Mapping crop evapotranspiration by integrating vegetation indices into a soil water balance model

Agricultural Water Management 143 (2014) 71–81

Contents lists available at ScienceDirect

Agricultural Water Management journal homepage: www.elsevier.com/locate/agwat

Mapping crop evapotranspiration by integrating vegetation indices into a soil water balance model S. Consoli ∗ , D. Vanella Dipartimento di Gestione dei Sistemi Agroalimentari e Ambientali (DiGeSA), Università di Catania, Catania, Italy

a r t i c l e

i n f o

Article history: Received 5 March 2014 Received in revised form 19 June 2014 Accepted 27 June 2014 Available online 18 July 2014 Keywords: Crop coefficient Eddy covariance FAO-56 model Soil water monitoring

a b s t r a c t The approach combines the basal crop coefficient (Kcb ) derived from vegetation indices (VIs ) with the daily soil water balance, as proposed in the FAO-56 paper, to estimate daily crop evapotranspiration (ETc ) rates of orange trees. The reliability of the approach to detect water stress was also assessed. VIs were simultaneously retrieved from Worldview 2 imagery and hyper-spectral data collected in the field for comparison. ETc estimated were analysed at the light of independent measurements of the same fluxes by an eddy covariance (EC) system located in the study area. The soil water depletion in the root zone of the crop simulated by the model was also validated by using an in situ soil water monitoring. Average overestimate of daily ETc of 6% was obtained from the proposed approach with respect to EC measurements, evidencing a quite satisfactory agreement between data. The model also detected several periods of light stress for the crop under study, corresponding to an increase of the root zone water deficit matching quite well the in situ soil water monitoring. The overall outcomes of this study showed that the FAO-56 approach with remote sensing-derived basal crop coefficient can have the potential to be applied for estimating crop water requirements and enhancing water management strategies in agricultural contexts. © 2014 Elsevier B.V. All rights reserved.

1. Introduction In arid and semi-arid regions, the availability of water is a major limitation on crop production due to insufficient rainfall to compensate crop water requirements. Improvements in water management in irrigated areas and adequate irrigation scheduling are essential also to increase the sustainability of irrigated agriculture (Hsiao et al., 2007; Padilla et al., 2011). Evapotranspiration (ET), i.e. the water transferred to the atmosphere by soil evaporation and plant transpiration, is one of the most relevant components of the soil water balance. In the Mediterranean agriculture, perennial crops like citrus species are predominant, with about 1 Mha of extension and consequent large water requirements for their sustainability. In Southern Italy, and in Sicily in particular, citrus species represent one of the most relevant components in the agricultural economy, as well as in the utilization of water resources (Capra et al., 2008; Consoli and Vanella, 2014). Therefore, the monitoring of citrus orchards water needs and consumption is a major challenge for developing a regionally sustainable irrigation strategy. In the region, in fact, the

∗ Corresponding author. Tel.: +39 095 7147547; fax: +39 095 7147600. E-mail address: [email protected] (S. Consoli). http://dx.doi.org/10.1016/j.agwat.2014.06.012 0378-3774/© 2014 Elsevier B.V. All rights reserved.

evaporative demand (i.e. around 1200 mm/year according to reference evapotranspiration estimates (Allen et al., 1998)) is very large when compared with rainfall, which mean is about 600 mm/year. Accurate estimation of ET constitutes a very important part of irrigation system planning and designing, and accurate spatial determination is crucial to achieving sustainable agriculture (ErRaki et al., 2007). Several techniques, such as eddy covariance (EC) , Bowen ration (BR), and weighted lysimeters provide ET measurements, but these are expensive, often limited to small experimental field scales (i.e. fetch requirements) and labourious. Numerous studies have evaluated remote sensing techniques for estimating crop ET on a large scale (Barbagallo et al., 2009; González-Dugo and Mateos, 2008; Teixeira et al., 2009; Padilla et al., 2011; Mateos et al., 2013) and several methodologies, integrating thermal and optical remote sensing data into energy and water balance models, have been developed to estimate crop evapotranspiration fluxes (Kustas and Norman, 1999; Allen et al., 2007). The use of remote sensing to estimate ET is presently being developed along two approaches: (i) land surface energy balance (EB) method, that uses remotely sensed surface reflectance in the visible (VIS) and near-infrared (NIR) portion of the electromagnetic spectrum and surface temperature from an infrared thermal band (Idso et al., 1975; Moran, 1989; Hatfield and Pinter, 1993; Norman et al., 1995; Chavez et al., 2005; Allen et al., 2007; González-Dugo

72

S. Consoli, D. Vanella / Agricultural Water Management 143 (2014) 71–81

et al., 2009) and (ii) reflectance-based crop coefficient and reference ET approach where the crop coefficient (Kc ) is related to vegetation indices (VIs ) derived from canopy reflectance values (Rouse et al., 1974; Huete, 1988; Neale et al., 1996; Chavez et al., 2005; Er-Raki et al., 2013; González-Dugo et al., 2013; Mateos et al., 2013). The first approach is based on the rationale that ET is a change of the state of water using available energy in the environment for vapourization. Remote sensing based EB models convert satellite sensed radiances into land surface temperature to estimate ET as residual of the surface energy balance equation. The current limited availability of high-resolution thermal satellite sensors hampers their use in the applied research, and thus evidences the importance of models based on readily available optical data as alternative options for ET estimate. In particular, in this second approach VIS and NIR reflectance measurements are used to compute vegetation indices such as the normalized difference vegetation index (NDVI, Rouse et al., 1974) or the soil adjusted vegetation index (SAVI, Huete, 1988) and these indices can be used to obtain rapid, non-destructive estimates of certain canopy attributes and parameters. One parameter of special interest for water management application is the crop coefficient (Kc ) which employed the FAO-56 model to derive actual crop ET. The crop coefficient is defined as the ratio of the crop ET (ETc ) to the reference crop ET0 (grass) and represents an integration of the effects of the following characteristics distinguishing a given crop from the reference one: crop height (affects aerodynamic resistance and vapour transfer), albedo (affects net radiation), canopy resistance (to vapour transfer), and evaporation from soil (Allen et al., 1998). The Kc can be calculated using a single method that combines the effects of crop transpiration and soil evaporation into a unique coefficient (Kc ) or a dual method that separates the plant transpiration, represented by a basal crop coefficient (Kcb ) and the soil evaporation coefficient (Ke ). The single model is commonly used because of its simplicity, requiring only phenological data and standard meteorological information to determine ET estimates. The dual model is more precise than the single approach and mainly oriented towards applied researches on irrigation scheduling for high-frequency water applications (Padilla et al., 2011). In the relatively recent years progresses have been make on the estimation of the Kc temporal evolution from remote sensing measurements of vegetation indices (VIs ). Some authors have in fact suggested that relationships between crop coefficient and VIs are linear (Neale et al., 1989), but others have found non-linear relationships (Hunsaker et al., 2005). These relationships have been studied for several crops and recently for potato (Jayanthi et al., 2007), cotton and sugarbeet (González-Dugo and Mateos, 2008), wheat (Er-Raki et al., 2007), grapes (Campos et al., 2010; Serrano et al., 2012; Er-Raki et al., 2013), and citrus orchard (Barbagallo et al., 2009; Consoli and Barbagallo, 2012; Consoli and Vanella, 2014). Among the more interesting studies, Neale et al. (1996) obtained reflectance-based Kc data that were related to the SAVI; Tasumi et al. (2005) showed a method to estimate Kc using a satellite-based model and a parameterization of Kc using NDVI to obtain daily ET; Er-Raki et al. (2007) obtained fairly good comparisons between estimates of actual ET for winter wheat obtained using NDVIbased Kcb , and measurements of ET by the eddy covariance (EC) system. In this study we adopted the FAO-56 “dual” crop coefficient approach, where the basal crop coefficient (Kcb ) is derived from vegetation indices obtained from a series of satellite images acquired during May–July 2012, and a daily water balance in the root zone of the crop. The combined methodology allows the calculation of the daily crop coefficient and crop ET of an orange orchard in the semi-arid region of Sicily (South Italy). A validation of the approach was performed using field soil moisture measurements and an eddy covariance system for ET direct measurement.

