**4. Assessing soil erosion rate in a catchment area in Cyprus using remote sensing and GIS techniques**

#### **4.1. Introduction**

The objective of this work was to develop and evaluate two different erosion models in the catchment area of Yialias in Cyprus. The first was an empirical multi-parametric model which is mainly based on expert's knowledge (Analytical Hierarchical Process - AHP) and the second (Revised Universal Soil Loss Equation - RUSLE) was the model which is consid‐ ered to be a contemporary simple and widely used approach of soil loss assessment.

#### **4.2. Methodology**

#### *4.2.1. RUSLE methodology*

The RUSLE equation incorporates five different factors concerning rainfall (R), soil erodibili‐ ty (K), slope length and steepness (L and S. respectively), support practice (P) and cover management (C):

$$\begin{array}{ccccccccc}\text{A} \bullet \text{R} & \text{K} & \text{L} & \text{S} & \text{P} & \text{C} & & & & & & & \text{C} \end{array} \tag{1}$$

seasonal distribution. Therefore the MFI was applied to take into account the monthly rain‐

Integrated Remote Sensing and GIS Applications for Sustainable Watershed Management: A Case Study from Cyprus

<sup>N</sup> N 12 <sup>j</sup> ij

NN P å åå (4)

http://dx.doi.org/10.5772/39307

105

(5)

J=1 J=1 I=1 i Fa 1 P F= =

rainfall total for the same year. After the calculation of R, a continuous surface was pro‐ duced using the ordinary Kriging method based on Gaussian function, which was found to be the most effective for the production of the final iso-erosivity map. The mean values of R range from 267 MJ mm ha year-1 in the most flat areas in Yialias watershed to 694 MJ mm ha

The soil erodibility factor (K) refers to the average long-term soil and soil profile response to the erosive power associated with rainfall and runoff. It is also considered to represent the

A digital soil map of the study area was used and the main soil formations were categorized in three different major classes: coarse sandy loam, sandy loam and silty clay. According to Prasannakumar et al. (2011) the estimated K values for the textural groups vary from 0.07 t ha h ha-1 MJ-1 mm-1 for coarse sandy loam, 0.13 t ha h ha-1 MJ-1 mm-1 for sandy loam and 0.26

The topographic factor is related to the slope steepness factor (S) and slope length factor (L) and is considered to be a crucial factor for the quantification of erosion due to surface run–off.

The combined topograpfic factor was calculated by means of ArcGIS spatial analyst and Hy‐ drotools extension tools. In this study, the equation derived from Moore & Burch (1986) has

The practice factor (P) is defined as the ratio of soil loss after a specific support practice to the cor‐ responding soil loss after up and down cultivation. In order to delineate areas with terracing practices, the two GeoEye-1 satellite images were used and the delineation was accomplished in GIS environment with extensive monitoring of the study area. Areas with no support practice

0.4 1.3 Flow Accumulation CellSize . sin(Slope) LS . 22.13 0.0896 å å å å å å <sup>=</sup> å ÷ å ÷ å ÷ å å å å

is the MFI index, pij is the rainfall depth in month / (mm) of the year j and P is the

fall distribution during each year for a period of 20 years, as follows:

f

rate of soil loss per unit of rainfall erosion index for a specific soil.

year-1 in the mountainous and generally steep areas.

where, Ff

*4.2.1.2. Soil erodibility (K)*

t ha h ha-1 MJ-1 mm-1 for silty clay.

*4.2.1.3. Topographic factor (LS)*

been adoped:

*4.2.1.4. Practice factor (P)*

AHP allows interdependences between decision factors to be taken into account and uses expert opinions as inputs for evaluating decision factors. The final weight of significance for each factor can be defined by using the eigen-vectors of a square reciprocal matrix of pair‐ wise comparisons between the different factors. Moreover, a specific grade is assigned to all the different pairs from 1/9, when the factor is "not important at all", to 9, when the factor is "extremely important".

