Evapotranspiration estimation of crops protected by windbreak in a Mediterranean region

Agricultural Water Management 104 (2012) 153–162

Contents lists available at SciVerse ScienceDirect

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

Evapotranspiration estimation of crops protected by windbreak in a Mediterranean region P. Campi ∗ , A.D. Palumbo, M. Mastrorilli CRA, Research Unit for Cropping Systems in Dry Environments, Via Ulpiani, 5 - 70125 Bari, Italy

a r t i c l e

i n f o

Article history: Received 26 November 2010 Accepted 12 December 2011 Available online 30 December 2011 Keywords: Agrometeorology Crop coefficients Dry-farming systems Tree barrier Bean Durum wheat Penman–Monteith model

a b s t r a c t This study is part of a number of studies designed to improve the applicability of the FAO56 model (Irrigation and Drainage Paper by the FAO); the aim of this study is to estimate the crop evapotranspiration (ETc ) protected by windbreaks. The study was conducted in three steps: (i) We parameterised the effects of the windbreak on the inputs to the FAO56 model. The calibration parameters were obtained from an experimental study performed on two typical crops of the Mediterranean environment, durum wheat and beans. The former is a rain-fed crop, and the latter is an irrigated crop. A new modified version of the FAO56 model, named the FAO56-wb, which takes into account the windbreak effects, is suggested. (ii) We validated the evapotranspiration (ETwb ) provided by the FAO56-wb in daily and seasonal scales. The validation was performed using an independent data set of measured soil-water balances that can determine the evapotranspiration (ET) at various distances from the windbreak. The calculated data of the ETwb and the measured data agree, both in the daily and in the seasonal scales, as demonstrated by the Relative Root Mean Square Error test (RRMSE, between 19.9 and 16.3 for durum wheat and between 12.3 and 16.7 for beans) and by the linear regression (R2 > 0.8 for durum wheat and R2 > 0.9 for beans). (iii) We evaluated the simulation of the ETwb for durum wheat and beans using a 25-year series of agro-meteorological data. The FAO56-wb simulates the evapotranspiration as a function of the porosity of the barrier. The simulations demonstrated the windbreak’s potential to contain the evapotranspiration of durum wheat and beans in a semi-arid environment, such as the Mediterranean. The effects on ET reduction were more clearly seen with the use of a windbreak with low porosity (20%) and 3 m in height. In this case, a maximum reduction of 31% of the evapotranspiration in a summer crop, such as beans, occurred within a distance from the windbreak equal to 15 times the barrier height. The results showed that the FAO56-wb is a useful tool to predict the effects of a windbreak even though the model requires only a few agro-environmental inputs. © 2011 Elsevier B.V. All rights reserved.

1. Introduction The incorporation of windbreaks in agricultural systems is becoming prevalent in many regions of the developed world, where the scarcity of land and water resources has resulted in a need for agriculture to consume less water. In many parts of Europe, hedgerows or low windbreaks are common features of the traditional rural landscapes. Recently, agricultural development has been lead to the establishment of a windbreak system that has been widely diffused in many regions of Australia and New Zealand, the heathlands of Denmark, the Great Plains of North America and the steppes of the former Soviet Union.

∗ Corresponding author. Tel.: +39 0805475014; fax: +39 0805475023. E-mail address: [email protected] (P. Campi). 0378-3774/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.agwat.2011.12.010

In the Mediterranean environment, the limited availability of water implies the need to search for agronomical strategies that can mitigate the consequences of water deficits and increase the efficiency of the irrigation supply. In dry farming, the use of windbreaks is one of various proposed solutions to reduce evapotranspiration (ET) by modifying the components of the energy balance (Burke, 1998). ET is the most important variable to know accurately to develop a rational irrigation management plan. The ET is particularly important in environments with long hot, arid periods, such as the Mediterranean region, where the lack of water resources is the main factor limiting agricultural development. The ET depends on many factors, such as the climate, the soil type and the species being cultivated. The crop evapotranspiration (ETc ) can be estimated by two approaches. In the ‘one step’ approach, the ETc is directly estimated by the environmental conditions and the physical, morphological and physiological features

154

P. Campi et al. / Agricultural Water Management 104 (2012) 153–162

of the soil-plant system. The ‘two step’ approach considers the ET of a reference crop (ET0 ), under known ecophysiological factors, and the crop coefficient (Kc) is used as an aggregation of the physical and physiological differences between the cultivated and reference crops. At instantaneous time scale, the estimate of ET is performed using the Penman–Monteith equation (Monteith, 1973): ET =

(Rn − G) + (Cp(ea − ed )/ra )  + (1 + (rc /ra ))

(1)

where ET is the latent heat flux (W m−2 ),  the curve slope expressing the saturated vapour tension as a function of the temperature (kPa ◦ C−1 ), Rn the net radiation (W m−2 ), G the flux density of soil heat (W m−2 ),  the average air density (kg m−3 ), Cp the specific air heat (MJ kg−1 ◦ C−1 ), ea the saturated vapour pressure at air temperature (kPa), ed the actual vapour pressure (kPa), ra the aerodynamic resistance (s m−1 ),  the psychrometric constant (kPa ◦ C−1 ), and rc is the crop canopy resistance to vapour flux (s m−1 ). Some parameters (, Cp , ) are obtained through the environmental characteristics and are almost constant; others (, Rn , G, ea , ed ) can be easily measured at agro-meteorological stations. The resistances (ra and rc ) that reduce the evapotranspiration flow (␭ET) are difficult to determine because they require specific crop characteristics. In particular, rc is mainly a function of the leaf area index (LAI), global radiation (Rg ), vapour pressure deficit (VPD), temperature and plant water status whereas ra requires the calculation of the vegetation roughness (Jarvis, 1976) and the measurement of the wind speed. Recently, some papers (Shuttleworth, 2006; Shuttleworth and Wallace, 2009; Rana and Katerji, 2009) have suggested alternative approaches to calculate ET by analytical one-step models using operational characteristics. However, the use of two-step models prevails in the calculation of ET for practical purposes. The ‘one step’ approach requires that the climatic variables be measured above the crops (Rana et al., 1994) while the ‘two step’ approach uses the measurements of variables found in standard meteorological stations. The ‘two step’ method proposed by Penman–Monteith in Irrigation and Drainage Paper No. 56 (FAO56 model), which was designed for operating purposes, can be written as ET = Kc ET0 . For practical purposes, Allen et al. (1998) suggested the use of the Penman–Monteith equation in a daily scale; in this case, the energy terms in Eq. (1) are the integral of time in instantaneous values, and the other inputs are the daily mean of the variables. The ET is estimated by the FAO56 model in nonstandard vegetation conditions (e.g., water stress, use of saline waters, non-pristine vegetation, mulching, intercropping of different species). Recent scientific research has suggested some changes in the ‘two step’ approach to improve the estimate of ET and to generalise it for specific agro-environmental conditions (Ko et al., 2009; Er-Raki et al., 2006; Liu and Luo, 2010). Among the changes performed to the FAO56 model, those concerning the wind speed and the windbreak effects have not yet been taken into account. The use of windbreaks can reduce the ET because they create an important action on the aerodynamic resistance of the water vapour flow (Burke, 1998). Windbreaks have been studied all over the world, in Northern China (Zao et al., cited by Brandle et al., 2004), Canada (Kort, 1988), New Zealand (Sturrock, 1984), USA (Brandle et al., 2004), South America (Luis and Bloomberg, 2002), Australia (Nuberg, 1998; Cleugh et al., 2002), and in the Mediterranean region (Ben Salah et al., 1989; Benzarti, 1990; Casa et al., 1994; Campi et al., 2009). However, these studies do not suggest any modelling of the effects