2. Materials and methods 2.1. Description of the model The FAO-56 dual crop coefficient approach, in the form popularized by the FAO 56 manual (Allen et al., 1998) describes the relationship between daily evapotranspiration of a given crop (ETc ) and reference evapotranspiration (ET0 ) by separating the single crop coefficient (Kc ) into the basal crop coefficient (Kcb ), soil water evaporation (Ke ) coefficient and water stress coefficient (Ks ). Crop transpiration, represented by the basal crop coefficient, Kcb , is separated from soil surface evaporation as follow: ETc = (Kcb Ks + Ke ) ET0

(1) mm d−1 .

where ETc and ET0 are in In the study, daily ETc was calculated by combining the FAO-56 dual crop coefficient model with spectral data provided by remote sensors. The ET0 was estimated using the Penman–Monteith equation with hourly data of solar radiation, wind speed, air temperature and relative humidity supplied by a weather station managed by the Sicilian Agro-meteorological Service (SIAS) and located close to the experimental site. The water stress coefficient, Ks , quantifies the reduction in crop transpiration due to soil water deficit, where Ks = 1 for non-stress conditions and Ks < 1 when there is a shortage of water in the root zone. Ke is the soil evaporation that describes the evaporative component of ETc . The procedure for calculating each coefficient is described as follows. 2.1.1. The basal crop coefficient, Kcb The dual crop coefficient approach calculates the actual increase in crop coefficient Kc for each day as a function of plant development (Kcb ) and the wetness of the soil surface (soil evaporation). The vegetation indices, VIs , are sensitive to leaf area index, LAI, and the crop ground cover fraction, fc (Choudhury et al., 1994), which has been used to estimate Kcb from VI (Bausch and Neale, 1987; Neale et al., 1989). Vegetation indices (VIs ) are transformation of two or more spectral bands designed to assess vegetation condition, land-cover classification, climate and land-use change detection (Glenn et al., 2008). SAVI, the soil adjusted vegetation index (Huete, 1988) is one of the most used indices able to minimize the effect of the soil on vegetation quantification. The relationship between SAVI and the ground-cover fration is in fact approximately linear in the range from bare soil to near full ground cover (Qi et al., 1994) and SAVI is less sensitive than other VIs to soil differences (Qi et al., 1994). In order to integrate the remote sensing data into the dual crop coefficient model, the parameter Kcb in Eq. (1) was derived from SAVI by an equation described by González-Dugo et al. (2009) and Mateos et al. (2013): Kcb =

Kcb, max fc,max

Kcb = Kcb,max

 SAVI − SAVI  min SAVImax − SAVImin if fc ≥ fc,max

if fc < fc,max

(2) (3)

where fc,max is the ground-cover fraction (fc ) at which Kcb is maximal (Kcb, max ); the subscripts max and min refer to values of SAVI for very large leaf area index (LAI, m2 m−2 ) and bare soil, respectively. The values adopted in the model are derived from field measurements and can be found in Table 1. The SAVI index was calculated as follows: SAVI =

(NIR − red ) (L + 1) (NIR + red + L)

(4)

NIR , red are the reflectance in the near-infrared and red spectra, respectively, and L is a soil normalization factor, generally taken to be 0.5 (Huete, 1988).

S. Consoli, D. Vanella / Agricultural Water Management 143 (2014) 71–81 Table 1 Parameters used in the soil water balance based on the FAO-56 methodology.

The water stress coefficient Ks is computed on the basis of the relative root-zone water deficit as (Allen et al., 1998):

Parameter

Value

Maximum crop height Maximum effective root depth (Zr, max ) Minimum effective root depth (Zr, min ) Average soil water content at field capacity ( FC ) Average soil water content at wilting point ( WP ) Total evaporable water (TEW) Readily evaporable water (REW) Fraction of soil exposed and wetted during irrigation (few ) SAVImax SAVImin Maximum basal crop coefficient (Kcb, max ) Ground cover fraction for Kcb, max (fc,max )

3.7 m 1.0 m 0.4 m 0.35 m3 m−3 0.12 m3 m−3 29 mm 10 mm 0.5 0.70 0.04 0.88 0.95

In the study, the fractional vegetation cover, fc , was determined as fc = (NDVI − NDVImin ) / (NDVImax − NDVImin ) (Gutman and Ignatov, 1998), with NDVI derived from WorldView 2 (WV 2) reflectance data, as NDVI = (NIR − red )/(NIR + red ) (Rouse et al., 1974). Mateos et al. (2013), in their study, have proved the validity of Eqs. (2) and (3), firstly developed for annual crops (i.e. cotton and garlic) (González-Dugo et al., 2009), also for crops with ground cover fraction pretty stable, like citrus, olive and peach. In particular, they concluded that the Kcb -SAVI based method is valid and robust for estimating spatially distributed evapotranpsiration in irrigated areas. 2.1.2. The water stress coefficient A daily water balance in the soil root zone was performed following Eq. (5), in order to calculate the water stress coefficient, Ks (Eq. (1)). In particular, the change in the root zone water content, Sw , was calculated as the difference between the water inflows and outflows. Sw = Swf − Swi = P + I − ETc

(5)

where Swf and Swi refer to the root-zone water content at the end and the beginning of the water balance period, respectively; I and P are terms relative to water inputs such as irrigation and precipitation, both during the water balance period; ET is the crop evapotranspiration. Capillary rise from groundwater table, deep percolation and runoff from the soil was assumed insignificant in Eq. (5) that may be expressed in terms of root-zone water deficit, calculated daily: RZWDi = RZWDi−1 + Pi + Ii − ETci

(6)

where the subscript i indicates a given day and RZWDi and RZWDi−1 are the root zone water deficits on day i and i−1, respectively. In particular, the balance equation does not account for water lost by drainage and surface water runoff due to the low rainfall regime occurring during the irrigation season 2012 and the characteristics of the localized micro-irrigation method adopted in the area. The calculation of these coefficients is described by Allen et al. (1998), taking into account the soil physical water retention properties as well as the characteristics of the orchard. In particular, the domain of RZWD is between 0 (i.e. the root zone is full of water), which occur when the soil water content is at field capacity ( FC , m3 m−3 ) and a maximum value (i.e. the root-zone is empty) when the water content reduces plants to the wilting point ( WP , m3 m−3 ). The root-zone total water holding capacity of available water, RZWHC (mm) is the depth of water between these two extremes:





RZWHC = 1000 FC − WP Zr where Zr (m) is the depth of the root system.

