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

## **3.1. Introduction**

258 Cartography – A Tool for Spatial Analysis

erosion ranges from moderate to high.

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

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

**Class Subclass Code Area** 

**Plains** Flat plains 11a 11b 11c 11d 142.93

**Plateaus** Low Plateaus 14c 14d 24c 24d 0.52

**Plains with hills** Plains with low hills 14a 14b 24a 24b 0.63

**Open hills** Very low open hills 31a 31b 31c 31d 3.84

**Hills** Very low hills 41a 41b 41c 41d 2.22

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

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

**2.5. Conclusions** 

Nearly flat plains 12a 12b 12c 12d 65.92 Smooth plains 13a 13b 13c 13d 16.04 Slightly irregular plains 21a 21b 21c 21d 27.51 Irregular plains 22a 22b 22c 22d 52.62 Very irregular plains 23a 23b 23c 23d 23.97

Plateaus 15c 15d 25c 25d 0.02

Plains with hills 15a 15b 25a 25b 0.01

Low open hills 32a 32b 32c 32d 32.33 Localized open hills 33a 33b 33c 33d 37.38 Moderate open hills 34a 34b 34c 34d 3.48 High open hills 35a 35b 35c 35d 0.13

Low hills 42a 42b 42c 42d 46.44 Localized hills 43a 43b 43c 43d 226.67 Moderate hills 44a 44b 44c 44d 128.85 High hills 45a 45b 45c 45d 14.45

**(km2)** 

complex cultural systems providing a low or moderate risk of soil erosion.

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

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, have been developed only during the last decades.

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

The main objective of this work is the development of a GIS to determine the risk of erosion in Serra de Grândola (Alentejo, Portugal).

#### **3.2. Study area**

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 the Grândola, Sines and Santiago do Cacém municipalities, and the mountain Serra de Grândola (maximum altitude: 383 m). Serra de Grândola extends up to the West coast, in a regular and continuous form, without major abrupt changes in topography. Annual rainfall ranges from 600 to 1,200 mm. The Atlantic-influenced climate is moderate, with average annual temperatures of 170 C. Lithologically there are three important groups (1:50,000 Portuguese Geological Map; DGM, 1984): (i) in the highlands, the Flysch formation of the lower Alentejo, (ii) in the highlands surrounding areas, sandstones and gravel of the littoral of Lower Alentejo and Vale do Sado and, (iii) in the coastal zone, the beach and sand dunes. Pedologically there is a predominance of Eutric Lithosols (highlands) and Podzols (coastal zone).

GIS-Based Models as Tools for Environmental Issues: Applications in the South of Portugal 261

LISS III, represents 44 classes of land use with a 150 m of positional accuracy resolution,

The soil protection map was obtained by crossing the land use map with the vegetation coverage map. According to the classification proposed by Zavala (2001) the soil protection was classified into the following classes: 1-Very High, 2- High, 3- Moderate, 4- Low , 5- Very

A DEM was obtained by ordinary kriging of 32,000 points, which were retrieved from a set of 526,770 points obtained from discretization of 10 m equidistant contour lines (military

The hillslopes map was obtained from the DEM (see Eq.(3)). The slopes were classified into

The lithofacies map was created from a 1:50,000 scale Portuguese Geological Map (DGM, 1984). Five lithofacies classes were defined based on the PAP/RAC, (1997) classification.