of windbreaks on the operating methods to calculate the ETc using the FAO56 model. Therefore, the aim of this study was to suggest the functions that need to be introduced into the FAO56 model to calculate ETc in cases using windbreaks (FAO56-wb model). After the calibration phase, the validation of the FAO56-wb model is performed by comparing its output (ETwb ) in a daily scale with the independent ET measurements that were performed in autumn-spring (durum wheat) and in summer (bean) crops. Finally, to draw a conclusion from the results, it can be assumed that the FAO56-wb can be used to simulate the benefits provided by the presence of a windbreak on crop water use. For this reason, the ETwb will be simulated over a 25-year period for the two crops, and the FAO56-wb will also be used to develop the appropriate dimensioning of the porosity of the windbreak.

2. Materials and methods The study is based on the analysis of the microclimate of durum wheat (cv. ‘Simeto’) and bean (cv. ‘Lingua di fuoco’) plots, both with a windbreak and without it. Both crops were cultivated in southern Italy at the experimental farm of the Agricultural Research Council – Research Unit for Cropping Systems in Dry Environments (C.R.A. – S.C.A.) in Rutigliano (lat: 40◦ 59 , long: 17◦ 01 , alt: 147 m a.s.l.). The environment is characterised by an average rain fall of 600 mm, with precipitation mainly concentrated during the autumn period and much more reduced or absent in the spring-summer period. Precipitation is around inadequate to meet the atmosphere’s ET requirements; the annual water deficit is 560 mm (Campi et al., 2005). In the location under study, the average annual wind speed is 2.8 m s−1 , which is higher than the average global wind speed (2.0 m s−1 ) (Allen et al., 1998). The prevailing wind direction is north. The experimental field (100 m × 200 m) has a windbreak barrier of Cupressus arizonica L. (3 m high, 20 years old), set perpendicular to the prevailing wind. The porosity is 40%, as calculated according to the approach suggested by Nelmes et al. (2001). This approach for estimating the porosity requires only two wind speed measurements, one located upstream and one located downstream. The soil has a clay texture with a field water capacity of 30% and a wilting point of 18% (measured through Richards plates on dry soil weight), and its bulk density is 1.15 g m−3 . Because the soil profile is shallow (0.6 m), it has a moderate available soil water capacity (83 mm). The durum wheat and beans were cultivated following conventional agro-techniques. The irrigation schedule for the beans was performed using the water balance FAO method, where the root development was considered to have a depth up to a maximum of 0.6 m. Irrigation was performed, distributing a total of 2800 m3 ha−1 , using mini-sprinklers uniformly distributed over the field. The irrigation depths were measured using volumetric counters installed in the main irrigation line. The microclimatic data were measured on the experimental field of durum wheat from 20/10/2004 to 4/06/2005 and then on the nearby bean field from 5/06/2005 to 22/08/2005, using the same microclimatic sensor set (Fig. 1). The wind speed was measured using cup anemometers (A100R, Vector Instruments, UK; sensitivity threshold: 0.2 m s−1 ) installed at the following positions (D/H, where D is the distance from windbreak and H represents the height of the windbreak): 2D/H, 2.7D/H, 3.3D/H, 4.7D/H, 6D/H, 7.3D/H, 10D/H, 12.7D/H, 18D/H and 23.3D/H. The air temperature and relative humidity were measured using thermo-hygrometers (Campbell model cs500; accuracy: ±0.2 ◦ C for air temperature and ±1.5% for RH) installed at the following positions: 2D/H, 2.7D/H, 3.3D/H, 4.7D/H, 12.7D/H and 23.3D/H. The

P. Campi et al. / Agricultural Water Management 104 (2012) 153–162

155

Fig. 1. Positions of the micrometeorological sensors and TDR probes as a function of the windbreak distance (D) respect to the height of windbreak (H = 3 m).

wind, air temperature and relative humidity were monitored every hour. A Fritschen-type net radiometer (REBS Q*6, USA) was set up at the 2.7D/H position. All the above measurements varied with the canopy height within the surface boundary layer (Wieringa, 1993). At the position farthest from the windbreak, the direction of the wind (measured with a weathervane sensor, Young model 03302) and the wind speed were measured at a 2 m height. The calibration of the anemometers and thermo-hygrometers was performed in an area unaffected by the windbreak. As a reference, high-accuracy sensors were used. The correlation between the data collected from each sensor and the reference provides a correction to the specific coefficients of each sensor.The thermohygrometers measure small differences in relative humidity at different distances from the windbreak. In addition, VPD, calculated using the measured air relative humidity and air temperature (according to the FAO56 procedure to derive VPD), shows small differences in the protected area. The highest difference was measured when the wind comes from the upwind sector. The highest difference in VPD, calculated as average daily values of the whole experimental period, ranged between 0.80 kPa, at the position closest to the windbreak (2D/H), and 0.76 kPa, at the farthest position (23.3D/H).To simplify the analysis of the microclimatic data, we did not consider the eight main directions of origin of the wind, but we proceeded to combine the wind directions in three classes: upwind (north, northeast, northwest), parallel (west and east) and downwind (south, southeast, southwest). The correction coefficients (experimentally determined) of the wind speed, air temperature and relative humidity were obtained by taking into account the microclimate modifications caused by the presence of the windbreak. The correction coefficients of the wind speed and the temperature were calculated as a function of the wind direction and of the distance from the windbreak. The FAO56-wb model development refers to the calculation of the coefficients needed to modify the agro-meteorological parameters in the FAO56 model. The correction coefficients take into account the effects of the windbreak to estimate the ETwb . The validation was performed by comparing the estimated values of the ETwb (as calculated by the FAO56-wb model) with the observed values of the durum wheat and bean ET measured on a daily (ETm ) and seasonal (ET) scale.

