![Free access]({"type":"image" , "src":"/pb-assets/3rdPartyLogos/FREE_ACCESS_ICON-1465051266437.png"}){kind=icon}
![ORCID Icon]({"type":"image","src":"/templates/jsp/images/orcid.png"} "ORCID Icon")
http://orcid.org/0000-0002-2241-9334![CrossMark Logo]({"type":"image" , "src":"/templates/jsp/images/CROSSMARK_Color_horizontal.svg"})
ABSTRACT
ABSTRACT
Eddy correlation measurements within the Nile Delta allowed the determination of evapotranspiration (E) for seven crops (rice, maize, cotton, sugar beet, berseem, wheat and fava beans) using basin irrigation (BI), furrow irrigation (FI), BI with increased intervals (BIi), FI with increased intervals (FIi), strip irrigation (SI) and drip irrigation (DI). Total E values over the cropping season for rice (BI, BIi) were the highest (>600 mm), while those for sugar beet (DI), maize (SI and DI) and berseem (BIi) were the lowest (<250 mm). The differences were due to a combination of atmospheric demand, soil moisture, the presence of surface standing water, root depth, and the length and timing of the cropping season. The DI and SI methods had the advantage for water saving, while the FIi and BIi methods were effective for crops with shallow root lengths. Estimated annual E was 566–828 mm/year (water-saving irrigation) and 875–1225 mm/year (conventional irrigation).
1 Introduction
Agriculture is the largest water consumer in many parts of the world (e.g. Turner et al. 2004). Although a certain portion of water withdrawn for agriculture is returned as surface water or groundwater, much is consumed by evapotranspiration. As such, efforts have been made to determine accurate evapotranspiration for various crops under different growing conditions. However, to date, evapotranspiration measurements for crop fields have largely been determined using the soil water balance method or the lysimeter method. In comparison to other methods, micrometeorological approaches such as the eddy correlation method have been used infrequently (Zwart and Bastiaanssen 2004, Farahani et al. 2007, Sugita et al. 2014, 2015), although the eddy correlation method is currently considered to be the most accurate method with the largest time resolution (Foken 2008) provided that careful attention is given to measurements and data processing. Also, to date, less attention has been paid to hydrological processes related to evapotranspiration within crop fields as compared to other types of surfaces. Such a lack of attention to crops has likely resulted because crop evapotranspiration has largely been addressed in agronomy subdisciplines such as crop science, agricultural meteorology, or agricultural engineering, where the main concern is generally crop yield and where evapotranspiration is regarded as a factor affecting crop yield. Another reason for the lack of investigation could be related to the mismatch of time scales for hydrological processes (often hours to 100 d) as compared to those for evapotranspiration, generally determined using the soil water balance approach (101 d) which makes it difficult to assess hydrological processes in relation to evapotranspiration.
For this study, we surveyed an area of the Nile Delta in Egypt (see Section 2.1), where crop fields occupy at least 70% of the area, agriculture water use accounted for 82% of total water use from 2011 to 2012 (Central Agency for Public Mobilization and Statistics 2014), and the evapotranspiration of various crops has been determined either by applying an estimation formula or by considering the water balance of a crop field or a lysimeter but not micrometeorological methods (see Table 1 for previous studies within the Nile Delta; also see Rana and Katerji 2000 for a review of evapotranspiration measurements in Mediterranean climate areas). For example, El-Shal (1966) applied the soil water balance method in order to estimate the evapotranspiration of various crops. Swelam et al. (2010) reported evapotranspiration for wheat using a weighing lysimeter installed in a crop field. In these studies, hydrological processes were not studied for the purpose of interpreting derived amounts of evapotranspiration. Also in these studies, available evapotranspiration was obtained using the conventional irrigation method. In today’s world, crop evapotranspiration when using newer irrigation methods, such as drip irrigation, is often needed for future water use planning and management, so updating and improving our knowledge of crop evapotranspiration based on micrometeorological methods, together with measurements of hydrological processes using various irrigation methods, is desirable and even necessary.
Table 1. Previous studies on crop evapotranspiration based on field scale measurements in the Nile Delta.
Due to the lack of micrometeorological measurements for crop evapotranspiration under various irrigation methods in the Nile Delta, we performed this study with the primary objectives of determining daily mean and total evapotranspiration (E) values for major crops cultivated using various irrigation methods by means of the eddy correlation method. Secondly, using observations and analyses of hydrological processes within crop fields, we clarified factors that cause E differences for various crops and irrigation methods. Thus, the motivation of our study was largely scientific. However, the outcome of this study should have a wide range of practical applications. For example, since they provide information on when, where and how water is consumed by evapotranspiration or lost to groundwater, investigations of hydrological processes are useful not only for evapotranspiration analyses but also for improving irrigation design in order to reduce water use. In a similar manner, based on precise estimates for the water consumption required for a particular crop or a particular irrigation method, actual water use in agriculture relative to water rights (often based on historic consumption) should become more clear, leading to more equitable use of water amongst competitors (e.g. Rice and White 1997).
2 Study area and observation sites
2.1 Nile Delta
The Nile Delta is located in an area of arid climate and has mean annual precipitation ranging from approximately 200 mm/year near the Mediterranean coast, rapidly decreasing inland, to 25 mm/year in Cairo, which is located on the southern edge of the delta (e.g. Griffiths 1972). The climate is quite uniform over most of the delta, except for areas near seas and deserts. Year-to-year changes in climate are small and, in general, quite stable due to its location within the sub-tropical high-pressure belt (Griffiths 1972; also see Supplementary material, Section 1).
According to Shalaby and Moghanm (2015), with the exception of the western and eastern desert fringes and areas along the coast, soils within the Nile Delta can be classified as Entisols (Vertic Torrifluvents). As determined by combining images from four satellites (Fujihira 2014), the area containing the same soil type corresponds to the distribution of croplands within the Nile Delta. Thus, not only climate but soils in croplands within the Nile Delta can generally be assumed to be uniform with the exception of boundary zones between the delta and surrounding deserts and seas.
2.2 Experimental fields
To obtain detailed evapotranspiration data, three level crop fields (Sakha-A, Sakha-B and Zankalon), with a size of 200 m × 200 m, were established at two locations. The Sakha-A field (31°5′54.70″N; 30°55′21.00″E) was located immediately north of the Sakha-B field (31°5′47.60″N; 30°55′21.20″E) and they are part of the experimental field of the Agricultural Research Center near the city of Kafr El-Shaikh, located in the central delta. The Zankalon field (30°34′50.04″N; 31°25′59.94″E), near the city of Zagazig in the southeastern portion of the delta, belongs to the Water Management Research Institute. All of the fields were located within agricultural areas that continuously extended at least 2 km (Sakha) and 0.8 km (Zankalon) in the dominant, northwest, wind direction.
Typical soil properties for Sakha were reported by Orii (2012) and Kubota et al. (2015); and the properties for Zankalon were reported by A. Kubtota (personal communication, 2014). Briefly, the clay content is approximately 50% throughout the soil profile, as deep as 1 m. The bulk density is high and is in the range of 1.4–1.7 g/cm3. These variables are approximately the same as for the crop fields of surrounding areas.
