Eddy Covariance fluxes versus satellite-based modelisation in a deficit irrigated orchard

Page created by Grace Jenkins
 
CONTINUE READING
Eddy Covariance fluxes versus satellite-based
              modelisation in a deficit irrigated orchard
              Daniela Vanella1*, Simona Consoli1

              Abstract: This study describes a combined approach to determine actual evapotranspiration (ETa) fluxes in a deficit
              irrigated orchard in Sicily, Italy. Different ET models mostly based on satellite data (i.e., surface energy balance
              models: one source ET1S - two source ET2S, and vegetation indexes derived crop coefficient model, ETVI) were compared
              with both direct ET measurements obtained in situ by the Eddy Covariance (EC) technique (ETEC) and crop ET (ETc)
              calculated by FAO-56 approach. Results of the comparisons show a general overestimation of the latent heat flux (or
              ET) by the satellite models and estimated ETc, respect to the direct ETEC data. This discrepancy can be the consequence
              of the deficit irrigation regimes (i.e. ranging from a minimum reduction of about 15% to a maximum of 45% of ETc)
              applied at the experimental orange orchard, which may have caused a high source of sensible heat flux, driven by the
              temperature gradient between plant canopy-atmosphere continuum.
              Keywords: Evapotranspiration, eddy covariance, remote sensing, deficit irrigation.

              Riassunto: Lo studio descrive un approccio combinato per determinare i flussi effettivi di evapotraspirazione (ETa) in
              un frutteto irrigato in condizioni di deficit idrico in Sicilia, Italia. Diversi modelli di stima di ET basati principalmente
              su dati satellitari (modelli di bilancio energetico superficiale: mono-dimensionale ET1S e bi-dimensionale ET2S e
              un modello basato sulla determinazione del coefficiente colturale mediante indici di vegetazione, ETVI) sono stati
              confrontati con le misurazioni dirette di ET ottenute in situ mediante la tecnica Eddy Covariance (EC). I risultati dei
              confronti mostrano una sovrastima generale del flusso di calore latente (o ET) da parte dei modelli satellitari rispetto
              alle misure dirette da Eddy Covariance. Questa discrepanza può essere la conseguenza dei regimi di irrigazione
              deficitaria (che impongono riduzione idriche dal 15% al 45% di ETc) applicati alle colture in studio, che può aver
              causato incrementi di calore sensibile e un elevato gradiente di temperatura nel continuum vegetazione-atmosfera.
              Parole chiave: Evapotranspirazione, eddy covariance, telerilevamento, irrigazione deficitaria.

              1. INTRODUCTION                                                   citrus orchards), together with the adoption of micro-
              Water deficit is a major limiting factor of crop                  irrigation methods, can allow a significant saving of
              productivity especially in semi-area climates (Singh et           water resources, while optimizing production, avoiding
              al., 2017). At global scale, irrigation sector is responsible     waste (Consoli et al., 2017).
              of 70% of water demand to sustain 40% of the food                 Applied researches have confirmed by an accurate
              production (FAO, 2003). Nowadays, the competition                 monitoring of the continuum soil-plant-atmosphere
              for water among the different users, and the greater              (SPAC) system, that the WUE deficit irrigation (DI)
              water resources depletion due to climate changes                  strategies are followed is greatly increased when (Capra
              conditions require the application of water optimization          et al., 2008; Abrisqueta et al., 2015; Consoli et al., 2017;
              measures (Wada et al., 2013), among which the                     Ruiz-Sánchez et al., 2018). In particular, among the

                                                                                                                                                 Italian Journal of Agrometeorology - 2/2018
              water saving measures in irrigation agriculture may               different fluxes of mass and energy exchanges within

                                                                                                                                                                                         Rivista Italiana di Agrometeorologia - 2/2018
              be included. These strategies offer opportunities for             the SPAC system, evapotranspiration (ET) is the main
              increasing the water use efficiency (WUE) in the                  one, which accounts for the water demand of the plant
              agriculture sector, reserving water amounts for other             system (Swank and Douglass, 1974).
              priority uses, including those related to the ecosystem           ET measures and/or estimates can be obtained at
              service needs, thus contributing to the sustainability of         different spatial and temporal scales with many levels
              the primary sector (Capra et al., 2013; Power, 2010).             of accuracy. As examples, direct ET measurements
              In this view, the application of planned water deficit            can be determined by using Eddy Covariance (EC)
              conditions (i.e. RDI –Regulated Deficit Irrigation,               technique, Bowen Ratio (BR), weighing lysimeters,
              PRD – Partial Root-Zone Drying) to specific crops                 or scintillometer systems. Indirectly, ET estimates
              of the Mediterranean environment (i.e. including                  may be retrieved from changes in soil water, via water
                                                                                balance or through the surface energy budget, using
                                                                                respectively the conservation of mass and energy
              * Corresponding author’s e-mail: d.vanella@unict.it               laws. In most practical situations, where the available
              1 Dipartimento di Agricoltura, Alimentazione e Ambiente (Di3A),

              Università degli Studi di Catania, Catania, Italy.                instrumentation or resources are not sufficient to direct
              Submitted 15 April 2018, accepted 23 May 2018                     measure ET or when a wide spatial scale estimation of
              DOI:10.19199/2018.2.2038-5625.041                                                                                                                        41