#### *4.2.1.1. Rainfall (R) factor*

The rainfall factor R is a measure of the erosive force of a specific rainfall value. For the cal‐ culation of the R factor with the use of the Modified Fournier Index (MFI), the following two different approaches suggested by Ferro et al. (1991) and Renard & Freimund (1994) for the areas of Sicily and Morocco were used respectively :

$$\mathbf{R}\_1 = 0.612 \text{ MFI}^{1.56} \tag{2}$$

$$\mathbf{R}\_2 = 0.264 \text{ MFI}^{1.50} \tag{3}$$

According to Kouli et al. (2009), MFI is well correlated with the rainfall erosivity. The specif‐ ic index is considered as an effective estimator of R because it takes into account the rainfall seasonal distribution. Therefore the MFI was applied to take into account the monthly rain‐ fall distribution during each year for a period of 20 years, as follows:

$$\mathbf{F\_{i}} = \sum\_{\substack{\mathbf{J}=1 \\ \mathbf{J}=1}}^{\mathbf{N}} \frac{\mathbf{F} \mathbf{a\_{j}}}{\mathbf{N}} \frac{\mathbf{1}}{\mathbf{N}} \sum\_{\substack{\mathbf{J}=1 \\ \mathbf{I}=1}}^{\mathbf{N}} \frac{\mathbf{P\_{ij}}}{\mathbf{P\_{i}}} \tag{4}$$

where, Ff is the MFI index, pij is the rainfall depth in month / (mm) of the year j and P is the rainfall total for the same year. After the calculation of R, a continuous surface was pro‐ duced using the ordinary Kriging method based on Gaussian function, which was found to be the most effective for the production of the final iso-erosivity map. The mean values of R range from 267 MJ mm ha year-1 in the most flat areas in Yialias watershed to 694 MJ mm ha year-1 in the mountainous and generally steep areas.

#### *4.2.1.2. Soil erodibility (K)*

**4. Assessing soil erosion rate in a catchment area in Cyprus using remote**

The objective of this work was to develop and evaluate two different erosion models in the catchment area of Yialias in Cyprus. The first was an empirical multi-parametric model which is mainly based on expert's knowledge (Analytical Hierarchical Process - AHP) and the second (Revised Universal Soil Loss Equation - RUSLE) was the model which is consid‐

The RUSLE equation incorporates five different factors concerning rainfall (R), soil erodibili‐ ty (K), slope length and steepness (L and S. respectively), support practice (P) and cover

AHP allows interdependences between decision factors to be taken into account and uses expert opinions as inputs for evaluating decision factors. The final weight of significance for each factor can be defined by using the eigen-vectors of a square reciprocal matrix of pair‐ wise comparisons between the different factors. Moreover, a specific grade is assigned to all the different pairs from 1/9, when the factor is "not important at all", to 9, when the factor is

The rainfall factor R is a measure of the erosive force of a specific rainfall value. For the cal‐ culation of the R factor with the use of the Modified Fournier Index (MFI), the following two different approaches suggested by Ferro et al. (1991) and Renard & Freimund (1994) for the

According to Kouli et al. (2009), MFI is well correlated with the rainfall erosivity. The specif‐ ic index is considered as an effective estimator of R because it takes into account the rainfall

A=R K L S P C (1)

1.56 R = 0.612 MFI <sup>1</sup> (2)

1.50 R = 0.264 MFI <sup>2</sup> (3)

ered to be a contemporary simple and widely used approach of soil loss assessment.

**sensing and GIS techniques**

104 Remote Sensing of Environment: Integrated Approaches

**4.1. Introduction**

**4.2. Methodology**

management (C):

"extremely important".

*4.2.1.1. Rainfall (R) factor*

areas of Sicily and Morocco were used respectively :

*4.2.1. RUSLE methodology*

The soil erodibility factor (K) refers to the average long-term soil and soil profile response to the erosive power associated with rainfall and runoff. It is also considered to represent the rate of soil loss per unit of rainfall erosion index for a specific soil.

A digital soil map of the study area was used and the main soil formations were categorized in three different major classes: coarse sandy loam, sandy loam and silty clay. According to Prasannakumar et al. (2011) the estimated K values for the textural groups vary from 0.07 t ha h ha-1 MJ-1 mm-1 for coarse sandy loam, 0.13 t ha h ha-1 MJ-1 mm-1 for sandy loam and 0.26 t ha h ha-1 MJ-1 mm-1 for silty clay.