3 Methods
3.1 Crops, irrigation methods and drainage system
Selection of the crops and irrigation methods used for our experiment, which began in the summer of 2010 and continued through the 2014 cropping season, is summarized in Table 2. Three major summer crops – rice, maize and cotton – and four major winter crops – wheat, berseem (Trifolium alexandrinum), fava beans and sugar beet – were chosen for our study based on their relevance in terms of cultivation area and historical significance (e.g. Brown 1955) in Egypt. These crops occupied 55% and 83% of the cropland in the Nile Delta during the summer and winter, respectively, of 2012 (see Fig. S1 in the Supplementary material, Section 8 for annual changes in the cultivation areas for summer and winter crops).
Table 2. Crop (first row) and irrigation method (second row) selected for each cropping season and field, with the cropping period in the third row. Underlined irrigation amounts include estimated values.
Surface irrigation and drip irrigation (DI) were employed. Surface irrigation methods included (a) furrow irrigation (FI), (b) basin irrigation (BI), (c) strip irrigation (SI), (d) FI with increased irrigation intervals (FIi) and (e) BI with increased irrigation intervals (BIi). The FI and BI methods are conventional and are currently in use in the Nile Delta (see e.g. Strelkoff et al. 1999), while the other methods were new to the area and were being tested for their capacity to save water (El-Kilani and Sugita 2017). With the exception of the DI method, the field was divided into two sections, each encompassed by a dyke. The design for each method, including the dimensions of planting beds and furrows, laterals and emitters, is provided in Table S1 in the Supplementary material, Section 5.
The applied amount of irrigation water is provided in Table 2, and was based on an experimental design (Supplementary material, Table S2) determined from Egyptian standards for each crop and adjusted for the specific requirements of the experiments (Maruyama, personal comm. 2015; Maruyama 2017). For actual irrigation implementation, the amount and timing of irrigation were modified from the experimental design, whenever necessary, in order to accommodate factors that changed for different years and locations, such as winter periodic rainfall and the availability of workers and water (see Satoh and El-Gamal (2017) for water management practices in the Nile Delta).
The standard tile drainage system for Egypt (e.g. Abdel-Dayem 1987, Amer and de Ridder 1989, Kubota et al. 2017) was adopted in the three fields. Lateral drains, buried at 1.35 m depth at an approximate horizontal interval of 20 m, extended to a length of 100 m from the main collectors in two opposite directions. Collectors were connected to drainage canals.
3.2 Eddy correlation measurements of evapotranspiration
To derive evapotranspiration, E, together with frictional velocity and sensible heat flux using the eddy correlation method (Table 3), turbulence measurements of wind velocities, humidity and temperature were made continuously using sensors installed at the top of a 5 m tower constructed at the centre of each field. A standard procedure (e.g. Lee et al. 2004, AsiaFlux Steering Committee 2007) was applied in order to produce flux data from raw turbulence measurements (for details, see Supplementary material, Section 2). In the discussion that follows, we mainly analyse daily evaporation.
Table 3. List of measurements and products obtained at each crop field.
3.3 Apparent crop coefficient
As mentioned, climate and soil conditions within crop fields in the delta are quite uniform and year-to-year climate variability is small. Thus, a direct comparison of E values measured for different years and locations is likely acceptable. Nevertheless, to further enhance the credibility of comparisons, reference crop evapotranspiration, E0, as defined by (Allen et al. 1998): 
(1)
was introduced, and the daily apparent crop coefficient, 
, was determined. The coefficient Kca is called apparent because it is different from the crop coefficient Kc as defined by the FAO. Coefficient Kc is generally, but not always, defined for E under optimum soil water conditions, while Kca reflects not only the difference due to a crop, but also the soil water condition reflecting the adopted irrigation method.
In Equation (1), Le is the latent heat of vaporization, ρ is the density of air; cp is the specific heat of air at constant pressure; rs is the surface resistance, which is set equal to 70 s/m; rav is the aerodynamic resistance formulated using the wind speed at 2 m, u2, as 208/u2 (s/m); Δ is the rate of change of the saturation vapour pressure, es, at air temperature, Ta; and ea is the atmospheric vapour pressure. To avoid a difference in E0 resulting from surface conditions, net radiation, Rn, was estimated from measured downward shortwave and longwave radiation, Rsd and Rld, respectively, the upward shortwave radiation, Rsu, estimated using a fixed albedo of α = 0.23, 
(Allen et al. 1998), and soil heat flux G from CR × Rn, with CR = 0.1 for a grass surface (e.g. Brutsaert 2005).
In addition to daily values of Kca, the mean value over a certain period of time, T, was defined as 
, where
and
over the period T. For the analysis, the following descriptions of T were employed: “tot” for the total cropping season (Table 1); “ini” for the initial stage; “dev” for the crop development stage; “mid” for the mid-season stage; “late” for the late season stage (Allen et al. 1998); “fallow, s” for the spring fallow season; “fallow, f” for the fall fallow season; and “annual” for one year.
Small differences in the length of the cropping season for the same crop but for different years and locations were considered using the following procedure. First, standard values of E0,tot and the cropping period for each crop were determined using the average reference crop evapotranspiration, 
, and the average cropping period,
(d), for all available E0,tot and the N values for each crop. These values were used to derive the adjusted Etot (mm) and the mean daily
(mm/d) used in the analysis based on:
(2)
(3)
When multiple results were obtained for a given combination of crop and irrigation methods, the means of Equations (2) and (3) were determined and used for the analysis.
3.4 Additional related variables for explaining the factors controlling evapotranspiration
Additional variables measured during the experiment are summarized in Table 3. Briefly, these variables were: (a) related to crop growth, including the mean crop height (h0), the leaf area index (LA), the canopy cover fraction (fv), the crop yield and the root zone depth (zrz); (b) related to the water availability for crops and evapotranspiration, such as the soil water content (θ), the groundwater level (zGW) and the irrigation amount (Pi); (c) related to general meteorological variables, including air temperature (Ta), relative humidity (r), wind speed (u) and atmospheric pressure; (d) the four radiation components; and (e) the energy balance components.
3.5 Annual evapotranspiration
Annual evapotranspiration, Eannual, was estimated as the sum of the following four terms: 
(4)
for a given combination of summer and winter crop irrigation method(s). The lengths of the two fallow seasons were determined using: 
since their average lengths during the observation period were determined to be the same (43 d).
4 Results
The values of Etot, 
and
for the various crops and for the various irrigation methods are compared in Figure 1(a)–(c), respectively. The left side provides information for summer crops and the right side provides information for winter crops.