AGRO 2-18_IIIA BOZZA.indb 41                                                                                                                   18/07/18 16.22
ET area required (i.e. regional), the adoption of remote      (RS), i.e. surface energy balance (SEB) models such
                                                                                          modelling is generally pursued. In particular, at wider       as one source (ET1S) (e.g. Barbagallo et al., 2009), two
                                                                                          scale of application than that of the single plant, in last   sources (ET2S) (e.g. Li et al., 2005); and VI based models
                                                                                          three decades numerous methods based on remote                as the dual Kc method (ETVI) (e.g. Gonzalez-Dugo, et
                                                                                          sensing (RS) data acquisition (e.g., empirical, residual,     al., 2009). Daily estimates of ET by RS approaches and
                                                                                          deterministic models and inference methods) have              measured ETEC were compared with crop ET (ETc)
                                                                                          been developed for deriving spatially distributed ET          obtained by the FAO-56 model.
                                                                                          estimates. Tab. 1 reports some of these methods with
                                                                                          their strengths and limitations (Courault et al., 2005).      2.1. Field site description
                                                                                          All these methods need several inputs that sometimes          The study was carried out in an orange orchard (Citrus
                                                                                          are not easily obtained at the appropriate space/             sinensis (L.) Osbeck) of around 1 ha, located in Eastern
                                                                                          temporal scale, and therefore increasing the difficulty       Sicily, Italy (latitude 37° 20ʹ N, longitude 14° 53ʹ E).
                                                                                          for their applicability. Among the Vegetation Index (VI)      Mature orange trees (Tarocco Sciara C1882 grafted on
                                                                                          based methods, the dual crop coefficient (Kc) approach        Citrange Carrizo), with spacing of 6 x 4 m, were irrigated
                                                                                          proposed by FAO (Allen et al., 1998) is extensively           with different strategies: (i) full irrigation (control, T1)
                                                                                          applied to derive ET in irrigated agriculture conditions      to replace 100% of ETc using a surface drip irrigation
                                                                                          (Gonzalez-Dugo et al., 2009; Consoli and Vanella,             system; (ii) sustained deficit irrigation treatment (SDI,
                                                                                          2014a, b). This approach is often preferred due to its        T2) irrigated to replace 75% of ETc using a sub-surface
                                                                                          simplicity and robustness for operational applications.       drip irrigation system; (iii) a regulated deficit irrigation
                                                                                          It requires fewer input data, and provides acceptable         treatment (RDI, T3) irrigated to replace 50% of ETc
                                                                                          ET estimates when compared to heavily parameterized           in those plant vegetative phases less sensitive to water
                                                                                          physically based models (Er-Raki et al., 2010).               stress conditions (Pérez-Pérez et al., (2008); (iv) partial
                                                                                          The motivation of this study was to evaluate how robust       root-zone drying treatment (PRD, T4) alternatively the
                                                                                          a satellite-based estimation of ET and estimates of ETc       irrigated side of the root-zone at 50% of ETc, while the
                                                                                          could be, for the specific conditions of the experimental     other side was kept dry (Consoli et al., 2014; Consoli et
                                                                                          site, when compared with ET data from Eddy                    al., 2017). The study was performed during the irrigation
                                                                                          Covariance (EC) during deficit irrigation conditions.         seasons (June-September) of the years 2016 and 2017.
                                                                                                                                                        The climate of the experimental site is typical semi-arid
                                                                                          2. MATERIALS AND METHODS                                      Mediterranean, with hot and dry summers. Maximum
                                                                                          This section focuses on the description of the materials      air temperature during the irrigation seasons reached
                                                                                          and methodologies applied to determine ET at the              about 30°C, with mean values of VPD and RH of 0.62
                                                                                          experimental field site by using EC technique (ETEC)          kPa and 69.8%, respectively; and mean value of rain
                                                                                          and by combining different ET satellite-based models          of 68 mm.

                                                                                                                Methods                                   Strengths                         Limitations
                                                                                                             RS data are introduced
                                                                                                             directly in semi-empirical
                                                                                          Empirical                                            Applicable from local             Need of ground ancillary data;
                                                                                                             models to estimate ET
                                                                                          direct                                               to regional scales                spatial variation of parameters
  Italian Journal of Agrometeorology - 2/2018

                                                                                                             (e.g., relationship using RS
                                                                                                             and meteorological data)
                                          Rivista Italiana di Agrometeorologia - 2/2018

                                                                                                             Integration of empirical
                                                                                                             relationships and physical
                                                                                                                                               Applicable if combined            Require detection of wet
                                                                                          Residual           models, using remotely
                                                                                                                                               with ground measurement           and dry pixels
                                                                                                             sensed data (e.g. surface
                                                                                                             energy balance models, SEB)
                                                                                          Vegetation         Use of surface reflectance
                                                                                                                                                                                 Require calibration
                                                                                          Index (VI)         data to compute reduction         Applicable if combined
                                                                                                                                                                                 for each crop type, Kc varies
                                                                                          or inference       factors (crop coefficient, Kc)    with ground measurement
                                                                                                                                                                                 according to water stress
                                                                                          method             for ET estimation
                                                                                                             Based on more complex models      Permit the estimation
                                                                                                             such as Soil–Vegetation–          of intermediate variable
                                                                                                                                                                                 Need of ground ancillary data;
                                                                                          Deterministic      Atmosphere Transfer (SVAT),       such as leaf area index (LAI);
                                                                                                                                                                                 require accurate RS data
                                                                                                             which computes the different      possible links with climate
                                                                                                             components of energy budget       and/or hydrological models
                                                                                          Tab. 1 - Remote sensing-based methods for ET estimation: strengths and limitations.
                                                                                          Tab. 1 - Metodi basati su dati satellitari per la stima di ET: vantaggi e limiti.

                        42

AGRO 2-18_IIIA BOZZA.indb 42                                                                                                                                                                                           18/07/18 16.22
ECH2O probe (Decagon, Inc.) located at different
              depths (0.15-0.40 m) of the soil profile were used
              to monitor soil moisture variations at the different
              treatments.

              2.2. ET estimates by the FAO-56 approach
              An automatic meteorological station at the
              experimental field was used to monitor hourly the
              weather conditions (i.e. solar radiation, Rs, W m-2, air
              temperature, Ta, °C, relative humidity, RH, %, wind
              speed, u, ms-1 and direction and rainfall, P, mm) during     Fig. 1 - Footprint of the EC surface energy fluxes at the
              the experimental period. Climatic data were used to          experimental site.
              calculate the reference evapotranspiration (ET0, mm          Fig. 1 - Impronta della torre Eddy Covariance press il sito
              d-1) rates, using the Penman-Monteith equation (Allen        in studio.
              et al., 1998; Allen et al., 2006).
              ETc was obtained by multiplying daily ET0, by the            surface flux layer and the surface energy balan-
              seasonal Kc for orange orchard. The seasonal Kc was          ce closure were also evaluated (Kaimal and
              assumed of 0.7 (Consoli et al., 2006, Consoli and            Finningan, 1994).
              Papa, 2013). Because tree size was smaller than fully        Eddy Covariance sensible heat flux (H, W m-2) was
              developed orange trees, a reduction coefficient (Kr)         computed as:
              was applied (Fereres et al., 1981). Kr was assumed
              equal to 0.66.                                                H        cp     wT                                       (1)
              Irrigation rates were determined by ETc and adju-
              sted according to rainfall. Irrigation was supplied          where ρ (g m-3) is the air density, cp (1004 J g-1 K-1) is the
              to the orchard early in the morning, 3 times per             air specific heat capacity at constant pressure and σwT
              week.                                                        (m ET
                                                                               s-1 K) is thewqcovariance between the vertical wind
                                                                           speed and air temperature.
                                                                             H        c p wT
              2.3. Eddy Covariance ET fluxes                               The vertical      flux of water vapour content, i.e.
              In May 2016, EC system (Fig. 1) was installed at the         the latent heat flux (lET, W m-2), was expressed as:
                                                                                    H ET
              experimental site on a micrometeorological tower at 7 m       CR
                                                                                    Rn G
              above the ground (about two times the canopy height).         HET      cp    wq
                                                                                           wT
                                                                                                                                     (2)
              The EC system was equipped by a three dimensional
              sonic anemometer (CSAT3-3D, Campbell Scientific              where λ (J g-1) is the latent heat of vaporization and σwq
              Inc.) and an infrared open-path gas analyzer (Li-7500,       (g m−2 s−1)) is the covariance between the vertical wind
              Li-cor Biosciences Inc.) to obtain high frequency               ET and
                                                                           speed   H waterETwq vapour density.
                                                                             CR
              measurements of the three wind components and the                     R n Genergy balance closure ratio (CR) is
                                                                           The surface
              H2O and CO2 concentrations, respectively. The sample         expressed as:
              frequency for the raw data was 10 Hz (high frequency          LE = R n − G − H
                                                                                    H ET