73

(7)

Ks =

RZWHC − RZWDi (1 − p) RZWHC

if RZWDi < (1 − p)RZWHC

Ks = 1 if RZWDi > (1 − p)RZWHC

(8) (9)

where p is a depletion coefficient, taking into account the crop tolerance to water stress conditions. In particular, p is the fraction of RZWHC below which transpiration is reduced as a consequence of water deficit. Van Diepen et al. (1988) proposed a simple empirical formulation for adjusting p for different conditions of atmospheric evaporative demand; p was chosen as 0.5 in our study. 2.1.3. The soil evaporation coefficient In general, the soil evaporation coefficient, Ke , reaches a maximal value after rainfall or irrigation events and falls to zero when the soil surface is dry and thus no evaporation occurs. The FAO-56 dual crop coefficient calculations were used to estimate the separate contribution to ETc due to soil evaporation, described by Ke . The soil evaporation coefficient Ke , is obtained by calculating the amount of energy available at the soil surface as: Ke = Kr (Kc,max − Kcb )

(10)

where Kr is a dimensionless evaporation reduction coefficient related to the topsoil water depletion (Allen et al., 1998) and Kc,max is the maximum value of Kc following rainfall or irrigation. Since evaporation is restricted at any moment by the energy available at the exposed soil fraction, the values of Ke cannot be higher than the product few × Kc,max , where few is the fraction of the soil surface not covered by vegetation and wetted by irrigation and precipitation (Allen et al., 1998). The reduction of actual evaporation when the amount of water in the surface soil layer decreases is accounted as: Kr =

TEW − De,i TEW − REW

if De,i > REW

Kr = 1 if De,i ≤ REW

(11) (12)

where De,i (mm) is the cumulative depletion of evaporation from the soil surface at the end of the day i−1. TEW (total evaporable water) is the maximum cumulative depth of evaporation from the soil surface layer; REW (mm) is the readily evaporable water, that is the maximum depth of water that can evaporate from the topsoil layer without restriction (i.e. values of REW were reported for soil type in the FAO-56 paper, Table 19). It is assumed that shortly following a major wetting event, the water content of the evaporation layer is at field capacity,  FC , and the soil can dry to a water content level that is halfway between oven dry and wilting point,  WP . The total evaporable water can be then estimated as (Allen et al., 1998):





TEW = 1000 FC − 0.5WP Ze

(13)

where Ze (equal to 0.1 m) is the depth of the surface soil layer that is subject to drying by way of evaporation; it is generally within the range 0.10–0.15 m (FAO-56). 2.2. Description of the experimental site and meteorological measurements The experimental site is a 20-ha orange orchard, planted with about 20 year-old trees (Citrus sinensis, cv Tarocco Ippolito) (Fig. 1). The field is located in Lentini (Eastern Sicily, Lat. 37◦ 16 N, Long. 14◦ 53 E) in a Mediterranean semi-arid environment, characterized by an annual average precipitation of around 550 mm, very dry summers and average air temperature of 7 ◦ C in winter and 28 ◦ C during the summer. For the period of interest from May 1 to late

74

S. Consoli, D. Vanella / Agricultural Water Management 143 (2014) 71–81

Fig. 1. WV 2 image of the study site with the area of footprint of the eddy covariance station.

July 2012, 15 mm of precipitation and an average air temperature of 28 ◦ C were measured. The site presented conditions of crop homogeneity, flat slope, dominant wind speed and fetch that are ideal for micrometeorological measurements (Fig. 2). The planting layout was 4.0 × 5.5 m2 and the trees were drip irrigated with four on-line drippers per plant with 16 L h−1 of total discharge; the crop was well-watered by irrigation supplied every day during the warm months (from May to October), with irrigation timing of 5 h d−1 . The study area had a mean leaf area index (LAI) of 4 m2 m−2 , measured by a LAI-2000 digital analyser (LI-COR, Lincoln, Nebraska, USA). The instrument was programmed to calculate a mean reading from 18 measurements (two above and 16 below the canopy). LAI data were collected with about 25 replications at selected ground sites and the percentage ground cover was estimated from the tree size relative to tree spacing in the orchards. The area shaded by a tree was estimated as the product of length of foliage within the row and the width of foliage across the row. This was divided by the area per tree within the orchard (distance between rows × distance between trees within a row) to estimate the percentage ground cover. The mean photosynthetical active radiation (PAR) light interception was 80% within rows and 50% between rows; the canopy height (hc ) was 3.7 m. The soil characterization was measured through textural and hydraulic laboratory analyses, according to USDA standards. In this study we used Van Genuchten’s (1980) analytical expression to describe soil water retention and a falling-head permeameter to determine the hydraulic conductivity at saturation. For each soil sample, the moisture content at standard water potential values was determined by a sandbox and a pressure membrane apparatus (Aiello et al., 2014). Soil and crop parameters used in this study for the model application are reported in Table 1. Soil parameters such as the depth of soil surface evaporation layer (Ze ), REW and TEW were adapted from values reported in Allen et al. (1998).

Fig. 2. Eddy covariance tower in the field.