Published online:
05 July 2017Figure 1. Comparison of (a) Etot, (b) 
and (c) Kca,tot over the cropping season for the selected crops and irrigation methods: summer crops (left), winter crops (right). The tabulated values can be found in Table 6.3 of El-Kilani and Sugita (2017). In (a), box-and-whisker plots are provided as an inset (for the same y-axes) and display the 10th, 25th, 50th, 75th and 90th percentile values determined from the Etot values reported in previous studies, compared with those of our study. Etot values were obtained from Figure 2(b) of Zwart and Bastiaanssen (2004) for rice, maize and cotton. The references for berseem, fava beans and sugar beet are provided in the Supplementary material, Table S3.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/thsj20/2017/thsj20.v062.i10/02626667.2017.1341631/20170802/images/medium/thsj_a_1341631_f0001_oc.jpg"})
Figure 1. Comparison of (a) Etot, (b) and (c) Kca,tot over the cropping season for the selected crops and irrigation methods: summer crops (left), winter crops (right). The tabulated values can be found in Table 6.3 of El-Kilani and Sugita (2017). In (a), box-and-whisker plots are provided as an inset (for the same y-axes) and display the 10th, 25th, 50th, 75th and 90th percentile values determined from the Etot values reported in previous studies, compared with those of our study. Etot values were obtained from Figure 2(b) of Zwart and Bastiaanssen (2004) for rice, maize and cotton. The references for berseem, fava beans and sugar beet are provided in the Supplementary material, Table S3.
4.1 Summer crops
4.1.1 Differences due to crop selection
It is clear from Figure 1 that rice consumed more water as evapotranspiration than other crops. The 
value of rice (BI) was 56% larger than that of maize (FI). In a similar manner, the Etot and
of rice (BI) were 81% and 42%, respectively, larger than those of maize (FI).
The difference in Etot was due partly to the longer cropping season of rice (
= 125 d) as compared to maize (
= 98 d). However, the main factor was the difference in E during the earlier cropping stage. As shown in Figure 2(f), the daily values of E and Kca for maize (FI) increased quickly versus E0 for each irrigation event and then gradually decreased with time during early stages of the cropping season, while they remained more or less constant during later stages almost regardless of irrigation events. This result likely occurred because soil evaporation, Eg, dominated E during the earlier stages of the cropping season due to little vegetation cover. Soil evaporation tends to be more easily affected by θ changes near the surface. Indeed, wetting and drying cycles can clearly be observed within the soil column, particularly near the surface, in response to irrigation events and groundwater level, zGW, increases (Fig. 3(f)). Additionally, the shallow root zone depth, zrz, during this stage (Fig. 3(f)) helped to suppress transpiration when θ decreased within the root zone. On the other hand, for later stages it can be speculated that Eg gradually decreased and that transpiration became the main component of E as fc and zrz increased (Fig. 3(f)). The larger zrz allowed plants to make use of soil water at greater depths, and, as a result, E was minimally influenced by soil moisture fluctuations near the surface.
Published online:
05 July 2017Figure 2. Changes in the daily values of E, E0, Kca and Pi for some of the selected summer crops: (a) rice (BI) in 2012, (b) cotton (SI) in 2014, (c) maize (DI) in 2013, (d) maize (SI) in 2013, (e) maize (FIi) in 2012, and (f) maize (FI) in 2013. The dotted horizontal lines with triangle heads at both ends indicate the cropping season. Thick horizontal lines indicate periods when daily total E values were gap-filled using the procedure explained in Supplementary material, Section 2.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/thsj20/2017/thsj20.v062.i10/02626667.2017.1341631/20170802/images/medium/thsj_a_1341631_f0002_oc.jpg"})
Figure 2. Changes in the daily values of E, E0, Kca and Pi for some of the selected summer crops: (a) rice (BI) in 2012, (b) cotton (SI) in 2014, (c) maize (DI) in 2013, (d) maize (SI) in 2013, (e) maize (FIi) in 2012, and (f) maize (FI) in 2013. The dotted horizontal lines with triangle heads at both ends indicate the cropping season. Thick horizontal lines indicate periods when daily total E values were gap-filled using the procedure explained in Supplementary material, Section 2.
Published online:
05 July 2017Figure 3. Time changes for θ, zrz (dotted line) and zgw (solid lines) for selected summer crops: (a) rice (BI) in 2012, (b) cotton (SI) in 2014, (c) maize (DI) in 2013, (d) maize (SI) in 2013, (e) maize (FIi) in 2012, and (f) maize (FI) in 2013 (see Supplementary material, Section 3 for details of the θ measurements and processing; also see Supplementary material, Section 4 regarding the root zone depth of maize for zrz). The contour line for θ = 0.45 is shown in orange, and is close to the field capacity.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/thsj20/2017/thsj20.v062.i10/02626667.2017.1341631/20170802/images/medium/thsj_a_1341631_f0003_oc.jpg"})
Figure 3. Time changes for θ, zrz (dotted line) and zgw (solid lines) for selected summer crops: (a) rice (BI) in 2012, (b) cotton (SI) in 2014, (c) maize (DI) in 2013, (d) maize (SI) in 2013, (e) maize (FIi) in 2012, and (f) maize (FI) in 2013 (see Supplementary material, Section 3 for details of the θ measurements and processing; also see Supplementary material, Section 4 regarding the root zone depth of maize for zrz). The contour line for θ = 0.45 is shown in orange, and is close to the field capacity.
In contrast, the values of E, Kca and θ remained high throughout the cropping season of rice (Figs. 2(a) and 3(a)) due to the presence of standing water on the soil surface resulting from the BI method. Although zGW measurements were not available for this cropping season, it appears that the soil column was completely saturated.
These differences are also clear in Table 4, where the Kca and 
values for various growth stages are listed for maize (FI) and rice (BI). Smaller
and Kca values for maize during the initial stage can clearly be seen as compared to those of rice. Comparisons are also provided in Table 4 for average values of the Bowen ratio (
), albedo (
), upward longwave radiation (
) and net radiation (
). A striking finding was that the
of maize for the initial stage was much larger than that for the other stages of maize and any of the growth stages of rice. This finding is another indication of soil surface dryness between irrigation events during the initial stage.
Table 4. Mean values for the different growth stages of maize (FI) and rice (BI).
Cotton (SI) consumed less daily water as 
than maize (SI) but more water as Etot (Fig. 1). The larger Etot of cotton was the result of the longer cropping season of cotton (approximately 169 d) as compared to other crops (98–125 d). On the other hand, the difference in
was mainly due to smaller Kca and daily
values for cotton than those of maize during the initial and development growth stages (Table 5, Fig. 2(b) and (d)). The
values for these stages of cotton were also quite large (Table 5), indicating surface soil dryness. Unfortunately, vegetation growth data, such as those for the cover fraction and the leaf area index of cotton, were not available for comparison. However, we suspect that the surface coverage of cotton was smaller than that of maize during the early stages and that the exposed soil surface tended to easily become dry. Such a hypothesis is consistent with general knowledge that initial growth for cotton is slow (e.g. National Cotton Council of America 2015).
Table 5. Mean values for the different growth stages of maize (SI) and cotton (SI).
4.1.2 Differences due to irrigation methods
For maize, a comparison was possible for all irrigation methods; and the Etot, 
and Kca,tot values all indicated similar differences (Fig. 1). One can immediately notice that FI and FIi did not produce markedly different Etot values, probably because the roots of maize (FIi) quickly developed (less than 1 month after seeding) to greater depths where θ > θf (θ at field capacity) (Fig. 3(e)). Therefore, with the exception of the first month, maize (FIi) made use of soil water available at greater depths even when the surface soil was dry between irrigation events. The fact that θf appeared at a relatively shallow depth of 0.25–0.5 m is likely due to the shallow groundwater level and the clay-rich soils common within the Nile Delta.