The erodibility map was obtained by crossing the slope and lithofaceis maps. Accordingly to Zavala (2001) five levels of erodibility were set: 1- Very Low, 2- Low, 3- Moderate, 4- High

The erosive status map was obtained by crossing erodibility and soil protection maps. According to Zavala (2001) five levels of erosive status were set: 1- Very Low, 2- Low, 3- Moderate, 4- High and 5- Very High. Soil erosive states are expressed in terms of protection

> **Classes** 1 2 3 4 5 1 1 1 2 2 1 1 2 3 4 1 2 3 4 4 2 3 3 5 5 2 3 4 5 5 **Unprotect** 3 4 5 5 5

The Modified Fournier Index (*MFI*), an improved version of the Fournier Index (*FI*); Fournier (1960), is used to estimate the rainfall aggressivity for all months as in this area it occurs along the year, since the *FI* is only used in regions characterized by dry seasons. The

> 1 . *<sup>n</sup> <sup>i</sup> i t*

*<sup>P</sup>* (7)

*<sup>P</sup> MFI*

**Protection Erodibility**

with a minimum mapping unit of 25 ha.

ranges: 0-3%, 3-16%, 16-21%, 21-31% and over 31%.

Low 6- Unprotected.

1:25,000 map).

and 5- Very High.

and erodibility (Table 5).

**Table 5.** Classification of soil erosive status.

*3.3.2. Modified Fournier Index model* 

*MFI* is calculated by Eq. (7) (*e.g.*, Arnouldus, 1978).

#### **3.3. Methodology**

Based on the intersection of the soil's erosive status with the rainfall aggressivity, the latter classified according to the Modified Fournier Index (MFI), the risk of erosion at the Serra de Grândola was assessed (Figure 3) using IDRISI Taiga software (Eastman, 2009).

**Figure 3.** Schematic of the methodology used in this application.

#### *3.3.1. Mapping of the erosive status*

The vegetation coverage map was obtained by applying vegetation indexes developed in order to simplify the number of parameters present in the multi-spectral measurements. These indexes, generated from remote sensing data, constitute an important way to include anthropic activity in the ecosystems. Although there are many vegetation index available in the literature, because vegetation has a high reflectance in Near Infra-Red (NIR) and a low reflectance in R (*e.g.*, Lillesand *et al.*, 2004), this study used the Normalized Difference Vegetation Index (NDVI), a technique introduced by Rouse *et al.* (1974) which enables to know the density and the state (greenness) of the vegetation cover.

The land use information was obtained from the CORINE Land Cover map (CORINE-CLC, 2006) at a 1:100,000 scale. The map, based on satellite images SPOT-4, SPOT-5 and IRS-P6 LISS III, represents 44 classes of land use with a 150 m of positional accuracy resolution, with a minimum mapping unit of 25 ha.

The soil protection map was obtained by crossing the land use map with the vegetation coverage map. According to the classification proposed by Zavala (2001) the soil protection was classified into the following classes: 1-Very High, 2- High, 3- Moderate, 4- Low , 5- Very Low 6- Unprotected.

A DEM was obtained by ordinary kriging of 32,000 points, which were retrieved from a set of 526,770 points obtained from discretization of 10 m equidistant contour lines (military 1:25,000 map).

The hillslopes map was obtained from the DEM (see Eq.(3)). The slopes were classified into ranges: 0-3%, 3-16%, 16-21%, 21-31% and over 31%.

The lithofacies map was created from a 1:50,000 scale Portuguese Geological Map (DGM, 1984). Five lithofacies classes were defined based on the PAP/RAC, (1997) classification.

The erodibility map was obtained by crossing the slope and lithofaceis maps. Accordingly to Zavala (2001) five levels of erodibility were set: 1- Very Low, 2- Low, 3- Moderate, 4- High and 5- Very High.

The erosive status map was obtained by crossing erodibility and soil protection maps. According to Zavala (2001) five levels of erosive status were set: 1- Very Low, 2- Low, 3- Moderate, 4- High and 5- Very High. Soil erosive states are expressed in terms of protection and erodibility (Table 5).


**Table 5.** Classification of soil erosive status.

260 Cartography – A Tool for Spatial Analysis

zone).