S. Consoli, D. Vanella / Agricultural Water Management 143 (2014) 71–81

75

Table 2 Characteristics of the WorldView 2 Sensor System. Spectral band

Band pass

Centre wavelength

Spectral irradiance [W m−2 ␮m−1 ]

KBand [W m−2 sr−1 count−1 ]

Coastal Blue Green Yellow Red Red edge NIR 1 NIR 2

401.4–453.2 447.5–508.3 511.3–581.1 588.5–627.0 629.2–688.5 703.8–743.6 772.4–890.2 861.7–954.2

427.0 478.3 545.8 607.7 658.8 724.1 832.9 949.3

1758.2229 1974.2416 1856.4104 1738.4791 1559.4555 1342.0695 1069.7302 861.2866

9.29×10−3 1.78×10−2 1.36×10−2 6.81×10−3 1.85×10−2 6.06×10−3 2.05×10−2 9.04×10−3

Three soil water content profiles were realized at the experimental field. Calibrated Campbell Scientific CS616 water content reflectomers (±2.5% of accuracy) were installed to monitor every 1 h the changes of volumetric soil water content (). The Time Domine Reflectometry (TDR) probe installation was designed to measure the variation in soil water content with time in a plane perpendicular to the orchard tree. For each site the TDR equipment consisted of two sensors inserted vertically at 0.25 m and 0.40 m of depth and of two sensors located horizontally at 0.25 m of depth and at 0.20 m of distance between. Soil water content was also determined as the difference between wet and dry weight of soil samples taken at intervals of 5 days throughout the satellite acquisition period. Ten randomly distributed samples were taken during each measurement day. The samples were taken at the same depths of the TRD profiles. The samples were dried in laboratory in a oven at 105 ◦ C for about 2 days to obtain the dry weight. Hyper-spectral data were collected during a field campaign through mid June–early August 2012. A 512-channel spectroradiometer, manufactured by Analytical Spectral Device (FieldSPEC UV/VNIR), with a range from 325 to 1075 nm at 1-nm intervals was used to gather spectral data of crop and soil at six selected ground sites chosen throughout the entire studied area. Sunny days were chosen for field campaigns and data collection time was bracketed between 2 h before and 2 h after solar noon. Measurements were made using a bare tip of the fiber optic cable, which had a 258 fieldof-view (FOV); the optic tip about 1.0 m above the selected targets (crop and soil) resulting in a FOV radius ranging between 0.2 and 0.9 m. Digital photographs provided valuable additional information on each site, for example, in identifying plant density and crop condition rating. Global Positioning System locations were noted in Universal Transverse Mercator projection. The upwelling raw hyper-spectral data (DN) acquired with the field handheld spectroradiometer for each of the selected target of vegetation and soil were converted to reflectance (r) using scans of a Spectralon reference panel (Consoli et al., 2006). Ground-level hyperspectral reflectance measurements were used to observe crop characteristics including LAI and vegetation indicators (NDVI, SAVI); all field spectra were re-sampled to match the wavelengths of the WV 2 data. The FAO-56 hydrological model was applied in a simulation period from May 1st (Day of the Year, DOY 122) to the end of July (DOY 213) 2012. The main hydrological parameters were derived from in-situ observation, as summarized in Table 1. During the monitoring, hourly meteorological data (incoming short-wave solar radiation, air temperature, air humidity, wind speed and rainfall) have been acquired by an automatic weather station located near the orchard (about 7 km far) and managed by SIAS (Servizio Informativo Agrometeorologico Siciliano). The wind data recorded from this station were considered during the tower setting up, with the prevailing wind direction determined to be west. For the dominant wind directions, the fetch was larger than 550 m. For the other sectors the minimum fetch was 400 m (SE) (Fig. 2).

2.3. Remote sensing data acquisition and processing Satellite remote sensing data were provided by the WorldView 2 (WV 2) satellite sensor, manufactured by ITT Space Systems Division for DigitalGlobe, during 2012. The WV 2 sensor provides a 0.5 m panchromatic spatial resolution band and 1.8 m multispectral resolution. The sensor is characterized by eight multispectral bands; four standard colours (red, green, blue, and near-infrared 1) and four new bands (coastal, yellow, red edge, and near-infrared 2). The WV 2 images were collected on May 8th and 12th, June 16th and 19th, and July 2nd and 8th, at 10.30 a.m. The images acquired for the purposes of our study were atmospherically corrected and geometrically rectified to a Universal Transversal Mercator projection system (UTM Zone 33N) by using a linear transformation of coordinates and the Nearest-Neighbour resampling method for pixel reflectance values (Jensen, 1986). The information that each pixel of the WV 2 images contains were converted into spectral radiance at the atmosphere top, LPixel,Band , by using the following expression: LPixel,Band =

KBand qpixel,Band

(14)

Band

with LPixel,Band the spectral radiance at the atmosphere top [W m−2 sr−1 ␮m−1 ], KBand a radiometric calibration factor for each multi-spectral band [W m−2 sr−1 count−1 ], qpixel,Band the radiance of each pixel in the raw image [count], and Band the effective amplitude of every satellite band [␮m]. KBand values were supplied by the producer for each image’s multispectral band (Table 2). The Top of Atmosphere (ToA) reflectance was obtained by applying the following formula: Pixel,Band =

2 · LPixel,Band · dES

 

Esun,Band cos S

(15)

with Pixel,Band , the ToA spectral reflectance for i-th band [adimensional]; LPixel,Band , the ToA spectral radiance [W m−2 sr−1 ␮m−1 ]; dES , the Earth–Sun distance measured in Astronomic Units [AU]; Esun,Band , the solar irradiance for each spectral band [W m−2 ␮m−1 ] showed in Table 2;  S , the zenith angle in radians. The information to derive dES and  S was included in the header file that the producer provides together with satellite images. In order to correctly compare the satellite images, they were corrected for atmospheric influence through the dark-object subtraction technique (Chavez, 1996). A comparison between satellite-derived and ground-based SAVI was performed to evaluate the ability of the satellite imagery to reproduce field measurements. 2.4. Micrometeorological and sap flow measurements used for model validation The hydrological FAO-56 model was validated using field measurements of ET obtained through micrometeorological

76

S. Consoli, D. Vanella / Agricultural Water Management 143 (2014) 71–81

measurements. In particular, the experimental site was equipped with eddy covariance (EC) systems mounted on a micrometeorological fluxes tower. Continuous energy balance measurements were made from the year 2009 at the study site. Net radiation (Rn , W m−2 ) was measured with two CNR 1 Kipp&Zonen (Campbell Scientific Ltd.) net radiometers at a height of 8 m. Soil heat flux density (G, W m−2 ) was measured with three soil heat flux plates (HFP01, Campbell Scientific Ltd.) placed horizontally 0.05 m below the soil surface. Three different measurements of G were selected: in the trunk row (shaded area), at 1/3 of the distance to the adjacent row, and at 2/3 of the distance to the adjacent row. The soil heat flux was measured as the mean output of three soil heat flux plates. Data from the soil heat flux plates were corrected for heat storage in the soil above the plates. The heat storage (S) was quantified in the upper layer by measuring the rate of temperature change. The net storage of energy (S) in the soil column was determined from the temperature profile taken above each soil heat flux plate. Three probes (TCAV, Campbell Scientific Ltd.) were placed in the soil to sample soil temperature. The sensors were placed 0.01–0.04 m (z) below the surface; the volumetric heat capacity of the soil Cv was estimated from the volumetric fractions of minerals (Vm ), organic matter (V0 ) and volumetric water content (). Therefore, G at the surface was estimated by measuring G at a depth of 0.05 m and the change in temperature over time of the soil layer above the heat flux plates to determine S. The air temperature and the three wind speed components were measured at two heights, 4 and 8 m, using fine wire thermocouples (76 ␮m diameter) and

Fig. 3. Surface energy balance closure from eddy covariance measurements during 2012.

sonic anemometers (Windmaster Pro, Gill Instruments Ltd., at 4 m, and a CSAT, Campbell Sci., at 8 m). A gas analyzer (CSAT, Campbell Sci.) operating at 10 Hz was deployed at 8 m. The raw data were recorded at a frequency of 10 Hz using two synchronized data loggers (CR3000, Campbell Sci.). Low frequency measurements were taken for air temperature and humidity (HMP45C, Vaisala), wind speed and direction (05103

Fig. 4. Maps (a–c) of SAVI for the study area during May–July 2012 and comparison (d) between satellite-based estimates of SAVI and NDVI and ground-based measurements of the same VI.

