007) in the catchment of Alto Rio Paraná in Brazil extracted information by applying the Principal Component Analysis (PCA) in multispectral satellite images. Oliveira and Santos (2009), applied spatial filters of high pass and low pass in DEM in Feira de Santana (state of Bahia, Brazil). Teng *et al.* (2009) extracted the landforms in Shaanxi Province (China) with hillslope units from DEM.

In the present study we intend to carry out the mapping of landforms in the catchment of the Ria Formosa located in the south of Portugal (Algarve), using the methodology followed by Jordán *et al.* (2005), modified in order to take in consideration the specific features of this region.

This study contributes to a better characterization of the region allowing the preparation of regional plans to control the processes of soil degradation, with an indication of possible uses and restrictions.

GIS-Based Models as Tools for Environmental Issues: Applications in the South of Portugal 255

The coefficients *a*, *b*, *c* and *d* are determined at the expense of the coordinates of the square-

The determining of slope of each pixel in line l and column *k* is based on elevation values of neighbouring pixels and the spatial resolution of the model, *E* (distance between pixels). The

From the analysis a smooth slope map was produced, using the "Moving Window" technique with a size of 4,900 m2 (*i.e.*, 7×7 matrix). For each window was determined a percentage of soft slope (considered below 4%), and that value was assigned to the central

> **Code Percentage of soft slope** More than 80% 50% at 80% 20% at 50% Less than 20%

The curvature is defined according to the rate of change of the slope, which is determined

*A* 2 2 *H H C H*

As we are working on a discrete space, it is possible to approximate the Laplacian in two dimensions, using the finite difference method, being the size equal to the unity (one cell):

> 010 141 010

This matrix 3×3 is called the Laplacian filter, which was used in a spatial convolution process on the MDT, wherein each central cell of the window was assigned a curvature value. The negative values indicate concavity (sedimentation basins, valleys, etc.), while

2

Therefore, the same expression in the matrix form, can be written as follows:

2 2

*X Y*

<sup>2</sup> *H Hl k Hl k Hlk Hlk Hlk* ( 1, ) ( 1, ) ( , 1) ( , 1) 4 ( , ) (5)

(4)

(6)

pixel of the window. After this process, the map was reclassified (Table 1).

**Table 1.** Reclassification of the slopes by percentage of soft slopes.

by the partial derivatives of second degree of surface *H*:

2 2 ( , 1) ( , 1) ( 1, ) ( 1, ) 2 2 *Hlk Hlk Hl k Hl k E E*

(3)

defined four vertices.

slope is calculated using Eq. (3).

*H aMP bM cP d* (2)

#### **2.2. Study area**

The Ria Formosa catchment is limited by the WGS84 coordinates 37º 15' N to 36º 57' N and 7º 28' W to 8º 4' W. It has an area of 864 km2 and a perimeter of 166 km, including a shallow coastal lagoon with an area of about 16,000 ha. It is protected by EU and Portuguese Laws, and is classified as a Wetland of International Importance under the RAMSAR convention (PORTUGAL Ramsar Site 212). It covers the municipal areas of Tavira, Faro, Olhão, São Brás, Loulé, Vila Real de Santo António and Castro Marim. The topography of the region is regular and continuous without abrupt changes in altitude. The average slope is 11% and the elevation varies between 0 and 530 meters above sea level. Mean annual rainfall of the catchment ranges between 400 and 800 mm. The mean annual temperature is 17 ° C.

#### **2.3. Methodology**

The DEM has been used as a basic source of information on the catchment of the Ria Formosa and was obtained from a geostatistical study with a resolution of 10×10 m2, which was based on cartography at the scale 1:25,000 from the Geographical Institute of the Portuguese Army (IGeoE, 2004). From this model, other information about other terrain features were obtained. The analysis and mapping of the data has been performed with the IDRISI Taiga software (Eastman, 2009). With this software, several terrain variables such as slope, curvature, relative position, and local relief were modelled for each point relatively to the DEM. Finally, the automatic classification of landforms was carried out, as established by Jordan *et al.* (2005).