**3.3. Methodology** 

the Grândola, Sines and Santiago do Cacém municipalities, and the mountain Serra de Grândola (maximum altitude: 383 m). Serra de Grândola extends up to the West coast, in a regular and continuous form, without major abrupt changes in topography. Annual rainfall ranges from 600 to 1,200 mm. The Atlantic-influenced climate is moderate, with average annual temperatures of 170 C. Lithologically there are three important groups (1:50,000 Portuguese Geological Map; DGM, 1984): (i) in the highlands, the Flysch formation of the lower Alentejo, (ii) in the highlands surrounding areas, sandstones and gravel of the littoral of Lower Alentejo and Vale do Sado and, (iii) in the coastal zone, the beach and sand dunes. Pedologically there is a predominance of Eutric Lithosols (highlands) and Podzols (coastal

Based on the intersection of the soil's erosive status with the rainfall aggressivity, the latter classified according to the Modified Fournier Index (MFI), the risk of erosion at the Serra de

The vegetation coverage map was obtained by applying vegetation indexes developed in order to simplify the number of parameters present in the multi-spectral measurements. These indexes, generated from remote sensing data, constitute an important way to include anthropic activity in the ecosystems. Although there are many vegetation index available in the literature, because vegetation has a high reflectance in Near Infra-Red (NIR) and a low reflectance in R (*e.g.*, Lillesand *et al.*, 2004), this study used the Normalized Difference Vegetation Index (NDVI), a technique introduced by Rouse *et al.* (1974) which enables to

The land use information was obtained from the CORINE Land Cover map (CORINE-CLC, 2006) at a 1:100,000 scale. The map, based on satellite images SPOT-4, SPOT-5 and IRS-P6

Grândola was assessed (Figure 3) using IDRISI Taiga software (Eastman, 2009).

**Figure 3.** Schematic of the methodology used in this application.

know the density and the state (greenness) of the vegetation cover.

*3.3.1. Mapping of the erosive status* 

#### *3.3.2. Modified Fournier Index model*

The Modified Fournier Index (*MFI*), an improved version of the Fournier Index (*FI*); Fournier (1960), is used to estimate the rainfall aggressivity for all months as in this area it occurs along the year, since the *FI* is only used in regions characterized by dry seasons. The *MFI* is calculated by Eq. (7) (*e.g.*, Arnouldus, 1978).

$$MFI = \sum\_{i=1}^{n} \frac{P\_i}{P\_t}.\tag{7}$$

where: *Pi* is the monthly rainfall at month *i* (mm) and *Pt* is the annual rainfall (mm).

Monthly rainfall data observed from 01.01.1911 to 31.12.2010 on 30 weather stations were used as an input to the *MFI* model.

Rainfall distribution is highly variable in space and time. In this study the precipitation data are insufficient and poorly distributed. Thus, for creating the MFI model other variables which are correlated with precipitation were used such as hillshading, aspect, distance to coastline, latitude and elevation. The software Statistica 6.0 (Statsoft, 2001) was used to establish a multilinear regression with t critical=1.711 for a significance of 95% ( = 0.05). Aspect model was created based on DEM, by Eq. (8).

$$Aspect = \text{atan2}\left(\frac{dz}{dy}, -\frac{dz}{dx}\right) \tag{8}$$

GIS-Based Models as Tools for Environmental Issues: Applications in the South of Portugal 263

**1 2 3 4 5**

Finally, the erosion risk map (Figure 4) was obtained by relating the erosive status and MFI classified according to CORINE-CEC (1992). According to Zavala (2001), five levels of erosion risk were set: 1- Very Low, 2- Low, 3- Moderate, 4- High and 5- Very High. Erosion

**Erosive status MFI (Corine-CEC)**

**Table 7.** Determination of the levels of erosion risk as a function of the erosive status and the rainfall

**Figure 4.** Digital model of the annual risk of erosion of the Serra de Grândola.

Each erosive status class was characterized according to the slope, soil and vegetation cover

**3.4. Results and discussion** 

*3.4.1. Erosive status map* 

(Table 8).

 1 1 1 2 3 1 2 2 3 4 1 2 3 4 5 2 3 4 4 5 3 4 5 5 5

risk was expressed in terms of erosive status and MFI (Corine-CEC, 1992) (Table 7).

