**4. Methodology**

An attempt has been made to statistically describe the crop canopy factors, namely crop height (CH) and LAI, through the vegetation indices (VI). Crop canopy factors are vital ele‐ ments in the procedure of estimating ETc. These indices were produced from spectroradio‐ metric measurements using a hand-held field spectroradiometer (GER 1500) and after this data were filtered through the Relative Spectral Response (RSR) filters of the corresponding Landsat TM/ETM+ bands. At the same time LAI and CH direct measurements were taken in situ. Hence, time series of LAI, CH and VI have been created and were used to model LAI and CH to VI. After applying the needed regression analysis and evaluating them, the best model for each crop, based on the determination factor (r2 ), was used in specific ETc algo‐ rithms in a procedure to adapt and modify the algorithms in the geo-morphological and me‐ teorological conditions of the island of Cyprus as a representative Mediterranean region.

Crop water requirements were inferred by applying the algorithms and it was tested to check if the specific modifications have assisted the algorithms to improve their precision

Remote Sensing for Determining Evapotranspiration and Irrigation Demand for Annual Crops

http://dx.doi.org/10.5772/39305

33

The overall methodology is described below. The intended purpose is to estimate ETc of

**•** Spectroradiometric measurements were undertaken for two years (2009-10) in order to collect spectral signatures of each crop included in the study. For each crop (Potatoes, Groundnuts, Beans and Chickpeas) the average spectral signature in each phenological stage was created based on the two cultivating periods (2009-2010). The purpose is to have the reflectance of each crop during their phenological stages after the data was fil‐

**•** Leaf Area Index (LAI) and Crop Height (CH) measurements were also taken simultane‐ ously to spectroradiometric measurements and following the same phenological cycle of each crop for the corresponding cultivating periods. The purpose was to create time series

**•** Development of vegetation indices (VI). Time series of VI were created based on the re‐

**•** Modeling VI to LAI and CH. Different models were tested in order to identify the best

**•** Preprocessing of satellite images was applied. Geometric rectification, radiometric correc‐ tion including atmospheric correction of satellite data were applied before main process‐

**•** Mapping LAI, CH and albedo was performed. The three crop canopy parameters were mapped using the ERDAS Imagine v.10 software. The satellite images were transformed into maps in order firstly to test in practice the models and secondly to be inserted as in‐

**•** Models verification. After inferring the best model describing LAI or CH using VI, an evaluation of this procedure was taking place. A priori knowledge of satellite over pass‐ ing over the area of interest has assisted the procedure of taking LAI and CH measure‐ ments in different plots and different cultivating period. These average measurements were compared to the predicted measurements arising from the models application found

**•** Application of ETc algorithms. Original and modified by previous equations, ETc algo‐ rithms have been applied to check, based on the reference method, if and how the models

specific crops in the area of interest using remote sensing techniques.

of these two parameters to correlate them to Vegetation Indices (VI).

possible model which better describes LAI and CH through VI.

have boosted accuracy on estimating ETc for each crop.

tered through the Relative Spectral Response filters.

flectance of each crop, in each phenological stage.

when estimating ETc.

ing of the data.

puts in ETc algorithms.

in the previous step.


**Table 1.** Landsat TM/ETM+ images used in this study (Papadavid, 2012)

Crop water requirements were inferred by applying the algorithms and it was tested to check if the specific modifications have assisted the algorithms to improve their precision when estimating ETc.

Landsat TM/ETM+ bands. At the same time LAI and CH direct measurements were taken in situ. Hence, time series of LAI, CH and VI have been created and were used to model LAI and CH to VI. After applying the needed regression analysis and evaluating them, the best

rithms in a procedure to adapt and modify the algorithms in the geo-morphological and me‐ teorological conditions of the island of Cyprus as a representative Mediterranean region.

), was used in specific ETc algo‐

model for each crop, based on the determination factor (r2

32 Remote Sensing of Environment: Integrated Approaches

**Table 1.** Landsat TM/ETM+ images used in this study (Papadavid, 2012)

The overall methodology is described below. The intended purpose is to estimate ETc of specific crops in the area of interest using remote sensing techniques.