The estimated values of the ETwb were obtained by introducing the following information into the FAO56-wb model: • Meteorological data recorded at the agro-meteorological station near the experimental farm; • FAO56 basal Kc (wheat: Kc ini = 0.15, Kc mid = 1.1, Kc late = 0.35; bean: Kc ini = 0.15, Kc mid = 1.1, Kc late = 0.2); • Correction coefficients for wind speed. In the validation test, the inputs to the FAO56-wb model remained unchanged except for the correction coefficient of the windbreak. The ETm values were obtained using the water balance method (Lhomme and Katerji, 1991) for any positions where the soil humidity was monitored at progressive distances from the windbreak barrier (2.7D/H; 4.7D/H; 6D/H; 7.3D/H; 10D/H; 12.7D/H; 18D/H; 23.3D/H): ETm = ±W + P − Dr

(2)

The simplified water balance (Eq. (2)) considers only three terms because experimentally the runoff and the capillary rise were shown to be negligible (Mastrorilli, 1999) at the Rutigliano site where the experiment was performed. In fact, the area is flat; the soil is superficial and lies on cracked rock that prevents a capillary rise from the deeper layers. The water supply (P) is represented by the sum of the daily measurements of precipitation and irrigation water. The drainage (Dr) is calculated as equal to the excess of water over the soil water capacity. The soil water content (W) was obtained by a Time Domain Reflectometry (TDR) method. Before sowing, the coaxial probes (0.3 m in length) were installed horizontally into the soil at a 0.3 m depth. It is the soil layer colonised by most of the root system. The vertical homogeneity of the soil’s hydrological characteristics allows to the moisture measured at a single depth in the site where this study is carried out, to be considered representative of the whole soil profile. Previous studies (Mastrorilli et al., 1998) demonstrated that the daily ET calculated through the soil water balance was equal to the ET measured by Bowen ratio methods.

156

P. Campi et al. / Agricultural Water Management 104 (2012) 153–162

The probes were linked to a TDR100 that was connected to a Campbell data logger CR10X. The soil dielectric constant (Ka), measured by the TDR instrument, was used as the input in the following polynomial equation to calculate the soil water content (W): W = −25.83 + 12.70Ka − 1.077Ka2 + 0.034Ka3

(3)

The equation was used in the same experimental site (Rutigliano) by Mastrorilli et al. (1998), and it represents the regression line between the Ka measured by the TDR and the W measured by a thermogravimetric method and corrected by the soil density. The seasonal water consumptions (ET) were measured at different distances from the windbreak using the sum of the daily values of the ETm . For the statistical validation, we used the methodology proposed by Loague and Green (1991). The Relative Root Mean Square Error (RRMSE) was calculated as follows:



RRMSE =

n (P i=1 i

n

− Oi )

2

·

100 ¯ O

(4)

where n is the number of observations, Pi is the ETwb value pre¯ is dicted by the FAO56-wb, Oi is the observed value (ETm ), and O the mean of the observed values. The RRMSE provides a percentage measurement of the difference between the simulated versus the observed data. The validation is considered to be excellent when the RRMSE is <10%, good if the RRMSE is between 10 and 20%, acceptable if the RRMSE is between 20 and 30%, and poor if >30% (Jamieson et al., 1991). The RRMSE test provides a value for the prediction model error rate by placing emphasis on a high level of errors. Other statistical indexes used here to evaluate the accordance between the measured and the simulated values are parameters a and b of the linear regression Pi = a + bOi, checking the null hypothesis H0 : a = 0, b = 1 when p < 0.05. Parameters a and b were obtained by applying the General Linear Model (SAS, 2001). The development of the FAO56-wb model is presented in the following steps: • Calculation of the relative agro-meteorological parameters for the wind speed, air temperature and relative humidity in the presence of the windbreak; • Parameterisation of the correction coefficients; • Formalisation of the FAO56-wb model.

The relative agro-meteorological parameters (R = 0: maximum variation, R = 1: no variation) were obtained using the data measured during the entire monitoring period (n), corresponding to the crop cycle of the durum wheat and beans. The data of the wind speed (u) and air temperature (t) were taken into account whereas the air relative humidity was not taken into account because it was not modified by the windbreak (Campi et al., 2009). Thus, the relative agro-meteorological parameters were defined only for wind speed (Ru) and air temperature (Rt). Because the meteorological data were measured in the surface equilibrated boundary layer, 0.15 m from the top of the canopy (Wieringa, 1993), the hourly values were reported at the standard height of 2 m from the soil surface, taking into account the wind profile. The local wind profile was calculated as (Rosemberg et al., 1983): ln(2 − d/z0 ) uz ln(z − d/z0 )

Rudir,D/H

n u i=1 i,D/H,dir =  n

(6)

u i=1 i,S,dir

The relative air temperature (Rt) was calculated using the same approach. In this case, only the hourly temperature values (ti ) from the upwind sector (up) were taken into account. The wind blowing from other directions (downwind and parallel) did not change the air temperature values (Campi et al., 2009): Rtup,D/H

 n  t i=1 i,D/H,up = n

(7)

t i=1 i,S,up

2.2. Parameterisation of the correction coefficients The correction coefficients (k) were obtained for the wind speed (u) and for the air temperature (t) using the relationships between Ru and Rt and the ratio between the distance from the windbreak (D) and the height of the windbreak (H). To calculate the correction coefficients for the wind speed, the linear regression was retained. Any correlation of a higher order than the first degree was excluded because these correlations do not significantly improve the relationship between both the correction coefficients and the D/H. In this way, the ‘General Linear Model’ and the analysis of residues (SAS, 2001) have shown that the linear regression (Ru vs D/H) is significant at a distance: ≤12D/H for the wind blowing from the North sector (Fig. 2a); ≤7.3D/H for the wind blowing from the South and East-West sectors (Fig. 2b). In accordance with the equation:

2.1. Relative agro-meteorological parameters

u2 =

of measurement above the ground surface (m), d the zero-plane displacement height (approximately d = 0.67hc , where hc is the crop height) (m), and z0 is the roughness length (approximately z0 = 0.1hc ) (m). The thermal profile was neglected because the air temperature measured from a standard agro-meteorological station (at a 2 m height) is similar to the air temperature measured above the crop canopy (Allen et al., 1998). The relative wind speed (Ru) was calculated for each wind sector (dir) as the ratio between the average values measured on an hourly time scale (ui ) at each position (D/H) and the values recorded at the agro-meteorological station (S) and not affected by the windbreak:

(5)

where u2 is the wind speed at a 2 m height (m s−1 ), uz the wind speed within the balanced boundary layer (m s−1 ), z the height

ku(dir) = a + b

D

(8)

H

where ku is the correction coefficient for the wind speed in each wind sector (dir); b is the angular coefficient experimentally calculated for each sector of the wind direction; a is the value of the intercept that can be assimilated to the ratio between the wind speed close to the barrier in the downwind area (u2,wb ) and the wind speed not affected by the windbreak, and therefore it is measured in the reference agro-meteorological station at the height of 2 m (u2,S ). The average wind speed takes into account the entire period of monitoring (n) in each wind sector (dir). Therefore, Eq. (8) can be rewritten as:

n D u2,wb,dir + b ku(dir) = i=1 n u i=1 2,S,dir

H

(9)

For the downwind and parallel sectors, the speed reduction occurs with the same entity and in the same area protected by the windbreak (<7.3D/H). Therefore, we cannot consider a single correction coefficient that was obtained from the linear regression of the wind speed data coming from both those sectors (Fig. 2b).

P. Campi et al. / Agricultural Water Management 104 (2012) 153–162

157

correction coefficient for the air temperature was obtained only for the upwind sector (kt(up) ) that affects the air temperature: kt(up) = a + b

D

(10)

H

where a is the value of the intercept that represents the increase of temperature that occurs near the windbreak; b is the slope experimentally drawn for each sector of the wind direction. In this experiment, b = −0.02. If the wind blows from a different direction, kt is 1. The value of the intercept (a) can be obtained using the ratio between the value of the average air temperature measured at a distance <1D/H (t2,wb ) and the air temperature not affected by the barrier, which can be assimilated with the average temperature measured in the reference agro-meteorological station at the height of 2 m above the soil (t2,S ), taking into account the entire monitoring period (n) in the upwind sector (up):

n D t2,wb,up +b kt(up) = i=1 n t i=1 2,S,up

H

(11)

2.3. Formalisation of FAO56-wb model

Fig. 2. Relationship between the relative wind speed (Ru), from upwind (a) downwind and parallel (b) directions, and the ratio between distance from windbreak (D) and height of the windbreak (H).

In conclusion: • a = 0.31, b = 0.054, for the upwind sector; • a = 0.69, b = 0.042, for the downwind and parallel sectors. The correction coefficient to be applied to the air temperature was obtained using the same approach used for ku(dir) . The ‘General Linear Model’ and the analysis of the residues (SAS, 2001) have shown that the linear regression of the relative temperature is significant at a distance less than 4.7D/H (Fig. 3). The

In the ‘two-step’ approach, the ET0 is calculated with the resistive standard parameters of the reference meadow (rc = 70 s m−1 ; ra = 208/u2 ) but does not take into account the effects of the windbreaks on the wind speed and temperature that influence the deficit of the air vapour pressure. Corrections are also required for dual crop coefficient (Kcdual ), which consists of splitting Kc into two separate coefficients, one for crop transpiration, the basal crop coefficient (Kcb), and one for soil evaporation (Ke). The FAO Kcdual uses typical values expected for an average Kc under a standard climatic condition, which is defined as a sub-humid climate with an average daytime minimum relative humidity (RHmin) of 45% and calm to moderate wind speeds averaging 2 m s−1 . Therefore, to estimate the ETwb , according to the FAO56-wb model, the correction coefficients ku(dir) and kt(up) were introduced into the two steps, corresponding with the determination of the ET0 and the Kcdual . 3. Results The results of the research are presented in 4 steps, as follows: 1. Validation of the model; 2. Sensitivity analysis; 3. The effect of the windbreak on the agro-meteorological parameters; 4. Simulations: • Seasonal ET in 25-year periods; • Dimensioning of the barrier. 3.1. Validation of the model

Fig. 3. Relationship between the relative temperature (Rt), from upwind direction, and the ratio between distance from windbreak (D) and height of the windbreak (H).

Table 1 indicates the statistics (R2 , a, b, RRSME) obtained by the comparison between the daily values of ETwb , estimated through the FAO56-wb model, and the values observed independently through the ‘water balance’ for the 2 crops (durum wheat and beans) and for each distance from the windbreak during the crop development and the mid-season stages (Allen et al., 1998). In particular, taking into account the following, we deem that the FAO56-wb model can correctly estimate the ETwb :

158

P. Campi et al. / Agricultural Water Management 104 (2012) 153–162

Table 1 Statistics (R2 , a, b, RRSME) of the comparison between the values of daily ETwb estimated by the FAO56-wb and the ET values measured through the ‘water balance’ at different distances from the windbreak. H

Wheat R2 a b RRSME Bean R2 a b RRSME * ** ***

2.7

4.7

6

7.3

10

12.7

18

23.3

0.85 0.35*** 0.87** 14.5

0.83 0.30*** 0.91* 15.3

0.87 0.37*** 0.92** 15.9

0.88 0.49*** 0.83** 14.1

0.86 0.31*** 0.89*** 13.7

0.87 0.38*** 0.93** 16.3

0.89 0.32*** 0.85** 12.9

0.87 0.40*** 0.87** 13.8

0.93 0.77*** 1.12** 14.0

0.92 0.75*** 1.17** 15.6

0.92 0.8*** 1.14** 15.4

0.9 0.65*** 1.11* 16.7

0.95 0.57*** 1.09* 12.7

0.96 0.65*** 1.12** 12.3

0.93 0.85*** 1.15** 15.7

0.94 0.84*** 1.15** 14.3

The level of significance respectively at 0.05, according to the General Linear Model. The level of significance respectively at 0.01, according to the General Linear Model. The level of significance respectively at 0.001, according to the General Linear Model.