*3.3.3. Erosion risk mapping* 

aggressivity.

where: *dz dy* is the variation of height in latitude and *dz dx* is the variation of height in longitude.

Models of hillshading created refer to solstice and equinox. These models represent the hillshading for a given solar declination and azimuth. These variables were calculated using the equations described by Díez-Herrero *et al.* (2006). Hillshading models were calculated with Eq. (9).

$$HS = 255 \left[ \left( \text{Sim}(\gamma) \text{Cost}(D) + \text{Cost}(\gamma) \text{Sim}(D) \text{Cost}(\phi - A) \right) \right] \tag{9}$$

where: *HS* is the hillshading, (º) is the elevation of the sun above the horizon, *D* (º) is the declination of the sun, (º) is the solar azimuth and *A* (º) is de aspect.

While mapping of the MFI various models were created and the respective outputs analysed such as Cook's distance, consistency, independence and normality. The best model found was that in which the independent variables were latitude, elevation, and distance to coastline, spring hillshading and aspect:

$$R = 1.1708E + 0.0009L + 708.8582SH - 0.1225A - 0.0032DC - 3.432.4703 \tag{10}$$

where: *R* (mm) is rainfall, *E* (º) is elevation, *L* (º) is latitude, *SHS* is spring hillshading, *A* (º) is aspect and *DC* (m) is a distance to coastline.


**Table 6.** *MFI* values according CORINE – CEC (1992).

#### *3.3.3. Erosion risk mapping*

262 Cartography – A Tool for Spatial Analysis

where: *dz*

with Eq. (9).

used as an input to the *MFI* model.

Aspect model was created based on DEM, by Eq. (8).

*dy* is the variation of height in latitude and *dz*

where: *Pi* is the monthly rainfall at month *i* (mm) and *Pt* is the annual rainfall (mm).

Monthly rainfall data observed from 01.01.1911 to 31.12.2010 on 30 weather stations were

Rainfall distribution is highly variable in space and time. In this study the precipitation data are insufficient and poorly distributed. Thus, for creating the MFI model other variables which are correlated with precipitation were used such as hillshading, aspect, distance to coastline, latitude and elevation. The software Statistica 6.0 (Statsoft, 2001) was used to establish a multilinear regression with t critical=1.711 for a significance of 95% ( = 0.05).

> atan2 , *dz dz Aspect dy dx*

Models of hillshading created refer to solstice and equinox. These models represent the hillshading for a given solar declination and azimuth. These variables were calculated using the equations described by Díez-Herrero *et al.* (2006). Hillshading models were calculated

*HS Sin Cos D Cos Sin D Cos A* 255 ( ( ) ( ) ( ) ( )

where: *HS* is the hillshading, (º) is the elevation of the sun above the horizon, *D* (º) is the

While mapping of the MFI various models were created and the respective outputs analysed such as Cook's distance, consistency, independence and normality. The best model found was that in which the independent variables were latitude, elevation, and distance to

where: *R* (mm) is rainfall, *E* (º) is elevation, *L* (º) is latitude, *SHS* is spring hillshading, *A* (º) is

**Classes Range Classification** <60 Very low 60-90 Low 90-120 Moderate 120-160 High >160 Very high

*R E L SHS A DC* 1.1708 0.0009 708.8582 0.1225 0.0032 3,432.4703 (10)

(9)

declination of the sun, (º) is the solar azimuth and *A* (º) is de aspect.

coastline, spring hillshading and aspect:

aspect and *DC* (m) is a distance to coastline.

**Table 6.** *MFI* values according CORINE – CEC (1992).

(8)

*dx* is the variation of height in longitude.

Finally, the erosion risk map (Figure 4) was obtained by relating the erosive status and MFI classified according to CORINE-CEC (1992). According to Zavala (2001), five levels of erosion risk were set: 1- Very Low, 2- Low, 3- Moderate, 4- High and 5- Very High. Erosion risk was expressed in terms of erosive status and MFI (Corine-CEC, 1992) (Table 7).