#### *4.2.1.3. Topographic factor (LS)*

The topographic factor is related to the slope steepness factor (S) and slope length factor (L) and is considered to be a crucial factor for the quantification of erosion due to surface run–off.

The combined topograpfic factor was calculated by means of ArcGIS spatial analyst and Hy‐ drotools extension tools. In this study, the equation derived from Moore & Burch (1986) has been adoped:

$$\text{LIS} = \frac{\text{Flow Acceleration} \cdot \text{CellSize}}{22.13} \pm \frac{^{0.4}}{0.0896} \pm \dots \tag{5}$$

#### *4.2.1.4. Practice factor (P)*

The practice factor (P) is defined as the ratio of soil loss after a specific support practice to the cor‐ responding soil loss after up and down cultivation. In order to delineate areas with terracing practices, the two GeoEye-1 satellite images were used and the delineation was accomplished in GIS environment with extensive monitoring of the study area. Areas with no support practice were assigned with a P factor equal to 1. However, the terrace areas which are considered to be less prone to erosion were assigned a 0.55 value, according to expert's opinion.

#### *4.2.1.5. Cover management factor (C)*

According to Prasannakumar et al. (2011), the C factor represents the effect of soil-disturb‐ ing activities, plants, crop sequence and productivity level, soil cover and subsurface biomass on soil erosion.

The NDVI (Normalised Difference Vegetation Index) extracted from the study area (applied to GeoEye-1 image) has values that range from -0.65 to 0.99. The NDVI is used along with the Equation 6 in order to calculate the *C* factor values of the study area in GIS environment.

$$\text{C=exp} \cdot \text{a} \frac{\text{NDVI}}{\text{(b-NDVI)}} \tag{6}$$

Elevation Model). Next, a buffer zone of 50m was constructed around each drainage net‐

Integrated Remote Sensing and GIS Applications for Sustainable Watershed Management: A Case Study from Cyprus

http://dx.doi.org/10.5772/39307

107

**Figure 6.** Map of the spatial distribution of soil loss after the application of RUSLE methodology in Yialias catchment

According to AHP methodology, a pair-wise comparison of the contribution of each factor was established. Specifically, answers of several experts were collected on the reciprocal ma‐ trix, and the appropriate eigenvector solution method is then employed to calculate the fac‐

The final soil erosion risk map (Fig. 7) was constructed by summing up (through Boolean operators) the product of each category (that had already been rated accordingly for its subcategories) with the corresponding weight of significance according to the following

The final erosion risk assessment map was reclassified to three soil erosion severity classes separated as low (pixel value 1), moderate (pixel value 2) and high risk (pixel value 3). The results denoted that 77.5% of the study area was classified as low potential erosion risk,

Where F1, F2,..., FN are the different factors incorporated in the model.

17.5% as moderate potential risk and only a 5% as high risk.

LS=F1 0.025+F2 0.09+F3 0.146+F4 0.059+F5 0.38+F6 0.3 (7)

work segment.

area

tor weights.

equation:

where, a and b are non-dimensional parameters that determine the shape of the curve relat‐ ing to NDVI and C factor.

According to the final results, C factor values ranged from 0 to 2.7.

#### *4.2.1.6. Application of RUSLE methodology for soil loss estimation*

The annual soil loss was calculated in a GIS environment (Fig. 6), according to Eq. 1. Ac‐ cording to the final results, the estimated soil loss ranges from 0 to 6394 t ha-1 yr-1 with a mean value of 20.95 t ha-1 yr-1. The maximum value of 6394 t ha-1 yr-1 cannot be considered as appreciable due to the fact that only one pixel in a total of 1199 was attributed with this val‐ ue. However, the mean value of 20.95 t ha-1 yr-1 is representative of the current soil loss re‐ gime of the basin.