The values of Etot, 
and Kca,tot obtained using the SI and DI methods were, respectively, 66% and 58% of those obtained using the FI method (Fig. 1). The differences can be explained by Figures 2 and 3. θ for DI and SI (Fig. 3(c) and (d)) was, in general, much smaller than θ for FIi and FI (Fig. 3(e) and (f)). A larger zGW value can also be observed for DI. These differences caused different behaviours of the magnitude and time changes of E and Kca. Those for SI (Fig. 2(d)) were, in general, smaller than those for FI and FIi (Fig. 2(e) and (f)), although the course of the time changes was quite similar for these three methods, with quick and strong responses to irrigation events during earlier stages and more stable and steady behaviour during later stages. In contrast, those for the DI method were different (Fig. 2(c)). The time changes were quite simple, with slow increases of E and Kca in response to root development (Figs. 2(c) and 3(c)), but without clear responses to irrigation events.
For rice, a comparison between BI and BIi was possible (Fig. 1), and Etot and Kca,tot estimates for BIi were 78% of those cultivated using the BI method. The figure showing the change in E, E0 and Kca for BIi (not shown) and a comparison to BI (Fig. 2(a)) indicated that the shapes of both time changes were similar but that the magnitude of E and Kc for BIi was smaller than that for BI. We suspect that longer intervals of irrigation under BIi reduced the frequency of the presence of standing water as compared to that for BI. Unfortunately, due to a technical problem with the soil moisture sensors in saline soil (Supplementary material, Section 3, Sugita et al. 2017), θ values were not available for BIi, so we could not verify this hypothesis.
4.2 Winter crops
4.2.1 Differences due to crop selection
The values of Kca,tot and 
for wheat (BI) were, respectively, 14–16% and 22–28% larger than those for other crops cultivated using the same FI/BI methods, while the difference in Etot was small, with the exception of fava beans (Fig. 1). The difference in root development is likely to have contributed, at least in part, to the differences in Kca,tot and
. The root system of wheat is known to develop vertically and to a greater depth (>1 m; e.g. Weaver et al. 1924, Thorup-Kristensen et al. 2009) as compared to berseem and fava beans. Root density measurements obtained at the end of each cropping season (H. Fujimaki, personal communication, 2014) indicated that the root density of fava beans and berseem were almost zero at the lowest measurement depth of 0.5 m, while this was not the case for wheat. Therefore, wheat had an advantage in that it could make use of greater soil moisture at greater depths, resulting in larger values for Kca,tot and
.
Since sugar beet is known to develop a main root down to 1.5–1.8 m (Dunham 1993, Hergert 2012), this explanation is not applicable to sugar beet, although, for our experiment, we did not conduct formal measurements of the root system of sugar beet. The smaller Kca,tot and 
values for sugar beet, as compared to those for wheat, appear to be due to slower initial growth for sugar beet (Seadh et al. 2013), which led to longer periods when the soil surface was not covered by vegetation. As shown in Table 6, the
values for sugar beet during the initial and development stages were large, implying surface dryness, likely due to little vegetation cover. Thus, the Kca,ini and Kca,dev (and also
for these two stages) for sugar beet were smaller than those of wheat (see also Fig. 4(c) and (d) for seasonal variations of Kca and E).
Table 6. Mean values for the different growth stages of wheat (BI) and sugar beet (FI).
Published online:
05 July 2017Figure 4. Changes in the daily values of E, E0, Kca and Pi for some of the selected winter crops: (a) fava beans (FI) in 2012/13, (b) wheat (BI) in 2012/13, (c) wheat (BI) in 2012/13, (d) sugar beet (FI) in 2013/14, (e) berseem (BI) in 2011/12, and (f) berseem (BIi) in 2013/14. See Figure 2 for explanation.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/thsj20/2017/thsj20.v062.i10/02626667.2017.1341631/20170802/images/medium/thsj_a_1341631_f0004_oc.jpg"})
Figure 4. Changes in the daily values of E, E0, Kca and Pi for some of the selected winter crops: (a) fava beans (FI) in 2012/13, (b) wheat (BI) in 2012/13, (c) wheat (BI) in 2012/13, (d) sugar beet (FI) in 2013/14, (e) berseem (BI) in 2011/12, and (f) berseem (BIi) in 2013/14. See Figure 2 for explanation.
The difference in Etot between wheat and berseem or sugar beet was smaller than that for 
and Kca,tot because the length of the cropping season for wheat (approx. N = 160 d) is shorter than that for berseem (N = 196 d) and sugar beet (N = 184 d). However, the difference in Etot was approximately the same, with values of
and Kca,tot between wheat and fava beans because the cropping period of fava beans (N = 162 d) is similar to that of wheat.
4.2.2 Differences due to irrigation methods
For sugar beet, the impact of adopting the FI and DI irrigation methods on evapotranspiration was considered. As Figure 1 indicates, when the DI method was employed, the values of Etot, 
, and Kca,tot were reduced by 46% as compared to the same values obtained using FI. For wheat and berseem, a comparison between BI and BIi was possible, and we determined large differences between these two crops. When berseem was cultivated under the BIi method, Etot values, as well as Kca,tot, were smaller by 47% as compared to those for BI, while they were larger by 7% for wheat. The differences appear to have been caused by the difference in their root systems. As previously mentioned, the root system of wheat extends down much deeper than that of berseem. Therefore, the difference in θ near the surface caused by the two different irrigation methods of BI and BIi (Fig. 5(b) and (c) for wheat and Fig. 5(e) and (f) for berseem) did not result in a meaningful impact on wheat water consumption, while it did for berseem.
Published online:
05 July 2017Figure 5. Time changes for θ, zrz (dotted line) and zgw (solid lines) for selected winter crops: (a) fava beans (FI) in 2012/13, (b) wheat (BIi) in 2012/13, (c) wheat (BI) in 2012/13, (d) sugar beet (FI) in 2013/14, (e) berseem (BI) in 2011/12, and (f) berseem (BIi) in 2013/14. See Figure 3 for explanation.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/thsj20/2017/thsj20.v062.i10/02626667.2017.1341631/20170802/images/medium/thsj_a_1341631_f0005_oc.jpg"})
Figure 5. Time changes for θ, zrz (dotted line) and zgw (solid lines) for selected winter crops: (a) fava beans (FI) in 2012/13, (b) wheat (BIi) in 2012/13, (c) wheat (BI) in 2012/13, (d) sugar beet (FI) in 2013/14, (e) berseem (BI) in 2011/12, and (f) berseem (BIi) in 2013/14. See Figure 3 for explanation.