                                                                                                                                              Italian Journal of Agrometeorology - 2/2018
              data). Low frequency data (30-min) were obtained for:         CR                                                       (3)
                                                                                    Rn G

                                                                                                                                                                                      Rivista Italiana di Agrometeorologia - 2/2018
              net radiation (Rn , W m-2,net radiometer CNR-1 Kipp
              & Zonen, located 7 m above the ground) and soil heat           H allows
                                                                           and         c p forwTthe determination of how well the
              flux (G, W m-2) obtained by self-calibrated soil heat flux
                                                                           turbulent fluxes of heat and water vapour account for
              plates (HFP01SC, Hukseflux) placed in the exposed,             LEavailable
                                                                                  = R n − Genergy.
                                                                                             − H The ratio, as suggested by Prueger
                                                                           the
              half-exposed and shadowed soil, at a depth of about
              0.05 m.                                                          ET (2005), was
                                                                           et al.,           wq
                                                                                                  performed only when Rn is greater
                                                                           than 100 W m-2.
              High and low frequency data were recorded and stored
                                                                           Thirty-minute fluxes data were aggregated to a daily
              in a CR1000 logger (Campbell Scientific Inc.).                 LE =and
                                                                           scale    R n latent
                                                                                        −G − H  heat fluxes, acquired in W m-2, were
              The standard EUROFLUX rules (Aubinet et al.,
                                                                                    H ET to equivalent depth of ET (mm d-1).
                                                                           then transformed
              2000) were adopted for EC measurements and                     CR
                                                                           Using the R nETG EC measures of actual evapotran-spiration
              data processing. Common errors in the measu-                 by the eddy covariance (ETa, EC), the Kc was estimated
              red high frequency data, such as running means               as in the follows:
              for detrending, three angles coordinate rotations
              and despiking, were removed during the post                                                                            (4)
              processing by quality checks. The stationary of the
                                                                                                                                                                    43
                                                                            LE = R n − G − H

AGRO 2-18_IIIA BOZZA.indb 43                                                                                                                18/07/18 16.22
2.4. Remote sensing approaches                                    DOY                      Acquisition date
                                                                                          The RS approaches adopted in this study were                      180                        28/06/2016
                                                                                          implemented by using Landsat 8 Operational Land                   203                        21/07/2016
                                                                                          Imager (OLI) and Thermal Infrared Sensor (TIRS)                   219                        06/08/2016
                                                                                          images (https://landsat.usgs.gov/landsat-8) (Tab. 2).             235                        22/08/2016
                                                                                          Tab. 3 reports the day-of-year (DOY) and the                      251                        07/09/2016
                                                                                          acquisition dates of the 10 selected images (based                150                        30/05/2017
                                                                                          on clear sky condition) within the monitoring period
                                                                                                                                                            166                        15/06/2017
                                                                                          (irrigation seasons 2016-17).
                                                                                                                                                            182                        01/07/2017
                                                                                          Row data were re-calibrated in radiance for visible
                                                                                                                                                            189                        17/07/2017
                                                                                          (VIS), near-infrared (NIR) and thermal infrared
                                                                                          (TIR) bands, on the basis of spectral specifications              214                        02/08/2017
                                                                                          and the gain settings (values given in metadata of each        Tab. 3 - Dates of the images selected during the irrigation
                                                                                          image). Reflectance in VIS-NIR bands was calculated            seasons 2016-17.
                                                                                                                                                         Tab. 3 - Date di acquisizione delle immagini satellitari utiliz-
                                                                                          by the declination angle correction (USGS, 2016).              zate nel corso della stagione irrigua 2016-17.
                                                                                          Vegetation indexes (VIs, i.e. Normalized Difference
                                                                                            H
                                                                                          Vegetation  c p Index
                                                                                                             wT
                                                                                                                 – NDVI; Soil Adjusted Vegetation                                            (1)
                                                                                                                                                         These methods estimate surface resistances     (surface
                                                                                          Index – SAVI; Leaf Area Index - LAI), Fraction cover           and aerodynamic resistances to heat or water vapor,
                                                                                          (Fc) and albedo (a) were assessed starting from VIS-           rah and rs) adopting various schemes (Zhang et al,
                                                                                          NIR reflectance bands (Consoli and Vanella 2014b).             2016).
                                                                                              ET
                                                                                          Radiometric       surface
                                                                                                             wq     temperature (Ts) was calculated                                          (2)
                                                                                                                                                         In the study, two SEB models were applied to estimate
                                                                                          from band 10 by using the model developed by Sobrino           ET: the one-source (ET1S) and the two-sources (ET2S)
                                                                                          et al., in (2004).                                             models. A full overview of the used algorithms is
                                                                                                   H ET                                                  described in Consoli and Vanella, (2014b). The main
                                                                                            CR Surface energy balance models
                                                                                          2.4.1                                                          difference between both SEB models  (3) consists in the
                                                                                                   R n GET
                                                                                          for estimating
                                                                                                                                                         radiometric surface temperature (Ts) schematization.
                                                                                          Surface energy balance (SEB) models combine the
                                                                                                                                                         In particular, the one-source (or one-dimensional)
                                                                                          use of satellite information (surface reflectance in VIS/
                                                                                          NIR and Ts) with ancillary data (weather data) to derive       model refers Ts to a single layer which include soil
                                                                                          surface energy fluxes and instantaneous ET, the latter         and vegetation together, while in   (4) the two-source
                                                                                          as a residual term of the energy balance equation:             model different Ts components were referred to soil
                                                                                                                                                         and vegetation separately. Figg. 2 and 3 report in flow
                                                                                           LE = R n − G − H                                       (5)    charts the main inputs (satellite and meteorological)
                                                                                                                                                         required for the application of the(5)
                                                                                                                                                                                              one (1S) and two-
                                                                                          where: LE is latent heat flux (W m2), Rn is net radiation      sources (2S) models.
                                                                                          (W m2), G is soil heat flux (W m2), H is sensible heat         Both the approaches (i.e. 1S and 2S) provided
                                                                                          flux (W m2).                                                   instantaneous values of the LE (or ET)
                                                                                                                                                                                             (6) corresponding
  Italian Journal of Agrometeorology - 2/2018