#### *2.3.1. Slope and curvature*

Maps of slopes and curvatures are commonly used to describe the hydrologic drainage structure of a region or a catchment. Soil properties and the characteristics of the hillslopes are factors that combined determinate a higher or lower resistance to soil erosion, especially due to rainfall. The inclination, the length and the shape of a slope are associated to the velocity of runoff and to the water infiltration into the soil.

The slope at point (*M*, *P*), in the azimuth direction *A*, is given by calculating the inner product between the gradient of the surface *H* and the unit vector *wA* with the components ( *SinA CosA* ; ):

$$\left\|\nabla H\right\|\left\|\vec{w}\_{A} = \frac{\partial H}{\partial \mathbf{X}} \alpha + \frac{\partial H}{\partial Y} \beta = \left\|\nabla H\right\| = \sqrt{\left(\frac{\partial H}{\partial \mathbf{X}}\right)^{2} + \left(\frac{\partial H}{\partial Y}\right)^{2}}\tag{1}$$

When the ground is represented by a matrix *H* (m, n), *X* and *Y* are coordinate axes, and the gradient of the surface *H* represents the topography of a square that includes point (*M*,*P*) defined by a bi-linear polynomial expression:

GIS-Based Models as Tools for Environmental Issues: Applications in the South of Portugal 255

$$dH = aMP + bM + cP + d\tag{2}$$

The coefficients *a*, *b*, *c* and *d* are determined at the expense of the coordinates of the squaredefined four vertices.

The determining of slope of each pixel in line l and column *k* is based on elevation values of neighbouring pixels and the spatial resolution of the model, *E* (distance between pixels). The slope is calculated using Eq. (3).

$$\delta = \sqrt{\left(\frac{H(l,k+1) - H(l,k-1)}{2E}\right)^2 + \left(\frac{H(l-1,k) - H(l+1,k)}{2E}\right)^2} \tag{3}$$

From the analysis a smooth slope map was produced, using the "Moving Window" technique with a size of 4,900 m2 (*i.e.*, 7×7 matrix). For each window was determined a percentage of soft slope (considered below 4%), and that value was assigned to the central pixel of the window. After this process, the map was reclassified (Table 1).


**Table 1.** Reclassification of the slopes by percentage of soft slopes.

254 Cartography – A Tool for Spatial Analysis

uses and restrictions.

**2.2. Study area** 

**2.3. Methodology** 

*2.3.1. Slope and curvature* 

 *SinA CosA* ; 

):

(

This study contributes to a better characterization of the region allowing the preparation of regional plans to control the processes of soil degradation, with an indication of possible

The Ria Formosa catchment is limited by the WGS84 coordinates 37º 15' N to 36º 57' N and 7º 28' W to 8º 4' W. It has an area of 864 km2 and a perimeter of 166 km, including a shallow coastal lagoon with an area of about 16,000 ha. It is protected by EU and Portuguese Laws, and is classified as a Wetland of International Importance under the RAMSAR convention (PORTUGAL Ramsar Site 212). It covers the municipal areas of Tavira, Faro, Olhão, São Brás, Loulé, Vila Real de Santo António and Castro Marim. The topography of the region is regular and continuous without abrupt changes in altitude. The average slope is 11% and the elevation varies between 0 and 530 meters above sea level. Mean annual rainfall of the

catchment ranges between 400 and 800 mm. The mean annual temperature is 17 ° C.

classification of landforms was carried out, as established by Jordan *et al.* (2005).

velocity of runoff and to the water infiltration into the soil.


defined by a bi-linear polynomial expression:

product between the gradient of the surface *H* and the unit vector *wA*

The DEM has been used as a basic source of information on the catchment of the Ria Formosa and was obtained from a geostatistical study with a resolution of 10×10 m2, which was based on cartography at the scale 1:25,000 from the Geographical Institute of the Portuguese Army (IGeoE, 2004). From this model, other information about other terrain features were obtained. The analysis and mapping of the data has been performed with the IDRISI Taiga software (Eastman, 2009). With this software, several terrain variables such as slope, curvature, relative position, and local relief were modelled for each point relatively to the DEM. Finally, the automatic