S. Consoli, D. Vanella / Agricultural Water Management 143 (2014) 71–81

77

Fig. 5. Maps (a–c) of evapotranspiration rates (ET) obtained using the FAO-VI based model during May–July 2012 and comparison (d) between daily measured ETc from eddy covariance technique and estimated ETc using a satellite-based Kcb ; the thin solid line represent the 1:1 line.

RM Young), and atmospheric pressure (CS106, Campbell Scientific Ltd.) at 4, 8 and 10 m. Rainfall (AGR 100 Waterra, UK) was measured nearby. The freely distributed TK2 package (Mauder and Foken, 2004) was used to determine the first and second order statistical moments and fluxes on a half-hourly basis following the protocol used as a comparison reference described in Mauder et al. (2007). Surface energy balance measurements at the experimental site showed that the sum of sensible and latent (LE) heat flux was highly correlated (r2 > 0.90) (Fig. 3) to the sum of net radiation and soil heat flux (Castellví et al., 2012; Consoli and Papa, 2012, 2013). A linear fit between the two quantities showed a certain energy balance un-closure for 2012. The percentage of un-closure (about 10%) is in the range reported by most flux sites (Wilson et al., 2002) and provides additional confirmation of the turbulent flux quality (Moncrieff et al., 2004). Measurements of water consumption at tree level (TSF ) were done by using HPV (Heat Pulse Velocity) technique that is based on the measurement of temperature variations (T), produced by a heat pulse of short duration (1–2 s), in two temperature probes installed asymmetrically on either side of a linear heater that is inserted into the trunk. For HPV measurements, two 4 cm sap flow probes with four thermocouples embedded (Tranzflo NZ Ltd., Palmerston North, NZ) were inserted in the trunks of the trees, belonging to the area of footprint of the micrometeorological eddy covariance tower. The probe was positioned at the North and South sides of the trunk at 50 cm from the ground and wired to a data-logger (CR1000, Campbell Sci., USA) for heat-pulse control and measurement; the sampling interval was 30 min. The temperature measurements were obtained by means of ultra-thin thermocouples that are located at 5, 15, 25 and 45 mm within the trunk.

Data of the probe were processed according to Green et al. (2003) to integrate sap flow velocity over sapwood area and calculate transpiration. In particular, the volume of sap flow (Qstem ) in the tree stems per unit time was estimated by multiplying the sap flow velocity by the cross sectional area of conducting tissue. To this purpose, fractions of wood (FM = 0.48) and water (FL = 0.33) in the sapwood were determined on the trees where sap flow probes were installed. Wound-effect correction (Green et al., 2003) was done on a per-tree basis. Scaling up the sap flow from a single tree to the field scale requires analysing plant size variability to determine the mean of those monitored. This was obtained by analysing the spatial variability of plant leaf area (Jara et al., 1998), including the analysis of soil evaporation contribution. Thus, scaling was done only on the basis of the ratio between orchard LAI and tree leaf area (Motisi et al., 2012).

3. Results The WV 2 images, acquired during May–July 2012 for the area under investigation, were used to derive the SAVI index as input to the FAO-56 model (see maps a–c of Fig. 4), as indicated in Eq. (2). The images processing evidenced very low differences on VIs data (i.e. SAVI, NDVI, LAI) between acquisition date differing by a satellite temporal resolution lower that 20 days; thus, in the follow we have included just maps reporting significant evolutions in time of VIs . Satellite-derived SAVI were verified by the comparison with ground-based SAVI estimates obtained during the field campaign in 2012; as shown in the graph of Fig. 4d data which were in a good agreement with a coefficient of determination (R2 ) of 0.84 and a

78

S. Consoli, D. Vanella / Agricultural Water Management 143 (2014) 71–81

RMSE (root mean square error) of 0.02; the same good correlation was found for NDVI data. Daily satellite-based estimates of ETc (maps a–c of Fig. 5) were compared with daily measured ETc using the eddy covariance system (see graph in Fig. 5d). The performance of the FAO-56 model was determined using the RMSE between estimated and measured ET values and the coefficient of determination (R2 ). The value of RMSE (i.e. about 0.11 mm d−1 ) results lower than those reported by other authors presented similar studies; in particular, Er-Raki et al. (2007) and González-Dugo et al. (2009) found differences close to 0.5 mm d−1 . These differences could be explained by the quite different climate of the areas under investigation (i.e. cold and semi-arid regions) and the adopted management practices. The coefficient of determination was about 0.85, slightly higher than the correlations presented by other authors using the same method for woody crops and wine (Campos et al., 2010; Cammalleri et al., 2012). The satellite-based FAO-56 model showed a general tendency to slightly overestimate daily ETc from eddy covariance by about 6%. Data reported in Fig. 5d show a low degree of dispersion, indicating that satellite estimates of SAVI were representative of the area covering the flux tower footprint under conditions of high fractional vegetation cover. Similar good correlations were presented by other author for both woody crops and orchard (Er-Raki et al., 2007; Campos et al., 2010). However, some differences between fluxes may be related to the eddy covariance feature of measuring the evapotranspiration over a large area, whereas the FAO-56 model simulates it locally. Fig. 6 depicts the comparison between the model-estimated root zone water deficit (RZWD, see Eq. (6)) and the measured deficit obtained from the soil sampling campaign and the TDR continuous measurements of soil moisture. In fact, the soil water content measurements carried out at the study site were used to validate the water balance adopted for the calculation of Ke and Ks . The trend in the estimated deficit matched reasonably well the measured data, with a RMSE of 0.5 mm d−1 , confirming the general good behaviour of the FAO-56 hydrological model to estimate the root-zone water deficit under irrigated conditions. However, some slight discrepancies between estimated and measured RZWD might indicate the need of further analysis on the model. The feature of Fig. 6 suggests

6.0 Sap flow measurements of transpiration TSF (mm d-1)

y = 0.92x 2

R = 0.872 RMSE=0.38 mm 4.0

2.0

0.0 0.0

2.0

4.0

Daily estimated Tc FAO-56 (mm d ) Fig. 7. Comparison between daily measured crop transpiration by the HP Sap flow technique TSF and estimated crop transpiration Tc by the FAO-56 model; the thin solid line represent the 1:1 line.