                                                                                          Bands                                                         Wavelength (μm)                       Resolution (m)
                                          Rivista Italiana di Agrometeorologia - 2/2018

                                                                                          Band 1 - Coastal aerosol                                        0.43 - 0.45                               30
                                                                                          Band 2 – Blue                                                   0.45 - 0.51                            (7)30
                                                                                          Band 3 – Green                                                  0.53 - 0.59                               30
                                                                                          Band 4 – Red                                                    0.64 - 0.67                               30
                                                                                          Band 5 - Near Infrared (NIR)                                    0.85 - 0.88                               30
                                                                                          Band 6 - SWIR 1                                                 1.57 - 1.65                               30
                                                                                          Band 7 - SWIR 2                                                 2.11 - 2.29                            (8)30
                                                                                          Band 8 – Panchromatic                                           0.50 - 0.68                               15
                                                                                          Band 9 – Cirrus                                                 1.36 - 1.38                               30
                                                                                          Band 10 - Thermal Infrared (TIR) 1                              10.60 - 11.19                         100 * (30)
                                                                                          Band 11 - Thermal Infrared (TIR) 2                              11.50 - 12.51                         100 * (30)
                                                                                          *TIR bands are acquired at 100 m resolution, but are resampled to 30 m in delivered data product.
                                                                                          Tab. 2 - Technical specifications of Landsat 8 (OLI and TIRS).
                                                                                          Tab. 2 - Specifiche tecniche delle immagini satellitari Landsat 8 (OLI and TIRS).

                        44

AGRO 2-18_IIIA BOZZA.indb 44                                                                                                                                                                                                18/07/18 16.22
Fig. 2 - Lay-out
                                                                                                       of the main input/output
                H         cp   wT                                                                               (1)
                                                                                                       parameters
                                                                                                       in the 1S model.
                                                                                                       Fig. 2 - Schema
                                                                                                       dei principali parametri
                                                                          H        cp   wT             di input/output utilizzati
                  ET           wq                                                                               (2)
                                                                                                       nel modello mono-
                                                                                                       dimensionale.
                                                                            ET          wq
                        H ET
                CR                                                                                            (3)
                        Rn G

                                                                                 H ET
                                                                          CR
                                                                                 Rn G
                                                                                                              (4)

              toLE     Rn − G
                 the= time       H satellite overpass (about 9:30 a.m.
                            of −the                                      meteorological and soil hydraulic characteristics
                                                                                                              (5)          (Fig. 4,
              local standard time). Daily ET values were computed,       Tab. 4). The FAO-56 dual crop coefficient approach, in
              as follows:                                                the form popularized by the FAO 56 manual (Allen et
                                                                           LE = R n − G − H
                                                                         al., 1998), describes the relationship between daily ET
                                                                   (6)   of a given crop (ETVI) and ET0 by separating
                                                                                                               (6)      the single
                                                                        crop coefficient (Kc) into the basal crop coefficient
              where ETinst and ET0 are the instantaneous crop            (Kcb), soil water evaporation (Ke) coefficient and water
              and reference ET values, respectively, and ET0 24 is       stress coefficient (Ks). Crop transpiration, represented
              daily ET0. Note that the ratio ETinst/ET0 in Eq. (6) is    by Kcb is separated from soil surface(7) evaporation as
              equivalent to the Kc for the considered day.               follow:

              2.4.2 Vegetation index RS approach                                                                                    (7)
              ETVI was obtained by combining a VIS/NIR-RS                                               
              procedure with ancillary ground based data on              where ETVI and ET0 are in mm d-1.
                                                                                                              (8)

                                                                                                       Fig. 3 - Lay-out
                                                                                                       of the main input/output
                                                                                                       parameters in the 2S model.
                                                                                                       Fig. 3 - Schema
                                                                                                       dei principali parametri
                                                                                                       di input/output utilizzati

                                                                                                                                            Italian Journal of Agrometeorology - 2/2018
                                                                                                       nel modello

                                                                                                                                                                                    Rivista Italiana di Agrometeorologia - 2/2018
                                                                                                       bi-dimensionale.

                                                                                                                                                                  45

AGRO 2-18_IIIA BOZZA.indb 45                                                                                                              18/07/18 16.22
H ET
                                                                                           CR                                                                                                  (3)
                                                                                                  Rn G

                                                                                                                                                                                     Fig. 4 - Lay-out
                                                                                                                                                                                             (4) input/output
                                                                                                                                                                                     of the main
                                                                                                                                                                                     parameters
                                                                                                                                                                                     in the VI-satellite
                                                                                                                                                                                     based model.
                                                                                           LE = R n − G − H                                                                          Fig. 4 - Schema
                                                                                                                                                                                             (5) parametri
                                                                                                                                                                                     dei principali
                                                                                                                                                                                     di input/output utilizzati
                                                                                                                                                                                     nel modello VI.