#### **5. Ground data**

#### **5.1. Spectral signatures of crops**

It is well established that the reflectance and transmission spectrum of leaves is a function of both the concentration of light absorbing compounds (chlorophylls, carotenoids, water, cel‐ lulose, lignin, starch, proteins, etc.) and the internal scattering of light that is not absorbed or absorbed less efficiently (Newnham and Burt, 2001; Dangel et al., 2003). Each crop has a dif‐ ferent spectral signature depending on the stage of its phenological cycle (Gouranga and Harsh, 2005; McCloy, 2010; Papadavid et al., 2011). A general view of the vegetation spectral signature is shown in Figure 6; there is strong absorption in blue and red part of the light spectrum while at green and infrared part there is light and strong reflectance, respectively.

nm - 2500 nm) is also a region of strong absorption, primarily by water in green leaves (Mai‐ er, 2000). More specifically, visible blue and red are absorbed by the two main leaf pigments, chlorophyll a and b in green-leaf chloroplasts. These strong absorption bands induce a re‐ flectance peak in the visible green. Thus most vegetation has a green-leafy color. Chloro‐

Remote Sensing for Determining Evapotranspiration and Irrigation Demand for Annual Crops

http://dx.doi.org/10.5772/39305

35

Apart from chlorophyll, other leaf pigments have a significant effect on the visible spectrum. Carotene, a yellow to orange-red pigment strongly absorbs radiation in the 350-500 nm range and is responsible for the color of some flowers and leaves without chlorophyll. Xan‐ thophyll, the red and blue pigment also strongly absorbs radiation in the 350-500 nm range, giving the distinctive color to the leaves in Autumn. In the near infrared range (700-1000 nm) of the electromagnetic spectrum, there is strong reflectance in the spongy mesophyll

Phenology can be defined as the study of the timing of biological events, the causes of their timing with regard to biotic and abiotic forces, and the interrelation among phases of the same or different species (Shaykewich 1994). As McCloy (2010) mentions the phe‐ nological cycle can be defined as the trace or record of the changes in a variable or attrib‐ ute over the phenological period (usually one agricultural year) and a phenophase is defined as an observable stage or phase in the seasonal cycle of a plant that can be de‐ fined by start and end points. Crop phenological stages are important indicators in agri‐ cultural production, management, planning, decision-making and irrigation scheduling (O' Leary et al., 1985; Gouranga and Harsh, 2005; Papadavid et al., 2011). Indeed, Food and Agriculture Organization (FAO) guidelines of estimating crop evapotranspiration for irrigation demands, take into account crop characteristics and the phenological stages of a crop; Crop coefficient (Kc )refers to crop growth stage and the length in time of this stage (Allen et al., 1998). Moreover crop phenology is difficult to be studied for large areas us‐

Recently, many studies have been performed in order to derive the crop phenological stages based on satellite images (Papadavid, 2011). These studies aim to validate vegetation indices for monitoring the development of the phenological cycle from times series data (Papadavid, 2011). For example, Sakamoto et al., (2005), Minaccapili et al., (2008) and Papadavid et al., (2011) used times series of remotely sensed data in order to develop a new systematic method for detecting the phenological stages of different crops from satellite data while Bradley et al., (2007) in their study have introduced a curve fitting procedure in order to derive inter-annual phenologies from time series of noisy satellite NDVI data. Moreover, Funk and Budde (2009) have used an analogous metric of crop performance based on time series of NDVI satellite imagery. Papada‐ vid et al., (2009; 2010; 2011) and Papadavid (2011) have shown that field spectroscopy and em‐ pirical modelling, when successfully integrated, can develop new models of Leaf Area Index

phyll pigments are also known as the green pigments.

cells that occur at the back of leaves.

ing traditional techniques and methods.

(LAI) and Crop Height, during the phenological cycle of crops.

**5.2. Phenology of the crops**

**Figure 6.** Vegetation spectral signature: Vegetation has low reflectance in the visible region and high reflectance in the near infrared (data analysis)

The domain of optical observations extends from 400 nm in the visible region of the electro‐ magnetic spectrum to 2500 nm in the shortwave infrared region. The strong absorption of light by photosynthetic pigments dominates green leaf properties in the visible spectrum (400-700nm). Leaf chlorosis causes an increase in visible reflectance and transmission. The near-infrared region (NIR, 700-1100 nm) is a region where biochemical absorptions are limit‐ ed to the compounds typically found in dry leaves, primarily cellulose, lignin and other structural carbohydrates (Wang et al., 2005). However, foliar reflection in this region is also affected by multiple scattering of photons within the leaf, related to the internal structure, fraction of air spaces, and air-water interfaces that refract light within leaves. The reflectance and transmittance in the middle-infrared also termed the shortwave-infrared (SWIR, 1100 nm - 2500 nm) is also a region of strong absorption, primarily by water in green leaves (Mai‐ er, 2000). More specifically, visible blue and red are absorbed by the two main leaf pigments, chlorophyll a and b in green-leaf chloroplasts. These strong absorption bands induce a re‐ flectance peak in the visible green. Thus most vegetation has a green-leafy color. Chloro‐ phyll pigments are also known as the green pigments.