a moderate depletion of water from the root zone layer, always driven by the high frequent irrigation rates application at the study site. Generally, irrigation should be applied when the readily evaporable water (RAW) is depleted, that coincides in this case with a RZWD of about 46 mm. In our case, the maximum water deficit in the root zone reached 18 mm, quite far from the permanent wilting point condition. In addition to the consideration that the FAO-56 dual crop coefficient approach allowed to incorporate remote sensing of Kcb into the method, which is not possible with a “single” crop coefficient method, the “dual” approach allowed to separately estimate the contributions of the soil and the vegetation to ETc . In particular, during the period of study in 2012 (from DOY 122 to DOY 213) a cumulative ETc of 457.5 mm was simulated, including components of soil evaporation and crop transpiration of about 122 and 335 mm, respectively. The comparison between measured crop transpiration rates by the Sap Flow Heat Pulse method and those simulated by the dual FAO-56 crop coefficient method shows a good agreement, with RMSE of 0.38 mm d−1 (Fig. 7). The simulated cumulative transpiration using the FAO-56 method is about 35 mm higher than the measured transpiration by the sap flow heat pulse method. A

20

0

10 12 15 8

Rainfall and Irrigation (mm)

5

16 Root zone water deficit (RZWD, mm)

6.0 -1

20 4

25

0

30 122

127

132

137

142

147

152

157

162

167

172

177

182

187

DOY Irrigation

Rainfall

RZWD from FAO-56

RZWD measured

Fig. 6. Daily measured and estimated root zone water deficit (RZWD) for orange orchard using the FAO-56 approach during the satellite acquisition period.

S. Consoli, D. Vanella / Agricultural Water Management 143 (2014) 71–81

79

Fig. 8. Maps (a–c) of Kc obtained using the FAO-VI based model during May–July 2012 and satellite estimation of basal crop coefficient (Kcb ), stress coefficient (Ks ) and crop coefficient (Kc ) for the orange orchard under study (d).

factor that may partly explain some of the difference between measured and simulated crop transpiration data is the crop stress and the misrepresentation of the rooting depth which directly influence the ability of the plant to extract water. An important step of this study was that related to the monitoring of the crop water stress conditions. In particular, the degree of water stress was analysed by following the development of the estimated Ks coefficient. Ks values lower than the unit indicates that the crop is suffering for water shortage conditions. The decrease of Ks may be attributed to the increase in water depletion at the root zone through a removal of water by transpiration and percolation losses that induced stress condition and diminution of soil moisture at the root zone. Fig. 8 (graph d) reports the stress and the crop coefficient (Kcb + Ke ) for the orange orchard under study during the irrigation season 2012. Several periods of mild stress can be observed in the graph of Fig. 6 for the irrigated orange orchard. The mean value of (Kcb + Ke ) (i.e. Kc under optimal conditions of water availability) was found to be 0.71, very close to the FAO-56 recommended Kcb value at mid-season (see maps a–c of Fig. 8). This may suggest a fairly good tolerance of the studied crop to a certain decrease in the soil water content and consequently the absence of negative impacts on the growth dynamic of the orange orchard under study. 4. Discussion and conclusions The results of our study have evidenced that the FAO-56 procedure integrated with remote sensed data is a predictive method in that it predicts ETc and Kc as a function of weather data, stage of crop development and water availability. Many studies have in fact proved how it is useful for operational applications where dayto-day estimates of Kc area needed (Allen, 2000). Furthermore, the

method may prove to be valuable for filling in between satellite flyovers, especially if it can be refined or calibrated in real time using direct measures of ETc . As outlined is this study, the FAO-56 method requires minimal data for application; weather data area daily measurements, soil and plant parameters can be taken from general tables provided in FAO-56, but ET estimates are improved when specific information on soil water holding capacity and dates for planting, harvest and irrigation is available. Numerous studies have reported comparisons between ETc predicted by the FAO-56 approach and other methods including micrometeorological or remote sensing-based estimates. The brief comment of these studies, herein reported, allowed us to better interpret our findings. In particular, among these studies, Padilla et al. (2011) have evaluated a reliable methodology for combing the basal crop coefficient derived from vegetation indices (VIs ) with a daily soil water balance in the root zone to estimate daily evapotranspiration rates for corn and wheat crops at field scale in Spain. The results obtained from these authors have evidenced a good degree of correspondence between estimated and measured root-zone water deficit conditions for both crops. Er-Raki et al. (2007) have analysed the performance of three methods based on the FAO-56 “dual” crop coefficient approach to estimate actual evapotranspiration for winter wheat under different irrigation treatments in the semi-arid conditions of Morocco. Their outcomes evidenced that the relationships Kc –VIs employed in the FAO-56 “dual” crop coefficient model hold great potential for estimating crop water requirements on an operational basis and consumptions at a regional scale. In 2013, Er-Raki et al. have developed and evaluated a relationship between NDVI and crop coefficient for estimating crop evapotranspiration of table grape vineyards in the semi-arid region of Northwest Mexico. The similarity between seasonal patterns of NDVI and Kc of the