4.3 Differences between summer and winter crops
In this section, we summarize the differences between winter and summer crops. Daily changes in the E, E0, Kca and Pi of winter crops (Fig. 4) were compared to those for summer crops (Fig. 2). A striking difference can be noticed regarding the impact of irrigation events and the values of E and Kca, particularly for the FI or BI methods. As mentioned above, sudden increases in E and Kca are clearly noticeable for maize (Fig. 2(d)–(f)) and, to a lesser extent, for cotton (Fig. 2(b)) during the early cropping stages in response to irrigation events. Corresponding increases in soil moisture and in the water table can also be observed in Figure 3. In contrast, this type of response for E and Kca to irrigation events was not observed for all of the winter crops tested during our experiment (Fig. 4), even though the responses of soil moisture and the water table were as clear as those obtained for the summer cropping season.
The results can be understood based on the difference in atmospheric demand for evapotranspiration between the two seasons as compared to soil water storage in crop fields within the Nile Delta. For example, daily mean values for July (corresponding to the early cropping stage during summer) were 
= 2.5 mm/d and
= 5.2 mm/d, while those in December (during the early cropping stages in winter) were
= 1.2 mm/d and
= 1.4 mm/d for 2013 at the Sakha-A field. Thus,
for the winter crops, while
<
for the summer crops.
Soil water storage, when the soil column is completely saturated just after irrigation, was determined from the saturated soil water content, θs, as 38 mm (for the soil layer over a depth range of 0–0.15 m), 193 mm (0–0.3 m), 368 mm (0–0.6 m) 730 mm (0–1.2 m). For θf, the values were 29, 141, 282 and 564 mm for the respective layers. Thus, if initial storage with θf is assumed, 21 d would be required for 
to completely deplete soil water storage from 0–0.15 m during the winter cropping season. In contrast, this period would be only 5.6 d for the summer cropping season. In reality, before soil water is completely consumed, evapotranspiration begins to decrease. Such a scenario was often observed for maize. However, for winter crops, water storage was large enough in comparison to
so that a decrease of
was not observed.
4.4 Annual evapotranspiration
The estimated annual evapotranspiration values are provided for six combinations of crops and irrigation methods (Table 7). To reflect current conditions within the Nile Delta, we selected either the FI or BI method from conventional irrigation methods. The SI, DI or BIi methods were adopted as water-saving irrigation methods.
Table 7. Estimated annual evapotranspiration for the selected crops and irrigation methods.
With conventional irrigation methods, maize (FI) and fava beans (FI) produced Eannual = 875 mm/year, which is likely closer to the minimum annual E value of crop fields within the Nile Delta. When rice (BI) and wheat (BI) are selected, the result (1225 mm/year) should be the largest annual E for conventional irrigation methods. As expected, the difference was as large as 350 mm/year. When water saving irrigation methods were introduced and berseem (BIi) or sugar beet (DI) were selected as winter crops, Eannual became 816–828 mm/year for rice (BIi) and 566–584 mm/year for maize (DI) as the selected summer crop.
An interesting finding was that the magnitude of Eannual was much larger than the global mean evapotranspiration for areas of land surfaces (approximately 420–540 mm/year, e.g. Brutsaert 2005) and was compatible with that of natural vegetation in humid areas (e.g. Mueller et al. 2011). Therefore, it appears that the Nile Delta evaporates water in a similar manner to surfaces in humid areas, a finding clearly due to the large amount of water introduced from the Nile River by irrigation practices.
Finally, since fallow evaporation accounts for as much as 10–20% of Eannual in Table 7, the importance of fallow evaporation for water balance considerations of crop fields should be estimated. In general, fallow evaporation is not considered in crop evapotranspiration studies in agronomy.
As mentioned in the methods section, the average 
for all fallow seasons during the experiment was used for estimating fallow evaporation. Since the time change for
and Kca during the fallow season can be different depending on the crop and irrigation method selected during the previous cropping season, this is a crude simplification. For example, a comparison between Figure 2(e) and (f) immediately verifies such differences for the spring fallow season. As seen in Figure 2(e),
kept decreasing while in Figure 2(f) the rate of decrease was much smaller due to the timing of the last irrigation application during the previous winter cropping season, which was 6 May for Figure 2(e) (see also Fig. 4(e) for the previous season) and 2 April for Figure 2(f) (see also Fig. 4(c)). Thus, for a more precise estimation of annual evapotranspiration, refinement of the fallow season treatment is necessary.
5 Discussion
The results presented above indicate that there are large differences in evapotranspiration between crop types, between summer and winter crops, and between different irrigation methods for the Nile Delta. As a result, depending on the selected combination of summer and winter crops and irrigation methods, annual evapotranspiration could also differ greatly (566–828 mm/year for water-saving irrigation methods and 875–1225 mm/year for conventional irrigation methods – a maximum difference of 659 mm/year). Differences in evapotranspiration resulted from the combinations of atmospheric demand, soil moisture status, the presence of standing water on the surface, root depth and the length and timing of the cropping season. Again, the relative importance of individual factors changed depending on the crop and the irrigation method.
Since our results are probably the first of their kind for the Nile Delta based on the eddy correlation method together with measurements of hydrological processes, they have the potential to be useful in various applications, including water resources assessment and planning. However, their limitations and reliability should be evaluated before they can be used with confidence. Therefore, below, we compare our results with those obtained in previous studies (Sections 5.1 and 5.2), which is useful for identifying the applicability of our results to other areas (Section 5.3).
5.1 A comparison with crop evapotranspiration from previous studies based in the Nile Delta
We compared our results to those of previous studies within the Nile Delta. Since this study represents the first study that has applied the eddy correlation method to the Nile Delta, our comparison required us to compare results obtained from more traditional methods. Among such studies (Table 1), crop evapotranspiration determined by applying the soil water balance method to a 50 m2 plot cultivated with maize, wheat or cotton (El-Shal 1966, also reported by Shahin 1985) was worth consideration. This comparison was beneficial because the experiment of El-Shal (1966) was conducted in Sakha, one of our study areas. Thus, a comparison of the two studies is a comparison between results obtained under the same soil and climate conditions. Results from this previous study were also provided for actual crop fields, which is also compatible with the results obtained from our study. Irrigation intervals and amounts were also, more or less, compatible with our results obtained using the FI method.
El-Shal (1966) obtained values of 
= 3.9–4.0 mm/d and Etot = 512–520 mm for late maize cultivated during the cropping season from 30 July to 8 December. For a direct comparison, the Kca,tot value of El-Shal (1966) was derived by estimating E0,tot = 422 mm for the cropping season of late maize, using our data obtained during 2011. The derived Kca,tot value was 1.2, which is comparable to our Kca,tot value of 0.84. Therefore, the evapotranspiration estimate by El-Shal (1966) is approximately 42% larger than that obtained during our study. In a similar manner, the results of El-Shal (1966) for cotton and wheat in the Sakha field were converted to Kca,tot in the range 0.89–0.91 and 0.80–0.85, respectively, by applying the same procedure. Thus, the Kca value (and thus Etot) was larger by 93–98% for cotton and smaller by 21–25% for wheat than our results obtained using the SI or BI irrigation methods. In other words, agreement was poor.