• The values of R2 (from 0.82 to 0.85 for durum wheat and from 0.91 to 0.94 for beans) indicate a high level of correlation between the observed and the measured data; • The parameters a and b of the linear regression Pi = a + bOi are statistically significant (p < 0.05); • The RRMSE test indicates a good level of acceptance of errors from the model (RRMSE varies between 10 and 20%). The statistical analysis performed on the comparison of the seasonal values of ET for both crops (Fig. 4) indicates a good degree of correlation (R2 = 0.9), with significant parameters a and b (p < 0.05).

The low values of the RRSME (7.6% for durum wheat and 7.8% for beans) on a seasonal time scale confirm an excellent relationship. The slope values (b) show slight differences in the ET, namely, in wheat, b is underestimated (b < 1), and in beans, b is overestimated (b > 1). The site-specific Kc observed during the mid-season (calculated as the ratio of the ETc measured in 23D/H position and the ET0 calculated by the Penman–Monteith method) confirms that the dual FAO coefficient during mid-season, adjusted by local climatic conditions in presence of windbreak (Kcdual(wb) ) and used in the estimation of the ETwb , is lower for wheat (FAO Kcdual(wb) = 1.08, observed Kc = 1.14) and higher for beans (FAO Kcdual(wb) = 1.11, observed Kc = 1.05). It was shown that the slope values reported in Table 1 were calculated using the daily ET during the mid-season whereas the estimated season values of ET were slightly underestimated for beans if the values of Kcdual(wb) are introduced in the FAO56-wb model. 3.2. Sensitivity analysis The sensitivity analysis of the FAO56-wb model, performed according to the approach presented by Saltelli et al. (2004) showed that the kt did not affect ET; and therefore, the formalisation of the FAO56-wb model considers only the reduction of wind speed. The correction coefficients of wind speed (ku(dir) ) were introduced to modify the values of the daily average wind speed (u2 ) required to calculate the reference evapotranspiration (ET0(wb) ): ETo(wb) =

0.408··(Rn −G)+(900/(T +273))u2 ku(dir) (es −ea ) +(1+0.34u2 ku(dir) )

(12)

For Kcdual(wb) , both basal crop coefficient (Kcb(wb) ) and soil evaporation coefficient (Ke(wb) ) are modified for taking into account the effect of windbreaks:

 0.3 Kcb(wb) = Kcbtab + [0.04(u2 ku(dir) − 2) − 0.004(RHmin − 45)] ·



Fig. 4. Comparison between the seasonal values of



ET, for durum wheat (a) and

ET measured with the water balbean (b) estimated by the FAO56-wb model, and ance model. D is the distance from windbreak (m) and H is the height of windbreaks (m).

hc 3

(13)

The daily Kcb value was determined by assuming Kcb to be constant during the initial and mid-season stages and assuming a linear relationship between the Kcb value at the end of the previous stage and the Kcb value at the beginning of the next stage during the crop development and late season stages. The soil evaporation coefficient, Ke, describes the evaporation component of ETc. Where the topsoil is wet, following rain or irrigation, Ke is maximal and can never exceed a maximum value (Kc max.) When the topsoil dries out, less water is available for evaporation and a reduction in evaporation begins to occur in

P. Campi et al. / Agricultural Water Management 104 (2012) 153–162

159

proportion to the amount of water remaining in the surface soil layer, or: Ke(wb) = Kr(Kc max(wb) − Kcb(wb) ) ≤ few Kc max(wb)

(14)

where Kc max(wb) is maximum value of Kc following rain or irrigation in presence of windbreak: Kcmax(wb) = max −45)] ·

h

c



0.3 

3

1.2 + [0.04(u2 ku(dir) − 2) − 0.004(RHmin



, (Kcb(wb) + 0.05)

(15)

Kr is dimensionless evaporation reduction coefficient dependent on the cumulative depth of water depleted (evaporated) from the topsoil. Soil evaporation from the exposed soil is presumed to take place in two stages (Ritchie, 1972; Doorenbos and Kassam, 1979): an energy limiting stage (stage 1) and a falling rate stage (stage 2). During stage 1, following rain or irrigation Kr is 1, and evaporation is only determined by the energy available for evaporation. When the soil surface dries, Kr becomes less than one (sage 2) and evaporation is reduced. Kr becomes zero when no water is left for evaporation in the upper soil layer. Kr is calculated as follows: Kr =

TEW − De,i−1 TEW − REW

for De,i−1 > REW

(16)

where TEW (total evaporable water) is the maximum depth of water (mm) that can be evaporated from the soil. The topsoil has been initially completely wetted, with TEW = 1000( FC − 0.5 WP ),  FC is the soil water content at field capacity and  WP is the soil water content at the wilting point, REW (readily evaporable water) is the cumulative depth of evaporation (mm) at the end of stage 1. The threshold REW is dependent on the physical properties of the soil. The threshold value of 10 mm is adopted here for the study, De,i−1 is the cumulative depth of evaporation (depletion) from the soil surface layer at the end of day (the previous day) (mm), and few is the fraction of the soil that is both exposed and wetted. These values are derived by the maximum value the fraction of soil surface covered by vegetation (0.95 for wheat and 0.80 for bean) and by the fraction of soil surface wetted by irrigation (0.3 for bean) or precipitation (1.0 for both the crops). The ETc affected by the windbreak action (ETwb ) can be estimated as: ETwb = ET0(wb) · (Kcb(wb) + Ke(wb) )

Fig. 5. Relative wind speed (uw /u0 , where uw is the wind speed at 2 m in height influenced by windbreaks, and u0 is wind speed not influenced by windbreaks) as a function of the ratio between distance from windbreak (D) and height of the windbreak (H). The relationships between relative wind speed and D/H are represented as function of wind directions during the experimental period (whole cycle of wheat and bean crops).