80

S. Consoli, D. Vanella / Agricultural Water Management 143 (2014) 71–81

investigated crop showed potential for establishing robust relationships between both parameters. Cammalleri et al. (2012) have adopted a modified version of the standard FAO-56 dual crop coefficient procedure, in which ET rates are obtained using the Penman–Monteith approach, with canopy characteristics derived from optical remote sensed data in an area covered by olive orchards. The results obtained in their study suggest the capability of the proposed method to correctly distinguish evaporation and transpiration components and to detect water stress conditions. Hunsaker et al. (2005) have also proved that the crop evapotranspiration estimates based on the Kc –VIs relationship were very close to direct measures of ET rate for cotton plots in Arizona. González-Dugo et al. (2013) have evidenced the utility of analysing the temporal trend in SAVI values for a successful identification and classification also of non-permanent crops in large scale basins. Mateos et al. (2013) have demonstrated the ability of the FAO dual-Kc method in combination with series of SAVI from satellite images for estimating the actual ET of different crops (i.e. cotton, garlic, mandarin, olive and peach). Furthermore, as outlined by Allen et al. (1998), Amayreh and Al-Abed (2004) and Er-Raki et al. (2007), the FAO-56 dual crop coefficient approach is well suited for real-time irrigation scheduling with highly frequent water application, as in the case of the drip irrigated orange orchard of our study case. In fact, the performance of the single crop coefficient approach is lower for orchards irrigated by drip technique due to an overestimation of soil evaporation, which is included implicitly in the single crop coefficient. The low Kc values, generally obtained when drip irrigation is practiced in citrus orchards, reflect the effect of practising localized irrigation that reduces soil evaporation. According to these studies, our results of daily ETc , obtained for orange trees with Kc values calculated using field and satellite derived remote vegetation indices, were quite consistent with measurements. The modelled outcomes compared well with both ET measurement system EC and sap flow heat pulse method, showing average overestimates of 6% on daily ET. The FAO-56 model integrated with remote sensed data was also able to detect a soil water deficit in accordance with point measurements of soil moisture. To conclude, we can assess that the employed methodology, integrating the FAO-56 approach with remote sensing-derived basal crop coefficient, can be used to plan irrigation strategies and perform water stress analyses. It could be a useful and low-cost tool for estimating crop water requirements and enhancing water management in those areas afflicted by limited water availability. Acknowledgements The authors would like to recognize the funding provided by the Italian M.I.U.R., Grant E61J12000200001, under the Project of Relevant Interest (PRIN) “I paesaggi tradizionali dell’agricoltura italiana: definizione di un modello interpretativo multidisciplinare e multiscala finalizzato alla pianificazione e alla gestione”, years 2010–2011. References Aiello, R., Bagarello, V., Barbagallo, S., Consoli, S., Di Prima, S., Giordano, G., Iovino, M., 2014. An assessment of the Beerkan method to determine the hydraulic properties of a sandy loam soil (Submitted to Geoderma). Allen, R.G., Pereira, L.S., Raes D., Smith M., 1998. Crop evapotranspiration. Guidelines for computing crop water requirements. FAO Irrigation and Drainage Paper No. 56. Rome, Italy. Allen, R.G., Tasumi, M., Trezza, R., 2007. Satellite-based energy balance for mapping evapotranspiration with internalized calibration (METRIC)-model. J. Irrig. Drain. Eng.—ASCE 133 (4), 380–394. Allen, R.G., 2000. Using the FAO-56 dual crop coefficient method over an irrigated region as part of an evapotranspiration intercomparison study. J. Hydrol. 229, 27–41.

Amayreh, J., Al-Abed, N., 2004. Developing crop coefficients for field-grown tomato (Lycopersicon esculentum Mill.) under drip irrigation with black lastic mulch. Agric. Water Manag. 81, 1–22. Barbagallo, S., Consoli, S., Russo, A., 2009. A one-layer satellite surface energy balance for estimating evapotranspiration rates and crop water stress indexes. Sensors 9, 1–21. Bausch, W.C., Neale, C.M.U., 1987. Crop coefficients derived from reflected canopy radiation: a concept. Trans. ASAE 30, 703–709. Cammalleri, C., Ciraolo, G., Minacapilli, M., Rallo, G., 2012. Evapotranspiration from an olive orchard using remote sensing-based dual crop coefficient approach. Water Resour. Manag. 27, 4877–4895. Campos, I., Neale, C.M.U., Calera, A., Balbontin, C., González-Piqueras, J., 2010. Assessing satellite-based basal crop coefficients for irrigated grapes (Vitis vinifera L.). Agric. Water Manag. 97, 1760–1768. Capra, A., Consoli, S., Russo, A., Scicolone, B., 2008. Integrated agro-economic approach to deficit irrigation on lettuce crops in Sicily (Italy). J. Irrig. Drain. Eng. ASCE 134 (4), 437–445. Castellví, F., Consoli, S., Papa, R., 2012. Sensible heat flux estimates using two different methods based on surface renewal analysis. A study case over an orange orchard in Sicily. Agric. For. Meteorol. 152, 58–64. Chavez, P.S., 1996. Image-based atmospheric corrections—revisited and improved. Photogramm. Eng. Remote Sens. 62, 1025–1036. Chavez, J.L., Neale, C.M.U., Hipps, L.E., Prueger, J.H., Kustas, W.P., 2005. Comparing aircraft-based remotely sensed energy balance fluxes with eddy covariance tower data using heat flux source area functions. J. Hydrometeorol. 6, 923–940. Choudhury, B.J., Ahmed, N.U., Idso, S.B., Reginato, R.J., Daughtry, C.S.T., 1994. Relations between evaporation coefficients and vegetation indices studied by model simulations. Remote Sens. Environ. 50, 1–17. Consoli, S., D’urso, G., Toscano, A., 2006. Remote sensing to estimate ET-fluxes and the performance of an irrigation district in southern Italy. Agric. Water Manag. 81 (3), 295–314. Consoli, S., Barbagallo, S., 2012. Estimating water requirements of an irrigated Mediterranean vineyard using a satellite-based approach. J. Irrig. Drain. Eng., ASCE 138, 896–904. Consoli, S., Papa, R., 2012. Estimates of sensible heat flux of heterogeneous canopy crop using different micrometeorological methods. Ital. J. Agrometeorol. 2, 37–46, ISSN: 2038–5625. Consoli, S., Papa, R., 2013. Corrected surface energy balance to measure and model the evapotranspiration of irrigated orange orchards in semi-arid Mediterranean conditions. Irrig. Sci. 31, 1159–1171. Consoli, S., Vanella, D., 2014. Comparisons of satellite-based models for estimating evapotranspiration fluxes. J. Hydrol. 513, 475–489. Er-Raki, S., Chehbouni, A., Guemouria, N., Duchemin, B., Ezzahar, J., Hadria, R., 2007. Combining FAO-56 model and groundbased remote sensing to estimate water consumption of wheat crops in semi-arid regions. Agric. Water Manag. 87, 41–54. Er-Raki, S., Rodriguez, J.C., Garatuza-Payan, J., Watts, C.J., Chehbouni, A., 2013. Determination of crop evapotranspiration of table grapes in a semi-arid region of Northwest Mexico using multi-spectral vegetation index. Agric. Water Manag. 122, 12–19. Glenn, E., Huete, A., Nagler, P., Nelson, S., 2008. Relationship between remotelysensed vegetation index, canopy attributes and plant physiological processes: what vegetation indices can and cannot tell us about the landscape. Sensors 8 (4), 2136–2160. González-Dugo, M.P., Mateos, L., 2008. Spectral vegetation indices for benchmarking water productivity of irrigated cotton and sugarbeet crops. Agric. Water Manag. 95, 48–58. González-Dugo, M.P., Neale, C.M.U., Mateos, L., Kustas, W.P., Prueger, J.H., Anderson, M.C., Li, F., 2009. A comparison of operational remote sensing-based models for estimating crop evapotranspiration. Agric. For. Meteorol. 149, 1843–1853. González-Dugo, M.P., Escuin, F., Cano, F., Cifuentes, V., Padilla, F.L.M., Tirado, J.L., Oyonarte, N., Fernández, P., Mateos, L., 2013. Monitoring evapotranspiration of irrigated crops using crop coefficients derived from time series of satellite images. II. Application on basin scale. Agric. Water Manag. 125, 92–104. Green, S., Clothier, B., Jardine, B., 2003. Theory and practical application of heat pulse to measure sap flow. Agron. J. 95, 1371–1379. Gutman, G., Ignatov, A., 1998. The derivation of the green vegetation fraction from NOAA/AVHRR data for use in numerical weather prediction models. Int. J. Remote Sens. 19, 1533–1543. Hatfield, J.L., Pinter Jr., P.J., 1993. Remote sensing for crop protection. Crop Prot. 12, 403–414. Hsiao, T.C., Stetuto, P., Fereres, E., 2007. A systematic and quantitative approach to improve water use efficiency in agriculture. Irrig. Sci. 25, 209–231. Huete, A.R., 1988. A soil-adjusted vegetation index (SAVI). Remote Sens. Environ. 25, 295–309. Hunsaker, D.J., Pinter, P.R., Kimball, B.A., 2005. Wheat basal crop coefficients determined by normalized difference vegetation index. Irrig. Sci. 24, 1–14. Idso, S.H., Schmuggr, T.J., Jackson, H.I., Reginato, K.J., 1975. The utility of surface temperature measurements for the remote sensing of surface soil water status. J. Geophys. Res. 80, 3044–3049. Jayanthi, H., Neale, C.M.U., Wright, J.L., 2007. Development and validation of canopy reflectance-based crop coefficient for potato. Agric. Water Manag. 88 (1–3), 235–246. Jara, J., Stockle, C.O., Kjelgaard, J., 1998. Measurement of evapotranspiration and its components in a corn (Zea mays L.) field. Agric. For. Meteorol. 92, 131–145.