                                                                                                                                                                                               (6)

                                                                                                                                                                                               (7)
                                                                                          The reduction of actual evaporation (Kr) when the             5). ET0 was well above 700 mm; ETc, obtained by a
                                                                                          amount of water in the surface soil layer decreases was       corrected Kc of 0.46, reached about 300 mm during
                                                                                          accounted as:                                                 both studied irrigation seasons.
                                                                                                                                                        Average values of soil moisture content at the different
                                                                                                                                                  (8)   irrigation treatments are reported in(8)Tab. 6.
                                                                                                                        
                                                                                                                                                        In all the investigated soil profiles, soil water content
                                                                                          where REW (mm) is the readily evaporable water                remained between the FC and WP (Fig. 5).
                                                                                          (10 mm, by Allen et al, 1998) , TEW is the maximum
                                                                                          cumulative depth of evaporation from the soil surface         3.2. Micrometeorological results
                                                                                          layer (0.1 m) and De,i (mm) is the cumulative depletion       Fig. 6 shows the surface energy balance closure from
                                                                                          at the end of the 1-th day. The soil parameters used in       EC measurements at the field site during the 2016 (a)
                                                                                          the calculation of Kr, as field capacity (FC, 0.28 cm3        and 2017 (b) irrigation seasons (June – September).
                                                                                          cm-3) and permanent wilting point (WP, 0.14 cm3 cm-3),        The slope of the regression forced through the origin
                                                                                          were determined by laboratory analyses for the soil-          was around 0.82 in both the study periods, with
                                                                                          type under study (Consoli et al., 2017).                      determination coefficients (r2) of about 0.88 and 0.90
                                                                                                                                                        in 2016 and 2017, respectively. The results confirmed
                                                                                          3. Results                                                    the good quality of the flux sources within the area of
                                                                                          3.1. Climatic data and soil moisture dynamics at              footprint of the micrometeorological EC tower (Twine
                                                                                          the experimental site during deficit irrigation               et al., 2000).
                                                                                          The climate of the site during the two irrigation             The net radiation, Rn, was as expected the dominant
                                                                                          seasons in 2016 and 2017 was quite typical of semi-arid       surface energy balance component, driving the energy
                                                                                          Mediterranean conditions. No significant variations           budget. Observing the energy fluxes reported in Fig.
                                                                                          were detected during the two periods of trial (Tab.           7, it can be noted that the most part of the available
  Italian Journal of Agrometeorology - 2/2018

                                                                                                                             wind
                                          Rivista Italiana di Agrometeorologia - 2/2018

                                                                                          Year         DOY        Ta                Tdew      ET0       RHmin      Rain      Kcb      Kc max         Kr     Ks
                                                                                                                            speed
                                                                                                                  °C         ms-1    °C     mm d-1        %        mm
                                                                                                       180       33.8         2.8   13.6     8.4         29.6      0.0       0.72     1.29           0.08   0.94
                                                                                                       203       34.7         1.5   13.2     7.8         27.4      0.0       0.72     1.25           0.07   1.00
                                                                                          2016         219       34.4         1.9   17.7     7.5         37.3      0.0       0.72     1.22           0.13   1.00
                                                                                                       235       34.5         2.1   16.7     7.4         34.7      0.0       0.72     1.24           0.29   1.00
                                                                                                       251       26.9         1.3   15.4     3.9         49.4      2.4       0.72     1.15           1.00   1.00
                                                                                                       150       21.8         1.5   10.4     6.8         48.0      0.0       0.55     1.16           0.20   1.00
                                                                                                       166       25.6         1.4   13.2     7.7         46.1      0.0       0.55     1.16           0.04   1.00
                                                                                          2017         182       26.4         2.3   20.0     6.9         68.0      0.0       0.55     1.11           0.04   1.00
                                                                                                       189       24.9         1.8   13.5     6.7         49.1      0.0       0.55     1.17           0.27   0.97
                                                                                                       214       32.0         1.8   16.4     7.9         39.1      0.0       0.55     1.21           0.64   1.00
                                                                                          Tab. 4 - Main inputs for the dual-Kc FAO-56 approach.
                                                                                          Tab. 4 - Principali input del modello dual-Kc FAO-56.

                        46

AGRO 2-18_IIIA BOZZA.indb 46                                                                                                                                                                                        18/07/18 16.22
Parameter                      2016               2017            is plausible to figuring out a significant reduction of
               ET0 (mm)                            723.3              782.9      latent heat flux, LE, (mean value of 60.2 ± 74.3 W m-2
               ETc=ET0*Kc (mm)                     334.2              361.7      in 2016 and 59.5 ± 71.2 W m-2 in 2017) in benefit of H
               ETEC (mm)                           237.8              235.0      increments (mean value of 86.4 ± 126.7 W m-2in 2016
               Rainfall (mm)                        99.6               36.8      and 98.0 ± 138.8 W m-2 in 2017).
               Tair (°C)                      24.9(±2.5)         24.9(±2.5)
                                                                                 Therefore, during the two irrigation seasons daily
                                                                                 measured ETEC values ranged from 0.8 to 3.6 mm
               RH (%)                        60.1(±10.1)        60.0(±10.1)
                                                                                 day-1, with a mean of 2.1(± 0.6) mm day-1. Total ETEC
              Tab. - 5 ET rates and meteorological data (mean ± standard         during the monitoring periods were 237.8 and 235.0
              deviation) for the study site in 2016 and 2017.
              Tab. - 5 Flussi di ET e grandezze meteorologiche (media ± devi-
                                                                                 mm in 2016 and 2017, respectively (Tab. 4).
              azione standard) per il sito in studio nel periodo in 2016-2017.   As expected, estimated daily ETc values by the FAO-
                                                                                 56 approach (Table 4) were higher than ETEC of about
                                                                                 36% and 46% in 2016 and 2017, respectively. Fig. 8 (a
               Irrigation                 Soil water content (m3 m-3)
               treatment                                                         and b) reports the comparison between the Kc value
                                              2016            2017
                                                                                 (about 0.5) adopted during the study to correct the
               T1:100% ETc                     0.26            0.26
                                                                                 reference ET and the crop coefficient derived from EC
               T2:75% ETc                      0.26            0.26
                                                                                 (Kc, EC) during the study. The average values of Kc, EC
               T3:RDI                          0.20            0.20              (ETEC/ET0) were of 0.34(± 0.1) and 0.31(± 0.1) in 2016
               T4: PRD                         0.21            0.17              and 2017, respectively. As expected these values of Kc,
              Tab. 6 - Mean values of soil water content (m3 m-3) at the         EC have been significantly different from the single Kc,
              different irrigation treatments during the years 2016 and          accounting for deficit irrigation conditions.
              2017 obtained by ECH20.
              Tab. 6 - Valori medi del contenuto idrico del suolo (m3 m-3)
              misurati da sonde ECH20 per i diversi trattamenti irrigui          3.3. Remote sensing results
              nel periodo 2016-17                                                Fig. 9 (a- d) compares the instantaneous surface energy
                                                                                 fluxes (W m-2) directly measured by EC and estimated
              energy (Rn-G) is transformed into sensible heat (H).               from SEB models (i.e. 1S and 2S). The RS-based
              This behaviour is typical of advective conditions, in              models tend to overestimate the available energy (Rn-
              which a high heat rate is exchanges by the plant system            G) and LE flux, and to underestimate H flux.
              to the atmosphere, corresponding to crop temperature               The comparison between daily surface energy
              increase. In our experimental conditions, where the                fluxes estimated by FAO-56, dual Kc (VI) and SEB
              footprint of the EC tower is mainly irrigated under                approaches (1S and 2S) and measured by EC system is
              deficit conditions (i.e. treatments T2, T3 and T4), it             reported in Fig. 10.