(17)

The calculation carried out using spreadsheet programs described by FAO Allen et al. (1998), which define steps for calculation of different components of water balance. 3.3. The effect of a windbreak on the agro-meteorological parameters

Fig. 6. Relative air temperature (tu /t0 , where tu is the air temperature inside the equilibrium boundary layer and t0 is air temperature not influenced by windbreaks) as a function of the ratio between distance from windbreak (D) and height of the windbreak (H) during the whole cycle of wheat and bean crops (upwind direction).

The reduction in ET is due to the alteration of climatic in the area affect by the windbreak. The production of wheat was significantly higher in the tested area than in the area not affected by the windbreak (Campi et al., 2009). Out of the protected area the soil humidity decreases at the end of the wheat cycle, after the flowering (Fig. 7).

The microclimate analysis has been reported in detail in a prevision paper (Campi et al., 2009). The following is evident for both crops: • The wind speed is mitigated at a barrier distance that varies with the wind direction. In particular, when the wind blows from the upwind sector (N, NE, NW), the windbreaks produce a reduction in the wind speed up to a maximum of 60% for a distance less than 12.7D/H; however, when the winds blow from the parallel sectors (E, W) and downwind (S, SE, SW), a maximum reduction (25%) occurs for a distance of 7.3D/H (Fig. 5). • The air temperature increases up to 1.2 ◦ C for a distance lower than 5D/H only when the prevailing wind blows from upwind sector (Fig. 6). • The relative humidity is not affected by the windbreak.

Fig. 7. Trend of soil water content measured by the TDR method at 23.3D/H during the wheat season 2005 (sowing date is 20/10/2004).

160

P. Campi et al. / Agricultural Water Management 104 (2012) 153–162 Table 3 Reduction (%) of the ETwb in wheat and bean crops (ETwb simulated at the distance 7.3D/H respect to ETwb simulated at 23.3D/H). The air maximum temperature (tmax ) and the wind speed (u) are also reported (mean values measured during the crop season). Years

Fig. 8. Simulations of the windbreak on the wheat (a) and bean (b) ET in a Mediterranean region in the period 1984–2008. D is the distance from windbreak (m) and H is the height of windbreaks (m).

3.4. Simulations 3.4.1. Seasonal ET in 25-year periods Using the FAO56-wb model, the effect of the windbreak on the seasonal ET of durum wheat and beans over a 25-year period was simulated (Fig. 8). For the simulation, the basal crop coefficient (Kcb) indicated in the FAO56 handbook was taken into account while the lengths of the crop development stages are typical for the Mediterranean region (Table 2). The depletion fraction (p) and the maximum crop height (hc ) for wheat and beans are reported in the FAO handbook (Table 2). The simulations (Table 3) showed that, in the 1984–2008 period, the windbreak would have caused an average reduction of the ETwb in the durum wheat crop by 60 mm in the vicinity of the barrier (D/H < 7.3), and the average reduction of the ETwb in the bean crop is 60 mm as well. In particular, in 2001 for the durum wheat and in 2005 for the beans, the reduction of the ET in the field area close to the windbreak would have reached 98 mm and 102 mm, respectively. No variations in the ETwb occurred during the crop cycle of the durum wheat in 1996 and during that of the beans in 1995. The action of the windbreak on the ET is influenced by the interannual variability Table 2 Lengths of the crops development stages (day), depletion fractions (p) and crops height (hc , in m) used for calculating ETwb . Crop

ini

dev

mid

late

p

hc

Bean Wheat

15 45

35 105

20 50

15 27

0.4 0.5

0.6 0.7

Wheat

Bean ◦

−1

ETwb (%)

tmax (◦ C)

u (m s−1 )

2.9 2.7 2.7 2.9 2.7 2.7 2.8 3.1 3.0 2.7 2.8 2.6 2.4 2.6 3.0 2.7 3.3 3.0 2.8 2.9 2.5 3.1 2.5 2.9 2.6

15.6 20.8 19.9 16.8 17.8 12.6 20.1 19.6 15.5 13.3 12.1 8.0 14.0 9.0 14.3 15.6 22.5 20.6 17.1 17.3 14.8 22.8 14.5 16.0 12.0

29.0 31.0 30.5 30.0 30.2 28.1 30.3 29.6 29.3 28.1 28.9 27.8 28.2 28.1 30.0 29.2 32.0 32.0 29.0 30.2 29.2 32.4 29.2 30.1 28.8

2.5 3 2.7 2.5 2.5 2.5 2.9 2.8 2.3 2.5 2.5 2 2.6 2.1 2.4 2.4 2.8 2.9 2.5 2.6 2.5 2.9 2.4 2.6 2.4

2.8

16.1

29.6

2.6

ETwb (%)

tmax ( C)

u (m s

1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

13.7 14.8 15.3 15.0 14.7 12.6 14.9 17.1 14.1 17.2 15.9 12.6 11.1 13.3 17.2 14.0 19.2 21.3 15.9 17.9 12.6 19.5 10.5 17.1 15.2

17.3 16.4 17.2 18.2 17.6 17.1 18.6 18.6 17.0 18.7 18.4 17.0 15.6 16.8 18.5 17.9 20.6 20.3 18.5 18.7 17.1 18.8 15.8 19.3 18.8

Average

15.3

18.0

)

of the wind speed and temperature (Table 3). In the presence of the windbreak, the reduction in the ET is more significant during the windier and hotter seasons. 3.4.2. Dimensioning of the barrier When the wind blows from the upwind direction, the wind speed in the downwind direction of the windbreak is due to the air flow that passes beyond the barrier and is characterised by a given porosity (ˇ). It should be noted that the porosity of the windbreak does not modify the extension of the protected area; it determines only the wind speed reduction (Heisler and De Valle, 1988; Cleugh and Hughes, 2002). The porosity is proportional to the windbreak density (1 − ˇ). For example, a porosity of 40% reduces the wind speed by 60% near the windbreak belt (Heisler and De Valle, 1988). Cleugh and Hughes (2002) provide a more accurate relationship between the wind speed reduction and the porosity: umin = 1.14ˇ − 0.16 uw

(18)

where umin is the minimum speed that occurs in a downwind position, and uw is the average of the speed in the upwind area. This relationship was confirmed by the experimental data from this study. In fact, through Eq. (15) we have: 1.14(0.4) − 0.16 = 0.3. The value 0.3 represents the intercept in Eq. (8) (Fig. 2a). If the porosity is taken into account, Eq. (9) can be rewritten as follows: ku(up) = 1.14ˇ − 0.16 +

b(1 − ˇ) 0.6

D H

(19)

where the denominator of 0.6 was introduced to take into account that the slope (b) of Eq. (9) which was experimentally obtained from a windbreak having a density equal to 0.6 (derived by 1 − 0.4). When the wind blows from the downwind direction, the porosity is negligible (Cleugh and Hughes, 2002). Therefore, the correction coefficient of the wind speed should be calculated as in Eq. (9).