#### **4.3. AHP methodology**

In the AHP methodology, interdependencies and feedback between the factors were consid‐ ered. The factors used in this methodology were: rainfall (R), soil erodibility (K), slope length and steepness (LS), cover management (C), support practice (P) and stream proximi‐ ty. Six out of seven factors had already been analyzed in the RUSLE methodology. The addi‐ tional agent to be analyzed was the proximity to rivers and streams.

#### *4.3.1. Proximity to rivers and streams*

According to Nekhay et al. (2009), an area of 50 m around rivers and streams was consid‐ ered to be prone to flooding and, consequently, to the detachment of particles of soil by floodwaters. Thus, initially with the use of ArcGIS 10 Hydrotools module, the drainage net‐ work of the basin was automatically extracted from the hydrological corrected DEM (Digital Elevation Model). Next, a buffer zone of 50m was constructed around each drainage net‐ work segment.

were assigned with a P factor equal to 1. However, the terrace areas which are considered to be

According to Prasannakumar et al. (2011), the C factor represents the effect of soil-disturb‐ ing activities, plants, crop sequence and productivity level, soil cover and subsurface bio-

The NDVI (Normalised Difference Vegetation Index) extracted from the study area (applied to GeoEye-1 image) has values that range from -0.65 to 0.99. The NDVI is used along with the Equation 6 in order to calculate the *C* factor values of the study area in GIS environment.

where, a and b are non-dimensional parameters that determine the shape of the curve relat‐

The annual soil loss was calculated in a GIS environment (Fig. 6), according to Eq. 1. Ac‐ cording to the final results, the estimated soil loss ranges from 0 to 6394 t ha-1 yr-1 with a mean value of 20.95 t ha-1 yr-1. The maximum value of 6394 t ha-1 yr-1 cannot be considered as appreciable due to the fact that only one pixel in a total of 1199 was attributed with this val‐ ue. However, the mean value of 20.95 t ha-1 yr-1 is representative of the current soil loss re‐

In the AHP methodology, interdependencies and feedback between the factors were consid‐ ered. The factors used in this methodology were: rainfall (R), soil erodibility (K), slope length and steepness (LS), cover management (C), support practice (P) and stream proximi‐ ty. Six out of seven factors had already been analyzed in the RUSLE methodology. The addi‐

According to Nekhay et al. (2009), an area of 50 m around rivers and streams was consid‐ ered to be prone to flooding and, consequently, to the detachment of particles of soil by floodwaters. Thus, initially with the use of ArcGIS 10 Hydrotools module, the drainage net‐ work of the basin was automatically extracted from the hydrological corrected DEM (Digital

å å (6)

NDVI C=exp -a (b-NDVI) å å å å

According to the final results, C factor values ranged from 0 to 2.7.

tional agent to be analyzed was the proximity to rivers and streams.

*4.2.1.6. Application of RUSLE methodology for soil loss estimation*

less prone to erosion were assigned a 0.55 value, according to expert's opinion.

*4.2.1.5. Cover management factor (C)*

106 Remote Sensing of Environment: Integrated Approaches

mass on soil erosion.

ing to NDVI and C factor.

gime of the basin.

**4.3. AHP methodology**

*4.3.1. Proximity to rivers and streams*

**Figure 6.** Map of the spatial distribution of soil loss after the application of RUSLE methodology in Yialias catchment area

According to AHP methodology, a pair-wise comparison of the contribution of each factor was established. Specifically, answers of several experts were collected on the reciprocal ma‐ trix, and the appropriate eigenvector solution method is then employed to calculate the fac‐ tor weights.

The final soil erosion risk map (Fig. 7) was constructed by summing up (through Boolean operators) the product of each category (that had already been rated accordingly for its subcategories) with the corresponding weight of significance according to the following equation:

$$\text{LS} \mathbf{F1} \quad 0.02 \mathbf{5} \mathbf{+} \mathbf{F2} \quad 0.09 \mathbf{+} \mathbf{F3} \quad 0.14 \mathbf{6} \mathbf{+} \mathbf{F4} \quad 0.05 \mathbf{9} \mathbf{+} \mathbf{F5} \quad 0.38 \mathbf{+} \mathbf{F6} \quad 0.3 \tag{7}$$

Where F1, F2,..., FN are the different factors incorporated in the model.