The exact reason(s) for these large differences is (are) not known since the finer details of the experiments conducted by El-Shal (1966) are not clearly documented (e.g. depth to groundwater), or not available (e.g. crop information such as plant density). However, several reasons are possible for the differences in results. The first reason is possible overestimation of Etot, due to the assumption of negligible drainage to deeper soil from the lowest measurement depth of 50 cm, by El-Shal (1966) for the soil water balance method. The second reason is random errors introduced by the use of only three soil samples of 100 cm3 for determining mean soil moisture in the 10 cm depth layer of a 50 m2 plot. The third reason, specific only for maize, is that the target was late maize (August to early December) in the study by El-Shal (1966), while our target was regular summer maize (June–September to early October). Also, wheat tends to have a deeper root system and could have extended its roots below the 50 cm level, taking water from deeper soils. For cotton, a comparison with our Etot value, obtained under the SI irrigation method, which presumably reduces water consumption, may have contributed to differences in the results. Finally, it is also possible that we over- or underestimated daily E values. Although we paid great attention to our measurements and post-data handling, our measurements are not perfect. In particular, the reliability of evapotranspiration data during the night-time and during some winter periods is a concern, and could be low since data gaps caused by the formation of dew on the sensor head of the gas analyser and by an electricity problem were filled in using various techniques (see Supplementary material, Section 2). However, a comparison with previous studies performed in other areas tends to indicate the general validity of our estimates (see Section 5.2 below). At any rate, given the poor agreement, performing independent eddy correlation measurements within the Nile Delta in order to formally validate our evapotranspiration measurements is desirable.
5.2 A comparison with crop evapotranspiration reported in previous studies for other regions
We also compared our values for crop evapotranspiration in the Nile Delta with those of other locations, although a wide range of conditions could affect the crop evapotranspiration values reported in these studies. For example, for the same crop, factors that are likely to affect evapotranspiration include climate, soil type, the irrigation method, the groundwater depth, the amount and timing of fertilizer application, weed/disease control, etc. Therefore, expecting a perfect match in conditions for comparing evapotranspiration values is not practical. Instead, we compared statistics that characterize the Etot of each crop to our values.
For this purpose, a large number of evapotranspiration studies that reported Etot values measured in a crop field or in a test plot, but not in a pot experiment, were gathered from recent papers published in the international literature. The Etot values were obtained from the review paper of Zwart and Bastiaanssen (2004) for wheat (number of Etot values, n = 325), rice (n = 101), cotton (n = 112) and maize (n = 198), and through a literature search for berseem (n = 27), fava beans (n = 132) and sugar beet (n = 54) (see Table S3 in the Supplementary material, Section 7 for the details of each study). Statistics (the 10th, 25th, 50th, 75th 90th percentile values) determined for each crop are provided in Figure 1 as a box-and-whisker plot.
First, the range of Etot values for each crop was quite large, and yet our Etot values, obtained under the conventional FI or BI irrigation methods, fell largely between the 25th and 75th percentiles. Thus, our Etot estimates are, in general, consistent with those of previous studies. Additionally, the Etot of rice, wheat and fava beans were on the higher side of the percentile range, while maize was on the lower side. The values are in agreement with comparisons of crop yield statistics for Egypt and the world. For 2013, crop yields (kg/ha) for Egypt ranked 2nd amongst 118 rice-producing countries, 11th amongst 125 wheat-producing countries, 11th amongst 59 fava bean-producing countries and 28th amongst 167 maize-producing countries according to FAOSTAT (FAO 2013). Agreement between values was natural since crop yield and evapotranspiration are known to be linearly correlated (e.g. Payero et al. 2009), although the ratio of crop yield to evapotranspiration is not exactly a constant and is affected by latitude, applied amounts of irrigation water and the amount of applied nitrogen (e.g. Zwart and Bastiaanssen 2004). Again, this analysis tends to indicate the general validity of our estimates.
The same comparison was also used to examine whether or not the water-saving irrigation methods we found effective for the Nile Delta were similarly effective on a global scale. A comparison of our Etot values indicated that the Etot of sugar beet, maize and cotton cultivated using the SI and DI methods within the Nile Delta were lower than the 10th percentile values, indicating the water-saving ability of these irrigation methods. The Etot of berseem obtained using the BIi method was also lower than the 10th percentile value, while that of wheat was closer to the 90th percentile value. Since the BIi and EIi methods are only effective for reducing water use for crops with shallow root zones, as discussed in Section 4.2.2, this result is reasonable.
5.3 Applicability to other cropping areas
The above comparison indicates that our Etot estimates are, in general, consistent with those of previous studies, although the lack of close agreement with the results from earlier studies in Egypt is still a concern. Therefore, until independent validation using the eddy correlation method becomes available for the Nile Delta, our results should be used with certain reservations.
Despite their newness, our results could be used for many practical purposes, not only for the Nile Delta but also for other areas. In addition to drought assessment, one immediate application for our results is water allocation planning. In many countries, to meet increasing water demand amongst competing sectors, water resources planning (e.g. Ministry of Water Resources and Irrigation, Egypt 2005) requires assessments of current water use and future water allocation. For estimations of water consumption by the agricultural sector, crop evapotranspiration is often derived by applying estimation schemes, such as Equation (1), using the crop coefficients of Allen et al. (1998) with possible adjustments for the local environment. To not only estimate water demand for irrigation but also provide estimates for the possible water withdrawal required from canals in relation to water availability in a watershed, in some cases, hydrological models are used together with crop models (e.g. McNider et al. 2015). For both cases, calibrating model parameters and, sometimes, modifying models or schemes to reflect local conditions is often necessary. Thus, the availability of Etot or Kca values, as well as information related to locally determined hydrological processes or values determined for similar environments, allows improved water allocation planning. The question then becomes: What variables should be similar so our results can be applied to other areas?
In general, climate, geomorphology, soils and the availability of water (i.e. depth to groundwater or closeness to rivers and lakes) are important because these parameters usually interact with one another to create a particular environment for a crop field. However, differences in climate alone can be mitigated, to some extent, by the use of Kca instead of E when applying our results to areas with different climates. However, soil conditions, geomorphic features and water availability are more difficult to understand. Soil types influence crop growth through the availability of soil water and nutrients. The geomorphology of the area where a crop field is located influences, and sometimes determines, soil type and water availability. Therefore, narrowing potential areas with similar geomorphology, soils and water availability where our results could safely be applied was a good idea. One such area with a similar environment, among others, would be a crop field developed on a delta that is characterized by extended flat surfaces, easy access to surface water, the presence of shallow groundwater and clay-rich soils. Deltas are not only important for agriculture, due to their fertile soils, they are also important for other industries. Therefore, competition for land and water is often a major problem in these areas (e.g. Bucx et al. 2010, Evans 2012) and yet, to our knowledge, there are no combined or detailed studies related to evapotranspiration and hydrological processes for a delta. Since equitable water allocation is essential for making better use of scarce water resources, our results could be useful for filling in such gaps in knowledge.