                                                                                                              Fig. 5 - Evolution of soil
                                                                                                              water content (m3 m-3)
                                                                                                              in the irrigation treatments
                                                                                                              and rainfall in 2016 (a)
                                                                                                              and 2017 (b) at the study
                                                                                                              site. FC = field capacity

                                                                                                                                               Italian Journal of Agrometeorology - 2/2018
                                                                                                              (0.28 m3 m-3), WP = wilting
                                                                                                              point (0.14 m3 m-3).

                                                                                                                                                                                       Rivista Italiana di Agrometeorologia - 2/2018
                                                                                                              Fig. 5 - Precipitazione
                                                                                                              e evoluzione del contenuto
                                                                                                              idrico del suolo (m3 m-3)
                                                                                                              nei diversi trattamenti
                                                                                                              irrigui nel 2016 (a)
                                                                                                              e nel 2017 (b). FC= capacità
                                                                                                              di campo (0.28 m3 m-3),
                                                                                                              WP = punto critico colturale
                                                                                                              (0.14 m3 m-3).

                                                                                                                                                                     47

AGRO 2-18_IIIA BOZZA.indb 47                                                                                                                 18/07/18 16.22
Fig. 6 - Surface energy balance closure of the EC measurements at the field site. Half-hourly data refer to irrigation seasons
                                                                                          (June – September) in 2016 (a) and 2017 (b).
                                                                                          Fig. 6 - Chiusura del bilancio energetico da misure EC presso il sito in studio. Dati alla mezz’ora relativi alle stagioni irrigue
                                                                                          (Giugno-Settembre) del 2016 (a) e del 2017 (b).

                                                                                          Fig. 7 - Example of hourly surface energy fluxes (W m-2) measured at the field site by the EC technique. Data refer to June 2017.
                                                                                          Fig. 7 - Esempio di flussi energetici orari (W m-2) da EC presso il sito in studio - Giugno 2017.
  Italian Journal of Agrometeorology - 2/2018
                                          Rivista Italiana di Agrometeorologia - 2/2018

                                                                                          Fig. 8 - Estimated Kc for the orange orchard using the FAO-56 approach, actual Kc, EC and rainfall (mm) during 2016 (a)
                                                                                          and 2017 (b) periods.
                                                                                          Fig. 8 - Stime di Kc da modello FAO-56, Kc effettivo da EC e precipitazione (mm) nel 2016 (a) and nel 2017 (b).

                                                                                          At the daily scale, ET fluxes obtained by RS                       measured ETEC and estimated fluxes (ET1S, ET2S,
                                                                                          approaches (ET1S, ET2S, ETVI) resulted greater than                ETVI).
                                                                                          the ETEC fluxes respectively of 38.2, 44.8 and 49.1%.              Some examples of ET maps obtained by applying the
                                                                                          Fig. 11 a shows the scatter plot between the daily                 RS techniques (dual Kc, 1S and 2S models) are shown
                        48

AGRO 2-18_IIIA BOZZA.indb 48                                                                                                                                                                                                   18/07/18 16.22
Fig. 9 - Instantaneous
                                                                                                              estimated versus measured
                                                                                                              surface energy fluxes
                                                                                                              (W m-2) at the field site
                                                                                                              in the study period.
                                                                                                              Fig. 9 - Stime istantanee
                                                                                                              e misure di flussi energetici
                                                                                                              (W m-2) presso il sito
                                                                                                              in studio.

              in Figg. 12 and 13, referring to a couple of DOYs in            conducted by the same authors (Barbagallo et al.,
              2016 and 2017, respectively.                                    2009; Consoli and Vanella 2014a; 2014b) in well-
                                                                              irrigated orange orchards revealed a quite good
              4. DISCUSSION                                                   accordance between the satellite estimates and the
              The most interesting result of this study is the                “truth” in the field. In this case, the deficit irrigation
              discrepancy between RS-based estimates of surface               conditions applied at the field during the two
              energy fluxes and the EC ET data. Previous studies              consecutive irrigation seasons played a strategic role in

                                                                                                                                                Italian Journal of Agrometeorology - 2/2018
                                                                                                                                                                                        Rivista Italiana di Agrometeorologia - 2/2018

              Fig. 10 - Daily ET fluxes during the monitoring period 2016-17. EC refers to ET measured by EC; ETc estimated by FAO
              – 56 approach, VI is the dual-Kc satellite approach; 1S and 2S refer to ET derived by SEB models.
              Fig. 10 - Flussi giornalieri di ET nel periodo 2016-17. EC indica la misura diretta di ET da Eddy Covariance; FAO indica
              le stime di ETc da modello FAO 56, VI è il modello dual-Kc da satellite; 1S and 2S sono le stime di ET da bilancio energetico
              satellitare (SEB).