The final erosion risk assessment map was reclassified to three soil erosion severity classes separated as low (pixel value 1), moderate (pixel value 2) and high risk (pixel value 3). The results denoted that 77.5% of the study area was classified as low potential erosion risk, 17.5% as moderate potential risk and only a 5% as high risk.

*4.3.2. Evaluation of AHP and RUSLE*

the two methodologies (Fig. 8).

**PALSAR radar images**

**5.2. Data and methodology**

*5.2.1. Data and methodology*

**5.1. Introduction**

**4.4. Conclusions**

In the same way that the AHP risk assessment map was reclassified, the estimated soil loss percentage map was separated in 3 different classes according to experts opinion (1st class 0-20 t ha-1 yr-1 for pixel value 1, 2nd class 20-100 t ha-1 yr-1 for pixel value 2, 3rd class 100-6391 t ha-1 yr-1 for pixel value 3). The two grid images were subtracted in GIS environment. A close look at the extracted grid image, it is obvious that there is a considerable similarity between

Integrated Remote Sensing and GIS Applications for Sustainable Watershed Management: A Case Study from Cyprus

http://dx.doi.org/10.5772/39307

109

This research demonstrated the potential for the integration of RS, GIS and precipitation da‐ ta to model soil erosion. The current research found that both RUSLE and AHP methodolo‐ gies can be efficiently applied at a basin scale with quite modest data requirements in a Mediterranean environment such as Cyprus, providing the end users with reliable quantita‐

**5. Flood mapping of Yialias river catchment area in Cyprus using ALOS**

ALOS (Advanced Land Observing Satellite) PALSAR data can be used to detect the water sur‐ face due to the L-band wave length. All SAR instruments share the advantages of day-night operability (as active sensors), cloud penetration, and the ability to calibrate without perform‐ ing atmospheric corrections. The longer L-band (~23.5 cm) SAR wavelength, and, to a certain extent, the C-band (~5.5 cm), have the ability to penetrate vegetation canopies to various de‐ grees depending on vegetation density and height, dielectric constant (primarily a function of water content), and SAR incidence angle. Variations in backscattering allow discrimination among non-vegetated areas (very low to low returns), herbaceous vegetation (low to moder‐ ate returns), and forest (moderate to high returns), and to some degree among different forest structures and regrowth stages. Where water is present beneath a forest canopy, enhanced re‐ turns caused by specular "double bounce" scattering between water surface and tree trunks

The purpose of this study is to explore the potential of ALOS-PALSAR imagery for observing flood inundation phenomena in the Yialias catchment area in Cyprus. Two PALSAR images (polarity: HH, pixel size 50 m) covering the study area before and after an extreme precipita‐ tion incident in 2009 were used (Table 1). A LULC map was also constructed with the use of high resolution images such as GeoEye -1 covering the study area. To analyze Radar backscat‐

tive and spatial information concerning soil loss and erosion risk in general.

makes it possible to distinguish between flooded and non-flooded forest.

**Figure 7.** Final erosion risk map constructed with AHP method

**Figure 8.** Image indicating the soil erosion severity class differences between AHP and RUSLE method

## *4.3.2. Evaluation of AHP and RUSLE*

In the same way that the AHP risk assessment map was reclassified, the estimated soil loss percentage map was separated in 3 different classes according to experts opinion (1st class 0-20 t ha-1 yr-1 for pixel value 1, 2nd class 20-100 t ha-1 yr-1 for pixel value 2, 3rd class 100-6391 t ha-1 yr-1 for pixel value 3). The two grid images were subtracted in GIS environment. A close look at the extracted grid image, it is obvious that there is a considerable similarity between the two methodologies (Fig. 8).

#### **4.4. Conclusions**

This research demonstrated the potential for the integration of RS, GIS and precipitation da‐ ta to model soil erosion. The current research found that both RUSLE and AHP methodolo‐ gies can be efficiently applied at a basin scale with quite modest data requirements in a Mediterranean environment such as Cyprus, providing the end users with reliable quantita‐ tive and spatial information concerning soil loss and erosion risk in general.