Finally, some implications of our study in terms of water-saving irrigation are also worth mentioning. As indicated above, the DI and SI irrigation methods and those with increased irrigation intervals were found to be effective (at least for some crops) for reducing water consumption. Amongst irrigation methods, the SI method appears promising because it does not require additional cost in relation to furrow irrigation. In contrast, the DI method requires initial and operating costs, and it is not feasible to expect the large-scale use of this method in developing countries. An increase in irrigation intervals does not require additional costs, but it does require well-coordinated crop field management, for which educated human resources are required. Quite often, a lack of human resources is the major obstacle in developing countries. Thus, in developing countries, the SI method is a good alternative to the furrow irrigation method. However, for this study, we only tested one version of a SI method. Therefore, finding optimum dimensions for the planting bed/furrow width for each crop is desirable. Our observations of hydrological processes during the SI method are useful for this purpose.
6 Concluding remarks
To determine crop evapotranspiration for major crops cultivated under different irrigation methods, we deployed eddy correlation systems from 2010 through 2014 for the first time in the Nile Delta. The general validation of the derived E values was made by comparing them with E values reported in previous studies. Our results indicated large differences for crop evapotranspiration depending on the type of crop and the corresponding irrigation method. Evapotranspiration for rice was by far the largest among the tested crops. Crop evapotranspiration found using the SI and DI methods was much smaller than that using the conventional irrigation methods, while increasing irrigation intervals in furrows or basin irrigation reduced evapotranspiration for crops with shallow root systems. Therefore, annual evapotranspiration in crop fields within the Nile Delta could differ by as much as 659 mm/year depending on crop selection and irrigation method.
Through detailed observations of hydrological processes, we were able to explain differences in evapotranspiration. The differences were due to a combination of atmospheric demand, soil moisture status, the presence of standing water on the surface, root depth and the length and timing of the cropping season. The relative importance of each of these factors changed depending on the crop and the irrigation method employed. For example, for rice, the longer cropping season, the presence of ponded water and the continuous high soil moisture condition were found relevant.
Our results indicated that the FIi/BIi, SI and DI methods were effective in reducing water consumption. When additional costs required in relation to conventional irrigation methods were considered, the SI method was identified as a promising choice. In order to consider overall water use efficiency of crop fields, however, other factors need to be considered. There are differences in the definition of efficiency among researchers in different disciplines such as irrigation scientists, economists or physiologists (Nair et al. 2013). Agronomists usually consider water use efficiency (WUE), defined as the yield per unit area divided by evapotranspiration. In this case, not only reducing evapotranspiration but also increasing crop yield is important when choosing optimum irrigation methods. In the case of the Nile Delta, our results on Etot for various crops and for different irrigation methods have been used together with corresponding yield measurements by Maruyama et al. (2017). They concluded that the DI and SI methods produced higher WUE values than the furrow irrigation method for maize because of a larger yield in the case of DI, and in spite of a smaller yield for SI. From the viewpoint of irrigation scientists, it is important to achieve higher irrigation efficiency ec, defined as the ratio of crop evapotranspiration to inflow water into the field. El-Kilani and Sugita (2017) examined this for maize with our Etot values and found ec = 0.91 for DI, 0.64 for SI and 0.77 for FI. Thus the DI method produced the highest irrigation efficiency followed by the FI method. Overall, DI could be chosen as the optimum method of irrigation if the additional costs can be accepted, while SI should be considered if costs, water-saving ability and water use efficiency are important.
Finally, the applicability of our results to other areas was discussed, particularly for the purpose of reducing water consumption in the agriculture sector and establishing more equitable water allocation within an area with increasing water demands in various sectors. Deltas were identified as potential regions where our results could be useful for various practical purposes such as drought assessment and water allocation planning.
| Reference | Study area/site | Target crops/size for spatial averaging | Irrigation method | Method | Main findings* |
|---|---|---|---|---|---|
| El-Shal (1966), also cited in Shahin (1985) | Four sites within the Nile Delta | Late maize, cotton and wheat | Conventional irrigation methods | Soil water balance method | Evapotranspiration of each crop over 10-day period was determined. |
| Swelam et al. (2010) | Two sites in the Nile Delta | Maize and wheat | Conventional irrigation methods | Soil water balance method (maize) Weighing lysimeter installed in a crop field (wheat) | Monthly evapotranspiration was obtained. Kca values for four growth stages were provided. |
| Amer (2010) | One site in the northern part of the Nile Delta | Maize | Conventional irrigation methods | Soil water balance method | Monthly and total evapotranspiration values were determined for different level of irrigation and salinity. |
| This study | Two sites in the Nile Delta | Maize, rice, cotton (summer crops) wheat, sugar beet, berseem, and fava beans (winter crops) | Conventional and water-saving irrigation methods | Eddy correlation method | Daily and total evapotranspiration values were obtained. Kca values for the cropping season were determined. |
*those related to our study purposes. Kca is the apparent crop coefficient defined in Section 3.3
| Calendar year | 2010 | 2011 | 2012 | 2013 | 2014 | ||||
|---|---|---|---|---|---|---|---|---|---|
| Cropping season | Summer | Winter | Summer | Winter | Summer | Winter | Summer | Winter | Summer |
| Sakha-A | Maize FI,680 mm 20 June–1 Oct. | Maize FI,632 mm 14 June–17 Sep. | Wheat BI,511 mm 25 Nov.–8 May | Rice BI, N/A 27 June–5 Nov. | Wheat BI,608 mm 11 Dec.–13 May | Maize FI,720 mm 1 July–10 Oct. | Sugar beet FI, 538 mm 7 Nov.–9 May | Rice BIi,610 mm 3 July–27 Oct. | |
| Sakha-B | Wheat BIi,270 mm 25 Nov.–8 May | Wheat BIi,297 mm 11 Dec.–13 May | Maize DI,245 mm 8 July–(10 Oct.) | Sugar beet DI, 287 mm 7 Nov.–9 May | |||||
| Zankalon | Berseem BI,549 mm 26 Oct.–25 May | Maize FIi,639 mm 20 June–7 Oct. | Fava beans FI,446 mm 8 Nov.–18 Apr. | Maize SI,495 mm 16 June–1 Oct. | Berseem BIi, 539 mm 21 Oct.–16 Apr. | Cotton SI,842 mm 1 May–16 Oct. | |||
| Variable | Method of measurement | Sensor(s) | Sensor height (m) | Time resolution |
|---|---|---|---|---|
| Sensible heat flux, H | Eddy correlation method | Gill Instruments, R3-50 and LI-Cor, LI-7500† | 5.78 | 30-minute averages, continuous |
| Friction velocity, u* | ||||
| Evapotranspiration, E | ||||
| Downward and upward shortwave radiation, Rsd, Rsu Downward and upward longwave radiation, Rld, Rlu | Four-component radiometer | Hukseflux Thermal Sensors, NR01 | 4.20 | |
| Soil heat flux, G | Soil heat flux plate | Hukseflux Thermal Sensors, HFP-01 | −0.03 | |
| Atmospheric pressure, p | Barometer | Vaisala, PTB210 | 0.8 | |
| Air temperature, Ta Relative humidity, r | Ventilated hygrothermometers with a radiation shield | Vaisala, HMP155 | 0.5, 1.0 and 3.0 | |
| Amount of irrigation, Pi | V-notch flow (surface irrigation) and flow meter (drip irrigation) measurements | N/A | Surface | At the time of irrigation event |
| Groundwater level, zgw | Pressure sensor | Solint Canada Ltd, Levelogger, Junior Edge 3001 | −1.35 | 30-minute, instantaneous, continuous (2012–2014) |
| Volumetric soil water content, θ | Time domain reflectometry (TDR) soil moisture sensor | Campbell Scientific, TDR-100 with CS630 probes | −0.03, −0.1, −0.2, −0.4 and −0.8 | 30-minute, instantaneous, continuous (A. Fujimaki, personal communication, 2010–2014; see also Supplementary material, Section 3) |
| Soil hydraulics parameters (θ at saturation, θs, and at field capacity, θf) | Laboratory test with soil samples taken from depths of 0–5 cm, 7.5–12.5 cm, 17.5–22.5 cm, and 77.5–82.5 cm at the Sakha-A field | N/A | N/A | One-time soil sampling and laboratory test (A. Hoshino, personal communication, 2010) |
| Crop height, h0 | In situ measurements | N/A | Periodically during cropping season in 2010–2012 (S. Maruyama, personal communication, 2010–2014) | |
| Leaf area index, LA | Canopy analyser or leaf sampling and measurement | Li-Cor, LAI2200 and LAI2000 | Above and below canopy | |
| Crop yield | In situ measurements | N/A | At the end of cropping season in 2010–2012 (S. Maruyama, personal communication, 2010–2014) | |
| Canopy cover fraction, fc | Nadir-looking digital camera and image processing | N/A | Above canopy | Summer, 2011 |
| Root zone depth, zrz | In situ measurements | N/A | One-time measurement in the summer cropping season in 2012 (Supplementary material, Section 4) |
† Calibrated twice during the experiment.