Maps of slopes and curvatures are commonly used to describe the hydrologic drainage structure of a region or a catchment. Soil properties and the characteristics of the hillslopes are factors that combined determinate a higher or lower resistance to soil erosion, especially due to rainfall. The inclination, the length and the shape of a slope are associated to the

The slope at point (*M*, *P*), in the azimuth direction *A*, is given by calculating the inner

*HH H H H w <sup>H</sup>*

When the ground is represented by a matrix *H* (m, n), *X* and *Y* are coordinate axes, and the gradient of the surface *H* represents the topography of a square that includes point (*M*,*P*)

 

*XY X Y*

with the components

2 2

(1)

The curvature is defined according to the rate of change of the slope, which is determined by the partial derivatives of second degree of surface *H*:

$$\mathbf{C}\_A = \nabla^2 H = \frac{\partial^2 H}{\partial \mathbf{X}^2} + \frac{\partial^2 H}{\partial \mathbf{Y}^2} \tag{4}$$

As we are working on a discrete space, it is possible to approximate the Laplacian in two dimensions, using the finite difference method, being the size equal to the unity (one cell):

$$
\nabla^2 H = H(l+1,k) + H(l-1,k) + H(l,k+1) + H(l,k-1) - 4H(l,k) \tag{5}
$$

Therefore, the same expression in the matrix form, can be written as follows:

$$
\begin{bmatrix} 0 & 1 & 0 \\ 1 & -4 & 1 \\ 0 & 1 & 0 \end{bmatrix} \tag{6}
$$

This matrix 3×3 is called the Laplacian filter, which was used in a spatial convolution process on the MDT, wherein each central cell of the window was assigned a curvature value. The negative values indicate concavity (sedimentation basins, valleys, etc.), while positive values indicate convexity (massifs, domes, peaks, upper parts of slopes, etc.). The values equal or very close to zero correspond to flat surfaces.

GIS-Based Models as Tools for Environmental Issues: Applications in the South of Portugal 257

**Code Class**

**Table 3.** Classes of the relative position.

**2.4. Results and discussion** 

**Figure 2.** Map of landforms in the Ria Formosa catchment.

carbonated rocks (Monte Figo, Malhão, Guilhim and Nexe).

Figure 2 and Table 4.

*2.3.3. Landforms* 

sub-classes.

**a** > 75% of smooth slopes on concaves hillsides **b** 50 – 75 % of smooth slopes on concaves hillsides **c** 50 – 75 % of smooth slopes on convexes hillsides **d** > 75% of smooth slopes on convexes hillsides

The mapping of landforms was constructed by crossing three levels of information: smooth slope, local relief and relative position. The five main forms considered were: plains, plateaus, plains with hills, and open hills. In turn, these classes were divided into twenty

The main forms and the respective subclasses, for the region of Ria Formosa are shown in

As can be seen, the *localized hills* and *moderate hills* (Table 4) cover most of the catchment area (43%), mainly on areas such as Mountain of Caldeirão, and the mountain formations of

In particular, in the Mountain of Caldeirão, there is a very dense drainage network due to the formation of Flysch from Baixo Alentejo region, consisting of turbidites (greywackes,

After extraction of the curvature, it was necessary to make a spatial convolution in order to filter the unhelpful and inconsistent information and highlight the most important formations. We used a Gaussian filter with 10×10 m2 cells within a 7×7 matrix.

#### *2.3.2. Local relief and relative position*

The local relief can be expressed as the vertical difference between the highest point and the lowest points, of a surface, within a certain horizontal distance or in a determined area of analysis (Figure 1; Left). The relative position sets up the flat shapes of the terrain, in uplands and lowlands, separating the plateaus from the plains with hills or mountains. In this study we considered that all concave and convex areas were, respectively, lowlands and highlands (Figure 1; Right).