                                                                                                                                                                      49

AGRO 2-18_IIIA BOZZA.indb 49                                                                                                                  18/07/18 16.22
5. CONCLUSION
                                                                                                                                                          The main objective of the study was to investigate the
                                                                                                                                                          potential of different satellite-based models (FAO-56
                                                                                                                                                          dual-Kc approach and SEB models) to provide accurate
                                                                                                                                                          ET estimates in a deficit irrigated orange orchard under
                                                                                                                                                          semi-arid Mediterranean climate conditions. Satellite-
                                                                                                                                                          based estimates of ET were compared with direct ET
                                                                                                                                                          measures from EC.
                                                                                                                                                          Results were generally unsatisfactory due to the
                                                                                                                                                          overestimation of the ET fluxes provided by the
                                                                                                                                                          satellite-based approaches. The approaches were, in
                                                                                                                                                          fact, not able to capture the relative high ratio H/LE
                                                                                                                                                          due to the induced water stress conditions.
                                                                                          Fig. 11 - Comparison between ETEC and ET estimated by
                                                                                          RS approaches.
                                                                                                                                                          The results deserve a certain interest for the need to
                                                                                          Fig. 11 - Confronto tra ETEC and le stime di ET da satellite.   deepen the nature of the advective phenomenon, that
                                                                                                                                                          may have a role in the physiological and hormonal (i.e.
                                                                                          altering the normal ration between the energy balance           abscisic acid and proline production) response of the
                                                                                          surface fluxes, determining a fairly high sensible heat         crop to sustain water stress.
                                                                                          and a ET reduction due to the imposed water stress
                                                                                          conditions. This behaviour was poorly captured by               ACKNOWLEDGMENTS
                                                                                          the remote estimates (except for few cases) and was             The authors wish to thank the personnel of Centro
                                                                                          exacerbated by the intrinsic nature of the semi-arid            di ricerca Olivicoltura, Frutticoltura e Agrumicoltura
                                                                                          climate characterizing the field site. In fact, errors          of the Italian Council for Agricultural Research and
                                                                                          in satellite ET estimates are generally large in these          Agricultural Economics Analyses (CREA-OFA,
                                                                                          climate conditions due to the presence of mixed                 Acireale) for their hospitality at the field site.
                                                                                          pixels, i.e. characterizing by fully irrigated fields           The authors thank the EU and the Italian Ministry
                                                                                          usually surrounded by dry landscape that accounting             of Education, Universities and Research for funding,
                                                                                          for differences in cover ground and moisture content            as part of the collaborative international consortium
                                                                                          (Liu and Luo, 2010).                                            IRIDA (‘‘Innovative remote and ground sensors,
                                                                                          Analysing the results of the satellite modelling                data and tools into a decision support system for
                                                                                          per se, it can be noted that in general the dual-Kc             agriculture water management”), financed under the
                                                                                          approach overestimates ET after irrigation events               ERA-NET Cofund WaterWorks 2014. This ERA-
                                                                                          and the estimates provided by the 2S model were in              NET is an integral part of the 2015 Joint Activities
                                                                                          good accordance with the results of the simplified              developed by the Water Challenges for a Changing
                                                                                          1S approach. Therefore, the use of 2S model can be              World Joint Programme Initiative (Water JPI).
                                                                                          questioning at the light of the required computation            The authors also acknowledge support from the
                                                                                          effort and inputs parameters.                                   ERANET-MED project WASA (‘‘Water Saving in
  Italian Journal of Agrometeorology - 2/2018

                                                                                                                                                                                       Fig. 12 - Examples
                                          Rivista Italiana di Agrometeorologia - 2/2018

                                                                                                                                                                                       of ET maps obtained
                                                                                                                                                                                       by applying the RS
                                                                                                                                                                                       approaches. Data refer
                                                                                                                                                                                       to DOY 180 and 219, 2016.
                                                                                                                                                                                       Fig. 12 - Esempio di mappe
                                                                                                                                                                                       di ET ottenute applicando
                                                                                                                                                                                       la modellazione da satellite.
                                                                                                                                                                                       Giorni dell’anno 180
                                                                                                                                                                                       and 219, 2016.

                        50

AGRO 2-18_IIIA BOZZA.indb 50                                                                                                                                                                                           18/07/18 16.22
Fig. 13 - Examples of ET
                                                                                                       maps obtained by applying
                                                                                                       the RS approaches.
                                                                                                       Data refer to DOY 150
                                                                                                       and 180, 2017.
                                                                                                       Fig. 13 - Esempio di mappe
                                                                                                       di ET ottenute applicando
                                                                                                       la modellazione da satellite.
                                                                                                       Giorni dell’anno 150
                                                                                                       and 180, 2017.

              Agriculture: Technological developments for the                irrigation on lettuce crops in Sicily (Italy). Journal
              sustainable management of limited water resources in           of Irrigation and Drainage Engineering, ASCE, 134
              the Mediterranean area”).                                      (4), pp. 437-445.
                                                                          Capra A., Consoli S., Scicolone B., 2013. Long-term
              REREFENCES                                                     climatic variability in Calabria and effects on
              Abdullah K.B., 2006. Use of water and land for food            drought and agrometeorological parameters. Water
                  security and environmental sustainability. Irrigation      Resources Management, 27(2), 601-617.
                  and Drainage, 55(3), 219-222.                           Consoli S., O’Connell N., Snyder R., 2006. Estimation
              Abrisqueta I., Conejero W., Valdés-Vela M., Vera J.,           of evapotranspiration of different-sized navel-
                  Ortuño M.F., Ruiz-Sánchez M.C., 2015. Stem                 orange tree orchards using energy balance. Journal
                  water potential estimation of drip-irrigated early-        of irrigation and drainage engineering, 132(1), 2-8.
                  maturing peach trees under Mediterranean                Consoli S., Papa R., 2013. Corrected surface ener-
                  conditions. Computers and Electronics in Agri-             gy balance to measure and model the evapo-
                  culture, 114, 7-13.                                        transpiration of irrigated orange orchards in
              Allen R.G., Pereira L.S., Raes D., Smith M., 1998. Crop        semi-arid Mediterranean conditions. Irrigation
                  evapotranspiration-Guidelines for computing crop           science, 31(5), 1159-1171.
                  water requirements-FAO Irrigation and drainage          Consoli S., Stagno F., Roccuzzo G., Cirelli G. L.,
                  paper 56. FAO, Rome, 300(9), D05109.                       Intrigliolo F., 2014. Sustainable management of
              Allen R.G., Pruitt W.O., Wright J.L., Howell T.A.,             limited water resources in a young orange orchard.
                  Ventura F., Snyder R., Itenfisu D., Steduto P.,            Agricultural water management, 132, 60-68.
                  Berengena J., Baselga Yrisarry J., Smith M.,            Consoli S., Stagno F., Vanella D., Boaga J., Cassiani
                  Pereira L.S., Raues D., Perrier A., Alves I.,              G., Roccuzzo G., 2017. Partial root-zone drying

                                                                                                                                         Italian Journal of Agrometeorology - 2/2018
                  Walter I., Elliott R., 2006. A recommendation              irrigation in orange orchards: effects on water use