| Stage | |||||
|---|---|---|---|---|---|
| Crop | Variable | Initial | Development | Mid-season | Late season |
| Maize (FI, 2011) |  | 2.6 | 4.5 | 5.2 | 4.1 |
| Kca | 0.51 | 0.95 | 1.20 | 1.10 | |
 | 1.69 | 0.55 | 0.36 | 0.42 | |
 | 201 | 195 | 198 | 166 | |
 | 0.11 | 0.13 | 0.14 | 0.15 | |
 | 465 | 467 | 450 | 449 | |
| Duration (d) | 22 | 48 | 8 | 18 | |
| Rice (BI, 2012) |  | 5.1 | 6.4 | 6.5 | 3.2 |
| Kca | 1.12 | 1.38 | 1.61 | 1.09 | |
 | 0.47 | 0.18 | 0.06 | 0.33 | |
 | 208 | 215 | 186 | 120 | |
 | 0.079 | 0.092 | 0.097 | 0.13 | |
 | 467 | 464 | 449 | 435 | |
| Duration (d) | 30 | 30 | 32 | 40 | |
| Stage | |||||
|---|---|---|---|---|---|
| Crop | Variable | Initial | Development | Mid-season | Late season |
| Maize (SI, 2013) |  | 2.4 | 3.1 | 2.6 | 2.5 |
| Kca | 0.45 | 0.62 | 0.57 | 0.58 | |
 | 1.72 | 0.97 | 0.88 | 0.73 | |
 | 188 | 203 | 200 | 177 | |
 | 0.088 | 0.13 | 0.13 | 0.14 | |
 | 481 | 468 | 452 | 446 | |
| Duration (d) | 21 | 48 | 16 | 15 | |
| Cotton (SI, 2014) |  | 1.2 | 1.8 | 3.4 | 2.1 |
| Kca | 0.22 | 0.31 | 0.67 | 0.56 | |
 | 2.38 | 1.70 | 0.44 | 0.50 | |
 | 178 | 193 | 194 | 167 | |
 | 0.10 | 0.13 | 0.16 | 0.16 | |
 | 457 | 472 | 465 | 445 | |
| Duration (d) | 26 | 43 | 52 | 48 | |
The duration of each stage was taken from Allen et al. (1998) and adjusted for the differences in total cropping period and crop growth.
| Stage | |||||
|---|---|---|---|---|---|
| Crop | Variable | Initial | Development | Mid-season | Late season |
| Wheat (BI, 2012/13) |  | 1.7 | 2.8 | 3.9 | 2.8 |
| Kca | 0.95 | 1.25 | 1.04 | 0.53 | |
 | 0.50 | 0.37 | 0.61 | 1.59 | |
 | 75 | 108 | 181 | 208 | |
 | 0.14 | 0.17 | 0.15 | 0.14 | |
 | 382 | 384 | 401 | 436 | |
| Duration (d) | 19 | 90 | 26 | 19 | |
| Sugar beet (FI, 2013/14) |  | 1.1 | 1.3 | 2.8 | 4.9 |
| Kca | 0.68 | 0.87 | 1.01 | 0.97 | |
 | 1.04 | 0.82 | 0.69 | 0.32 | |
 | 82 | 85 | 142 | 189 | |
 | 0.08 | 0.12 | 0.18 | 0.17 | |
 | 403 | 380 | 393 | 424 | |
| Duration (d) | 36 | 60 | 64 | 24 | |
The duration of each stage was taken from Allen et al. (1998) and adjusted for the differences in total cropping period and crop growth.
| Crop and irrigation selection |  |  |  |  |  | Remarks |
|---|---|---|---|---|---|---|
| Maize (FI) + fava bean (FI) | 368 | 341 | 77 | 90 | 875 | Conventional irrigation |
| Rice (BI) + wheat (BI) | 667 | 431 | 58 | 68 | 1225 | |
| Maize (SI) + berseem (BIi) | 245 | 226 | 52 | 61 | 584 | Water-saving irrigation |
| Maize (DI) + sugar beet (DI) | 215 | 220 | 61 | 71 | 566 | |
| Rice (BIi) + berseem (BIi) | 520 | 226 | 33 | 38 | 816 | |
| Rice (BIi) + sugar beet (DI) | 520 | 220 | 41 | 48 | 828 |
Supplementary Information
Download MS Word (249 KB)Related Research Data
Acknowledgments
The authors are grateful to H. Fujimaki (Tottori University) for providing us with his soil moisture, root profiles and soil physics data; to A. Hoshino for the soil hydraulic parameters; to A. Kubota and S. Maruyama (University of Tsukuba) for sharing their results on plant physiology and observation; to T. Fukuda, K. Tsuchihira, Y. Irigaki and I. Tsuji (University of Tsukuba) for their participation in field observations; and to A. Osada, T. Kamitani, Sayed El Nehlak, Hassan Mohamed Abd El Baki and Hussein Al Nadar (WAT project office in Cairo, Japan International Cooperation Agency), among others, for supporting the field observations for this study. Finally, the authors also express their appreciation to M. Satoh (University of Tsukuba), without whose initiative and leadership the experiment for this study would not have been possible.
Disclosure statement
No potential conflict of interest was reported by the authors.
Supplemental data
Supplemental data for this article can be accessed here.