**Figure 1.** Left: Local relief; Right: Relative position.

The determination of the local relief was calculated directly from the MDE, through two spatial convolution processes of a (7×7) matrix. The maximum and minimum elevations were determined and replaced in the respective central cells. At the end the results were subtracted. After this process, the map was reclassified as follows (Table 2).


**Table 2.** Classes of the local relief.

To calculate the relative position, the curvature and the smooth slopes (<4%) were integrated. The classification was used to define the classes presented in Table 3.


**Table 3.** Classes of the relative position.

#### *2.3.3. Landforms*

256 Cartography – A Tool for Spatial Analysis

*2.3.2. Local relief and relative position* 

**Figure 1.** Left: Local relief; Right: Relative position.

**Table 2.** Classes of the local relief.

highlands (Figure 1; Right).

positive values indicate convexity (massifs, domes, peaks, upper parts of slopes, etc.). The

After extraction of the curvature, it was necessary to make a spatial convolution in order to filter the unhelpful and inconsistent information and highlight the most important

The local relief can be expressed as the vertical difference between the highest point and the lowest points, of a surface, within a certain horizontal distance or in a determined area of analysis (Figure 1; Left). The relative position sets up the flat shapes of the terrain, in uplands and lowlands, separating the plateaus from the plains with hills or mountains. In this study we considered that all concave and convex areas were, respectively, lowlands and

The determination of the local relief was calculated directly from the MDE, through two spatial convolution processes of a (7×7) matrix. The maximum and minimum elevations were determined and replaced in the respective central cells. At the end the results were

> **Code Class Classification** 0 - 15 m Very smooth 15 - 30 m Smooth 30 - 90 m Localized 90 - 150 m Moderate 150 - 220 m Rough

To calculate the relative position, the curvature and the smooth slopes (<4%) were

integrated. The classification was used to define the classes presented in Table 3.

subtracted. After this process, the map was reclassified as follows (Table 2).

formations. We used a Gaussian filter with 10×10 m2 cells within a 7×7 matrix.

values equal or very close to zero correspond to flat surfaces.

The mapping of landforms was constructed by crossing three levels of information: smooth slope, local relief and relative position. The five main forms considered were: plains, plateaus, plains with hills, and open hills. In turn, these classes were divided into twenty sub-classes.

#### **2.4. Results and discussion**

The main forms and the respective subclasses, for the region of Ria Formosa are shown in Figure 2 and Table 4.

**Figure 2.** Map of landforms in the Ria Formosa catchment.

As can be seen, the *localized hills* and *moderate hills* (Table 4) cover most of the catchment area (43%), mainly on areas such as Mountain of Caldeirão, and the mountain formations of carbonated rocks (Monte Figo, Malhão, Guilhim and Nexe).

In particular, in the Mountain of Caldeirão, there is a very dense drainage network due to the formation of Flysch from Baixo Alentejo region, consisting of turbidites (greywackes, silts and shales) which, because of its layered structure, hamper the infiltration of surface runoff. On the other hand the vegetation is typically mediterranean, composed by quercíneas and sclerophyllous over Eutric Leptosol. Therefore on more rugged slopes, there is little capacity for water storage, making it difficult to sustain the vegetation. Furthermore, poor agricultural practices have been destroying the natural vegetation, causing in the rainy season a deterioration of these soils and turning them into skeletal. So the predictable risk of erosion ranges from moderate to high.