                                                                                                                                                                                 Rivista Italiana di Agrometeorologia - 2/2018
                  on standardized surface resistance for hourly              and crop production characteristics. European
                  calculation of reference ETo by the FAO56                  Journal of Agronomy, 82, 190-202.
                  Penman-Monteith method. Agricultural Water              Consoli S., Vanella D., 2014a. Mapping crop eva-
                  Management, 81(1-2), 1-22.                                 potranspiration by integrating vegetation indices
              Aubinet M., Grelle A., Ibrom A., Rannik U., Moncrieff          into a soil water balance model. Agricultural water
                  J., Foken T., Vesala T., 2000. Estimates of the            management, 143, 71-81.
                  annual net carbon and water exchange of European        Consoli S., Vanella D., 2014b. Comparisons of satellite-
                  forests: The EUROFLUX methodology. Advances                based models for estimating evapotranspiration
                  in Ecological Research, 30, 113-175.                       fluxes. Journal of hydrology, 513, 475-489.
              Barbagallo S., Consoli S., Russo A., 2009. A one-layer      Courault D., Seguin B., Olioso A., 2005. Review on
                  satellite surface energy balance for estimating            estimation of evapotranspiration from remote
                  evapotranspiration rates and crop water stress             sensing data: From empirical to numerical modeling
                  indexes. Sensors 9, 1-21.                                  approaches. Irrigation and Drainage systems, 19(3-
              Capra A., Consoli S., Russo A., Scicolone B., 2008.            4), 223-249.
                  Integrated agro-economic approach to deficit            Er-Raki S., Chehbouni A., Boulet G., Williams D.G.,
                                                                                                                                                               51

AGRO 2-18_IIIA BOZZA.indb 51                                                                                                           18/07/18 16.22
2010. Using the dual approach of FAO-56 for                 Prueger J.H., Hatfield J.L., Parkin T.B., Kustas W.P.,
                                                                                             partitioning ET into soil and plant components for             Hipps L.E., Neale C.M.U., Cooper D.I., 2005.
                                                                                             olive orchards in a semi-arid region. Agricultural             Tower and aircraft eddy covariance measurements
                                                                                             water management, 97(11), 1769-1778.                           of water vapor, energy, and carbon dioxide fluxes
                                                                                          FAO, 2013. Food wastage footprint: Impacts on natural             during SMACEX. Journal of Hydrometeorology,
                                                                                             resources. Rome                                                6(6), 954-960.
                                                                                          Fereres E., Aldrich T., Schulbach H., Martinich D.,            Ruiz-Sánchez M. C., Abrisqueta I., Conejero W.,
                                                                                             1981. Responses of young almond trees to late-                 Vera J., 2018. Deficit irrigation management in
                                                                                             season drought. California Agriculture, 35(7), 11-12.          early-maturing peach crop. In Water Scarcity and
                                                                                          Gonzalez-Dugo M.P., Neale C.M.U., Mateos L., Kustas               Sustainable Agriculture in Semiarid Environment
                                                                                             W.P., Prueger J.H., Anderson M.C., Li F., 2009. A              (pp. 111-129).
                                                                                             comparison of operational remote sensing-based              Singh S., Boote K.J., Angadi S.V., Grover K.K., 2017.
                                                                                             models for estimating crop evapotranspiration.                 Estimating water balance, evapotranspiration
                                                                                             Agric. For. Meteorol. 149, 1843-1853.                          and water use efficiency of spring safflower
                                                                                          Kaimal J.C., Finnigan J.J., 1994. Atmospheric boundary-           using the CROPGRO model. Agricultural Water
                                                                                             layer flows: Their structure and measurement.                  Management, 185, 137-144.
                                                                                             (289 pp.) Oxford University Press.                          Swank W.T., Douglass J.E., 1974. Streamflow greatly
                                                                                          Li F., Kustas W.P., Prueger J.H., Neale C.M.U.,                   reduced by converting deciduous hardwood stands
                                                                                             Jackson J.T., 2005. Utility of remote sensing based            to pine. Science, 185(4154), 857-859.
                                                                                             two-source energy balance model under low and               Twine T.E., Frustas W.P., Norman J.M., Cook D.R.,
                                                                                             high vegetation cover conditions. J. Hydrometerol.             Houser P.R., Meyers T.P., Prueger J.H., Starks P.J.,
                                                                                             6, 878-891.                                                    Wesely M.L., 2000. Correcting eddy covariance
                                                                                          Liu Y., Luo Y., 2010. A consolidated evaluation of the            flux underestimates over grassland. Agric Forest
                                                                                             FAO-56 dual crop coefficient approach using the                Meteorol 103:279-300.
                                                                                             lysimeter data in the North China Plain. Agriculture        USGS, 2016. Landsat 8 (L8) Data users
                                                                                             Water Managent, 97, 31-40.                                     handbook, LSDS-1574 Version 2.0 https://
                                                                                          Monteith J.L., Unsworth M.H., 1990. Radiation                     landsat.usgs.gov/sites/default/files/documents/
                                                                                             environment. Principles of environmental physics,              Landsat8DataUsersHandbook.pdf
                                                                                             2nd ed., Edward Arnold, London, 36-57.                      Wada Y., Wisser D., Eisner S., Flörke M., Gerten D.,
                                                                                          Pérez-Pérez J.G., Romero P., Navarro J.M., Botia P.,              Haddeland I., Hanasaki N., Masaki Y., Portmann
                                                                                             2008. Response of sweet orange cv ‘Lane late’ to               F.T., Tessler Z., Schewe J., 2013. Multimodel
                                                                                             deficit irrigation in two rootstocks. I: water relations,      projections and uncertainties of irrigation water
                                                                                             leag gas exchange and vegetative growth. Irrigation            demand under climate change. Geophysical
                                                                                             Science, 26 (5), 415-425.                                      Research Letters, 40(17), 4626-4632.
                                                                                          Power A.G., 2010. Ecosystem services and agriculture:          Zhang K., Kimball J.S., Running S.W., 2016. A review
                                                                                             tradeoffs and synergies. Philosophical transactions            of remote sensing based actual evapotranspiration
                                                                                             of the royal society B: biological sciences, 365(1554),        estimation. Wiley Interdisciplinary Reviews: Water,
                                                                                             2959-2971.                                                     3(6), 834-853.
  Italian Journal of Agrometeorology - 2/2018
                                          Rivista Italiana di Agrometeorologia - 2/2018

                        52

AGRO 2-18_IIIA BOZZA.indb 52                                                                                                                                                                                       18/07/18 16.22
You can also read