Apart from chlorophyll, other leaf pigments have a significant effect on the visible spectrum. Carotene, a yellow to orange-red pigment strongly absorbs radiation in the 350-500 nm range and is responsible for the color of some flowers and leaves without chlorophyll. Xan‐ thophyll, the red and blue pigment also strongly absorbs radiation in the 350-500 nm range, giving the distinctive color to the leaves in Autumn. In the near infrared range (700-1000 nm) of the electromagnetic spectrum, there is strong reflectance in the spongy mesophyll cells that occur at the back of leaves.

#### **5.2. Phenology of the crops**

**5. Ground data**

**5.1. Spectral signatures of crops**

34 Remote Sensing of Environment: Integrated Approaches

the near infrared (data analysis)

It is well established that the reflectance and transmission spectrum of leaves is a function of both the concentration of light absorbing compounds (chlorophylls, carotenoids, water, cel‐ lulose, lignin, starch, proteins, etc.) and the internal scattering of light that is not absorbed or absorbed less efficiently (Newnham and Burt, 2001; Dangel et al., 2003). Each crop has a dif‐ ferent spectral signature depending on the stage of its phenological cycle (Gouranga and Harsh, 2005; McCloy, 2010; Papadavid et al., 2011). A general view of the vegetation spectral signature is shown in Figure 6; there is strong absorption in blue and red part of the light spectrum while at green and infrared part there is light and strong reflectance, respectively.

**Figure 6.** Vegetation spectral signature: Vegetation has low reflectance in the visible region and high reflectance in

The domain of optical observations extends from 400 nm in the visible region of the electro‐ magnetic spectrum to 2500 nm in the shortwave infrared region. The strong absorption of light by photosynthetic pigments dominates green leaf properties in the visible spectrum (400-700nm). Leaf chlorosis causes an increase in visible reflectance and transmission. The near-infrared region (NIR, 700-1100 nm) is a region where biochemical absorptions are limit‐ ed to the compounds typically found in dry leaves, primarily cellulose, lignin and other structural carbohydrates (Wang et al., 2005). However, foliar reflection in this region is also affected by multiple scattering of photons within the leaf, related to the internal structure, fraction of air spaces, and air-water interfaces that refract light within leaves. The reflectance and transmittance in the middle-infrared also termed the shortwave-infrared (SWIR, 1100

Phenology can be defined as the study of the timing of biological events, the causes of their timing with regard to biotic and abiotic forces, and the interrelation among phases of the same or different species (Shaykewich 1994). As McCloy (2010) mentions the phe‐ nological cycle can be defined as the trace or record of the changes in a variable or attrib‐ ute over the phenological period (usually one agricultural year) and a phenophase is defined as an observable stage or phase in the seasonal cycle of a plant that can be de‐ fined by start and end points. Crop phenological stages are important indicators in agri‐ cultural production, management, planning, decision-making and irrigation scheduling (O' Leary et al., 1985; Gouranga and Harsh, 2005; Papadavid et al., 2011). Indeed, Food and Agriculture Organization (FAO) guidelines of estimating crop evapotranspiration for irrigation demands, take into account crop characteristics and the phenological stages of a crop; Crop coefficient (Kc )refers to crop growth stage and the length in time of this stage (Allen et al., 1998). Moreover crop phenology is difficult to be studied for large areas us‐ ing traditional techniques and methods.

Recently, many studies have been performed in order to derive the crop phenological stages based on satellite images (Papadavid, 2011). These studies aim to validate vegetation indices for monitoring the development of the phenological cycle from times series data (Papadavid, 2011). For example, Sakamoto et al., (2005), Minaccapili et al., (2008) and Papadavid et al., (2011) used times series of remotely sensed data in order to develop a new systematic method for detecting the phenological stages of different crops from satellite data while Bradley et al., (2007) in their study have introduced a curve fitting procedure in order to derive inter-annual phenologies from time series of noisy satellite NDVI data. Moreover, Funk and Budde (2009) have used an analogous metric of crop performance based on time series of NDVI satellite imagery. Papada‐ vid et al., (2009; 2010; 2011) and Papadavid (2011) have shown that field spectroscopy and em‐ pirical modelling, when successfully integrated, can develop new models of Leaf Area Index (LAI) and Crop Height, during the phenological cycle of crops.