GIS-Based Models as Tools for Environmental Issues: Applications in the South of Portugal 259

by a number of geomorphologic characteristics, which distinguishes them from neighbouring areas. The localized and moderate hills and flat plains are the classes with the highest representation. Allied to information on soil type and vegetation cover, the former, appears to have a moderate to high erosion risk and the latter might have a low to moderate risk. However, in future research, it is intended to create a more accurate map of erosion risks, by matching the satellite images, climatic data, mapping of land use, geological and pedological features in an appropriate scale. Moreover, methodology used in this study for landform mapping, can also be validated by elaborating a descriptive mapping of a sample area, based in photo interpretation and field observations. The aerial photography, with 60% overlap, allows the creation of stereoscopic pairs which facilitate the characterization of the terrain. Fieldwork will also be useful for add and/or confirm the information obtained by stereo restitution. Comparison of the pixels in each unit of land allows validating the model.

Erosion is a global scale threat to sustainability and productive capacity of the soil (*e.g.*, Yang *et al.*, 2003; Feng *et al.*, 2010). It is estimated that about 10 million hectares of farmland

Climate change may have a great influence in soil erosion (Pruski and Nearing, 2002). Changes in the erosive power of rainfall can be hazardous in terms of soil erosion (Favis-Mortlock and Savabi, 1996; Williams *et al.*, 1996; Favis-Mortlock and Guerra, 1999; Pruski and Nearing, 2002). Erosion, the most common type of soil degradation, should be considered as the main symptom of desertification. Since the first half of the XX Century numerous studies have been carried out and gave a strong contribution to the knowledge on the mechanical processes leading to erosion and how these processes interact in the environment. However, studies on how social, economic, political and institutional factors are affected by erosion,

According to the digital Soil Map of the World (FAO, 1989) and a climate database Eswaran *et al.* (2001) the vulnerability to desertification of the Mediterranean area countries, it is considered that more than 600,000 km2 of the Mediterranean basin are at risk of desertification. Project DesertWatch, presented at the 10th Conference of the Parties to the United Nations Convention to Combat Desertification, states that the 33% of the Portuguese

The main objective of this work is the development of a GIS to determine the risk of erosion

The study area is delimited by the UTM coordinates: Zone 29S, Mmin=512,930.44 m, Pmin=4,205,893.13 m, Mmáx=540,965.44 m, Pmáx=4,230,328.13 m. Its area, with 675 km2, includes

territory is at risk of desertification, being the Alentejo the most affected area.

are lost annually in the world due to soil erosion (Yang *et al.*, 2003; Pimentel, 2006).

**3. Mapping of rainfall erosion in Serra de Grândola** 

have been developed only during the last decades.

in Serra de Grândola (Alentejo, Portugal).

**3.2. Study area** 

**3.1. Introduction** 

The flat plains are the second class of the highest representation (25.3%), distributed on the littoral, next to the Ria Formosa and along the leeward coast. These flat surfaces, with less dense dendritic drainage systems, are composed mainly of alluvium, sand dunes and pebbles (slightly cohesive soils and sediments). This system flows in parallel form to the arms of the estuary and may be subject to flooding. These areas are fluvisol, luvisols and cambisols and are fertile for agriculture. They have dense vegetation, with orchards and complex cultural systems providing a low or moderate risk of soil erosion.


**Table 4.** Classification of 20 classes of landforms and their cover area (Ria Formosa).

#### **2.5. Conclusions**

Using an automatic hierarchical method for classification the Ria Formosa drainage basin has been subdivided in twenty landforms. The area included in each class is characterized by a number of geomorphologic characteristics, which distinguishes them from neighbouring areas. The localized and moderate hills and flat plains are the classes with the highest representation. Allied to information on soil type and vegetation cover, the former, appears to have a moderate to high erosion risk and the latter might have a low to moderate risk. However, in future research, it is intended to create a more accurate map of erosion risks, by matching the satellite images, climatic data, mapping of land use, geological and pedological features in an appropriate scale. Moreover, methodology used in this study for landform mapping, can also be validated by elaborating a descriptive mapping of a sample area, based in photo interpretation and field observations. The aerial photography, with 60% overlap, allows the creation of stereoscopic pairs which facilitate the characterization of the terrain. Fieldwork will also be useful for add and/or confirm the information obtained by stereo restitution. Comparison of the pixels in each unit of land allows validating the model.