P. Campi et al. / Agricultural Water Management 104 (2012) 153–162

161

The use of complex crop models requires experimental data to calibrate the model. These agro-climatic and productive data describing the effects of the windbreak are difficult to find. Instead, the approach suggested in this study can be validated only through the measurement of the soil humidity at different distances from the windbreak. These measures are required for the estimation of the ET through the soil water balance. The modifications made by the windbreak were included in the FAO56 model using simple correction coefficients of the wind speed. In fact, to measure these coefficients, only the wind speed measurement is required at the distance of 3D/H, where the mitigating effect of the windbreak is maximised (Burke, 1998; Nelmes et al., 2001; Cleugh et al., 2002; Brandle et al., 2004). The knowledge of the ETwb at the seasonal level is indispensable in the farming design to predict the technical benefit, in terms of water savings, that can be derived by adopting the use of the windbreak. The simulation results showed that the water savings strongly depend on the barrier porosity. With a proper porosity (20%), a 20–30% reduction in the ETwb can be obtained, which is not negligible in a hot, arid environment characterised by limited water resources. Within a distance of 45 m from a 3 m barrier with a porosity of 20%, 130 mm of seasonal ET can be saved in the case of durum wheat and 140 mm in the case of beans. 5. Conclusions Fig. 9. Simulations of the effect of windbreak, with different porosity, on seasonal  ETwb ) of wheat (2001 season) (a) and bean (2005 season) evapotranspiration ( (b). D is the distance from windbreak (m) and H is the height of windbreaks (m).

By introducing these modifications in the FAO56-wb model, the seasonal ETwb at different porosities (20, 40, 80%) of the windbreak was simulated (Fig. 9) for two crops, durum wheat and beans, in the years 2001 and 2005, respectively, when the effect of the windbreak was the highest. The simulation showed that through a low porosity barrier (20%), the effect on the ET is more remarkable. In particular, the maximum reduction of the ETwb is 20% for durum wheat and 31% for beans. 4. Discussion The modifications of the microclimatic conditions and the relevant effects on the ET of crops protected by windbreaks can only be identified through complex experimental devices that require the installation of agro-meteorological equipment (anemometer, wind direction and thermo-hygrometer) or batteries of lysimeters or the use of micrometeorological techniques at different distances from the windbreak, if the fetch distance from the windbreak is enough to make such methodologies reliable. The relevant effects depend on the temporal and spatial variability of the climate (Carberry et al., 2002). The calculation of the ETc in cases with windbreaks was performed using various crop models, such as SCAM (Cleugh, 2002), APSIM (Carberry et al., 2002), CERESMaize and CROPGRO-Soybean (Brandle et al., 2004). Of course, these crop models, in addition to the ETwb estimate, also simulate other growth and productive parameters of the crop. However, some models can often be used to assess the likely effects of windbreaks on crops in a range of locations throughout the experimental area (Cleugh, 2002). For example, the APSIM approach was used throughout Australia (Carberry et al., 2002). However, to indicate only the effect of the barriers on the ETwb , a simplified approach is essential to allow such applications to be used with the available long-term meteorological data (Meinke et al., 2002).

The experimental results showed that agro-meteorological data that can be easily measured can provide some correction coefficients for the ETc used in the FAO56 ‘two step’ approach, which is a universally recognised ETc calculation method (Allen et al., 1998) The simulations performed using the FAO56-wb model could quantify the potential of a windbreak to contain the ET of durum wheat and beans in a hot, arid environment, such as the Mediterranean. The FAO56-wb model can be easily used in various environments and for different crops. In fact, its application requires few inputs: • Crop coefficient, obtained locally or tabulated in the paper n.56 FAO and corrected by the windbreak effects on the wind speed; • Agro-meteorological data that can be easily measured, namely, the wind speed near the windbreak (≤3D/H) and the wind speed, wind direction, air temperature, relative humidity, and global radiation measured at the agro-meteorological stations; • Windbreak porosity, for which it is important to assess the uniformity of the porosity because any opening causes a reduction in the protective effect (Rosemberg et al., 1983; Sudmeyer and Scott, 2002). Among the various models used to estimate ET, the FAO56-wb model is a useful tool to evaluate the windbreak effect on a crop’s water consumption and, at the same time, to provide an accurate water schedule for crops protected by a windbreak. References Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop evapotranspiration: guidelines for computing crop water requirements. Irrigation and Drainage Paper No. 56. FAO, Rome, 300 pp. Ben Salah, H., Beji, M.A., Salah, H.B., 1989. Effects of windbreak protection on faba bean production. Annales de l’Institut National de la Recherche Agronomique de Tunisie. Numero Speciale, 125–137. Benzarti, J., 1990. Effects of windbreack and speciality on microclimate and agricultural production in Tunisian irrigated lands. In: Xiang, K., Shi, J., Baer, N.W., Sturrock, J.W. (Eds.), Protective Plantation Technology. Northeast Forestry University, Harbin, pp. 218–224.