Tables indicating the phenology of each crop can be found in the FAO internet site (www.fao.org). Table 2 indicates the phenological stages of each crop and the number of *in situ* measurements (spectroradiometric and LAI/CH) taken at each stage.

labelled 'Sunscan measurements' the LAI measurements are presented for each phenologi‐ cal stage as for the second column which were taken simultaneously. The number of each spectroradiometric measurement is not random. There should be a change in the reflectance in the specific phenological stage to have another measurement meaning that the crop re‐ flectance during two consecutive days could be the same so the measurement would not en‐

Remote Sensing for Determining Evapotranspiration and Irrigation Demand for Annual Crops

http://dx.doi.org/10.5772/39305

37

Changes in the phenological cycles of crops may occur from different parameters, such as weather conditions, soil and crop characteristics, and changes in the climate of an area (Minaccapili et al., 2008; Kross et al.,2011). Between years, phenological markers (such as length of growing season) may respond differently, a phenomenon which can be associat‐ ed with short-term climate fluctuations or anthropogenic forcing, such as groundwater extraction, urbanization (Bradley et al., 2007). However, the interpretation of phenological changes based on a large dataset volume for a period of many years can turn to be very

**6. Semi-empirical modelling of satellite data (vegetation indices) to**

The commonly accepted equation for estimating evapotranspiration, according to the sche‐ matization of Monteith (Monteith and Unsworth, 1990), is a function of climate data such as temperature (T), humidity (RH%), solar radiation (Rs) and wind speed (U) and crop parame‐ ters, such as the surface albedo (*a*)*,* the leaf area index (LAI) and the crop height (CH):

Remote sensing techniques can be used for monitoring these vegetation characteristics. An analytical elaboration performed on Landsat reflectance values evidenced the possibility of retrieving the surface albedo (Brest and Goward, 1987), the leaf area index (Price, 1992) and the crop height (Moran and Jackson, 1991). Since these parameters directly affect the reflec‐ tance of cropped areas, it has been demonstrated that it is possible to establish a correlation between multispectral measurements of canopy reflectance and the corresponding canopy parameter's values (Bausch and Neale, 1987). In this study, the required crop parameters, *a,* LAI, CH have been derived from direct measurements and were correlated to reflectance

Many studies have illustrated the need and the know-how for modeling or correlating LAI and Crop Height to remote sensing data and mainly to the vegetation indices inferred from handheld sensors. Leaf Area Index is an important structural property of crop canopy. High correlations were found between reflectance factor and LAI by Ahlrichs et al., (1983). Strong correlations between spectral data from crops and various characteristics of crops have been elucidated in numerous studies (Serrano et al, 2000; Goel et al., 2003; Lee et al., 2004). Dar‐

*ETc f a LAI ch T RH Rs U* = ( , , , , %, , ) (1)

ter the table as different one.

**ground data (crop canopy parameters)**

measurements of the crops in each case.

complicated.


**Table 2.** Phenological stages of each crop (Papadavid, 2012)

In each sub-table, the phenological stages of each crop can be seen in the first column. In the 'GER 1500' column the number of spectroradiometric measurements taken at each stage are presented. For example, for Potatoes the measurements begun at the stage of 'closed lines' and there were 3 measurements during that stage (each measurement in the table is the average measurement from 25 measurements well spread in the plot. In the third column labelled 'Sunscan measurements' the LAI measurements are presented for each phenologi‐ cal stage as for the second column which were taken simultaneously. The number of each spectroradiometric measurement is not random. There should be a change in the reflectance in the specific phenological stage to have another measurement meaning that the crop re‐ flectance during two consecutive days could be the same so the measurement would not en‐ ter the table as different one.

Tables indicating the phenology of each crop can be found in the FAO internet site (www.fao.org). Table 2 indicates the phenological stages of each crop and the number of *in*

In each sub-table, the phenological stages of each crop can be seen in the first column. In the 'GER 1500' column the number of spectroradiometric measurements taken at each stage are presented. For example, for Potatoes the measurements begun at the stage of 'closed lines' and there were 3 measurements during that stage (each measurement in the table is the average measurement from 25 measurements well spread in the plot. In the third column

*situ* measurements (spectroradiometric and LAI/CH) taken at each stage.

36 Remote Sensing of Environment: Integrated Approaches

**Table 2.** Phenological stages of each crop (Papadavid, 2012)

Changes in the phenological cycles of crops may occur from different parameters, such as weather conditions, soil and crop characteristics, and changes in the climate of an area (Minaccapili et al., 2008; Kross et al.,2011). Between years, phenological markers (such as length of growing season) may respond differently, a phenomenon which can be associat‐ ed with short-term climate fluctuations or anthropogenic forcing, such as groundwater extraction, urbanization (Bradley et al., 2007). However, the interpretation of phenological changes based on a large dataset volume for a period of many years can turn to be very complicated.