**Table 7.** Determination of the levels of erosion risk as a function of the erosive status and the rainfall aggressivity.

**Figure 4.** Digital model of the annual risk of erosion of the Serra de Grândola.

#### **3.4. Results and discussion**

#### *3.4.1. Erosive status map*

Each erosive status class was characterized according to the slope, soil and vegetation cover (Table 8).

Table 8 shows that most representative classes have low risk (28%) and moderate risk (44%) of soil erosion, usually on the highlands or at the coastal areas. The coastline consists mainly of cliffs with non-cohesive materials (sand and gravel). In the highlands the material is more resistant (Flysch group from lower Alentejo) but the amount of rainfall is higher and the steep slopes favour soil erosion processes. Soils with material from class D (soil or poorly resistant or deeply altered rocks) and E (soils or sediments that are poorly cohesive or detritus materials) are the most representative, with slopes ranging between 0-3% (slow runoff) and 3-16% (moderate runoff).

GIS-Based Models as Tools for Environmental Issues: Applications in the South of Portugal 265

Environmental biophysics requires knowledge of the resources and processes affecting ecological systems conservation, as well as planning and land management. In completion of the erosion cartography it was possible to develop topographic models (slope, orientation of slopes and distance to the coastline), climate (rainfall, hillshading, Modified Fournier index) and lithofaces. These models represent quantitatively the environmental variables

Moderate erosive status is the most frequent class in the study area. Highlands, where the soil material have moderate resistance (flysch formation) and the precipitation is higher, and the coastline, essentially composed of cliffs with low cohesive material (sand gravel), are the most sensitive areas to erosion. However, using this erosive model it is difficult to justify the risk of water erosion at the coastline. In coastal areas infiltration prevails (sand dunes) and the wind action is the most important factor in the erosion process. Therefore, in future

The largest amount of precipitation is falling in December and January and the lowest in July and August. Through the mapping of the rainfall erosivity, it was found that the

This case study illustrates the use of GIS as a tool to establish hydrologic regional parameters for urban flood mapping purposes. Cartographic elements, hydrological and hydraulic models, and boundary conditions used to establish the maximum flood levels of a 10- and a 100-years return period flood and a 100-years climate change scenario are described. Cartographic information was completed by *in-situ* measurements and

Hydrologic regional parameters are extensively used in flood simulation, since drainage basins are characterized by natural variability in land-surface features (*e.g.*, Wooldridge and Kalma, 2001). Prasad (1997) refers that the improved accuracy of GIS-based hydrologic simulation comes from the capability that these models have to integrate hydrologic regional parameters; updating or modifying GIS data to study the impact of changes in a

This application is focused on the simulation of fluvial-originated urban flooding. The area selected for this study is the town of Tavira. This town is situated in the southernmost region of Portugal – Algarve. The Séqua/Gilão River crosses throughout the Tavira urban area until it flows into the Ria Formosa coastal lagoon. As the Séqua/Gilão River is intrinsically connected with the urban fabric, an overtopping of the margins always has negative consequences to people and assets. An example of a severe flood event was the 3rd

aggressiveness of rain in the coastline and in the highlands is higher in these months.

studies, the model should be capable of including wind erosion.

**4. Flood delimitation mapping of the Tavira urban area** 

drainage basin (*e.g.*, land use) becomes a relatively easy task.

December 1989 flood which caused extensive damage in the city.

**3.5. Conclusions** 

that affect the process of erosion.

**4.1. Introduction and objective** 

observations.

#### *3.4.2. Modified Fournier Index*

All study areas, classified according to CORINE - CEC (1992), present values of *MFI* of low risk (71%) and moderate risk (29%) of erosion, which means a small aggressiveness of rainfall.


**Table 8.** Characterization of the erosive status.