162

P. Campi et al. / Agricultural Water Management 104 (2012) 153–162

Brandle, J.R., Hodges, L., Zhou, X.H., 2004. Windbreaks in North American agricultural systems. Agroforest. Syst. 61, 65–78. Burke, S., 1998. Windbreaks. Inkata Press, 129 pp. Campi, P., Colucci, R., Mastrorilli, M., 2005. Andamenti meteorologici stagionali e gestione irrigua. Atti 36◦ Convegno della SIA: “Ricerca erd innovazione per le produzioni vegetali e la gestione delle risorse agro-ambientali”, FG 20-22 Settembre 2005, 113–114. Campi, P., Palumbo, A.D., Mastrorilli, M., 2009. Effects of tree windbreak on microclimate and wheat productivity in a Mediterranean environment. Eur. J. Agron. 30, 220–227. Carberry, P.S., Meinke, H., Poulton, P.L., Hargreaves, J.N.G., Snell, A.J., Sudmeyer, R.A., 2002. Modelling crop growth and yield under the enviromental changes induced by windbreaks. 2. Simulation of potential benefits at selected sites in Australia. Aust. J. Exp. Agric. 42 (6), 887–990. Casa, R., Valentini, R., Scarascia Mugnozza, G., 1994. Influenza di barriere arboree frangivento sul microclima e sul bilancio energetico di colture retrostanti in ambiente mediterraneo. Monti e Boschi 5, 41–47. Cleugh, H.A., Hughes, D.E., 2002. Impact of shelter on crop microclimates: a synthesis of result from wind tunnel and filed experiments. Aust. J. Exp. Agric. 42, 679–701. Cleugh, H.A., 2002. Parameterising the impact of shelter on crop microclimates and evaporation fluxes. Aust. J. Exp. Agric. 42, 859–874. Cleugh, H.A., Prinsley, R., Bird, P.R., Brooks, S.J., Carberry, P.S., Crawford, M.C., Jackson, T.T., Meinke, H., Mylius, S.J., Nuberg, I.K., Sudmeyer, R.A., Wright, A.J., 2002. The Australian National Windbreaks Program: overview and summary of results. Aust. J. Exp. Agric. 42, 649–664. Doorenbos, J., Kassam, A.H., 1979. Yield response to water. FAO Irrigation and Drainage Paper No. 33. FAO, Rome, Italy. Er-Raki, S., Chehbouni, A., Guemouria, N., Duchemin, B., Ezzahar, J., Hadria, R., 2006. Combining FAO-56 model and ground-based remote sensing to estimate water consumptions of wheat crops in a semi-arid region. Agric. Water Manage. 87, 41–54. Heisler, G.M., De Valle, D.R., 1988. Effect of windbreak structure on wind flow. Agric. Ecosyst. Environ. 22/23, 41–69. Jamieson, P.D., Porter, J.R., Wilson, D.R., 1991. A test of the computer simulation model ARCWHEAT1 on wheat crops grown in New Zealand. Field Crop Res. 27, 337–350. Jarvis, P.G., 1976. The interpetration of the variation in leaf water potential and stomatal conductance found in canopies. Philos. Trans. R. Soc. B 273, 593–610. Ko, J., Piccinni, G., Marek, T., Howell, T., 2009. Determination of growth-stage-specific crop coefficients (Kc) of cotton and wheat. Agric. Water Manage. 96, 1691–1697. Kort, J., 1988. Benefits of windbreaks to field and forage crops. Agric. Ecosyst. Environ. 22/23, 165–190. Liu, Y., Luo, Y., 2010. A consolidated evaluation of the FAO-56 dual crop coefficient approach using the lysimeter data in the North China Plain. Agric. Water Manage. 97, 31–40. Loague, K.M., Green, R.E., 1991. Statistical and graphical methods for evaluating solute transport models: overview and application. J. Contam. Hydrol. 7, 51–73.

Lhomme, J.P., Katerji, N., 1991. A simple modelling of crop water balance for agrometeorological application. Ecol. Model. 57, 11–25. Luis, P.P., Bloomberg, M., 2002. Windbreaks in southern Patagonia, Argentina: a review of research on growth models, wind speed reduction, and effects on crops. Agroforest. Syst. 56, 129–144. Mastrorilli, M., Katerji, N., Rana, G., Ben Nouna, B., 1998. Daily actual evapotranspiration measured with TDR technique in Mediterranean conditions. Agric. For. Meteorol. 90, 81–96. Mastrorilli, M., 1999. Sviluppo di modelli idrologici per ambienti mediterranei. Bollettino SISS 48, 245–250. Meinke, H., Carberry, P.S., Cleugh, H.A., Poulton, P.L., Hargreaves, J.N.G., 2002. Modelling crop growth and yield under the environment changes induced by windbreaks. 1. Model development and validation. Aust. J. Exp. Agric. 42, 875–885. Monteith, J.L., 1973. Principles of Environmental Physics. Edward Arnold, London, UK, p. 241. Nelmes, S., Belcher, R.E., Wood, C.J., 2001. A method for routine characterization of shelterbelts. Agric. For. Meteorol. 106, 303–315. Nuberg, I.K., 1998. Effect of shelter on temperature crops: a review to define research for Australian conditions. Agroforest Syst. 41, 3–34. Rana, G., Katerji, N., Mastrorilli, M., El Moujabber, M., 1994. Evapotranspiration and canopy resistance of grass in a Mediterranean region. Theor. Appl. Climatol. 50, 61–71. Rana, G., Katerji, N., 2009. Operational model for direct determination of evapotranspiration for well watered crops in Mediterranean region. Theor. Appl. Climatol. 97 (3), 243–253. Ritchie, J.T., 1972. Model for predicting evaporation from a crop with incomplete cover. Water Resour. Res. 8, 1204–1213. Rosemberg, N.J., Blad, B.L., Verma, S.B., 1983. Microclimate, The Biological Environment. John Wiley & Sons, New York. Saltelli, A., Tarantola, S., Campolongo, F., Ratto, M., 2004. Sensitivity Analysis in Practice. Wiley, New York, NY, USA, 219 pp. SAS, 2001. SAS/STAT Software Release 8.2. SAS Institute Inc., USA. Shuttleworth, W.J., 2006. Towards one-step estimation of crop water requirements. Trans. ASABE 49 (4), 925–935. Shuttleworth, W.J., Wallace, J.S., 2009. Calculating the water requirements of irrigated crops in Australia using the Matt–Shuttleworth approach. Trans. ASABE 52 (6), 1895–1906. Sturrock, J.W., 1984. The role of shelter in irrigation and water use. In: Sturrock, J.W. (Ed.), Shelter Research Needs in Relation to Primary Production: The Report of the National Shelter Working Party Ministry of Works and Development, vol. 94. Water and Soil Misc, Wellington, New Zealand, pp. 79–86. Sudmeyer, R.A., Scott, P.R., 2002. Characterisation of a windbreak system on the south coast of Western Australia characterisation of a windbreak system on the south coast of Western Australia. 2. Crop growth. Aust. J. Exp. Agric. 42, 712–727. Wieringa, J., 1993. Representative roughness parameters for homogeneous terrain. Bound.-Layer Meteorol. 63, 323–363.