S. Consoli, D. Vanella / Agricultural Water Management 143 (2014) 71–81 Jensen, J.R., 1986. In: Cliff, N.J. (Ed.), Introductory Digital Image Processing: A Remote Sensing Perspective. Prentice-Hall, Englewood, NJ, U.S.A. Kustas, W.P., Norman, J.M., 1999. Evaluation of soil and vegetation heat flux predictions using a simple two-source model with radiometric temperatures for partial canopy cover. Agric. For. Meteorol. 94, 13–29. Mateos, L., González-Dugo, M.P., Testi, L., Villalobos, F.J., 2013. Monitoring evapotranspiration of irrigated crops using crop coefficients derived from time series of satellite images. I. Method validation. Agric. Water Manag. 125, 81–91. Mauder, M., Foken, T., 2004. Documentation and Instruction Manual of the Eddy Covariance Software Package TK2. Universität Bayreuth, Abt. Mikrometeorologie, Arbeitsergebnisse, pp. 26–44. Mauder, M., Oncley, S.P., Vogt, R., Weidinger, T., Ribeiro, L., Bernhofer, C., Foken, T., Kosiek, W., De Bruin, H.A.R., Liu, H., 2007. The energy balance experiment EBEX-2000. Part II. Intercomparison of eddy-covariance sensors and post-field data processing methods. Bound.-Layer Meteorol. 123, 29–54, http://dx.doi.org/10.1007/s10546-006-9139-4. Moncrieff, J., Clement, R., Finnigan, J., Meyers, T., 2004. Averaging, detrending, and filtering of eddy covariance time series. In: Xuhui, Lee, William, Massman, Beverly, Law (Eds.), Handbook of Micrometeorology. Kluwer Ac. Publ., The Netherlands, pp. 7–31. Moran, M.S., 1989. A window-based technique for combining Landsat Thematic Mapper thermal data with higher-resolution multi-spectral data over agricultural lands. Photogramm. Eng. Remote Sens. 56, 337–342. Motisi, A., Consoli, S., Rossi, F., Minacapilli, M., Cammalleri, C., Papa, R., Rallo, G., D’urso, G., 2012. Eddy covariance and sap flow measurement of energy and mass exchange of woody crops in a Mediterranean environment. Acta Horticol. 951, 121–127. Neale, C.M.U., Bausch, W.C., Heerman, D.F., 1989. Development of reflectance-based crop coefficients for corn. Trans. ASAE 32 (4), 1891–1899. Neale, C.M.U., Almed, R.H., Moran, M.S., Pinter, P.J., Qi, J., Clarke, T.R., 1996. Estimating seasonal cotton evapotranspiration using canopy reflectance. In: Proceedings of the International Conference on Evapotranspiration and Irrigation Scheduling, American Society of Agricultural Engineers, San Antonio, USA, pp. 173–181.

81

Norman, J.M., Kustas, W.P., Humes, K.S., 1995. A two-source approach for estimating soil and vegetation energy fluxes from observations of directional radiometric surface temperature. Agric. For. Meteorol. 80, 87–109. Padilla, F.L.M., Gonz´ıalez-Dugo, M.P., Gavilan, P., Domìnguez, J., 2011. Integration of vegetation indices into a water balance model to estimate evapotranspiration of wheat and corn. Hydrol. Earth Syst. Sci. 15, 1213–1225. Qi, J., Chehbouni, A., Huete, A.R., Kerr, Y.H., Sorooshian, S., 1994. A modified soil adjusted vegetation index. Remote Sens. Environ. 48, 119–126. Rouse, J.W., Haas, R.H., Schell, J.A., Deering, D.W., Harlan, J.C., 1974. Monitoring the Vernal Advancement and Retrogradation of Natural Vegetation. NASA/GSFC, Type III. Final Report. Greenbelt, MD, pp. 1–371. Serrano, L., Gonzàlez-Flor, C., Gorchs, G., 2012. Assessment of grape yield and composition using the reflectance based Water Index in Mediterranean rainfed vineyards. Remote Sens. Environ. 118, 249–258. Tasumi, M., Trezza, R., Allen, R.G., Wright, J.L., 2005. Operational aspects of satellitebased energy balance models for irrigated crops in the semi-arid U.S. Irrig. Drain. Syst. 19, 335–376. Teixeira, A.H.D.C., Bastiaanssen, W.G.M., Ahmad, M.D., Bos, M.G., 2009. Reviewing SEBAL input parameters for assessing evapotranspiration and water productivity for the Low-Middle S˜ao Francisco River basin, Brazil: Part B: Application to the regional scale. Agric. For. Meteorol. 149, 477–490. Van Diepen, C.A., Rappoldt, C., Wolf, J., Van Keulen, H., 1988. Crop Growth Simulation Model WOFOST. Doc v4.1. Centre for World Food Studies, Wageningen, The Netherlands. Van Genuchten, M.T., 1980. A closed form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Sci. Soc. Am. J. 44, 892–898. Wilson, K., Goldstein, A., Falge, E., Aubinet, M., Baldocchi, D., Berbigier, P., Bernhofer, C., Ceulemans, R., Dolman, H., Field, C., Grelle, A., Ibrom, A., Law, B.E., Kowalski, A., Meyers, T., Moncrieff, J., Monson, R., Oechel, W., Tenhunen, J., Valentini, R., Verma, S., 2002. Energy balance closure at FLUXNET sites. Agric. For. Meteorol. 113, 223–243.